Narrow linewidth semiconductor laser has been widely used in high-speed coherent optical communication, distributed sensing, lidar and other fields due to its high monochromatic, low frequency noise and high tunability. With the development of high quality factor (Q) optical resonator, silicon optical isomerization integrated chip and other technologies, the narrow linewidth semiconductor laser has experienced revolutionary development in the last decade. The linewidth has been compressed to the order of kHz or even sub-kHz. In this paper, the latest developments of kHz narrow linewidth semiconductor laser are described, and the narrow linewidth lasers with different linewidth compression mechanisms are classified. The influence of optimizing coupling coefficient and reducing external cavity loss on the performance of narrow linewidth laser is discussed. The development trends such as further compressing the line width and improving the output optical power are prospected.

Mode multiplexing technology based on orbital angular momentum (OAM) in optical fiber has great potential in increasing the capacity of optical communication. The optical fiber suitable for OAM mode transmission has gradually realized the characteristics of more stable OAM signal transmission and longer transmission distance, as well as support for more OAM transmission modes through the optimized design of materials and structure. Starting from the optical fiber mode superposition theory, it analyzes and compares the research results, advantages and disadvantages, and suitable application scenarios of various optical fibers that can transmit OAM modes, including ring fibers and super-mode fibers designed based on traditional fibers, as well as hexagonal and circular photonic crystal fibers designed based on photonic crystal fibers, etc. Finally, a further prospect is made for the development of the optical fiber that suitable for the transmission OAM mode.

Aiming at the problem of zero bias error of MEMS gyroscope caused by temperature change, a zero bias compensation method based on particle swarm optimization (PSO) and radial basis function (RBF) neural network was proposed. RBF neural network was used to build the model of the temperature error of gyro zero bias after pretreatment. After searching the optimal parameters of RBF neural network with PSO to improve its generalization ability, the optimal parameters of PSO-RBF neural network are used to compensate zero bias of gyro. Experimental results demonstrate the effectiveness of the proposed algorithm. After the compensation by PSO-RBF neural network algorithm, the maximum error of MEMS gyroscope zero bias decreases from 0.046(°)/s to 0.0034(°)/s, and the standard deviation decreases from 0.0427(°)/s to 0.0013(°)/s, which can effectively improve the stability of zero bias of MEMS gyroscope.

To satisfy the needs of system miniaturization and integration, a Bragg grating-based RGB waveguide multiplexer/de-multiplexer was proposed, and the relationship between its performance and grating characteristic parameters such as grooving depth and period number were systematically studied. The results show that a SU-8 RGB multiplexer/de-multiplexer has a compact structure with cross section of only a few hundred nanometers and a length of dozens of microns, and exhibits better transmittance and flexibility than a traditional silicon materials component. The optical properties of the RGB multiplexer/de-multiplexer are mainly affected by its grooving depth and period number. With increase of the groove depth, the central wavelength of the reflection spectrum exhibits red shift first and blue shift later, the width of the reflection spectrum increases almost directly, and the peak value of the reflection spectrum tends to saturate gradually. However, with the increase of the period number, the width of the reflection spectrum does not change significantly, and the peak value of the emission spectrum tends to saturate gradually. Based on such characteristics, the proposed design realizes the RGB multiplexing/de-multiplexing, laying a solid foundation for the miniaturization, integration, and flexibility of RGB multiplexer/de-multiplexer.

The back-lighting structure of InGaAs FPA is blocked by its InP substrate, thus it has no response to visible light. For wide spectral detection application requirements, an InGaAs FPA with the pitch of 25μm and the array size of 640×512 has been developed. After the flip chip bonding process, the thinning polishing process combining with the dry and wet methods were performed on the chip, the photodiode array chip eventually retains a thickness of about 5μm. The visible and SWIR response of 400~1700nm spectrum was realized, the peak quantum efficiency and detectivity exceed 85% and 8.0×1012cm·Hz1/2·W-1, respectively, the non-uniformity of the FPA is less than 6% and the imaging effect of the device is good.

A 25Gb/s variable-gain amplifier (VGA) based on 0.18μm SiGe BiCMOS process was introduced for high-speed communications. The proposed VGA consists of a VGA core, an output buffer and a bias circuit. The core circuit of the VGA adopts an improved Gilbert structure to increase the gain dynamic range of the circuit. Meanwhile, the VGA achieves the broadband characteristic by utilizing inductive peaking technique to overcome the large parasitic capacitances. The post-simulation results show that the VGA exhibits a maximum gain of 20.15dB and a -3dB bandwidth (BW) of 26.8GHz, which can support a data rate up to 25Gb/s. The VGA consumes a power of 26.4mW from 3.3V supply voltage and occupied an area of 1120μm×1167μm.

A terahertz metamaterial sensor based on double opening ring is proposed, which can be used to detect the refractive index and thickness of materials. The structure of the sensor consists of two parts: a metamaterial structure with two open square rings and a ring nested inside, and a polyimide substrate. When the terahertz wave is perpendicular to the surface of the metamaterial, three high-Q resonance peaks are formed in the range of 0.8~1.8THz. By discussing the relationship between the surface current distribution and the formation of three resonance peaks, it is observed that the different response characteristics of the metamaterial structure to the incident terahertz wave lead to different surface current distribution. In addition, the applications of the sensor in refractive index sensing and thickness sensing are also explored. The sensing sensitivities of 170, 103 and 119GHz/RIU can be realized for the resonant frequencies f1, f2 and f3, respectively, at a given thickness, all of which have superior sensing characteristics and can be used for high-sensitivity refractive index sensing with their multiple resonant peaks. This highly sensitive refractive index sensor can detect small changes in the analyte to be measured and has promising applications in the field of biochemical detection.

In order to ensure the stability of electron emission from cathode filament of vacuum fluorescent display (VFD), the simplified model of filament deformation was established, the analytical formulas of deflection and maximum stress under an impact load were deduced, and the reliability of the analytical solution was verified by comparing with the finite element results. The criterion of structural parameter design was proposed. The results show that the analytical formula accurately describes the mechanical properties of VFD cathode filament under different impact loads. The maximum deflection and the maximum stress occur at the position where the impact load and the midpoint of VFD filament coincide, both of them increase with the filament lengthening or thinning, and the diameter has more obvious effect on their increase. The design criterion related to length and diameter of VFD filament is accurate, efficient, simple and feasible. The research results can offer qualitative support to the reliability design of VFD filament, and provide quantitative basis for the optimization design of structural parameters to improve its performance.

In this paper, Silvaco TCAD was used to establish the model to simulate the optical and electrical properties of graphene, and the experimental bipolar curves of graphene with different doping degrees were calculated. Furthermore, modeling and simulation of graphene-silicon photoconductive photodetector based on top gate regulation were carried out. The simulation results show that the top gate voltage can change the built-in potential of graphene-silicon heterojunction by regulating the type and concentration of graphene carriers, so as to improve the gain of graphene-silicon photodetector. Then the gate-controlled graphene-silicon photodetector was prepared and it is consistent with the gain enhancement of the graphene-silicon photoconductive photodetectors fabricated at 1550nm band.

In order to explore the temperature adaptability of modulated grating Y-branch (MG-Y) laser in optical fiber sensing application, a test system with wavelength stability of 0.6pm was designed and built. The wavelength characteristics of MG-Y laser at different internal and external temperatures were explored respectively, and the experiments were carried out with a look-up table with an interval of 20pm at 25℃. The results show that 92.30% of the wavelength drift is within ±7pm when the internal temperature is 25℃ and the external temperature is in the range of -20~50℃ and steps by 10℃, and the maximum wavelength drift is 3pm with the internal and external temperature difference of ±5℃. When the external temperature is 25℃ and the internal temperature is in the range of 25~30℃ and steps by 1℃, the mode hopping phenomenon will appear at the switching of the wavelength tuning section. The tuning coefficient range is 85~115pm/℃. It is proved that the optimal temperature of MG-Y laser should be controlled within the range of internal and external temperature difference of ±5℃, which provides a reference for solving the problem of wavelength drift caused by temperature change.

In order to reduce the temperature of the active region of the microdisk cavity semiconductor laser and improve the reliability of the package, the thermal characteristics of limacon microdisk cavity semiconductor lasers packaged with AlN, WCu10, SiC, graphene and CVD diamond transition heat sink respectively were analyzed with Ansys Workbench, and the temperature distribution, thermal stress and thermal strain distribution of the device were obtained. The results show that the active region temperature of SiC package is 2.18 and 3.078℃ lower than that of AlN and WCu10 package respectively, and it shows the lowest thermal stress and minimum thermal strain in the five kinds of transition heat sink packaging devices. SiC packaging can not only effectively reduce the temperature of the active region, but also can reduce the package stress and device strain, so as to improve the reliability and heat dissipation ability of the device. Theoretical calculation results have guiding significance for single tube heat dissipation and array integrated heat dissipation of semiconductor lasers.

The optical fiber Fabry-Perot (FP) interference microphone with dual wavelength demodulation is demonstrated. By adopting the normalization algorithm and differential cross multiplication (DCM) algorithm, the acoustic signal is accurately restored. In the normalization algorithm, through ellipse fitting, the optical signal with two-way wavelength is normalized, which can reduce the influence of the output fluctuation of laser on the output of optical fiber FP interference microphone. In the DCM algorithm, through signal processing and filtering, the accurate output of acoustic signal is realized, and the influence of temperature and other environmental factors on the output of optical fiber FP interference microphone is reduced. In the experiment, the characteristics of optical fiber FP interference microphone based on dual wavelength demodulation are tested by contrast method. The acoustic signal with sensitivity of 210mV/Pa and frequency response of 100~3150Hz is realized, which can be well applied in the fields of speech recognition, noise measurement and airborne acoustic detection.

Using polyvinylidene fluoride (PVDF) as the sensing element of the contact force sensor, the influence of the types and thickness of the upper and lower substrate materials on the sensitivity of the sensor was explored by controlling variables. Through theoretical calculation of piezoelectric equation of piezoelectric material, the relationship between contact force input and voltage output was obtained. The results of finite element analysis show that the thickness and elastic modulus of the upper and lower substrate materials has great influence on the sensitivity of the contact force sensor. When 0.025mm PDMS is used as the upper and lower substrate material, the sensitivity of the contact force sensor can reach 0.527mV/N.

The fitting formulas of variable physical parameters of four new p-type thermoelectric materials (Bi0.5Sb1.5)Te3 prepared by Ingot method, BM method, S-MS method and Te-MS method were obtained by point method. The influence of temperature on thermoelectric materials prepared by different methods was analyzed. The relation curve between dimensionless merit figure and absolute temperature of thermoelectric materials was obtained. In addition, the influence of different preparation processes on the maximum coefficient of performance (COP) of thermoelectric cooler composed of thermoelectric materials was studied from the thermodynamic aspect. The results show that: compared with the other three preparation methods, the new p-type thermoelectric material (Bi0.5Sb1.5)Te3 prepared by Te-MS method has the largest coefficient of merit and the best thermoelectric performance. The thermoelectric refrigerator with the new p-type thermoelectric material (Bi0.5Sb1.5)Te3 prepared by Te-MS method has the maximum COP of 2.49, which is 34.59%, 37.57% and 25.76% higher than that prepared by other three methods, respectively.

Single crystalline β-Ga2O3 thin films were deposited on 5.08nm (2inch) c-plane sapphire substrates by magnetron sputtering and then were annealed. The effects of oxygen partial pressure during sputtering on the structural and optical properties of films were systematically investigated. In this study, a single crystalline β-Ga2O3 thin film with a crystalline size of 138nm, a transparency larger than 80% and a wide bandgap of 5.12eV, has been accomplished by modifying oxygen partial pressure. And a β-Ga2O3 thin film has been improved with a roughness as low as 0.401nm and a refractive index of 1.94 at 800nm. The results show that, reducing oxygen partial pressure during sputtering is helpful to improve crystalline quality and increase the transparency and optical band gap by increasing the numbers of sputtered particles as well as their kinetic energy. However, increasing oxygen partial pressure is beneficial to improve the film surface evenness and density.

Bi2O2Se nanosheets were synthesized by molten salt method by using Bismuth nitrate pentahydrate as bismuth source, selenium powder as selenium source and hydrazine hydrate as reducing agent at different concentrations of NaOH (0~3mol/L). The morphology, structure and composition of the samples were characterized by XRD, SEM, TEM and XPS. Then, a wide spectrum self-powered detector was prepared by using Bi2O2Se as the working electrode, and their photoelectric detection performance was investigated. The test results show that the Bi2O2Se self-powered detector has the best photoelectric detection performance under the condition of 1.1mol/L NaOH. In the UV-Vis-IR band, it has a high response. Under 365nm UV light, the photocurrent can reach 7.8μA, and the rise and fall time are 30 and 21ms, respectively. The responsivity and detectivity are 4.2×10-4A/W and 1.02×109Jones, respectively.

Designed is a novel broadband micro-strip antenna with the shape similar to Chinese character ‘回’. The proposed antenna structure mainly consists of two square rings, and four rectangle radiation patches are printed to connect them. The material of the substrate is glass fiber epoxy resin, with the thickness of 1.6mm and area of 30mm×45mm. Printed circuit board technology is adopted to form the antenna structure, and the reference ground is printed on the back of the substrate. Sweeping frequency from 1~8GHz, the designed antenna can cover the range of 2.1~4.3GHz by Ansoft HFSS 13.0 with the relative bandwidth of 68%. The simulation results of reflection coefficient, voltage standing wave ratio, input impedance and gain pattern are given, and also the influence of structural parameter on radiation performance is analyzed. The actual measurement results are basically consistent with the simulation results.

Helical drilling experiments of SiC ceramic were carried out by femtosecond laser. The effects of energy density on the processing morphology and shape of micro-hole in material were studied. The laws of variation of diameter, circularity and taper of micro-hole with laser machining parameters were analyzed. The experimental results show that micro-hole at entrance is less affected by energy density in comparison with that at exit. Higher energy density is conducive to improve the processing quality of micro-hole. Near cylindrical micro-hole of SiC ceramic with the diameter less than 250μm, the circularity more than 0.95 and the taper less than 1° was obtained. It was found that a certain degree of oxidation existed on the surface of material by detecting the machining zone. The reasons for elemental contents change in different processing areas were analyzed.

In order to effectively improve the heat dissipation performance of the micro-channel radiator, a microchannel radiator with droplet-like microstructure was designed, and the influence of the number and height of the droplet-like microstructure on the pressure loss and heat dissipation performance of the microchannel was studied by simulation method. In this paper, under the conditions as the heat flux of 100W/cm2 and the fluid velocity of 1m/s at the inlet, 9 groups of different micro-channels with droplet-like microstructure were designed. Five groups were studied by changing the number of droplet-like microstructures in a single microchannel. It was found that when the number of microstructures was 7, the comprehensive heat dissipation performance of the microchannel was the best, and the average temperature of the bottom of the microchannel decreased by 18.42K, the heat dissipation coefficient increased by 37.63%. At the same time, four groups of micro-channels were designed based on the number of 7 microstructures to study the influence of the height of the microstructures on the heat dissipation performance of the micro-channels. When the height of each microstructure in the micro-channels increased along the flow direction, the heat dissipation coefficient of the micro-channels basically unchanged, while the pressure loss was reduced by 11.93%.

Ga2O3 thin films were grown on quartz substrate by radio-frequency magnetron sputtering. The structure and optical bandgap of Ga2O3 films were characterized by X-ray diffraction and UV-Vis-IR spectroscopy, respectively. The photoluminescence (PL) spectra of Ga2O3 films were measured at room temperature. The results show that the deposited Ga2O3 films are amorphous. As the pressure increases, the optical absorption edge firstly shows blue shift and then switches to red shift. The optical bandgap value ranges from 4.96 to 5.30eV. When the sputtering pressure is 1Pa, the deposited Ga2O3 film has the largest optical band gap. Under the excitation of 325nm laser, luminescence peaks related to defect energy levels appear near 400 and 525nm.

The detection model of sucrose content in tea was established by using near-infrared spectroscopy technology and a back propagation neural network algorithm. The quality of model prediction is improved by introducing a genetic algorithm. The prediction model was established by using Fourier transform diffuse reflectance spectroscopy data of 120 tea samples mixed with sucrose. The prediction results of another 42 samples show that the correlation coefficient based on the traditional back-propagation neural network algorithm model is 0.7380, the root mean square error of prediction is 3.0754, and the prediction accuracy is 83.3%. The correlation coefficient increases to 0.9419 after the introduction of the genetic algorithm. The root mean square error of prediction is 1.3176, and the prediction accuracy is 88.1%, thus the training error is reduced by more than one order of magnitude. The experimental results show that the back-propagation neural network model can be used to detect the sucrose content in tea. Simultaneously, the introduction of the genetic algorithm can optimize the initial weights and thresholds of the neural network, so as to diminish the prediction error.

In this paper, a photon calculation method with variable coefficients is proposed to solve the first order photonic differential equation a micro-ring resonator on silicon substrate with. The ultrafine solution of the photonic differential equation with different coefficients is realized. Based on the coupled mode theory, the transfer function of micro-ring resonator is analyzed in time domain and frequency domain. The etching depth is designed to be 220mm, and the silica-based ridge waveguide is designed to be 500mm. The shape of the micro-ring resonator is track type, which is composed of two semi-circle sandwiches with two long straight waveguides. The radius of the semi-circle is 30μm, and the length of the straight waveguide is 60.75μm. The coefficients of a and b of the differential equation can be adjusted, and the adjustment range of a0 is 1.5007×1010~1.5562×1010 and b0 is 4.5343×109~5.6473×109. The simulation realizes the numerical solution process of the differential equation with different coefficient and the adjustment and electric control of temperature. And it is compared with the real differential equation curve after normalization, and the error is less than 0.02.

Aiming at the disturbance of training samples with large aspect gap in SAR target recognition, a kernel function transformation collaborative algorithm based on adaptive atom selection is proposed. This method improves the representation dictionary in the traditional collaborative representation, and gets the adaptive dictionary which is more adaptable to the current test samples and can reduce the influence of the unrelated atoms to the system. The experiments of SAR target recognition based on MSTAR datasets were carried out. The experimental results show that the kernel collaborative representation based on adaptive atom selection is more effective than the kernel collaborative representation model based on all training sample dictionary, which reduces the harmful effect of the interference atoms and further improves the reliability and robustness of the system.

As a research hotspot in the field of computer vision, expression recognition plays an important role in emotion recognition, human-computer interaction, intelligent security and other fields. Aiming at the problem that VGG19 is easy to overfit due to the large number of fully connected layer parameters in the process of data training, CapsNet is used to replace the fully connected layer of VGG19 to realize the cascade of VGG19 and CapsNet, so as to improve the problem of overfitting during training. At the same time, the accuracy of the cascaded model on the RAF-DB dataset is improved by 5.28%. In view of the problem that the MaxPool of VGG19 feature extraction network is easy to lose information of face feature map, the SoftPool is used to replace MaxPool, so as to retain the fine-grain features of face to the maximum extent. The experiments show that the accuracy of the two improved models is 84.21% on the RAF-DB dataset and 73.16% on the FER2013 dataset, which has a good expression recognition effect.

In order to solve the problem of the worse decoding performance of the successive cancellation (SC) decoding algorithm under the medium/short code lengths, an improved successive cancellation list flip (SCLF) decoding algorithm is proposed by adding the path list and bit-flipping method on the basis of the SC decoding algorithm. The algorithm uses the bit-flipping to build the most unreliable information bit sets called as the flipping set (FS), and a new measurement criterion is proposed to reduce the range of the FS and improve the accuracy of the FS. The simulation results show that the proposed SCLF decoding algorithm can greatly improve the block error rate (BLER) with the increase of the signal-to-noise ratio (SNR). The gain of the SCLF (N=256, L=8) decoding algorithm is 0.55dB higher than that of the SC (N=256) decoding algorithm at BLER=10-3, the gain of the SCLF (N=256, L=8) decoding algorithm is 0.22dB higher than that of the CA-SCL (N=256, L=8) decoding algorithm at BLER=10-4, and the gain of the SCLF (N=256, L=16) decoding algorithm is 0.17dB higher than that of the CA-SCL (N=256, L=16) decoding algorithm at BLER=10-5.

For the inter-cell wireless resource management in macro-micro co-networking, proposed is a joint transmission and dynamic spectrum allocation strategy based on micro cell clustering in massive MIMO system. The strategy is implemented in two steps to optimize the network weighted sum rate. First, the proposed algorithm performs micro-cell clustering according to the current channel state information (CSI) of users, so as to reduce the inter-cell interference as much as possible and improve the system capacity. Then, the macro base station (MBS) and each micro-cell cluster dynamically allocate sub-carriers to its service users according to the current traffic request information to maximize the network weighted sum rate and improve the resource utilization. The simulation results show that the proposed micro-cell clustering and dynamic spectrum allocation strategy can not only effectively improve the system throughput but also reduce the interference among users in macro-micro collaborative networking.

The hard edge aperture effect of cats eye optical system is studied by matrix decomposition and the hard edge aperture window function is expanded into the superposition of complex Gaussian functions. The reflection law of laser is analyzed by numerical calculation. The study reveals that the farther the waist of the incident beam is from the cats eye optical system, the greater the aperture effect of the optical system. Meanwhile, adjusting the proper amount of positive defocus can make the echo light intensity of cats eye optical system higher. Theoretical analysis and numerical calculation have certain reference significance for the study of cats eye system.

Aiming at the problems of low artificial processing efficiency and low recognition rate of fuzzy image in the process of loading petrochemical dangerous goods, an intelligent repair and detection method of monitoring fuzzy image based on the combination of generative adversarial network (GAN), convolutional neural network (CNN) and extreme learning machine (ELM) is proposed. Firstly, using the deep learning network as the target detection framework, the fuzzy image is restored by using the zero sum game between the generator and the discriminator in the generative adversarial network to obtain a clear and complete job image. Secondly, using the ability of convolutional neural network to adaptively learn image features, the autonomous features of the repaired image are extracted. Finally, the extracted features are input into the extreme learning machine classifier for target recognition and classification to judge whether there are violations in the operation process. The experimental results show that the proposed method has fast image restoration speed, natural visual effect, high accuracy of target recognition and good generalization ability.

Stereo matching is an important research direction in the field of binocular vision. In order to ensure the matching accuracy of image texture regions and reduce the mismatch ratio of weak texture regions. A stereo matching method is proposed based on guided filtering and disparity map fusion. First, the image is divided into areas with rich texture and areas with weak texture according to the similarity of image colors. Secondly, two disparity maps are obtained by cost aggregation and disparity calculation by using guided filtering with different parameters. Then, the two obtained disparity maps are merged according to the result of texture region division. Finally, the final disparity map is obtained through disparity optimization steps such as left and right consistency detection and weighted median filtering. Experiments performed on standard image pairs on the Middlebury test platform show that the average mismatch rate of this method on 6 sets of weak texture images is 9.67%, the matching accuracy is much higher than that of the traditional guided filter stereo matching algorithm.

According to the theory of semiconductor optoelectronics, the transient transport mechanism of light-injected non-equilibrium carriers (electron-hole pairs) was analyzed, the saturation effect of THz wave radiation power and radiation intensity in SI-GaAs photoconductive dipole antenna under different conditions was studied. The main reason is that, under the action of DC bias electric field, the space charge electric field shielding and radiation electric field shielding will appear in optical injected carriers, thus the radiation power and intensity of THz wave are contained. The two shielding effects are different for antennas with different electrode gaps, the shielding effect is small when the light spot is large under the triggering light energy is certain.

Lane line detection is an important part of the vehicles intelligent driving system. Aiming at the problems of low accuracy and poor real-time performance in traditional lane line detection methods, an accurate lane line detection algorithm is proposed based on machine vision. The algorithm uses the inner edge line of the lane to represent the lane line, which improves the accuracy and real-time performance. The algorithm mainly includes two parts: preprocessing and lane line extraction. The preprocessing part includes grayscale, Sobel edge detection, region of interest setting and binarization, and finally the binary image of the lane line part is obtained. The lane line extraction part includes the image Slicing, improved Hough line detection, DBSCAN line clustering and straight line fitting, and finally accurate lane edge line information is obtained. Finally, the algorithm is applied to road condition tests in various scenarios. The experimental results show that the average accuracy of the algorithm is 94.9%, and the average processing time pre frame is 25.6ms. The algorithm has good real-time and robustness.