Surface-enhanced Raman spectroscopy(SERS) has unique advantages such as ultra-high sensitivity, fingerprint information, small samples, and non-destructive detection. The design and preparation of a SERS substrate with excellent reproducibility and stability are key factors in the further development of SERS detection technology. Hydrogel is a new type of encapsulation material; its cross-linked polymer network has a three-dimensional layered structure that can retain a large amount of water and has a good blocking effect on impurities and strong anti-interference ability. Hydrogel SERS substrate has many advantages, such as low cost, high sensitivity, rapid detection, and high throughput. In this review, the research process of SERS substrate, the advantages of hydrogel SERS substrate, and the application of hydrogel SERS substrate in the field of food, biology, and environmental detection are mainly reviewed to provide a new reference for the preparation of hydrogel SERSsubstrate. First, for the SERS substrate research process, early fixed metal nanoparticles(MNPs) rigid solid substrate, precious MNPs tend to oxidate and aggregate. SERS substrate has poor reproducibility, so it cannot analyze the surface rough samples. In the SERS substrate with MNPs modified on the flexible support material, MNPs are easy to separate in the detection process, and the sensitivity and stability of the SERS substrate are poor. The hydrogel is combined with a plasma nanostructure to obtain a SERS substrate with good uniformity and stability. Hydrogel, as the protective layer of MNPs, provides reliable size and charge selectivity for SERS analysis while maintaining high permeability.Furthermore, hydrogel SERS substrate can achieve in situ detection of the target without sample pretreatment, biocompatible composite hydrogels can be directly detected in vivo, and DNA hydrogel substrate can be accurately recognized, hydrogel modified with multiple antibodies allow for the detection of multiple analytes simultaneously, microgel with adjustable mesh size have a selective sieving effect on the target, and 3D nanostructured hydrogel provides more adsorption hotspots. The prepared hydrogel SERS particles, chips-, and patches show great potential for on-site trace analysis in food safety, biomedicine, and environmental monitoring. In conclusion, preparing hydrogel SERS substrates has good development prospects and can provide new references for future analysis and detection fields.

The polymer glass transition, as a transition between the glassy and highly elastic states of amorphous polymers, has an important impact on material properties, especially mechanical properties. Terahertz spectroscopy, as a novel spectroscopic analysis technique with the advantages of non-contact, rapidity, and high sensitivity, shows important potential for application in polymer glass transition studies. By summarizing and analyzing the findings in the literature in the last 15 years, this paper aims to provide a comprehensive understanding of the current status of the application of terahertz spectroscopy in this field and to look forward to its future development. Firstly, the properties of the polymer glass transition are introduced, and traditional measurement methods and the limitations they face, such as thermal analysis, dynamic mechanical analysis, and infrared spectroscopy, are explored. Researchers have begun to look for new study methods to compensate for these limitations. Terahertz spectroscopy can provide both spectral and phase information, allowing direct measurement of refractive index and dielectric constant and reflecting the free volume change of the polymer chain, which is why the THz spectroscopy can detect the glass transition of polymers. Subsequently, the progress in applying terahertz spectroscopy in studying polymer glass transition in recent years is summarized. Results of these studies show that terahertz spectroscopy can accurately determine the glass transition temperatures, provide microscopic information about the structure and conformation, and reveal the glass transition mechanism and dynamics behavior of polymers, including poly formaldehyde, polyamide, poly (-caprolactone), and polylactic acid. Moreover, this paper points out the existing problems of terahertz spectroscopy, including the bandwidth limitations and high cost of terahertz spectrometers. Therefore, future research needs further to improve the performance of terahertz sources and instruments, develop more efficient data analysis methods, and explore the potential of terahertz spectroscopy for industrial applications. Overall, terahertz spectroscopy, as an emerging research tool, has made positive advances in the field of polymer glass transition and can enable rapid and sensitive detection of polymers as well as precise structural analysis.

Composite materials, due to their advantages such as high strength, low weight, and corrosion resistance, are widely used in various industries, including aerospace, construction, and marine. However, during the production or use of composite materials, it is inevitable that they may suffer damage and degradation, leading to a decrease in material performance and potential safety hazards. Therefore, researching non-destructive testing (NDT) methods to detect the types and extents of damage inside the materials has become a hot topic in recent years. The advantage of Terahertz (THz) technology lies in its sensitivity to the chemical composition and crystal structure of materials. Detailed information about the internal conditions of materials, such as defect types, sizes, and distributions, can be obtained by analyzing the frequency spectrum and phase information of THz waves. This is crucial for quality control and performance assessment of composite materials. First, this paper outlines the basic principles of composite materials, which commonly use NDT and THz techniques. Secondly, it categorizes the defects such as delamination, inclusions, porosity, impact damage, thermal damage, etc., which occur during the manufacturing or use of composite materials and focuses on applying THz technology in detecting different defect types. Then the challenges faced by THz technology in composite defect detection are summarized, including the limited resolution of THz imaging, which makes it difficult to provide sufficient detailed information, the non-uniformity of composite structure that leads to the complexity and variety of THz wave propagation inside the composite material, and the existing defect identification techniques and equipment, which limit the application of THz in composite defect detection; Finally, an outlook on the future development direction is given, which includes improving the imaging system to increase the THz imaging resolution while reducing the detection time, combining artificial intelligence and deep learning techniques to improve the fast and accurate identification of different types of defects, improving the real-time data processing to achieve online inspection of composites, and optimizing the composite defect detection platform. These development directions will further promote the application of THz technology in composite materials, improve detection efficiency and accuracy, and promote the development of composite material manufacturing and application.

Cavity Ring-Down Spectroscopy (CRDS) is a highly sensitive trace gas concentration measurement technique in which the processing of ring-down time is crucial. This paper adopts the Kalman filter to process the cavity ring-down spectroscopy to reduce the measurement error introduced by noise during the collection and real-time measurement process. This method preprocesses with the traditional filtering method to obtain the observation noise covariance v2(R) of the Kalman filter parameters, adjusts the process excitation noise covariance W2(Q), and evaluates the filtering effect to optimize the measurement results. Using simulated ring-down signals with white noise, the linear regression summation method (LRS) fits the background rendering-downtimesto perform Kalman filtering. From four aspects of mean, standard deviation, residual standard deviation (RMSE), and different noise levels, the appropriate Q value range is obtained, which is less than 1×10-7 and 0.001, respectively. An experimental gas detection system based on CRDS technology is constructed, using a 405 nm center wavelength diode laser and a high-reflectivity mirror with a reflectivity of over 99.99%, with NO2 as the target gas, and the background ring-down time and ring-down time are processed and analyzed using Kalman filtering. The experimental results show that: (1) Selecting a Q value less than 1×10-7 for Kalman filtering of the background ring-down time increases the lowest detection limit by 9.12 times and reaches 4.9×10-11 after filtering; (2) Taking Q value of 0.001 for processing the ring-down time retains the time response information and achieves significant noise reduction; (3) The system’s time resolution is 1 s, and compared to the method of reducing time resolution to improve detection limit in the past, the Kalman filtering method improves the system’s sensitivity. The agreement between experimental and simulated results verifies the effectiveness of Kalman filtering in stability and noise reduction. Applying the Kalman filtering method in the CRDS spectroscopic detection of gases is practical and provides methods and references for optimizing other gas measurement results.

The liquid’s optical constants (extinction coefficient and refractive index) can be determined by the spectral inversion method, among which the double-thickness transmission is most representative. Since the liquid itself cannot form a definite shape, it needs to be stored in a transparent container (liquid cell), so the spectral transmittance obtained through experimental measurement includes the influence of the optical constant of the liquid cell, which makes the spectral transmittance equation established based on the double-thickness transmission method extremely complex and difficult to obtain an analytical solution. Usually, the inversion method is used to calculate the optical constant of the liquid. The existing inversion methods have the following problems: first, the inversion iteration consumes time; second, the inversion iteration will introduce errors; and third, the liquid refractive index obtained by the inversion method has a binary problem. To solve the above problems, based on the three-layer medium structure (liquid cell), considering the multiple reflections of light on the interface of the two media, a set of spectral transmittance equations satisfying the integral ratio of liquid thickness is established. The polynomial equation related to the extinction coefficient is obtained through algebraic operation, and the extinction coefficient is calculated by solving and selecting the real number root greater than 0 and less than 1. In addition, the quadratic equation about the reflectance of the optical window of the liquid cell is solved. The reflectance of the interface between the liquid and the container is calculated with the root greater than 0 and less than 1, and two values of the liquid refractive index are obtained. Then, the liquid cell made of another material is used to measure the spectral transmittance of the liquid, and then combined with the extinction coefficient that has been obtained for related calculation, two other values of the liquid refractive index are obtained, and the refractive index of the liquid is the same one by selecting from the four values. As an application example, this paper selects the optical constant of water in the literature at 0.5～1.0 m as the “theoretical value”, and the quartz and polymethyl methacrylate glasses with known optical constants are taken as the liquid cell materials. Without considering the instrument measurement error, the above literature data are substituted into the spectral transmittance equation, and the calculated transmittance is taken as “experimental data”. Then, the optical constants of water are determined by finding the roots of polynomials, and the results are in full agreement with the “theoretical values”. The simulation process and calculation results show that the new method is available and solves the problems of the inversion method such as time-consume, iteration error, and the binary of refractive index, and provides a new option for determining the optical constants of liquids.

During spacecraft reentry into Earth’s atmosphere, the temperature of the high-speed air at the front of the spacecraft can reach over 10 000 K, leading to changes in the aerodynamic performance of the spacecraft. Radiative heating is one of the main sources of surface heat transfer for spacecraft, and measuring the radiation spectra of air under high-speed shock waves is crucial for studying radiation effects and dynamic behavior during the flight of hypersonic vehicles. There is still a lack of measurement data for high-temperature air radiation spectra, and a large amount of experimental data is required to improve and optimize computational models. This paper uses a hydrogen-oxygen detonation shock tubecombined with a transient spectroscopic measurement system to obtain the radiation spectra of high-temperature air in the ultraviolet-visible region at velocities ranging from 3 259～8 218 m·s-1. The identification and simulation of typical high-temperature air radiation spectra characteristic spectra were carried out. The simulation results were in good agreement with the experimental results, validating the accuracy and reliability of the spectral identification. The impact of shock wave velocity on the radiation characteristics of high-temperature air was also analyzed, and the dynamic behavior of characteristic spectra was analyzed. The results showed that characteristic spectra mainly exist in the 225~675 nm wavelength range and the region with wavelengths below 500 nm. Different characteristic spectra were observed in the radiation spectra at different velocities. With increasing shock wave velocity, OH(A-X), NO(, , ), NH(A-X), N2(C-B), N2+(B-X), and H spectral lines appeared successively. The relative intensities of the characteristic spectra vary with shock wave velocity. Finally, the relaxation time of post-shock air excitation was determined based on the time-varying curve of radiation intensity. The relaxation time gradually shortened with increasing velocity and exhibited an exponential relationship with velocity. This study provides experimental data and references for spacecraft thermal protection design and the validation and optimization of computational models for hypersonic reentry into the atmosphere.

Absorbing materials are an effective supplementto external stealth technology. After electromagnetic waves enter the interior of absorbing materials, they should be rapidly and maximally attenuated. Therefore, when we design the electromagnetic parameters of absorbing materials, impedance matching should be considered to maximize the entry of electromagnetic waves into the material. On the other hand, it is necessary to ensure that all electromagnetic waves entering the material are attenuated as much as possible. For absorbing materials, it means that the material must have a large imaginary part of the dielectric constant and magnetic permeability, which can cause attenuation of electromagnetic wave energy. Traditional microwave-absorbing materials mainly focus on microwaves. With the development of technology, terahertz radar has been put into use. It is necessary to study and explore whether traditional microwave absorbing materials can effectively attenuate in the terahertz and even infrared frequency bands and evaluate these absorbing materials in the terahertz frequency band. Two types of absorbing material plates are studied in-depth by the terahertz time-domain spectrum, Fourier transform spectral system, and microwave anechoic chamber measurement system based on a vector network analyzer. The microwave, terahertz, and infrared bands measured the reflectivity of radar-absorbing materials from 1 GHz to 180 THz.The measurement results indicate that as the frequency increases, the surface morphology and roughness of the absorbing material have a certain impact on the reflectivity. Further research has been conducted on the changes in the reflection characteristics of the absorbing material with angles in the terahertz and infrared bands. To further study the absorption principle of the absorbing material in the terahertz and infrared frequency bands, the absorbing layer was peeled off from the metal lining plate and measured. The reflectivity and transmittance of the absorbing coating at 2~180 THz were obtained, and the equivalent dielectric parameters of the composite absorbing material were calculated based on the Kamers-Kronig relation. The test results showed that the absorbing material still has a certain absorption ability in the terahertz and infrared. Infrared is not sufficient to penetrate the reflective coating of the absorbing material to reach the metal lining plate, and its absorption mainly comes from reflection absorption and scattering from the surface of the absorbing material.

The 13C urea breath test is widely used as the “gold standard” for detecting Helicobacter pylori domestically and internationally. Accurate measurements of carbon and oxygen isotope characteristics in CO2 are significant for disease diagnosis. Tunable diode laser absorption spectroscopy (TDLAS) has been widely used in multiple fields due to its simple structure, fast response speed, and high sensitivity. It is also fully suitable for the measurement and research of gas isotopes. In this study, a quantum cascade laser with a central wavelength of 4.32 m and a small volume gas absorption cellwith an optical path of 14 cm and volume of 44 mL was used to simultaneously measure the volume concentrations of 16O12C16O，18O12C16O and 16O13C16O in CO2 based on direct absorption spectroscopy. In addition, the noise interference caused by the stability of the light source in the direct absorption spectrum system and gas sample fluctuations were reduced using the Back Propagation (BP) neural network model. The results showed that the measurement accuracy and stability of isotope abundance based on the BP neural network model were better than those of the absorbance peak ratio method. The concentration measurement accuracy of 16O13C16O and 18O12C16O increased by about 1.27 and 1.58 times, respectively. Meanwhile, the Allan variance analysis also showed that when the integration time was 106 s, the accuracy of 13C and 18O isotope abundance using the BP neural network model was 0.97‰ and 1.47‰, respectively, which improved about 2.1 times and 1.2 times compared to the absorbance peak ratio method. This fully proves the feasibility of the isotope abundance measurement method based on the BP neural network model, laying the foundation for developing high-precision isotope abundance sensors.

Currently, the traditional measuring methods ofgrain quality are mainly the traditional separation and manual inspection, which take a long time and have low efficiency. Near Infrared (NIR, 780~2 500 nm) spectral analysis technology has the advantages of a wide range of applicable samples, high accuracy of quantitative measurement, high measurement efficiency, and non-destructive testing, which is widely used in agriculture online or rapid measurement. Currently, the existing NIR instruments measuring grain quality are expensive, which prevents a wider application of this kind of device. Moreover, the predicting model is limited in applicability due to the differences ingrains in different seasons and regions. To solve these problems, in this study, a new type of NIR spectrometer system is developed to measure wheat quality. The system uses a control system developed with Python. By setting and modifying the acquisition parameters, the three steering gears and weight sensors are integrated to control the spectra data acquisition. The spectral data are preprocessed and substituted into the model to calculate the quality parameters of the target wheat samples. The principal component analysis (PCA) method removes the outlier’s spectral data. Then, the selected spectral data are preprocessed by recursive mean filtering and standard normal transformation (SNV). Finally, the optimized model is obtained with the partial least squares regression (PLS) method after competitive adaptive reweighting sampling (CARS) wavelength selection. The prediction model is currently developed for moisture, wet gluten, and whiteness of wheat. The results show that this model can effectively reduce the error caused by stray light, sample uniformity, and other effective factors. The developed NIR spectrometer system can satisfy the requirements of grain acquisition and storage.

Electron density is one of the key fundamental parameters of plasma discharges. H is the most used spectral line for spectroscopic diagnosis based on the Stark broadening method. The van der Waals broadening, which is related to the gas temperature, makes an important contribution to the broadening of the H line at atmospheric pressure. To extract the Stark broadening width, the gas temperature should be determined in advance from the rotational temperature of molecules, resulting in inevitable errors in measuring. During the nonlinear parameters fitting processes of a spectral line, the errors in gas temperature will transfer to electron density measurement. This work proposes combining a random forest regression model based on machine learning and a Stark broadening method based on optical emission spectroscopy. Compared with the error characteristic of the traditional least square method, this method is found to have a good performance in robustness and generalization capability so that it could diagnose the electron density of plasma more precisely and quickly. Because of the different states of plasma discharges, the training set of H standard theoretical line used for the machine learning is simulated by the model of spectral line broadening, in which the random errors are introduced into the gas temperature. A sample set, combined with the spectral line’s intensity distribution with each group’s temperature deviation and the corresponding electron density, is employed to train the random forest model. The hyperparameters (i.e., the minimum number of leaf nodes and the number of decision trees) that minimize the mean square error of the model are set to 2 and 100, respectively. It is found that the average relative error between the results predicted by the random forests regression model, which is well-trained, and the actual values are less than 3%. The model was evaluated by a test set of spectral data with a temperature error range of 0~±10%. With the increase in temperature error, the prediction results of the random forest model are better than those of the least squares method. When the error of gas temperature is ±10%, the mean squared error of predicted electron density is reduced by more than 30% compared with the least squares method. In the training set of spectral data, when the error of gas temperature introduced into the training set is in the range of 0~±10%, the minimum mean squared error of electron density is achieved, and the robustness of the model is better than that of the least squares method. However, the prediction results of the model become inaccurate when the temperature error introduced into the training set is beyond ±10%. In addition, the time spent analyzing the spectral line by the model, which is well-trained, is much less than that by the least square method.

The study of natural organic binders in historical objects is one of the research highlights in cultural heritage. However,identifying and classifying modern binding media used for restoration is also very important in heritage conservation. It is necessary for the subsequent restoration of ancient ceramics to understand the composition of the binding media used in the restoration of ancient ceramics. To address this problem, a batch of modern binding media for restoring ancient ceramics was analyzed by Fourier Transform infrared spectroscopy (FTIR), Pyrolysis-gas chromatography/mass spectrometry (Py-GC/MS) and thermally assisted hydrolysis-methylation pyrolysis-gas chromatography-mass spectrometry (THM-Py-GC/MS) techniques. The results showed that the characteristic absorption peaks of the infrared spectrum of the binding media at 1 252 and 829 cm-1 were in line with the range of characteristic absorption peaks of epoxy compounds and matched with the standard spectrum of epoxy resin in the infrared spectrum library, which indicated that the binding media was an epoxy adhesive. Moreover, additives such as pyridine compounds, benzoate compounds and primary alcohol compounds are also added to the binding media. Then, the chemical composition of the binding media was analyzed by applying Py-GC/MS and THM-Py-GC/MS techniques, and the binding media was detected as a bisphenol A-type epoxy resin. Moreover, to improve the performance of the epoxy resin adhesive, amine curing agent, diisooctyl phthalate plasticizers and diisooctyl phthalate inactive diluent were also detected to be added to the binding media. THM-Py-GC/MS technique affects the analysis and identification of amine curing agents in this binding media due to the presence of methylation reagents, whereas the Py-GC/MS technique is more suitable for the analysis and characterization of epoxy adhesives. In this study, the use of the Py-GC/MS technique can identify the types of modern binding media used for the restoration of ancient ceramic relics, providing a new approach for the analysis and identification of epoxy resin adhesives, as well as a new perspective for the identification of unknown restoration materials for cultural relics.

Fourier transforms infrared (FT-IR) spectra in the range of 400~4 000 cm-1 was ollected for 2,5-dichloropyrimidine (2,5-DCP) in solid phase as well as in liquid phase using four sample preparation methods: KBr pressed (KBr), mineral oil (Nujol), attenuated total reflection (ATR) and melting (Liquid), while Fourier transforms Raman (FT-Raman) and laser Raman (Laser-Raman) spectra in the range of 80~3 200 cm-1 was also recorded. To correctly interpret the experimentally obtained vibrational spectra, the geometry of 2,5-DCP was first optimized by applying 14 methods from density function theory (DFT) as well as second-order perturbation method (MP2), based on which its harmonic frequency, infrared intensity and Raman activity were obtained, followed by the conversion of Raman activity to Raman intensity. To consider the anharmonic effect, the perturbation calculation is performed near the equilibrium geometry to obtain the third and fourth-order force fields in normal coordinates, and the anharmonic vibration frequency and intensity of 2,5-DCP are obtained based on the vibration second-order perturbation (VMP2) theory. It is found that the anharmonic vibration frequencies calculated by B3LYP and B3PW91 have the smallest difference from the experimental values. Based on the preferred B3LYP method, the effect of the basis sets on the vibration frequency continued to be investigated, eight basis sets were adopted, and it was found that the difference between the 6-311++G(2pd, 2df) level and the experimental values was the smallest, with a root-mean-square error(RMSE) of 6.75 cm-1 (4.63 cm-1 under 22 vibration modes), the 6-311++G(d, p) greatly reduced the calculation time, while the accuracy of 6-311++G(d, p) is not much lost (6.79 cm-1). In summary, the anharmonic vibrational spectra calculated based on the B3LYP method combined with the 6-311++G(2df, 2pd) basis set are the best choice for assigning the experimental vibrational spectra of the 2,5-DCP. Then, according to the optimal calculation results and the vibrational fundamental frequencies obtained by the scaling factor method, combined with the anharmonic vibrational intensity, the schematic diagram of the normal coordinates analysis, the potential energy distribution (PED) of the vibrations, and compared to the experimentally acquired infrared and Raman spectra, all fundamental frequencies and some overtones of the 2,5-DCP were assigned, and two vibrational couplings were found, One is caused between 3 054 cm-1 and the combination tones of 1 554 and 1 540 cm-1; the other is from the coupling between 1 132 cm-1 and the sum frequency 793+351 cm-1 and the difference frequency 1 370～230 cm-1. Finally, the anharmonic vibrational spectra of 2,5-DCP under multiple isotopic substitutions were expected and the correctness of the attribution was checked again.

To quickly and non-destructively monitor the degree of cadmium contamination in lettuce, visible-near infrared spectroscopy is used to classify cadmium contamination in lettuce. Lettuce leaves’ visible-near infrared reflectance spectra were collected to analyze the variation in reflectance spectra under different cadmium pollution levels (0, 5, 10, 20 mg·kg-1) in soil, with lettuce as the research subject. The spectral analysis reveals that within the wavelength range of 510 to 730 nm, the reflectance of lettuce leaves in the visible-near infrared spectrum decreases and then increases with the increase in cadmium content in the soil. Within the wavelength range of 730 to 799.53 nm, the reflectance of lettuce leaves under 5 and 20 mg·kg-1 cadmium stress is higher than the CK, while under 10 mg·kg-1 cadmium stress, the reflectance is lower than the control group. Additionally, an absorption valley was observed at 762.199 nm. In establishing a cadmium pollution monitoring model for lettuce, various preprocessing methods were applied to the raw spectra to improve the signal-to-noise ratio. These methods include smoothing (SG), multiplicative scatter correction (MSC), standard normal variate (SNV), mean normalization (MN), SG+MSC, SG+SNV, SG+MN, SG+first derivative (FD), and SG+second derivative (SD). Based on the principal component analysis (PCA), dimensionality reduction was performed on the original spectra and spectra subjected to various preprocessing methods. Subsequently, the reduced data was divided into training and testing sets in a 4∶1 ratio. These sets were then used to establish classification monitoring models for cadmium pollution in lettuce by combining particle swarm optimization-random forest (PSO-RF), genetic algorithm-optimized support vector machine (GA-SVM), backpropagation neural network (BP-NN), extreme learning machine (ELM), and Naive Bayes, followed by analysis and comparison. The results demonstrate that among the different models, the PSO-RF (SG) model achieves the best recognition performance, followed by the GA-SVM (SG+FD) model and the ELM (MSC) model. The training accuracy of the PSO-RF (SG), GA-SVM (SG+FD), and ELM (MSC) models is 100%, while their testing accuracies are 100%, 83.33%, and 79.17% respectively. On the other hand, the BP-NN model and the Naive Bayes model perform relatively poorly. The training accuracy of the BP-NN (SNV) model is 42.72% with a testing accuracy of 50%. The Naive Bayes (SG+FD) model achieves a training accuracy of 71.84% and a testing accuracy of 83.33%. It indicates that applying visible-near infrared spectroscopy combined with particle swarm optimization random forest modeling can provide a novel approach for studying the monitoring of heavy metal contamination in lettuce.

Traceability of apple origin and prediction of apple SSC is of great practical significance, and the purpose of origin discrimination and SSC prediction is achieved by modeling. To overcome the limitations of a single model, the overall prediction performance is improved by combining the prediction results of multiple models. Near-infrared spectroscopy (NIRS) detection technology combined with a multi-model decision fusion strategy is utilized for traceability identification of apple origin and prediction of apple SSC to verify the feasibility of the theoretical method.The spectra of apple samples were collected using a handheld near-infrared detector, and apple origin discrimination models were established using the sample spectra in combination with the random forest (RF) method, the partial least squares discriminant analysis (PLS-DA) method, and the support vector machine (SVM) method. The predictions from the three discrimination models are then used in a voting system decision fusion method to generate new discriminant results. Actual values of SSC were collected for all apple samples, and SSC prediction models were developed using the sample spectra and actual values of SSC combined with the random forest (RF) method, the partial least squares regression (PLSR) method, and the support vector regression (SVR) method. Using the outputs of the three regression models, the new SSC prediction is output through the weighting method decision fusion strategy. When the voting decision-making method was not used, the discrimination modeling using the RF method was the most effective among the three qualitative modeling methods, with a prediction accuracy of 88.71%. The worst prediction was made using the SVM method, with a prediction accuracy of 77.43%. After using the voting decision method, the accuracy of apple origin identification reached 93.42%, and its prediction precision and recall also reached a double high, both above 85%. All three quantitative modeling methods gave good results in predicting apple SSC without using the weighted decision fusion method. All three methods predicted coefficients of determination around 0.87 and root mean square errors of prediction (RMSEP) around 0.78. The prediction of the SSC level was improved after using the weighted decision fusion method. The prediction coefficient of determination was 0.91, and the RMSEP was 0.66. The feasibility of the proposed method was confirmed by using the multi-model decision fusion method in the identification of apple origin and the prediction of apple SSC to improve the accuracy of apple origin discrimination and the precision of the prediction of apple SSC. Meanwhile, the handheld NIR detector combined with the multi-model decision fusion method provides a new high-precision prediction approach for on-site non-destructive testing analysis.

The quality of the bacterial strain cultivation in the seed tank during the citric acid fermentation process directly affects the fermentation level. Hence, it is crucial to accurately and rapidly detect the quality parameters of the culture solution in the seed tank. However, these parameters are currently largely measured manually, which does not meet real-time monitoring and precise control requirements. This paper builds a chemometric model for measuring the total acidity (TA) and reducing sugars (RS) in the seed tank’s culture solution, based on near-infrared spectroscopy. Initially, the original spectra were analyzed, and to eliminate random noise and reduce batch variability effects on the sample spectra, the SG-DT method of smoothing (SG) and detrending (DT) were sequentially used for spectral preprocessing. Then, the Interval Partial Least Squares (iPLS) method was used for feature wavelength selection, the effect of different division intervals on the selection result was discussed, and the optimal division interval number for the target quality parameter of TA was determined to be 21, with 495 feature wavelengths. For RS, it was 20, with 361 feature wavelengths. Subsequently, the correlation between spectral variables and quality parameter variables was analyzed. A BP network was introduced to establish the calibration model for TA, and both PLSR and BP networks were used to establish the calibration model for RS, and model prediction effects were compared to determine the optimal model. Finally, the optimal prediction model for TA based on the BP network had an Rp2 of 0.808 5 and an RMSEP of 0.123 4. The model prediction effect of RS based on the BP network was superior to the PLSR model, with an Rp2 of 0.964 7 and RMSEP of 0.173 9. This paper has realized online prediction of multiple quality parameters during the bacterial strain cultivation process in the complex citric acid fermentation system, providing a basis for real-time intelligent control of the fermentation process.

X-ray emission spectroscopy is an in situ non-destructive method to obtain chemical species of elements, and the development of laboratory-type XES devices worldwide is still in the exploratory stage. Energy-Dispersive X-Ray Fluorescence Spectrometer (EDXRF) and Wavelength-Dispersive X-Ray Fluorescence Spectrometer (WDXRF) are widely used in geological, environmental and archaeological fields. However, although EDXRF has a simple structure and can realize rapid, non-destructive detection of multiple elements, the resolution is not ideal, the spectral line interference is serious, and the detection limit is poor. Although WDXRF can distinguish most elements with spectral overlapping characteristics in conventional applications, the structure and cost are complicated. To explore the development of laboratory-type XES devices, this study synthesizes the performance advantages of the Dual Dispersions and Dural Focuses X-Ray Spectrometer (DDF-XRS) design concept and successfully develops the principle prototype. The experimental data and analysis results show that this DDF-XRS spectrometer combines the advantages of both micro EDXRF and WDXRF with a simple structure, good signal-to-noise ratio, high resolution, and low detection limit characteristics. Through the wavelength-energy double dispersion technology, X-rays are firstly diffracted by the crystal to undergo wavelength dispersion, thus obtaining monochromatic light, and at the same time, the energy dispersion of the silicon drift detector can be used to observe the degree of monochromaticity, reduce the risk of misjudgment of spectral line interference, and improve the accuracy and reliability of the analysis results, which overcomes the limitations of the complex structure for WDXRF and the insufficient energy resolution of the EDXRF, and highlights the necessity and superiority of the double dispersion. This technology overcomes the limitations of WDXRF multi-element peak determination and EDXRF energy resolution, emphasizing the necessity and superiority of double dispersion. At present, the resolution of the DDF-XRS spectrometer is 45 eV, which can reduce the spectral line overlap of transition metal K to K peaks; at the same time, it significantly reduces the background of the continuum spectrum, and the optimal signal-to-noise ratio is >1 000; and the detection limit of Cr in the determination of geologic samples can be up to 0.26 mg·kg-1. With the application of DDF-XRS, the transition metal K1 and K2 spectral lines can be resolved to a certain extent, and it is expected that the resolution can be further improved by combining with linear or two-dimensional array detectors to realize the determination of X-ray emission fingerprint spectra to obtain the chemical forms of the analyzed elements. Since the current crystal properties cannot fully resolve the overlapping spectral lines of transition metals, the search for curved crystals with high fractionation capacity and diffraction intensity will be the next research focus.

Laser-induced breakdown spectroscopy has showngreat-potential for application in the online analysis of molten steel composition. However, detecting the important C, P, and S component elements in molten steel has been challenging because their effective emission lines cannot be transmitted over long distances in the air. This paper studies the influence of the argon environment on detecting C, P, and S elements in steel under a 1.5-meter long probe gun; it is found that an argon flow rate that is too large or too small is not conducive to spectral measurement. When the gas flow rate is set to 11 L·min-1, the spectrum obtained is the most stable, and the spectral line intensity is the largest. Using 14 standard steel samples, the three elements C, P, and S were detected and quantitatively analyzed under optimal gas flow. After the internal standard calibration, the limit of detection (LOD) of the three elements were 0.009%, 0.04%, and 0.015%, the relative standard deviations (RSD) were 2.34%, 1.05%, and 1.01%, and the correlation coefficients (R) were 0.998, 0.997 and 0.987, respectively. The root mean square errors (RMSE) were 0.02%, 0.02% and 0.03%, respectively. The research results of this paper verify the effectiveness of the design of the probe gun and provide an important design basis for the online analysis of C, P, and S in the composition of molten steel.

With the policy of “carbon peaking and carbon neutrality” put forward in China, carbon emissions in the mining area have become the focus of attention. The study was based on soil samples taken from the mine areas, combined with 6 mathematical transformation methods (R, R, Log(1/R), 1st, MSC, SNV) and spectral feature screening methods (CC-SPA). This study explored the hyperspectral response characteristics of soil carbon emissions under different land use types in Hongshaquan Open-pit Coal Mine in Xinjiang; combined with soil temperature (ST), Soil moisture (SM) and 6 kinds of spectral indexes (NDVI, RVI, NGLI, SMM, SI-T, ATI), using partial least squares (PLSR), support vector machine (SVM), random forest (RF), genetic optimization neural network (GA-BP) algorithm to obtain the optimal remote sensing of soil carbon emissions inversion model. The main conclusions are as follows: (1) The reflectance of soil in the non-mining affected area is higher than that in the mining affected area under natural conditions, and the southern line is the most affected by coal mining and has the lowest reflectance, which proves that mining activities have an impact on the mining area soil; (2) Spectral characteristics In terms of screening, the number of carbon emission characteristic bands extracted based on the correlation coefficient-continuous projection algorithm (CC-SPA) is much smaller than that of the correlation coefficient method (CC) and the continuous projection algorithm (SPA), and the characteristic bands present a certain clustered distribution. In the wavelength range of 1 600～2 200 nm, the number of characteristic bands during the day is much higher than at night. Compared with the daytime, the characteristic bands at night have the characteristics of obviously shifting to long waves. (3) Adding the spectral index based on reflectivity and the inversion model of ST and SM can significantly improve the accuracy of estimating soil carbon emission rate. The support vector machine (SVM) model based on the first-order differential transformation (1st) can invert the mining area. Comprehensive land use types have the best effect on soil carbon emissions (validation set R2=0.813, RMSE=0.116); the optimal combination of soil carbon emission indices for five different land use types is different, and the introduction of different spectral indices has a significant effect on soil carbon emission rates. The estimation accuracy has been improved to varying degrees (the verification set R2 is above 0.8), and the optimal soil carbon emission inversion model can more accurately estimate the carbon emission rate of different land use types in the Hongshaquan mining area. This study can provide a basis for the remote sensing inversion of soil carbon emissions in desertified mining areas, quantitatively identify the carbon source-sink effect of soil under different land use types and realize the non-destructive detection of carbon emissions in mining areas, providing support for my country’s “30·60” double carbon goal. Data support.

Integer-order derivative methods (such as 1st or 2nd order) are traditional preprocessing methods for soil heavy-metal inversion models, which ignore the fractional-order spectral reflectance information associated with the target variable. Fractional order derivative (FOD) can flexibly select the differential order to enhance the spectral signal effectively. This study focused on the farmland soil in Mojiang Hani Autonomous County, Pu’er City, Yunnan Province, China. Sixty-one soil hyperspectral reflectance information and soil heavy metal content data (zinc and nickel) were measured. The spectral reflectance information underwent 0 to 2 fractional-order derivative preprocessing with intervals of 0.05. The preprocessed spectral reflectance information at each order was input into the Successive Projections Algorithm (SPA) to select characteristic bands. Subsequently, three soil heavy metal prediction models were separately established using Partial Least Squares Regression (PLSR), Random Forest (RF), and Bagging methods. The results show that after the fractional order derivative processing from 0 to 2 orders (41 orders in total with an interval of 0.05), the overall spectral intensity gradually weakens and gradually approaches zero with the increase of fractional orders. The spectral absorption band gradually narrows, and the differences between different spectral curves gradually decrease. As the derivative order increases, more abundant peaks and valleys are produced. The best-order models based on fractional derivatives are better than the original spectral model and the integer order model, and most of the better orders of the model are concentrated in low-order fractional orders. For heavy metal zinc, the best prediction model accuracy was achieved by the RF model of 0.75 order (R2=0.675, RMSE=6.149, RPD=1.755), followed by the Bagging model of 0.75 order (R2=0.633, RMSE=6.534, RPD=1.652), and the lowest was achieved by the PLSR model of 0.25 order (R2=0.551, RMSE=7.230, RPD=1.493). For the heavy metal nickel, the best prediction model accuracy was the RF model of order 0.80 (R2=0.854, RMSE=127.823, RPD=2.618), the Bagging model of order 0.80 was the next best (R2=0.841, RMSE=133.304, RPD=2.510), the PLSR model of order 0.40 lowest (R2=0.762, RMSE=163.162, RPD=2.051). Visible, the nonlinear models (RF and Bagging) constructed based on FOD preprocessing and SPA dimensionality reduction in this study have certain applicability in estimating heavy metal content in farmland soil. They can be a reference for predicting heavy metal content in similar regions.

To solve the problem of low data classificationaccuracy and poor model stability of functional near-infrared brain functional imaging signals in brain-computer interfaces, this paper proposes to study the changes of cerebral blood oxygen concentration in the prefrontal brain region during the stimulation period and carry out a study of the state binary classification identification of task and rest on the experimental data of the prefrontal brain region, and take the extracted single feature and multi-feature fusion as the inputs of the model, respectively, and validate the results of the classification through the model The conjecture that multi-feature fusion can improve the classification accuracy to some extent. Firstly, this paper designs an experimental paradigm for the prefrontal brain region: word fluency cognition experiment. The device used to collect the data is NirSmart, a Danyang Huichuang device from China, with a sampling rate of 11 Hz and a spatial resolution of 3 cm. The device uses avalanche diode and ultramicro photo detection technology, and the sensitivity is up to 0.1 pW. After that, the collected data are preprocessed using the Homer2 toolbox, the features are extracted using MATLAB, and the Random Forest model is constructed using two metrics of feature importance and error curve to evaluate the classification results. Importance and error curves were used to evaluate the performance of the model. Finally, the average of 20 runs of the random forest model was used as the final classification result. The experimental results show that multiple features can improve the final classification result compared with a single feature. The best second classification result for the case of three-feature fusion is 93.84%, 2.32%, 4.25%, and 5.33%, higher than the single-feature mean, slope, and peak-to-peak value, respectively. From the results of the experimental data, it can be seen that multi-feature fusion can improve the accuracy of near-infrared brain functional imaging classification to a certain extent. The performance of the random forest model is stable, which is expected to solve the previous problems of low classification accuracy and poor model stability and promote the development and application of brain-computer interface systems.

The regulation of circadian rhythm by light is a non-imaging visual function.Non-imaging visual effect cells, mainly intrinsically photosensitive retinal ganglion cells, significantly impact the body’s biological rhythm, metabolism, cognition, etc., primarilymediated by blue light. This study aimed to explore the effects of night light exposure on liver circadian rhythm and inflammation. The gerbils were exposed to white and blue light for 1 or 3 hours every night for 10 weeks. The real-time reverse transcription-polymerase chain reaction was used to detect the expression rhythm of liver clock genes; transcriptomics was used to analyze the expression level of all liver genes; hematoxylin-eosin staining was used to detect liver tissue morphology, and immunofluorescence staining was used to detect the level of inflammatory factors in liver tissue. The results showed that light exposure at night significantly disrupted the liver’s expression rhythm of Arntl, Clock, Cry1, Nr1d1, Per2, and Rora. Light exposure at night had the most significant effect on liver immune system gene expression. Light exposure at night induced hepatocyte edema and up-regulated the expression of pro-inflammatory factors interleukin 17A and granulocyte-macrophage colony-stimulating factor. In short, exposure to light at night disrupted liver circadian rhythms and triggered an inflammatory response in the liver. This study highlights the critical role of non-imaging visual pathways in hepatic metabolic homeostasis.

Plant leaf physical and chemical parameters, such as leaf chlorophyll content, carotenoid content, water content, protein content, and Carbone-based constituents content, are crucial for accurately monitoring plant growth status. In recent years, with the rapid development of deep learning technology in vegetation remote sensing, the combined use of deep learning and hyperspectral remote sensing for plant leaf parameters estimation has been widely applied; however, currently, few leaf parameters estimation works based on the combination of deep learning and hyperspectral remote sensing technology have been conducted. This study explores the possibility of estimating leaf chlorophyll, carotenoid, water, protein, and Carbone-based constituent content by combining hyperspectral remote sensing and deep learning techniques. The main work of this paper is to propose a leaf physical and chemical parameter estimation method based on hyperspectral remote sensing and deep learning. Firstly, this study determines the sensitive spectral regions of multiple vegetation leaf physical and chemical parameters based on the PROSPECT-PRO radiative transfer model. Then, we designed a LeafTraitNet deep learning model; the LeafTraitNet model is trained and tested based on the lobex93 dataset, and a high-precision leaf parameter estimation result is obtained. The conclusions of this study are as follows: (1) It is vital to select leaf spectral absorption features based on the PROSPECT-PRO radiative transfer model. The leaf chlorophyll (434 and 676 nm) and carotenoids (445 nm) spectral absorption regions are located in the visible bands. However, the absorption regions with the most significant correlation coefficients (absolute values) are not their maximum spectral absorption bands, which the mutual influence of leaf chlorophyll and carotenoid absorptions may cause. (2) The leaf water spectral absorption regions are mainly located in the bands 950～2 500 nm, which overlaps with the spectral absorption regions of leaf protein and carbon-based component content, thus weakening the hyperspectral remote sensing estimation accuracy of the latter. The correlation coefficients between leaf protein (and carbon-based component content) and the spectral reflectance in the 950～2 500 nm range are notably lower than the leaf water. The correlation coefficients analysis results of leaf parameters and hyperspectral for the PROSPECT-PRO radiative transfer model and lobex93 dataset show similar correlation coefficients. (3) The three traditional methods and the LeafTraitNet model can be ranked as LeafTraitNet (total nRMSE=0.84)<RF (total nRMSE=1.59)<MLP (total nRMSE=1.73)<MLR (total nRMSE=1.74), which means the leaf parameters estimation performance is notably higher than RF, MLP, and MLR. However, further experiments are needed to validate the LeafTraitNet model at the canopy scale.

Dissolved organic matter (DOM) is the most active component in soil organic matter, affecting heavy metal’s migration and transformation. Studying their correlation is of great value for environmental monitoring and pollution assessment. In this study, the topsoil (0~30 cm) was collected from the typical coniferous and broad-leaved mixed forest in Heilongjiang Province Maolan’gou National Nature Reserve, located in Xiaoxing’anling, China, and three-dimensional fluorescence spectroscopy- parallel factor analysis method was used to reveal the characteristics of forest soil DOM fluorescence spectra, and further to analyze the correlation between DOM fluorescence component and heavy metal content. The results showed that the forest soil DOM fluorescence index ranged from 1.468 to 1.635, with an average value of 1.531, indicating the DOM source had both autogenic and exogenous characteristics. The biological index ranged from 0.563 to 0.646, with an average value of 0.603, indicating a low contribution rate of recent autogenic source; The humification index ranged from 4.607 to 8.993, with an average value of 6.491, indicating a low degree of humification. Threetypes of fivekinds of fluorescent components were identified from the forest soil DOM fluorescence spectrum, including humus-likesubstances ［ultraviolet fulvic acid-like component (C1) and visible fulvic acid-like component (C2)］, humic acid-like substance (humic acid component, C3), and proteinlike substance ［tryptophanlike component(C4) and tyrosinelike component (C5)］. Humus-like substances accounted for the largest proportion of total components (60.12%), significantly higher than humic acid-like and protein-like substances, while humic acid-like substances accounted for the smallest proportion (11.25%) of total components. C1, C2, and C3 had a significant positive correlation, while C5 was significantly negatively correlated with the other four fluorescent components. There was a significant correlation between the fivekinds of fluorescent components and the fluorescence index, respectively, with only C5 showing a positive correlation with the fluorescence index.The forest soil heavy metal contentswere significant differences in the spatial distribution, with significant positive correlations between As and Cr, Cu and Zn, Ni and Zn, Hg and Pb, and significant negative correlations between Hg and Cu, Ni and Zn, Cr, and Pb, respectively.The correlationsamong the three types of fluorescent substances, Cr and Pb, were not significant, but all fluorescent substances had a significant correlation with Zn. Moreover, there was a significant correlation between humus-like substances and Cu and Hg; humic acid-like substances and Cu, Hg, and Ni; protein-like substances and Ni, respectively.The results of this study provide basic data for monitoring the forest soil environment of Maolan’gou National Nature Reserve and provide references for the assessment of heavy metal pollution in the soil of typical coniferous and broad-leaved mixed forests in Xiaoxing’anling.

Timely and accurate crop yield estimation is crucial for making informed decisions regarding crop management and assessing food security. This study aims to develop a method that combines continuous wavelet transform (CWT) with machine learning to predict wheat yield accurately. This research is based on the spectral data of canopy height and yield data obtained from two-year field trials conducted during wheat growth’s flowering and filling stages in 2020—2021. Initially, CWT is employed to extract three wavelet features (WFs), namely Bortua-WFs based on the Bortua method, 1% R2-WFs representing WFs along with the top 1% determination coefficient for wheat yield, and SS-WFs encompassing all WFs under a single decomposition scale. Subsequently, three machine learning algorithms Random Forest (RF), K-nearest neighbor (KNN), and extreme gradient Lift (XGBoost) are utilized to construct the yield prediction model. Finally, optimal spectral features are selected using the same methodology for modeling and comparison purposes. The results demonstrate that: (1) all three WFs models combined with machine learning methods perform well, with higher accuracy and stability observed in the model built based on Boruta-WFs. (2) Compared to the spectral characteristic model, improved accuracy was achieved by utilizing Bortua-WFs at each growth stage; specifically, an increase in R2 accuracy by 17.5%, 4%, and 39.6% during flowering stage, as well as an increase by 8.4%, 5.6%, and 16.9% during filling stage respectively were observed across different models.(3) The estimation model at the grouting stage outperformed that at the flowering stage; particularly noteworthy was the performance of XGBoost when combined with Bortua-WFs, which yielded an R2 value of 0.83 accompanied by an RMSE value of 0.78 t·ha-1. This study compared the performance of different characteristics and methods. It determined the best model accuracy under different schemes, which can provide technical references for the accurate wheat yield prediction by spectral technology.

Spectral confocal displacement sensor is a new geometric precision measurement sensor with high accuracy, high efficiency, and non-contact technical advantages. It is now widely used in measuring micro or macro geometric quantities. Conventional geometric measurement sensors use contact mechanical probes, which cause damage to the surface of the object, making it difficult to meet the needs of non-destructive measurement in modern manufacturing. Unlike conventional optical systems that require correction for axial chromatic aberration, spectral confocal displacement sensors use axial chromatic aberration to establish the relationship between displacement and wavelength. However, most current research on spectral confocal displacement sensing technology has focused on point sweep. This technique can only obtain the geometric information of a single point, which greatly limits the efficiency in the practical application of precision measurement of larger areas and requires high back-end data processing and cumbersome data reconstruction. This study designs a line-swept spectral confocal displacement sensor system with submicron resolution to address this technical drawback. This study analyzes the principle of the line-swept spectral confocal displacement sensor and the detailed design of a large-range dispersive objective lens and a high spectral resolution spectral spectroscopic unit is carried out. By optimizing the optical path structure of the dispersive objective and spectral spectroscopy unit and balancing the aberrations, the RMS radius of each field of view of the full system is less than 5.5 m, and good imaging quality is obtained. The results show that the full system has a resolution of 0.8 m at a scan line length of 10 mm and an axial range of 3mm. This study has a broad application prospect in high-efficiency and high-precision geometric precision measurement.

Based on tunable semiconductor laser absorption spectroscopy (TDLAS) and frequency division multiplexing (FDM) method, a near-infrared multi-component trace gas identification and detection system based on support vector machine (SVM) classification was studied. When laser spectroscopy technology characterizes gas absorption spectral lines, the absorption capacity of gas in the near-infrared band is lower than that in the far-infrared band. The absorption signal of gas detected by single-band laser spectrum is weak, and each gas component interferes with each other greatly. To improve detection accuracy, accurately identify gas components and perform multi-component detection at the same time, based on tunable semiconductor laser absorption spectroscopy technology, the frequency division multiplexing near-infrared TDLAS technology method is used, and the SVM classification algorithm is used to perform the real-time detection process of mixed gases. It effectively avoids cross-interference of various gases and realizes trace detection of eight gas markers: nitric oxide NO, hydrogen sulfide H2S, ammonia NH3, nitrogen dioxide NO2, acetylene C2H2, carbon dioxide CO2, methane CH4, and hydrogen chloride HCl. When eight lasers work simultaneously, the system controls the band-pass filter to perform time-sharing filtering. It sequentially transmits the second harmonic data after differential phase locking to the host computer for real-time display. The recognition rate is over 96.3%, and the average content prediction accuracy is higher than 99.6%. It has achieved high-precision detection results with the lowest detection limit of CH4 being 0.01 L·L-1, NO2 being 0.05 L·L-1, and C2H2 being 0.03 L·L-1, and the detection limits of other gases are below 5 L·L-1. Conduct anti-interference analysis and detection lower limit analysis on the multi-channel detection of the system to verify that the system can achieve high-precision concentration detection of mixed gases when the system is operating stably. This system uses a distributed feedback laser drive and lock-in amplifier combined with the SVM algorithm model of data processing to realize multi-component trace gas identification and detection of near-infrared TDLAS technology, which can meet the trace level detection of trace gases and provide ultra-low performance for the future. The detection of concentration mixed gases is of very important significance.

Hyperspectral Imagery (HSI), based on its high-resolution spatial and spectral information, has important applications in military, aerospace, civil, and other remote sensing fields, which has great research significance. Deep learning has the advantages of strong learning ability, wide coverage, and strong portability, which has become a hot spot in the research of high-precision hyperspectral image classification. Convolutional Neural Networks (CNN) are widely used in the research of hyperspectral image classification because of their powerful feature extraction ability and have achieved effective research results. Still, such methods are usually based on 2D-CNN or 3D-CNN alone. For the single feature of hyperspectral image, the complete feature information of hyperspectral data cannot be fully utilized. Secondly, the local feature optimization of the corresponding extraction network is good, but the overall generalization ability is insufficient. There are limitations in the deep mining of spatial and spectral information of HSI. Because of this, this paper proposes a Hybrid Spectral Convolutional Neural Network Attention Mechanism (HybridSN~~AM) based on attention mechanism. The principal component analysis method is used to reduce the dimension of hyperspectral images, and the convolutional neural network is used as the main body of the classification model to screen out more distinguishable features through the attention mechanism so that the model can extract more accurate and more core joint space-spectral information, and realize high-precision classification of hyperspectral images. The proposed method was applied to three datasets: Indian Pines(IP), the University of Pavia (UP), and Salinas (SA). The experimental results show that the overall classification accuracy, average classification accuracy, and kappa coefficient of target images based on this model are higher than 98.14%, 97.17%，and 97.87%. Compared with the conventional HybridSN model, the classification accuracy of the HybridSN~~AM model on the three data sets increased by 0.89%, 0.07%，and 0.73%, respectively. It effectively solves the problem of hyperspectral image joint space-spectral feature extraction and fusion, improves the accuracy of HSI classification, and has strong generalization ability. It fully verifies the effectiveness and feasibility of the attention mechanism combined with a hybrid convolutional neural network in hyperspectral image classification, which has important theoretical value for developing and applying hyperspectral image classification technology.

Early blight of chili peppers is a common biological disaster that affects the safe production of chili peppers. It is characterized by suddenness and susceptibility and can easily cause significant economic losses. During the growth process of chili peppers, scientific monitoring and early warning of disease infestation during the incubation period is an important prerequisite for ensuring the healthy growth of crops. This paper establishes the spectral characteristics discriminative model of crop disease incubation period using the hyperspectral image with the working band of 400~1 000 nm and the spectral similarity measurement method. Continuous and dynamic monitoring of hyperspectral images of pepper leaves and healthy leaves inoculated with early blight pathogens at different infection stages using a hyperspectral imager.Extract the average spectrum of the region of interest from a series of hyperspectral images collected in the experiment and preprocess it using convolutional smoothing, multivariate scattering correction, and maximum minimum normalization method (SG-MSC-MMN). Then, two measures, spectral angle cosine correlation coefficient, and Chebyshev distance, are proposed as spectral characteristics evaluation parameters for the incubation period of early blight. Finally, principal component analysis (PCA) was used to verify the results of the spectral characteristics discriminative model of the incubation period to realize the visual distribution of the incubation period of samples.The experimental results show that it is feasible to use the spectral angle cosine correlation coefficient and Chebyshev distance as the spectral characteristic evaluation parameters of the incubation period of early blight of pepper and establish the corresponding discriminative model, respectively, and the earliest identifiable time of incubation period of early blight of pepper obtained from these two discriminative models is 24 hours after inoculation.According to the PCA drawing, the spatial distribution of health vaccination samples during inoculation for 24 hours was verified by the two spectral characteristics-based discriminative models for the incubation period proposed in this paper.The discriminative model of the incubation period of early blight of pepper established in this paper can be extended to monitor and identify the incubation period of other crop diseases and provide theoretical reference and method reference for scientific control of the incubation period of crop diseases.

In this study, we employ near-infrared spectroscopy with Stacking ensemble learning to perform non-destructive sugar content analysis in kiwifruit. Our research focuses on the “Yunhai No.1” kiwifruit variety from Hubei. Using an infrared analyzer, we gathered spectral data from 280 samples, spanning 1 557 wavelengths in the 4 000～10 000 cm-1 range, and measured sugar content with a refractometer. Outliers were identified and excluded using a singular sample identification algorithm that combines Monte Carlo random sampling with a T-test. The SPXY algorithm was then employed to split the data into training and testing sets in a 4∶1 ratio. Data preprocessing involved multiple scattering corrections (MSC), Savitzky-Golay smoothing (SG), de-trending (DT), vector normalization (VN), and standard normal variable (SNV) transformations. Feature wavelengths were initially selected using uninformative variable elimination (UVE), competitive adaptive reweighted sampling (CARS), and interval variable iterative space shrinkage approach (iVISSA), followed by a secondary selection with the successive projections algorithm (SPA) to remove collinear variables. To address the limitations of single models in generalization, we designed an integrated learning model using the Stacking algorithm. This model incorporated Bayesian ridge regression (BRR), partial least squares regression (PLSR), support vector regression (SVR), and artificial neural networks (ANN) as base learners, with linear regression (LR) serving as the meta-learner. We assessed the performance of various ensemble model combinations and analyzed the influence of base learners on ensemble performance using the Pearson correlation coefficient. Experimental results indicated that vector normalization was the most effective among the five preprocessing methods. The VN-CARS-PLSR model demonstrated superior performance, with Rp2 of 0.805 and RMSEP of 0.498, identifying 177 feature wavelengths and reducing data volume by 88.6% compared to the original spectrum. Comparisons of different base learner combinations in the Stacking algorithm revealed that the PLS+SVR+ANN integrated model achieved the highest predictive accuracy, with Rp2 of 0.853 and RMSEP of 0.433. The study concludes that the stacking ensemble model offers more comprehensive modeling capabilities and superior generalization than single models, providing valuable technical support for non-destructive sugar quality detection in kiwifruit.

Soil Organic Matter (SOM) content is one of theimportant indicators used to measure soil fertility, and it is of great significance in accurately predicting SOM content from hyperspectral remote sensing images. Traditional machine learning methods require complex feature engineering. Still, they are not highly accurate, while deep learning methods represented by Convolutional Neural Networks (CNNs) are less studied in soil hyperspectral, and the modeling accuracy of small sample data is poor. The spatial feature extraction of spectral data is insufficient. This paper proposes a one-dimensional convolutional network model using a channel attention mechanism (SE Dilated Convolutional Neural Network, SE-DCNN). Taking 207 soil samples collected from Guangxi State-owned Huangmian Forest Farm and State-owned Yachang Forest Farm as research objects, this paper compares and analyzes the modeling effects of 3 machine learning and 4 deep learning methods under different spectral preprocessing. The results show that the SE-DCNN model, because of the use of dilated convolution and channel attention mechanism, expands the receptive field, extracts multi-scale features, and has good modeling accuracy and generalization fitting ability. The best prediction model in this paper is the SE-DCNN model established based on the spectral preprocessing method of Savitaky-Golay denoising (SGD) and first-order derivative (DR), the determination coefficient (R2) of the validation set is 0.971, the root mean square error (RMSE) is 2.042 g·kg-1, and the relative analysis error (RPD) is 5.273. Therefore, SE-DCNN can accurately predict the organic matter content of red soil in Guangxi forest land.

Metal phthalocyanines with different peripheralsubstituents were doped into silica (SiO2) gel glass matrix by the Sol-gel process to prepare metal phthalocyanine doped composite gel glass. The effects of the types and positions of different cyclic substituents in lead phthalocyanine series, aluminum phthalocyanine series, and zinc phthalocyanine series on the UV visible absorption spectra of phthalocyanine in DMF solution and silica gel glass matrix were systematically studied to detect the existence status of phthalocyanine in different matrices and the influencing factors of the types and positions of different cyclic substituents. Firstly, the influence of the types of pericyclic substituents was studied: in DMF solution, the introduction of different pericyclic substituents -CP, -OAr, t-Bu can lead to the formation of a larger conjugated system of large electrons, resulting in a red shift in the maximum electron absorption wavelength; In the composite gel glass, the doped metal phthalocyanine exists in the form of polymer, which makes the characteristic absorption peaks in the ultraviolet-visible spectrum become dispersed. Secondly, the influence of the position of cyclic substituents was studied: for the aluminum phthalocyanine series, the cyclic substituents OAr were studied, respectively and Bit time, - AlTPOPcCl and Ultraviolet-visible absorption spectrum of -AlTPOPcCl in DMF solution and silica composite gel glass. The results indicate that in DMF solution, due to the electron conjugation effect, Compared to the substituent of position , The characteristic peak has a redshift. In composite gel glass, The UV visible absorption spectra of - AlTPOPcCl are almost identical to those in DMF solution, indicating that - AlTPOPcCl exists as a monomer in these two matrices. The absorption spectra of -AlTPOPcCl dispersed in the composite gel system, indicating that the phthalocyanine mainly existed in aggregate in the composite system. The above research results lay the experimental foundation for further research on the optical properties of phthalocyanine-doped composite gel glass and also laid the theoretical foundation for the research on nonlinear optical limiting materials and devices.

The monazite flotation system often has many associated minerals such as calcite, fluorite, and dolomite. These associated minerals will dissociate a large amount of Ca2+ in the monazite ore, and Ca2+ often affects the pulp flotation environment. In this paper, the surface properties of monazite were studied by FTIR and XPS when Ca2+ affected the flotation of monazite by octyl hydroxamic acid. The research method is mainly through the solution chemical calculation of Ca2+, the flotation test of monazite, and the surface infrared spectroscopy ( FTIR ) and X-ray photoelectron spectroscopy ( XPS ) of monazite under flotation conditions. The solution chemical calculation of Ca2+ shows that an aqueous solution with increased pH value, Ca2+ exists in an ionic state, hydroxyl complex, and hydroxide compound, respectively. At the same time, the dominant components are Ca2+ and Ca(OH)+ at pH between 7 and 8. The results of monazite flotation showed that OHA could not fully float monazite without adding Ca2+, and the recovery rate was 75.37%. When the pH is (8±0.5), and the dosage of Ca Ca2+ is 3×10-4 mol·L-1, the flotation performance of OHA on monazite can be significantly improved, and the recovery rate reaches 96.48%. According to the solution chemistry calculation, Ca(OH)+ is the dominant component of activated monazite. When the Ca2+ dose is greater than 3×10-4 mol·L-1, the flotation recovery rate decreases greatly, indicating that the further increase of Ca2+ dose inhibits the flotation of monazite. Only a certain dose of Ca2+ effectively promotes the flotation of monazite, which may be attributed to the consumption of OHA concentration by Ca2+ dose, which in turn affects the flotation of monazite. Infrared spectroscopy analysis showed that under the action of Ca2+, two key new peaks appeared in the spectrum, one was the N—O—H bending vibration peak at 1 454 cm-1, the other was the O—N stretching vibration peak at 880 cm-1, and the organic peaks at 2 974 and 2 928 cm-1. The —CH3 and —CH2— peaks were significantly enhanced, and these groups appeared to indicate that chemical adsorption occurred and the adsorption intensity was greater than that of pure monazite adsorption of OHA. XPS analysis showed that the relative content of N element on the surface of monazite was 0.61 % when only OHA was used to adsorb monazite. In comparison, the relative content of N element on the surface of monazite was 2.36% when OHA was used to adsorb monazite after Ca2+ treatment. It can be concluded that Ca2+ will promote the adsorption of OHA on the surface of monazite. It can be seen from the peak fitting that the addition of Ca2+ reacts with Ca(OH)+ to form O—Ca—OH groups on the cleavage surface of monazite to form O—Ca—OH groups, which can be used as a new adsorption site for adsorbing OHA. At the same time, the cerium atom of monazite and the two oxygen atoms on OHA form a five-membered chelate, which also acts as an adsorption site. It is concluded that there can be two adsorption sites on the surface of monazite. The active sites of Ca and Ce atoms on the surface of monazite can adsorb OHA, which is beneficial to the adsorption of OHA on the surface of monazite and forms a more uniform and dense OHA hydrophobic adsorption layer. This is why the performance of Ca2+ in OHA flotation of monazite is improved. This study helps to enrich the activation theory of metal ions in pulp and also confirms that effective mineral flotation separation depends not only on the strength of collector-mineral interaction but also on the chemical properties of the flotation solution to a large extent. Utilizing or controlling the surface reaction should be the main goal of developing a more efficient and economical flotation process.

Beef is an important edible meat in China. In recent years, with the increasing demand for beef, the phenomenon of pork being impersonated or added to beef for sale has become increasingly serious, and there is an urgent need for simple and rapid detection methods to monitor adulteration behavior. This study analyzed the three-dimensional fluorescence spectra of beef and pork to determine the wavelength difference of synchronous fluorescence. A synchronous fluorescence spectrum with a fixed wavelength difference of 160 nm was used to qualitatively distinguish and quantitatively analyze the doping of beef with pork. The discriminant accuracy of the test set, verification set, and prediction set samples are taken as the evaluation index of the qualitative Discriminative model: Correlation coefficient (r), corrected Root-mean-square deviation (RMSEC) and predicted Root-mean-square deviation (RMSEP) were used as the evaluation indicators of the quantitative analysis model. The experimental results show a significant difference in the three-dimensional fluorescence spectra between beef and pork. Beef has fluorescence peaks at Ex/Em values of 270/320, 330/400, 350/500, 430/515 and 410/570 nm, while pork has three fluorescence peaks at Ex/Em values of 270/320, 330/400 and 430/515 nm. By setting the synchronous fluorescence wavelength difference to 160 nm, three fluorescence peaks of beef can be collected, with two of them located at the peak. The correction set accuracy of the SVM qualitative Discriminative model for beef, pork, and adulterated meat was 97.56%, and the prediction accuracy was 92.31%. The PLS prediction model for pork addition in beef without treatment, MSC treatment, and SNV treatment was compared. The PLS model without treatment was the best, with rc, rp, RMSEC, and REMSP reaching 0.978 6, 0.959 0, 0.059 7, and 0.092 7, respectively. Therefore, the qualitative discrimination and quantitative analysis detection model for beef adulterated pork based on synchronous fluorescence technology combined with SVM and PLS has a high recognition rate and detection accuracy, which can accurately and quickly detect whether pork is adulterated in beef.

With the continuous prosperity of China’s economy, people have put forward higher requirements for material living standards. Food that prevents diseases and improves physical function has become a “hot spot” in the consumer market. Oil can provide energy for the human body. Edible oil is one of the major ways for human beings to get oil, and high-quality vegetable oil contains substances that are more beneficial to human health, such as monounsaturated fatty acids, polyphenols, squalene, vitamin E, and other nutrients. Because of the physical cold pressing process, extra virgin olive oil keeps almost all the nutrients in its olive fruit, and the oleic acid content is as high as 70%. Therefore, although an “imported product”, extra virgin olive oil has been a “favorite” in the vegetable oil market since it entered the Chinese market, and its price is also significantly higher than ordinary vegetable oil on the market. Driven by interests, the phenomenon of making and selling fake super virgin olive oil has been repeatedly banned, and the means of making and selling fake olive oil have been constantly updated and iterated, resulting in the repeated prohibition of fake and inferior products in the domestic olive oil market. Adulterated oil products will not only harm the lives and property of consumers but also affect the production and sales of legitimate operators, disrupt the sales market, destroy the market order, and affect the public’s recognition of super virgin olive oil. The adulteration of vegetable oil is one of the urgent problems facing food safety at present. To realize the rapid, accurate, and low-cost identification and detection of the adulteration of vegetable oil and extra virgin olive oil, a method for qualitative and quantitative analysis of vegetable oil based on a generalized regression neural network and UV-Vis spectrum is proposed. Because the generalized regression neural network performs well in learning speed and nonlinear mapping ability, and the diffusion factor is the only optimization parameter of the network, it does not need backpropagation and repeated iteration. Compared with other detection technologies, UV-Vis technology has overwhelming advantages in terms of the detection cycle, stability, and low maintenance cost. This method has achieved 100% discrimination in the qualitative identification of vegetable oil and achieved the results that the determination coefficient (R2) is better than 0.988 75 and the root mean square error (RMSE) is better than 0.038 33 in the quantitative detection of extra virgin olive oil adulteration. The results showed that the model showed excellent predictive ability in identifying vegetable oils and in the quantitative detection of adulteration in extra virgin olive oil. Therefore, the method based on a generalized regression neural network algorithm and UV-Vis spectrum has important potential for application in vegetable oil’s qualitative and quantitative detection.

The traditional method for determining the content of Eburicoic acid is HPLC, which is inefficient and cumbersome to operate. To achieve rapid and non-destructive monitoring of Eburicoic acid, this paper attempted to establish a partial least squares (PLS) regression model based on near-infrared spectroscopy (NIR) to predict the Eburicoic acid content in Fomes officinalis Ames decoction pieces. Firstly, the traditional HPLC method was used to test the content of Eburicoic acid in Fomes officinalis Ames, and the test results were used as indicator values. Secondly, near-infrared data was collected, and five spectral transformation methods were used to preprocess spectral data, namely Multiplicative Scattering Correction (MSC), Normalized Normal Variation (SNV), Savitzky Golay Smoothing (7 points), First Derivative Transformation (FD), and Second Derivative Transformation (SD). Finally, wavelength selection was performed through competitive adaptive reweighted sampling (CARS) and the PLS model was optimized, greatly reducing the number of spectral variables and significantly improving the performance of the PLS model, especially the SNV-CARS-PLS model, which only accounted for 5.53% of the total spectral wavelength. The R2 value for prediction sets 0.982 3. The root mean square error (RMSEP) value for prediction is 0.103 7%, and the residual prediction deviation (RPD) value is 5.34. The t-tests indicated no significant difference in precision and accuracy between the results of the optimal model and that of the traditional HPLC method. The research results indicate that it is feasible to establish PLS models based on near-infrared spectroscopy combined with a competitive adaptive reweighting algorithm for non-destructive detection of Eburicoic acid content in Fomes officinalis Ames decoction pieces.

The spectral reflectance of an object completely determines its surface color; therefore, studying the spectral reflectance is of great significance for industries with high requirements for color information. Direct acquisition of spectral reflectance requires precise and expensive equipment. However, the cost of obtaining spectral reflectance can be greatly reduced by establishing a model that predicts spectral reflectance from RGB response values obtained from low-cost devices such as digital cameras. Spectral reflectance reconstruction algorithms based on regression methods have received widespread attention, and their core goal is to establish a mapping relationship between RGB vectors and spectral reflectance vectors. For most objects, the spectral reflectance curves of their surfaces have the property of smoothing. Therefore, there is a certain correlation between the spectral reflectance components. However, the existing algorithms have built prediction models for each dimension of the spectral reflectance vector separately, without taking advantage of the correlation between the spectral reflectance components. Unlike traditional single-output regression methods, the multi-target stacking regression method utilizes the correlation between outputs by reinjecting the first predicted output values into the inputs, and this paper studies spectral reflectance reconstruction based on multi-target stacking regression. However, the traditional multi-target stacking regression method is susceptible to the influence of errors in the first predicted output values. To address this problem, this paper proposes a screening method for the first predicted output value, selecting the part with less error as input to ensure the accuracy of the next model-building step. This screening method can preserve the samples with lower errors to a great extent, even without knowing the true values. The experimental data set in this paper is sourced from the ICVL hyperspectral image database, and the evaluation metrics are root mean square error and chromaticity error. The experimental results indicate that the proposed multi-target screening stacking regression can overcome the problems of multi-target stacking regression and achieve smaller errors than without stacking. Therefore, the proposed method in this paper can better utilize the correlation between spectral reflectance components.

Chlorophyll content is a key indicator of the physiological status of plants, and accurate estimation of chlorophyll content is important for characterizing its component content traits and quantifying its physiological status. In this paper, the hyperspectral reflectance and chlorophyll content (SPAD) of Spartina alterniflora in the Duliu-river wetland were used as the data source, the original spectrum was mathematically transformed and processed with continuous wavelet transformation (CWT). The spectral features were extracted using Sequential Projection Algorithm (SPA). And the hyperspectral estimation model of leaf chlorophyll content of Spartina alterniflora was developed based on random forest regression (RFR) algorithm. The results showed that: (1) CWT had more accurate time resolution and higher frequency in the low scale spectra, corresponding to a narrow wavelet function, which could better distinguish the differences between the spectra and highlight the characteristic spectral information. (2) Except for reciprocal and logarithmic first derivative spectrals, the spectral mathematical transform and CWT methods could effectively respond to the spectral detail features. CWT was generally better than the spectral mathematical transform, and the correlation between L10 scale and first derivative spectral reached 0.78 and 0.77. (3) First derivative spectral, reciprocal first derivative spectral, logarithmic derivative spectral and CWT could enhance the ability of spectral estimation of Spartina alterniflora chlorophyll content. The RF models based on first derivative spectral (R2=0.776, RMSE=0.510, RPD=1.893) and CWT with the multiscale of L2, L3 and L4 (R2=0.871, RMSE=0.305, RPD=3.846) were the optimal models. This study shows that hyperspectral techniques could be used as a non-destructive means of detecting chlorophyll content in leaves of Spartina alterniflora, and that the hyperspectral estimation model built by combining multiple scales after continuous wavelet decomposition could more estimate chlorophyll content in leaves of Spartina alterniflora.