Computer simulation technology is an effective tool to improve the quality of large-diameter InSb single crystals grown by Czochralski method and reduce crystal preparation costs. There are many types of simulation software currently available on the market, but there are very few simulation cases used for InSb crystal growth. By analyzing the existing cases, the characteristics of each type of software can be obtained, and their advantages and disadvantages as well as applicable growth methods can be compared. Based on the growth characteristics of InSb crystals, the most suitable simulation software for the InSb Czochralski growth can be selected.

In order to solve the problem that the heat capacity laser pumped by laser diode generate a large amount of waste heat inside the crystal during the pumping stage, a thermal model of Yb∶YAG composite crystal without cooling in the pumping stage and heat exchange between the face and air is established by the working characteristics of the LD end-pumped laser crystal. Based on the heat conduction theory, the temperature field, stress field and thermal deformation field of Yb∶YAG composite crystal are calculated by finite element method. The effects of bonding length and bonding mode on the temperature field and thermal deformation field are quantitatively analyzed. The results show that if a laser diode with a pumping power of 50 W and a pumping optical Gaussian radius of 300 m is used to pump the end face of Yb∶YAG composite crystal with a crystal size of 3 mm×3 mm×4 mm, a bonding length of 1 mm, a bonding mode of single end bonding, and a doping concentration of 5.0 at. %, and the working time is 2 s, The maximum temperature rise of the internal field is 248 ℃, the maximum thermal shape variable is 5.9515 m, and the maximum stress is 3.70395×109 N/m2. The research results provide a theoretical basis for the design of Yb∶YAG heat capacity laser pumped by laser diode end face.

In order to obtain better beam quality output, the gain distribution uniformity in diode side-pumped laser is studied. The gain field model of LD single-side pumped Nd∶YAG laser is established by using simulation software through theoretical analysis of the gain distribution in the working substance. In the simulation, the uniformity of gain field and absorbed light field of LD single-side pumped Nd∶YAG laser under different pumping power is compared. The result shows that the distribution uniformity of gain field and absorbed light field under different pumping power is not consistent, which is not a simple and intuitive equivalent linear relationship. Theoretical reference is provided for the high beam quality and high efficiency laser output of out-of-the-box LD side-pumped Nd∶YAG solid-state laser in high and low temperature environment.

Aiming at the problems that the Crust algorithm based on Delaunay triangulation is not smooth enough, time-consuming, and has low accuracy when reconstructing complex surfaces from laser and image point clouds, an improved 3D point cloud reconstruction method is proposed. Firstly, the voxel barycentric near feature point algorithm is used for down-sampling. After that, the moving least squares algorithm is used to fit the function and determine the quadratic basis function and Gaussian weight function to complete the data smoothing and optimization. Then, the Crust algorithm based on the adaptive extrinsic circle Delaunay triangulation method is used to reconstruct the coarse triangular mesh. Finally, the ratio of the outer radius of the tetrahedron to the shortest side length of the tetrahedron is used to eliminate the unqualified tetrahedron and complete the reconstruction and optimization of the model. The experimental results show that this method can reduce the time of holes and reconstruction, and build a smooth 3D model with more accurate topology of point cloud.

The surface morphology of the ablation region of femtosecond laser finishing surface gear material 18Cr2Ni4WA is directly related to the laser process parameters. In order to obtain a better surface topography of the ablation region, a three-factor, three-level orthogonal test is designed, and the effects of laser repetition frequency, energy density and scanning speed on the depth of the ablation region and surface roughness are analyzed by signal-to-noise ratio, and the optimal combination of process parameters is obtained with a single response target. Combining the grey correlation method to optimize the two response targets, the optimal process parameter combinations are obtained as 200 kHz repetition frequency, 3.5 J/cm2 energy density, and 110 mm/s scanning speed, and the optimal process parameter validation test shows that the optimized surface topography has a better overall quality, which proves the reliability of the grey correlation method of multi-response optimization, and provides a method to improve the quality of surface topography for the femtosecond laser micromachined face gears. This proves the reliability of the grey correlation method and provides an effective method to improve the surface topography of femtosecond laser micromachined gears.

With the rapid development of high-speed railway and urban rail transit systems, research on traffic safety technology is becoming increasingly urgent. The 3D point cloud of railway line environment generated by applying laser scanning technology can achieve accurate perception and monitoring of operating environments. In this paper, the three-dimensional point cloud data of railway scenes is taken as the research object, and a large-scale point cloud semantic segmentation dataset for railway scenes is constructed for the first time. The existing point cloud semantic segmentation models are mainly applicable to small-scale scenes, and large scenic point clouds need to be segmented first. However, three-dimensional point cloud data of railway line environments have the characteristics of high data acquisition frequency and large data scale. Therefore, a large-scale point cloud semantic segmentation method for semantic perception of railway scenes is proposed in this paper. During the coding stage, an adaptive local feature fusion module based on self-attention is proposed in the encoding stage, which can better aggregate local features of different scales and solve the problem of category imbalance. In the decoding stage, an up-sampling method guided by high-dimensional semantic information is proposed to compensate for the information loss caused by large-scale down-sampling in the coding stage. The proposed method achieves excellent segmentation performance on both railway scene datasets and public indoor datasets.

Laser cladding is favored in the fields of parts repair and coating preparation due to its significant advantages. The quality of laser cladding is affected by a number of process parameters, among which the laser power determines the heat input of the molten pool and directly affects the heat transport. In order to explore the influence of laser power on heat transport, a three-dimensional heat transport model of laser cladding is established, and the reliability of the model is experimentally verified, and the grid independence of the model is analytically demonstrated. The results show that under different laser powers, the time required for the molten pool to reach dynamic equilibrium is consistent, and the trend of temperature change in the molten pool is approximately the same. The peak temperature rises with the increase of laser power, and the peak temperature at 600 W is about 11% and 22% lower than that at 700 W and 800 W, respectively. The molten pool under the three powers has an annular flow mode from the inside to the outside, and the velocity of liquid metal increases with the increase of laser power.

In this paper, Laser Doppler Velocimetry (LDV) is applied to taxi meter checking through the Doppler effect, and the system design is completed by using a double oblique radiation asymmetric structure laser velocimeter, bispectral peaks signal processing technology, portable external tooling and a well-interacted operating system. Theoretical analyses and experiments demonstrate that the LDV-based taxi meter calibration system performs effectively. The pneumatic attachment of the apparatus enhances portability, allowing for broader application scenarios. The innovative dual oblique, asymmetric optical path design minimizes the impact of road irregularities on measurement accuracy, thus enhancing the system's reliability and precision. The dual spectral peak signal processing technology improves the speed of signal processing, enabling real-time processing of velocity signals. The system also features a custom-designed user interface, which is simple, user-friendly, and has a high error tolerance, greatly simplifying the meter calibration process. Consequently, the LDV-based taxi meter calibration system can meet the real-world demands of the market and solve prevalent challenges in meter calibration, providing substantial practical value.

In the industrial environment of laser cutting, there are a large number of complex backgrounds and other interfering elements such as equipment. The current combination of category learning methods lacks a separate target alignment process for the highlighted areas in the abstract target graphics, resulting in weak correlation between target features and inaccurate annotation results. Propose a visual image object annotation method for laser cutting robots based on improved YOLOv5s. By utilizing input terminals, pooling layers, and shared fully connected layers, an improved YOLOv5s model is built. This network uses max pooling and average pooling to generate two laser cutting robot visual images. Based on the channel dimension, the image features are connected to achieve rough target localization in the laser visual image, and target feature alignment is achieved by combining modulation factors and target detection losses. After aligning the target features, the key highlighted area frames of the laser cutting robot's visual image are determined. By implementing semi supervised training on the historical annotated images of the laser cutting robot, the spatial region associations of the images are determined, and laser visual image target annotation is performed based on the region associations. The experimental results show that the proposed method for laser cutting robot visual image target annotation has high intersection to union ratio, high accuracy, fast speed, and strong robustness.

With the development of emerging technologies such as dynamic infrared stealth and intelligent thermal control, it is gradually difficult for a single low-emissivity material to meet the complex needs of practical applications. The infrared emissivity of variable emissivity materials can be reversibly changed under certain conditions. In this paper, the research progress of thermo-emissivity, electro-emissivity, and other variable emissivity materials such as stress-induced and humidity-induced materials is systematically summarized according to the different applied fields that induce the infrared emissivity transition. In this paper, the principles and practical applications of vanadium dioxide (VO2) in thermally variable emissivity materials and tungsten trioxide (WO3) and polyaniline (PANI) in electro-variable emissivity materials are discussed, and the performance characteristics of various materials are summarized and analyzed. The research on variable emissivity materials has greatly promoted the development of adaptive infrared stealth technology, and has been widely used in the fields of intelligent thermal control and urban energy conservation, and will be developed in the direction of fast response, long life and multi-spectral modulation in the future.

Active laser jamming is an effective means to counter infrared imaging seekers, and the saturation threshold under different incident conditions is the core parameter for studying the jamming effect on infrared imaging seekers. To obtain the saturation threshold of infrared imaging seekers under strong laser irradiation, the modeling and simulation of continuous laser jamming on short-wave, medium-wave, and long-wave infrared HgCdTe detectors is conducted based on the COMSOL multi-physics simulation platform. Firstly, models of HgCdTe material and typical planar junction HgCdTe pixel structures are constructed. Then, the electromagnetic wave and semiconductor coupling multi-physics interface is used to calculate the photovoltaic effect, and the effects of different laser wavelengths, powers, operating temperatures, and other factors on the photogenerated electromotive force of zero-bias HgCdTe infrared detector pixels are analyzed to demonstrate the feasibility of simulating laser irradiation effects using the COMSOL multi-physics simulation platform. The simulation results can provide a reference for semi-physical simulation of laser jamming effects.

In the field experiment, due to the lack of obvious defect characteristics in infrared detection of civil aircraft composite skin defects, resulting in low detection accuracy and complex model leads to the slow detection speed. To solve these problems, an improved SSD algorithm is proposed to enhance the detection precision and realize the model lightweight. Firstly, U-Net network is used for image preprocessing to reduce the interference of irrelevant feature information and enhance the detectability of defects. Secondly, Mobilenetv2 is used as the backbone network to reduce the memory size of the model and improve the efficiency of defect detection. Then, the inverse residual module of Convolutional Block Attention Module (CBAM) serves as an auxiliary convolution layer to further lightweight the model and address precision reduction. The ablation experiments and comparison experiments show that the mAP accuracy of the proposed algorithm is as high as 96.8% on the defect data set of civil aircraft composites, with a detection speed of 72.74 f/s (FPS). Compared with the traditional SSD algorithm, AP0.5 improves by 8.3%, the number of parameters counts (Params) is reduced to 3.966 M, and the number of floating points (GFLOPS) is reduced by 42 times. This algorithm has a good application prospect in the field of infrared detection of aircraft composites.

To improve the recognition accuracy and speed of point cloud targets, a target recognition system combining infrared image acquisition and laser scanning is designed, and a point cloud target extraction algorithm based on infrared boundary constraints is proposed in this paper. Firstly, edge enhancement and infrared features are used to grayscale the infrared image. Secondly, the target infrared image area is utilized to provide boundary constraints for point cloud target extraction, and the projection of 2D images to 3D point clouds is achieved by mapping scale functions, achieving coordinate system alignment. Finally, target recognition is realized using a set of point clouds which meets boundary constraints. The experiments are tested for vehicle targets in complex contexts and the results are compared between the traditional algorithm and the present algorithm. As the total amount of point clouds increases, the detection rate of traditional algorithm increases from 83.7% to 97.6%. The algorithm in this paper increases from 96.2% to 98.8%, and is less affected by the total amount of point clouds. This algorithm is effective in rejecting pseudo-targets, so its accuracy is more stable. The detection time of this algorithm is only 1/3 to 1/4 of the traditional algorithm, and this design has improved both target recognition accuracy and detection time, and has better practicality.

Using spectral emissivity measurement equipment to detect the emissivity of infrared stealth coatings is an important means of monitoring the status of aircraft infrared stealth coatings. During the calibration cycle of the measuring equipment, due to factors such as usage environment, usage frequency, usage method and so on, the condition of the equipment occasionally becomes worse, and the measured values deviate from the reference value which poses a certain risk for timely detection of infrared stealth coating defects and may affect the overall infrared stealth characteristics of the aircraft. To address the issue of measurement deviation during the calibration cycle, a Gramian Angular Field (GAF) and Parallel Convolutional Neural Network (PCNN) calibration status prediction model is established. By collecting one-dimensional time-series data from the device and feeding it into the GAF-PCNN mode, a prediction model for the calibration status of infrared emissivity measurement equipment is trained through deep learning. The experiment shows that the average recognition accuracy of the calibration state prediction model reaches 95%, and the convergence speed is fast and stable, which can be applied to equipment calibration state prediction, prompting early calibration or use beyond the calibration cycle. While ensuring good equipment condition, it reduces equipment calibration activities and improves guarantee efficiency.

Ultra-high contrast imaging in space is essential for the direct imaging detection of Earth-like planets, the achievement of which depends on the precise control of the optical wavefront by the space coronagraph system, and therefore requires the development of specialized on-orbit algorithms dedicated to wavefront detection and correction. While such algorithms have been widely used in ground-based adaptive optics systems, but in space, they cannot be designed based on pure CPU computing due to the limitations of space CPU performance and selection. Based on the hybrid architecture of FPGA and CPU, the wavefront correction is realized, which is capable of locking the high-contrast imaging dark region required for exoplanet detection while taking into account the hardware resources and computing accuracy. The algorithm of the above hybrid architecture also has a significant speed advantage in large-scale adaptive optics systems, and the wavefront processing delay is shortened by 1281.826 s for a 100×100 sub-aperture adaptive optics system, which can meet the demand for high-speed parallel computation of the adaptive optics systems, such as the ExAO, the GLAO, and the MCAO, which are equipped with the ground-based large-aperture telescopes.

The solar irradiance absolute radiometer (SIAR) adopts the thermopile-type detector to convert incident solar light power into electrical power, achieving the measurement of total solar irradiance. However, the thermopile detector has the following limitations. Firstly, it is difficult to further improve the sensitivity due to the limitation of the number of thermocouples. Secondly, the consistency of the main and backup chambers is difficult to be ensured due to the limitation of the complex fabrication process. To solve these problems, this paper develops a thermoelectric bridge absolute radiometer with high-sensitivity and high-precision based on the high interchangeability and high-resolution characteristics of the Wheatstone bridge, and adopts a new detector cone cavity structure and a new thermal connection method. The test results show that the time constant of the detector is 9.69 s, the sensitivity is 0.078 W/m2, and the measurement repeatability of tungsten light source is better than 0.07%, which can meet the observation requirements of total solar irradiance with a precision of 0.1%.

The extraction of center lines from light fringes is a crucial step in line-structure light measurement, as the extraction accuracy directly affects the measurement speed and precision. In response to the low extraction accuracy issue in traditional Steger algorithm due to large variations in light fringe width and excessive redundant points, an improved Steger-based fast and high-precision method for extracting center lines from light fringes is proposed in this paper. First of all, the original image of light fringes is preprocessed to remove noise points and retain the region of interest. Then, the width of light fringes is rapidly obtained by directly traversing pixel points, followed by Canny edge detection on the fringe area and calculation of pixel initial points using the geometric center method. Finally, a one-dimensional convolution is performed on the initial points to construct the Hessian matrix, enabling further precise extraction of the center line along the normal direction. Experimental results demonstrate that this algorithm improves accuracy by 42.6% compared to the traditional Steger algorithm and reduces processing time by 64.5%, thus achieving precise sub-pixel level extraction of light fringes while meeting the real-time requirements of industrial measurement.

Pure rotational Raman Lidar is widely used for detecting vertical atmospheric temperature profiles due to its good collimation and high spatiotemporal resolution. In this paper, the uncertainty of temperature measurement caused by the central wavelength and bandwidth of the interference filter by designing a spectral path based on the interference filter. The ratio method is used to simplify the calculation of leakage errors caused by insufficient suppression of elastic scattering signals by filters. The system parameters are calibrated by means of different bandwidths, and the impact of system parameters on fitting errors during temperature inversion was analyzed. The random error introduced by the fixed angle error when inverting the temperature is analyzed. The results indicate that the error generated by the center wavelength of the filter and the bandwidth of 0.5nm/0.3nm actually selected by the system is less than 1.2K, and a suppression ratio greater than 6 is appropriate. The fitting error generated by the use of lower order functions can be ignored, and the fixed angle error of the filter should be less than 0.5°. And based on the pure rotational Raman Lidar equation, the simulated echo signal is simulated and the temperature profile is inverted. The error distribution of temperature measurement is analyzed, and the total error is less than 1.2 K.

3D Gaussian splash substitutes traditional point clouds with Gaussian bases, utilizing their smooth characteristics. While maintaining data accuracy, it effectively processes and visualizes scattered data in three-dimensional space, achieving a more continuous and natural image rendering effects. This has made significant achievements in the field of optical imaging and has become a recent research hotspot. In the application domain of SAR imaging, the attribute scattering center model is often used as the basis for images. In this paper, an attempt is made to use Gaussian bases instead of attribute scattering center model. Through experimental comparisons, the performance of Gaussian bases, the attribute scattering center model, and the commonly used simplified attribute scattering center model in SAR image reconstruction tasks are analyzed. The results show that the Gaussian basis method offers superior image reconstruction quality, with significantly better performance in terms of speed and stability compared to the attribute scattering center model. These findings provide new insights for feature and target information extraction in SAR images.

In order to solve the problems of serious feature loss, low accuracy and complex model in the process of infrared UAV target recognition, an improved infrared UAV target detection algorithm based on YOLOv8 is proposed. Firstly, deformable convolution is introduced into the backbone network to enhance the feature representation capability of the target region. Secondly, a lightweight feature pyramid network structure SOD-FPN for small targets is proposed for small targets, which avoids the information loss of small targets by reducing the number of network layers and deleting large target detection headers. Moreover, the multi-scale feature fusion capability of the model is enhanced by cross-scale connection and weighted feature fusion method. Finally, NWD Loss based on Wasserstein distance is selected as the bounding box loss function to further improve the convergence and detection accuracy of the model. The experimental results show that the mAP50 of the improved algorithm is 99.4%, which is 2.2% higher than YOLOv8n, and the number of parameters is 72.8% lower. Meanwhile, compared with other advanced target detection algorithms, the accuracy and speed of the improved algorithm are improved, which proves the effectiveness and advancement of the improved algorithm.

In view of the problem that the traditional ICP algorithm has a long registration time and tends to fall into the local optimum when the initial positions of the two-point clouds differ greatly, a point cloud registration method based on the feature improvement of 3DHarris key points combined with fast point feature histogram is proposed to improve the point cloud alignment. Firstly, the input point cloud is streamlined using voxel down sampling, and then the 3DHarris algorithm is applied to extract key points from the streamlined two-piece point cloud, and the 3DHarris-FPFH feature points are formed by the FPFH, and then the Random Sample Consensus (RANSAC) algorithm is used to coarsely align and output the initial transformation matrix. Finally, the refined alignment is performed by the improved Iterative Closest Point (ICP) algorithm. The algorithm is simulated on open data set, and the results show that the algorithm can improve the operation speed while maintaining the accuracy, and has certain practicability.

To avoid natural disasters affecting the normal transmission of power lines and to ensure the operation safety of transmission lines, a distributed optical fiber sensing monitoring technology for uneven icing on single transmission lines is designed for power conservation. The distributed fiber optic sensing structure is constructed using components such as seed light sources and Brillouin lasers, and the distributed fiber optic sensing structure is used to generate Brillouin frequency shifts and obtain the distribution of temperature and strain over the fiber sensing length. The mechanism of Brillouin frequency shifts and their relationship with temperature and strain are analyzed, providing a reliable basis for subsequent calculations. Through a two-stage, three-tower model, the operating parameters of a transmission line at a single stage are measured, and the line stress equation is applied to calculate the specific load of uneven icing at a single stage. By combining the specific load of uneven icing at a single stage with the temperature and strain conditions applied to the transmission line, the uneven icing thickness of the transmission line at a single stage is calculated to ensure its safety. After experimental verification, it can be concluded that this technology can accurately monitor stress changes in the line during icing, and the temperature error of the monitored line is relatively low, and the thickness of ice-covering can be monitored in real time on each file distance of the line.