Abstract:In this paper, a 10×10 metasurface reflector antenna array based on PIN diodes for beam scanning is designed. The surface of the antenna element of the antenna array is composed of a circular metal patch and four parasitic elements. The current on the antenna element is changed by controlling the working state of the parasitic element, so that the antenna unit theoretically form a phase difference of 180° in different states.In this paper, CST software is used to simulate and analyze the antenna. The operating frequency of the antenna is 15 GHz. The reference phase is added to the beam scanning calculation principle based on discrete amplitude weighting, and the optimization of the pattern reconstruction is carried out. Through the analysis of the simulation results The reference phase that can best restore the pattern is found, which improves the beam scanning capability of the antenna. Since the traditional phased array unit needs to be equipped with a phase shifter, and the antenna unit only needs a diode, its structure is relatively simple, the manufacturing cost is low, and it is suitable for forming a large-scale array. The CST simulation results show that the metasurface antenna has good beam scanning results in the range of ±50° on the ZOX plane.

Abstract:Focus on the problems that the airflow in the traditional radiation shield is not easy to circulate and the energy consumption of its work is relatively large, a design of a meteorological temperature sensor with low radiation error is proposed, which combined natural ventilation and forced ventilation. Firstly, Computational Fluid Dynamics (CFD) simulation is used to analyze the radiation error optimization design under different environmental variables and fan speeds. Secondly, the support vector regression (SVR) algorithm is used to train the simulation results to obtain the prediction model. Finally, an outdoor experimental test platform is built to verify the feasibility of the design and the measurement accuracy of the prediction model. The experimental results show that the proposed meteorological temperature sensor can reduce the measurement radiation error to less than 0.05 ℃, which has a significant effect on reducing radiation and the accuracy of the prediction model is high, and the RMS error between the experimental value and the measurement value of the 076B forced ventilation radiation shield is 0.185 ℃, and the RMS error with the predicted value is 0.129 ℃.

Abstract:The measurement method with correction for the pulse modulated radiation field using root mean square detection in frequency domain is put forward to solve the restriction caused by the resolution bandwidth. First,the measurement principle is proposed and the correction coefficients for field strength based on root mean square detection are defined. Then, the measurement result for the signal power and the correction coefficients for field strength are modeled theoretically and expressed by the power factor whose value can been worked out. Also the optimum selections of the resolution bandwidth and dwell time are discussed to simplify the correction method with the deviation precisely controlled. Finally, the experiment is carried out and it shows that the deviations between computation and measurement results of the correction coefficients are less than 1.00 dB. It proves that the measurement method is accurate and feasible. The peak field strength and average field strength can be accurately obtained by using any resolution bandwidth. So it is not limited by the performance of the instrument. The established method provides a certain support for ensuring the electromagnetic safety of equipment and also is helpful for researching measurement methods for pulse radiation field with other modulation modes.

Abstract:It is one of the important tasks of battery management system (BMS) to realize battery charge state (SOC) estimation. The identification of battery model parameters is the precondition of SOC estimation for lithium-ion batteries, which is also the key factor determining the estimation accuracy of SOC. This paper took 18650 lithium-ion battery as the research object, and used the recursive least square method with time-varying forgetting factor (TVFFRLS) to identify the battery parameters online, so as to realize the automatic optimization of forgetting factor adaptation and improve the stability of parameter online identification. On this basis, the adaptive cubature Kalman filter (ACKF) was used to realize the estimation of SOC of lithium-ion batteries, and the covariance of process noise and measurement noise was updated in real time.The algorithm was verified under various working conditions. The results show that the algorithm has good noise suppression performance and can realize the estimation of SOC. The maximum estimation error of SOC is no more than 1.5%, and ACKF algorithm has strong robustness.

Abstract:The millimeter-wave frequency band of 5G wireless communication has become the research direction of the future mobile communication field because of its large bandwidth and high speed. However, in the millimeter wave frequency band, the signal generator needs to realize the configuration of different signal bandwidths. In response to this problem, this paper designs a parallel interpolation structure combining a fast FIR filter and a polyphase interpolator, which can expand the transmission bandwidth range of the 5G millimeter-wave signal generator, and has high interpolation efficiency, low power consumption, and low resource consumption. Advantages such as low occupancy rate. The experimental verification results after simulation and optimization show that, without affecting the demodulation results, the interpolation structure proposed in this paper reduces the use of fast FIR interpolator DSPs by 25% compared with the traditional 4-channel parallel 4-times interpolator. The use of LUTs is reduced by 15.2%, and the power consumption of DSPs is reduced by 23.8%, which meets the actual needs of low system resource occupancy and low power consumption. The results have been applied to domestic 5G base station comprehensive testers.

Abstract:As one of the most basic geometric quantities, the accuracy of angle directly affects the product quality of manufacturing industry. Aiming at the technical bottleneck that the number of scribed lines of grating sensor is difficult to be further improved, an angle sensor measurement system based on pyramid is proposed in this paper. Through the regular pyramid reflection structure, multiple laser beams are incident on the regular pyramid prism at 45°, and combined with multiple position sensitive detectors set in dislocation, the continuous measurement of angle is realized. The experimental results show that the principle of the angle sensor based on pyramid is correct. Within the range of 1.2°, the measurement error of a single position sensitive detector is ±3.5″; When four position sensitive detectors measure continuously and alternately, the measurement error is ±4″ within the range of 4.8°. The pyramid angle sensor proposed in this paper does not need grid scribing, and the accuracy is related to the distance of the position sensitive detector and its measurement accuracy. With the improvement of the accuracy of the position sensitive detector, the measurement accuracy of the system is bound to be further improved.

Abstract:Aiming at the problems of poor versatility, complex design and low code rate of traditional missileborne telemetry communication system equipment, this paper adopts QPSK (Quadrature Phase Shift Keying) modulation and demodulation algorithm with high frequency spectrum utilization and strong anti-interference ability. The integrated telemetry system of software radio zero intermediate frequency transceiver of AD9364 and ZYNQ-7000 is built. The system performs QPSK modulation on the data collected and programmed by PCM coder, and transmits the data from PL(Programmable Logic) terminal to AD9364 through LVDS data interface. After mixing, upconversion, filtering and other processing in the chip, the system sends the data through antenna. Antenna can receive telemetry data back, after AD9364 down conversion, filtering, transmission to PL terminal for demodulation, carrier synchronization, complete the whole telemetry system transceiver process. In order to realize the accurate tracking of frequency skew signal, the second order frequency locked loop is used to assist the third order phase-locked loop and a new COSTAS loop is proposed to improve the phase detection accuracy. By building a telemetry test system and testing the transmission distance and analysis results of the telemetry system, the code rate of 10 Mbps high transmission is realized, and the transmission distance can reach more than 13 km.

Abstract:sampling oscilloscope is widely used in the field of broadband measurement. As the key device of sampling oscilloscope, sampler converts high-frequency signal into low-frequency signal based on comb wave convolution theory, and the output signal of sampler is transient value, and the symmetry of two output signals is poor. Based on the analysis of the principle of microwave broadband sampler, this paper will analyze the impedance integration circuit of IF signal output by sampler The offset control circuit is designed. Experimental verification shows that based on the designed sampler and impedance integral amplification circuit, 20 GHz RF signal can be converted into 50 kHz IF signal, and the offset control circuit can realize the calibration of signal offset, which can be used in the analog front end of sampling oscilloscope.

Abstract:In order to obtain the relationship between the position of the transmission peak of the photonic crystal defect and the refractive index (concentration) of the solution，a one-dimensional photonic crystal sensor model of defect layer constructed by sodium chloride solution is proposed. The relationship between defect peak wavelength and concentration of sodium chloride solution is calculated and analyzed by transfer matrix method. The results show that the wavelength of defect peak is proportional to the refractive index of solution and the concentration of sodium chloride. The sensitivity of solution concentration measured by the sensor model is 0.57 nm/%. The one-dimensional defect photonic crystal sensor model proposed in this paper has the advantages of high sensitivity, low temperature influence and real-time detection, which provides a theoretical reference for the design of practical photonic crystal sensors.

Abstract:Aiming at the problems of complex parameter identification and poor adaptability of traditional PID control algorithm in temperature control of moxibustion robot, reinforcement learning is introduced into the field of temperature control of moxibustion robot, and an improved reinforcement learning algorithm is proposed. First, the offline training simulation environment of the agent is jointly built by multi-physics simulation software and neural network to solve the problem of low efficiency of online training of the agent; then, an improved reinforcement learning algorithm combining reward guidance and cosine annealing strategy is proposed to improve the convergence and success rate of the algorithm; finally, the model trained in the simulation environment is transferred to the real environment for experimental verification. The experimental results show that the temperature overshoot is 0.2 ℃, and the steady-state temperature is kept within 43.1±0.4 ℃. The improved reinforcement learning algorithm has better temperature control ability than the traditional PID control algorithm.

Abstract:The accuracy of locating R wave in ECG signal is the basis of other waveform location, and it is in the primary position in the study of ECG signal characteristic waveform. According to the most obvious characteristic of R wave in the whole ECG waveform, this paper proposes a method to extract R wave based on wavelet theory. First, select an appropriate method to remove the noise from the ECG signal, and then perform 5-layer wavelet decomposition on the denoised ECG signal to obtain approximate signals and detail signals of each order. The average value of the superimposed signal is used as the search threshold, and the search starting point is determined. Finally, map the abscissa of the search starting point to the pure ECG signal, and use the method of comparing the amplitude point by point to search backward, until the R wave is found. Using the data in the MIT-BIH database for multiple experimental verifications, the experimental results show that this algorithm can accurately locate the R wave, and the accuracy of the inspection is 99.65%, the recall rate is 99.86%.

Abstract:Device to device (D2D) communication can be based on cellular facilities to improve resource utilization, user throughput and save battery energy. In D2D network, mode selection and resource allocation are the key issues. In order to improve the sum rate and spectrum efficiency of D2D communication, a scheme of joint mode selection, power and resource block allocation is proposed. Firstly, the mode selection criteria are selected according to the user′s geographical location to help the user select the corresponding communication mode; Then, for the multiplexing communication mode, the asynchronous dominant action evaluation (A3C) algorithm based on deep reinforcement learning is used to allocate resource blocks and power to different D2D users. The simulation results show that the joint optimization scheme based on A3C algorithm proposed in this paper has fast convergence speed and better performance than other algorithms.

Abstract:Aiming at the requirement of real-time feedback of test status in the storage-type explosion shock wave overpressure test system, a wireless multi-channel shock wave overpressure acquisition and storage system based on data transmission radio network is designed. The system takes XC6SLX16 as the core, which can simultaneously collect and store four channels of shock wave overpressure data. The sampling rate of each channel is 1 MSPS. The trigger threshold can be set. The integrated design of the system has been carried out to improve the adaptability of the system, and it is suitable for various experimental environments. A small-equivalent static explosion test was carried out at the shooting range, and the system successfully obtained accurate shock wave overpressure data. Compared with the HBM standard equipment, the error was within 5%. The test results prove the accuracy of the data obtained by the system, and provide a test technology for the damage assessment of new weapons.

Abstract:Long-term shielding will affect photovoltaic panel photoelectric conversion and produce hot spots, and the output U-I curve of the corresponding photovoltaic array will also change, and the corresponding U-P curve will contain multiple peaks. In order to accurately locate hot spots, it is necessary to analyze the mechanism of hot spots generation and build a universal pv array model with shadow occlusion. In view of the traditional engineering model cannot accurately describe the electrical parameters of complex PV array such as CTCT, this paper proposes a multi-point flat throw motion pv array mathematical model with shadow occlusion. In order to verify the accuracy of the model, Matlab/Simulink modeling and simulation was used to build a hot spot simulation platform, and the light intensity was changed to simulate shadow occlusion and the connection mode of photovoltaic panels. The U-I and U-P characteristic curves of photovoltaic array were obtained and analyzed, compared with the engineering model, the proposed model reduces the operation time to 2.5%, and can describe the output characteristics of the complex photovoltaic array under local shadow more comprehensively.

Abstract:Detecting and identifying cyber attacks is crucial to prevent cyber attacks such as advanced sustainable threats, promote the healthy development of network infrastructure, and guarantee the safe and stable operation of network facilities. In this paper, the key characteristics of network attacks in network traffic data are selected by using mutual information theory, a gray wolf boosting algorithm is proposed by improving the gray wolf optimization algorithm, and a GWB-LSSVM model is provided based on this algorithm and least squares support vector machine. The model shows good detection performance for the current main forms of network attacks. The experimental results based on NSL-KDD data set show that its detection precision, detection rate and detection accuracy reached 99.7%, 99.3% and 99.1% respectively. Compared with some existing research work, its detection precision is improved by up to about 2.58%, detection rate by up to about 3.98%, detection accuracy by up to about 3.78%, and the training time of the model by up to 55.9%.

Abstract:Deep neural network is one of the mainstream surface defect detection methods, a large number of samples are needed for model training, but the defect samples of the same type of ceramic tile are limited with the diversification of ceramic tile products. In this paper, a ceramic tile surface defect detection method based on improved domain countermeasure neural network (MDANN) is proposed. Referring to the traditional DANN structure, the network parameters are pretrained on the ImageNet to improve the training speed. Then, the bottleneck layer is added to the original network, and the maximum mean difference index is used to optimize the field distribution difference, which improves the ability of the original DANN network to screen the source domain and realizes the defect detection of small sample tiles. The experimental results show that the effective detection rate of MDANN for ceramic tile surface defects achieves 98.77%, which is 3.53% higher than the original DANN network. It can be quickly applied to the detection of different types of ceramic tiles with good generalization.

Abstract:In view of the current detection and judgment of the environmental mode of subway air conditioning system, there is still the problem of low efficiency and low intelligence degree. BP neural network optimized by Adam is designed to detect the environmental mode of subway air conditioning system. Choose three key variables: smoke concentration, carbon dioxide concentration, temperature, as a condition of environment characteristics of pattern recognition, Adam algorithm are used to optimize the gradient descent of the traditional BP neural network model, first-order moment estimation and second-order moment estimation are used to dynamically adjust the learning rate of each parameter, to speed up the model learning, improve the identification accuracy of the network, and reduce the oscillation during convergence. The experimental results show that the convergence speed of the optimized BP neural network subway environmental mode detection model is improved by 98.88%, the average number of prediction errors is reduced by 45.6%, and the oscillation is greatly reduced in the convergence process. At the same time, compared with other machine learning multi-classification models, the accuracy of the optimized BP neural network model is 99.88%, the detection running time is 12 ms, and the overall performance is better.

Abstract:In order to improve the reliability and transmission speed of the communication system, increase the transmission distance, reduce the BER of the system and achieve the purpose of localization, the innovative design of dual redundancy and CRC combined algorithm and the use of domestic devices, including domestic FPGA, to complete a set of high reliability high-speed communication system of the fully localized system design. After a variety of complex scenarios and tests, the transmission speed was increased from 0.4 Gbit/s to 1.25 Gbit/s, the single channel transmission was upgraded to dual channel redundancy transmission to improve the system reliability, the transmission distance was increased from 40 m to 100 m, and the BER was reduced to 0. The system achieves high reliability, high speed, long distance, zero BER and localization in a complex environment. The purpose is to maintain high reliability, high speed, long distance, zero BER and localization in complex environment. It has been applied in a project.

Abstract:In order to solve the problem of dangerous gas leakage detection, this paper proposes an outdoor leakage gas detection algorithm based on VIBE algorithm. Aiming at the problems in the detection process of the current baseline algorithm, such as low gray contrast between the leaked gas and the surrounding background imaging, the existence of non-gas motion foreground interference, and the serious noise of the detection results, this paper uses the following methods to solve them. Firstly, the infrared source image is enhanced by gas enhancement and background blur to enhance gray contrast through multiple cascade enhancement algorithms. Then, the gas enhancement detection image and the background blur detection image are subjected to differential operation to remove the non-gas motion foreground interference. Finally, the connected domain filtering is performed. Eliminate residual noise and obtain leak gas detection results. This paper conducts comparative experiments and robust detection experiments on a large number of real data. The experimental results show that this paper has stronger detection performance than the baseline algorithm, in which the detection rate and false detection rate are increased by 13.21% and decreased by 5.62%, respectively. The algorithm is time-consuming About 25 ms/frame, meeting real-time requirements. In this paper, the robust detection of outdoor gas leakage is realized by multiple modules such as cascade image enhancement, gas detection and false alarm filtering, which improves the gas detection accuracy and generates intuitive visual information to accurately locate the gas leakage source and escape range., and solves the problem of outdoor dangerous gas leak detection.

Abstract:A fully convolutional dual-path neural network model with improved cross-entropy loss function is proposed to solve the problem of identifying partial discharge maps of electrical equipment. Using the partial discharge map as the model input, the deep and shallow features of the map are extracted by two channels using different size convolution kernels, and then performing feature fusion. The convolutional layer is used instead of the fully connection layer to preserve more spatial correlation between PD features. The improved cross-entropy loss function can make the model more suitable for the situation of imbalanced datasets. The experimental results show that the accuracy of the improved FCN dual-path feature fusion recognition method reaches 99.31%, which can accurately identify the partial discharge map, and the amount of model parameters is smaller.

Abstract:The use of unmanned aerial vehicles for intelligent inspection of transmission lines has become the mainstream of the industry. Insulator defect detection is a key link in intelligent inspection operations. Aiming at the problem of low accuracy of insulator defect detection in complex environment, this paper proposes an improved YOLOv5s insulator defect detection algorithm. Firstly, the existing data sets are enhanced by random rectangle occlusion, horizontal flip, random pixel zeroing and adding random pixels, and the K-means algorithm is used to cluster the data sets to obtain the optimal anchor frame size, which effectively improves the generalization ability and positioning accuracy of the model. Secondly, GAM attention module is added to the end of the main network of YOLOv5s and the last three convolution networks of different scales, so that the model can be noticed on a larger network to solve the influence of invalid features on the recognition accuracy. Finally, based on the feature pyramid structure FPN, the adaptive feature fusion ASFF module is introduced to enhance the feature extraction ability of the network. The experimental results show that the accuracy and mAP0.5 of the improved YOLOv5s model are 2.4% and 2.2% higher than those of the original YOLOv5s network, respectively.

Abstract:Transformer-based target detectors proposed in recent years have simplified the model structure and demonstrated competitive performance. However, most of the models suffer from slow convergence and poor detection of small objects due to the way the Transformer attention module handles feature maps. To address these issues, this study proposes a Transformer detection model based on a pre-filtered attention module. Using the target point as reference, the module only samples a part of the feature points near the target point, which saves training time and improves detection accuracy. A newly defined directional relative position encoding is also integrated in the module. The encoding compensates for the lack of relative position information in the module due to the weight calculation that is more helpful for the detection of small objects. Experiments on the COCO 2017 dataset show that our model reduces the training time by a factor of 10 and improves the detection accuracy, especially on small object detection by 26.8 APs.

Abstract:The method of analyzing human behavior based on human skeleton data is highly interpretable and has obvious advantages in the research of human behavior analysis based on vision. However, viewing angle interference and target occlusion seriously affect the calibration of human skeleton joints. This paper proposes a human skeleton joint point estimation algorithm based on human pose features under the constraints of human structure, and recognizes human behavior based on skeleton data. Firstly, according to the steady-state trend and transient changes of human motion, feature extraction models are established based on decision tree and weighted linear regression, respectively, to estimate missing or confused joint points. Then, an action recognition network model combining lightweight temporal convolution and attention graph convolution is designed to optimize the model for the time scale of action samples. The occlusion condition was established in the NTU RGB+D 60 dataset for experiments, and the accuracy rates were 90.28%(CV) and 81.95%(CS), respectively, and 98.2% in the UTD-MHAD dataset, which were better than those of the existing methods.

Abstract:Aiming at the problem of large error in manual inspection of welding gap of Y-type pipe joint set, a visual inspection method for welding gap of Y-type pipe joint set based on improved Canny algorithm was proposed. Firstly, the Y-joint clearance image is enhanced with multi-scale details to reduce the influence of image details missing caused by filtering. Secondly, Otsu maximum inter-class variance method is used for threshold selection, which reduces the defects caused by manual selection of threshold. Finally, an algorithm is designed to extract the characteristic points and calculate the clearance, which can measure the clearance of Y-type pipe joint. The experimental results show that the method used in this paper can accurately and directly detect the clearance amount of Y-type pipe joint pairs, which meets the requirements of joint clearance quality detection in actual production.

Abstract:To address the problems of false detection and low detection accuracy of traditional wind turbine paddle detection algorithms in complex environments, a wind turbine paddle defect detection method integrating multi-scale features and attention mechanism is proposed. Firstly, the improved backbone network L-ResNet50 is used for feature extraction to retain more effective information. Then the attention mechanism module is embedded for different scale feature layers to enhance the focused semantic information. Finally, the extracted deep features and shallow features are fused with multi-scale features to further improve the model accuracy. Through the defect detection experiments on the wind turbine paddle images captured by UAV aerial photography, the results show that the average accuracy of the proposed method in the detection of wind turbine paddle defects in complex environments is improved by 8.2% compared with the original Faster R-CNN model.