Abstract:For the mower motor torque ripple control process exists, and poor dynamic performance, the design of the vector system speed brushless DC motor mower. Instead of using the PI control sliding mode controller for adjusting the speed of the motor, and a position estimation system were designed based on the high frequency square wave oscillator pulse injection SMO proposed a revolution in the motor mechanical angle analysis Within the two errors, the strategy of switching the position estimation method realizes the position estimation of the motor at full speed. Simulation in MATLAB, and take advantage of STM32F302 chip built hardware test platform. The results show that the two rotor position estimation methods can be switched smoothly, which improves the accuracy of the rotor position estimation. The use of the sliding mode controller reduces the speed adjustment time and improves the robustness of the system.

Abstract:In flight test, the aircraft will produce mixed combustible gas, and the change of gas concentration will bring great security risks to flight test safety. In view of the limitations of traditional test methods, such as long response time and low accuracy, the new aircraft and new mission put forward higher requirements for gas concentration test. According to the requirements of the model task, through the analysis of the test environment, gas composition and current gas concentration measurement technology, the overall technical scheme of mixed gas concentration test is determined, and a new airborne mixed combustible gas concentration test system is developed, which realizes the real-time and high-precision measurement of gas concentration. This paper mainly expounds the principle and design scheme of the test system, and introduces in detail the background noise removal technology based on time division multiplexing. The ground and flight tests are carried out respectively. After data analysis, the system is stable and functions normally, and all indexes of the test data can meet the requirements of the model task test.

Abstract:In order to meet the performance requirements of the solar power supply system for stable output power, high solar conversion rate, low cost, miniaturization and easy deployment, a small-scale solar power supply system with track the sun using Kalman filtering algorithm and MPPT technology is proposed. The design adopts photoelectric method to track the position of the sun. The control process is modeled and simulated in SIMULINK. It is designed to add a Kalman filter to the output of the classic PID control to make the system achieve the best control effect. The MPPT control circuit is added to the power output end to make the solar panel in the maximum power output state. The experimental results show that the system has high solar energy utilization, good power output effect, and high practical application value.

Abstract:Faced with different sea conditions, the shipboard coherent S-band wave radar digital receiver needs to support 6 different baseband sampling rates and bandwidths. In order to meet the multi-functional integration of radar modules and the radar system's requirements for multi-rate and multi-bandwidth, an FPGA-based multi-bandwidth digital down-conversion（DDC）solution is proposed, which can dynamically load decimation multiples and filter coefficients according to the receiver operating parameters and suitable for different sea condition. A 16-stage pipeline numerically controlled oscillator（NCO）module is realized through the coordinate rotation digital computer（CORDIC）; the distributed algorithm（DA）is improved to realize a fully parallel 128-order FIR filter, and the DA channel is time-division multiplexed; the asynchronous FIFO module guarantees that data is transmitted without distortion across clock domains, and the data is appropriately truncated to save hardware resources. The final result shows that the baseband I/Q signal frequency is accurate, and the signal-to-noise ratio is as high as 60 dB, which meets the design goals and radar requirements. This structure can save a lot of multipliers and RAM resources. After the DA structure is time-division multiplexed, it can save an additional 42.8% Look-up table resource. The designed DDC is reasonable and feasible, which can effectively switch the decimation factor and filter coefficient according to different sea conditions, and provide a stable and effective data source for the subsequent baseband signal processor. This has important reference significance for the multi-functional integration of radar modules.

Abstract:At present, most of the existing simulation test systems are aimed at the user, mainly simulating the actual operation process, test interface, etc., lack of simulation test for the parameters of the control equipment. In view of the current situation, a comprehensive test platform integrating system simulation and equipment test simulation is developed. The simulation platform not only tests the actual application interface, but also tests the control parameters of the control equipment. After testing, the effect is good, which meets the needs of system simulation test and control equipment parameter test, and the simulation test results are more comprehensive and complete. This paper mainly introduces its design and implementation, including the overall design, structure, working principle, test and application results.

Abstract:A 16-electrode electrical impedance imaging system was designed to facilitate the detection of defects in the aircraft skin structure and to make up for the lack of manual testing. The sinusoidal signal was used to stimulate the sample in a relative excitation mode. Compared with the relative reference amplitude obtained by forward operation, measured value demodulation information and the image reconstruction algorithm were used to obtain the conductivity distribution image of the skin structure based on open source software Octave and package EIDORS. So the defect status of skinning structure is characterized by the changing trend of its conductivity. Commonly used material 2024 T-3 aluminum alloy plate in aviation manufacturing industry was selected as a sample for real-time resistance anti-imaging experiments, and the reconstruction image quality is evaluated by PE and other evaluation indicators. The results show that the change of conductivity caused by defect of the sample can be measured by using the system and the system can effectively monitor the structural defects of the metal skin.

Abstract:In order to further improve the performance of motor imagery electroencephalogram (EEG) decoding, a new common spatial pattern (CSP) improvement method is proposed to address the problems of the CSP feature extraction method, that is, the logarithmic band power feature extraction method based on CSP transform and filter bank. First, the original EEG signals are preprocessed; then the preprocessed signals are spatially filtered using CSP transform; after that, the spatially filtered signals are decomposed into multiple sub-bands using filter bank, and the logarithmic band power of each sub-band signal is extracted as a feature; finally, the least absolute shrinkage and selection operator (LASSO) is used for feature selection, and the support vector machine (SVM) is used for classification. Experiments were conducted on the data set IIa of the brain-computer interface (BCI) competition IV, the proposed method achieved the highest average classification accuracy, and the result was 81.97%. The experimental results show that the classification performance of the proposed method is better than the existing improved CSP method, and the feature extraction time also has a greater advantage.

Abstract:The harmonic detection method based on FBD power theory was widely used in active power filters because of its simple implementation and strong stability. However, the detection accuracy of traditional FBD method is reduced due to the phase-locked loop unlocking under the voltage distortion of the power grid, and there is inherent contradiction between the steady-state error and the response speed when using the low-pass filter.To solve this problem, an improved method is proposed. In this method, the positive sequence voltage component is extracted quickly by shifting the three-phase voltage by 600 to replace the reference voltage signal to participate in the operation, so as to avoid the detection error caused by voltage asymmetry.An adaptive filter with variable step size based on hyperbolic tangent function was constructed to obtain the equivalent active conductance of fundamental wave. The adaptive algorithm and FBD detection algorithm were integrated organically, and the dynamic response speed of the algorithm was improved effectively.Theoretical derivation and simulation results confirm the correctness and effectiveness of the improved algorithm.

Abstract:Because a lot of jamming signals are often included in the bearing vibration signals, the key to bearing diagnosis is to extract fault features efficiently and classify them. Traditional methods of fault feature extraction often need a variety of index sets to represent different faults. In this paper, a method of bearing fault diagnosis based on Laplacian matrix weighted by Mahalanobis distance and improved k-means clustering is proposed. Firstly, the time domain discrete signal of the bearing is mapped to the graph domain to obtain the graph signal, and the set of characteristic indexes representing different bearing fault states is obtained by using Laplacian matrix in an algebraic form of the graph signal weighted by Mahalanobis distance. Then, the improved k-means clustering idea is applied to evaluate and classify the set of characteristic indexes, to realize the classification and recognition of different bearing fault states in the case of single index. The experimental results show that the method of bearing diagnosis based on Mahalanobis distance weighted by Laplacian matrix and improved K-means clustering can effectively extract and precisely classify the characteristic indexes of different bearing faults. At the same time, the accuracy of this method in single index classification is much higher than that of traditional fault feature extraction methods.

Abstract:Coal mine locomotive transportation plays the role of transporting materials and gangue in the daily production of coal mine. As the carrier of transportation, whether the locomotive can safely and efficiently complete the transportation task assigned by the dispatching center is undoubtedly very important for coal mine production. Reasonable scheduling optimization of locomotives can not only improve the transportation efficiency, but also greatly reduce the collision and other safety accidents. In this paper, an improved optimization algorithm combined with cross factor and simulated annealing algorithm is used to solve the scheduling optimization problem of underground locomotive transportation in coal mine. The improved algorithm introduces cross factor to update the individual position, and selects the individual to enter the next iteration according to the fitness increment of simulated annealing algorithm. In this way, the parameters of optimization algorithm are easy to adjust and the information sharing process is retained. It also solves the problem of premature and low precision of Drosophila algorithm. Through the simulation of underground locomotive transportation in coal mine, it is proved that the path planned by the algorithm is more reasonable and the efficiency of locomotive transportation is higher.

Abstract:In order to improve the accuracy of sign language recognition, this paper proposes a multi-sensor sign language recognition method based on Hybrid Particle swarm algorithm Support vector machine (HPSO-SVM). In the raw data collection stage, the ZTEMG-2000 EMG sensor is used to collect the EMG signal of the human arm surface, and the MPU6050 sensor is used to collect the right-hand acceleration and angular velocity signals.In the pretreatment phase of the experiment, short-term energy method, optimized by altering the adaptive fault tolerance length, is introduced to improve the extraction accuracy of the active segment. in the classification method stage, the optimal combination of the penalty factor of the support vector machine (SVM) and the kernel function parameter is found through the hybrid particle swarm algorithm (HPSO), and the SVM model is optimized. In experiments, the five Chinese sign languages performed by each subject were recognized, and the average recognition rate reached 96.78%. This method uses a relatively small number of more economical sensors to recognize sign language, and the recognition accuracy is 5% higher than that of the traditional SVM algorithm, demonstrating the superiority of this method in sign language recognition.

Abstract:In order to address the problem that the traditional calibration method will face degraded calibration performance or even complete failure in non-uniform background fields, this paper designs a new calibration method for vector magnetometer arrays in non-uniform background fields, establishes a new error correction model based on the rotationally invariant property of the tensor invariant at a point in space, and uses a numerical method to estimate the calibration parameters of the system. After simulations and field tests, the scaling of the magnetic gradient tensor matrix corrected by this tensor-invariant-based correction (TI) method is approximated to be a constant under various background fields, which is consistent with the theoretical analysis, indicating that the TI correction algorithm has satisfactory correction performance under different background fields. The results show that this correction method can be applied not only in non-uniform fields but also in uniform field environments, with better real-time performance and better correction effect than the conventional method.

Abstract:The traditional single fuzzy classifier method, which uses fixed de-fuzzification rules, is easy to case the problem of text ambiguity in emotional data. To solve the problem, a text classification method based on deep neural networks and fuzzy rules is proposed. The method is divided into two main stages. In the first stage, we use fuzzy rule formation algorithm and different fuzzy norms, feature extraction method (word bag method, word embedding vector), and feature subset selection method based on association to prepare features, so as to train multiple fuzzy classifiers. Then, we fuse classifiers to identify ambiguous instances. In the second stage, disambiguation instances are sorted out, a second training set is generated, and KNN is used to classify new disambiguation instances. Compared with the current advanced methods, the proposed method has better classification performance in most cases, and the Wilcoxon rank test is statistically significant.

Abstract:Aiming at the problem of long time and low success rate of multi-robot cooperative rounding in unknown dynamic environment, a new multi-robot cooperative rounding method based on biologically inspired neural network is proposed. First, a multi-robot collaborative rounding model is built, and the dynamic alliance strategy is used to realize the linkage of multiple robots. Second, a tracking strategy based on biologically inspired neural networks is constructed to dynamically guide all robots in the alliance to track. Finally, a formation strategy is used to achieve the target rounding. The experimental results show that the average capture time of the proposed method is 12.7s, 22.3s, 34.2s, 17.7s and 28.5s under the conditions of single target, multiple targets, partial robot failures, obstacles of different shapes, and different regular environments. The average capture success rate is 97.4%; compared with other multi-robot cooperative hunting algorithms, the algorithm proposed in this paper has advantages in capture time and capture success rate.

Abstract:As the degree of informatization continues to deepen, the application of robots is becoming more and more extensive. However, in many cases, robots need to work in a constantly changing and complex environment. Because of the inability to obtain environmental information in advance, it is often difficult to plan a suitable path for a robot. To solve this problem, this paper proposes an method for robot path planning. This method uses the grid method to establish an environmental model, uses the number of exploration steps to define the return value, and continuously optimizes the path through reinforcement learning. At the same time, aiming at the problem of the balance between the exploration and utilization of the environment in reinforcement learning, a variable ε-decreasing action selection strategy and learning rate selection method are proposed to make the exploration factor dynamically change as the agent explores the environment, thereby accelerating the convergence speed of the learning algorithm. Simulation results show that this method can realize autonomous navigation and fast path planning of mobile robots in complex environments, compared with traditional algorithms, under the premise of obtaining the same path length, the number of iterations is reduced by approximately 32%, effectively speeding up the convergence speed.

Abstract:Abnormal locomotive roof equipment status and foreign matter are one of the factors that affect the normal operation of locomotives. Research and application of locomotive roof status and foreign matter detection technology is of great significance to the safety of railway. This paper first discusses the three detection technologies in the status detection of locomotive roof equipment, which are mechanical contact type, non-contact type based on linear images, and comprehensive dynamic monitoring type. The application scenarios suitable for these technologies and the advantages and disadvantages are explained and compared; On the basis of traditional manual foreign matter detection, this paper focuses on the roof foreign matter detection technology based on image processing, summarizes the technical methods used in the locomotive roof imaging, image preprocessing, and foreign matter detection, and points out the development trend of the locomotive roof foreign matter detection technology and the main content to be studied.

Abstract:In the electromagnetic radiation information leakage of computer system, video signal can be intercepted and recovered easily. Aiming at the problem of video signal recovery of VGA cable radiation in computer system. The transmission and electromagnetic radiation model of computer analog video signal in VGA transmission channel are established. The imaging rules of analog video signal and the time domain characteristics of analog video signal in each stage during transmission are analyzed. The recovery steps of analog video signal radiated by VGA cable are obtained. Due to the poor quality of the quality of single frame video signal recovered at low sampling rate, according to the image presentation rules of video signal, multiple consecutive recovered images are overlapped, and the recovered image with high quality can be obtained at low sampling rate. Simulation results show that this method can effectively improve the quality of video image restoration, especially at low sampling rate. In order to verify the correctness of the simulation results, the actual intercepted VGA cable radiation information is restored, and the results are in accordance with the simulation.

Abstract:For catalytic materials, the size and shape of nanoparticles and other structural information have an important impact on the catalytic performance. The identification and statistics of nanoparticles based on transmission electron microscope images are the main means to obtain these information. In this paper, a deep separable convolutional U-Net architecture based on deep learning is proposed. Taking the core-shell nanomaterial as dataset, cross-entropy loss function, weighted cross-entropy loss function, IoU loss function and Dice loss function are adopted as optimization objectives to train the network respectively. The segmentation results show that the performance of IoU（Intersection over Union）loss function and Dice loss function is better for the dataset of core-shell structure nanoparticles with unbalanced positive and negative samples. Finally, the trained network is used to segment and conduct statistics on TEM images to obtain structural information such as particle size and perimeter distribution, which provides feasibility for the application of deep learning in the field of catalytic materials.

Abstract:Vehicle to Everything (V2X) communication, which can effectively improve traffic safety and mobility, is one of the key technologies in vehicle deployment scenarios. V2X communication links need to meet different Quality of Service (QoS) requirements for different applications, such as the latency and reliability requirements of Vehicle to Vehicle (V2V) links. Considering rapid changes in wireless channels due to vehicles high mobility, while ensuring QoS constraints for different vehicle links and improving the robustness of dynamic networks, a joint optimization framework based on Federal Deep Reinforcement Learning (FDRL) for spectrum allocation and power control is proposed. The framework first proposes the corresponding optimization problems according to different vehicle link requirements, and defines the state space, action space and reward function for reinforcement learning. Then the joint deep learning reinforcement learning training framework is given. Finally, the optimal spectrum allocation and power control strategies are found by distributed vehicle-side reinforcement learning and base station aggregation averaging training. Simulation results show that the proposed framework can improve the total transmission rates of all the vehicle-to-infrastructure (V2I) users and guarantee the robustness of the network when new vehicles are connected to the network compared with other comparative algorithms.

Abstract:In order to solve the problems that maximum simplex volume-based endmember extraction algorithms (EEAs) involve processing entire pixels and are sensitive to noises, this paper proposes a robust maximum simplex volume-based EEA for quickly selecting endmembers from hyperspectral images. The proposed algorithm first applies principal component analysis to reduce the hyperspectral image into p-1 subspace. It then detects convex hull points from each component pair by employing a convex hull algorithm. Next, it iteratively specifies p points and their simplex volume until they can provide a maximum simplex volume. Finally, it transforms p points into original dimensionality and obtains p denoised endmembers. Experiments conducted on synthetic and real hyperspectral images demonstrate that the proposed algorithm can quickly extract endmembers from the denoised hyperspectral. The proposed method can perfectly meet the requirements of high endmember accuracy and real-time in the field of hyperspectral endmember extraction.

Abstract:Facial expression recognition has a very important application prospect in computer vision fields such as human-computer interaction and emotion calculation. Aiming at the challenges of complexity, diversity, occlusion and illumination of facial expression recognition, a new end-to-end network is proposed, and attention mechanism is applied to automatic expression recognition. The new network architecture consists of four parts: feature extraction module, attention module, reconstruction module and classification module. By extracting image texture information from LBP features, the tiny motion of face is captured and the network performance is improved. attention mechanisms can make neural networks pay more attention to useful features. we combine LBP features and attention mechanisms to improve the attention model to improve the recognition accuracy. applying the newly proposed method to two representative expression datasets, namely JAFFE、CK、FER2013 and. the experimental results show that the accuracy of facial expression recognition on three data sets is 98.95%,98.95% and 79.89%, respectively. It is proved that the method is beneficial to improve the recognition rate of facial expression and is advanced.

Abstract:Due to the limitation of frequency band, polarization and ground antenna resources, data reception may have potential interference risks. This paper expounds the situation of three satellite digital frequency, bandwidth and the polarization of antenna as the channel design, establishes double satellite signal interference simulation model, and analyzes the influence of the multi-satellite in-orbit network observation on the data reception. The analysis results show that the multi-satellite data transmission signal characteristics and channel parameters is reasonable and feasible. In the case that multiple satellites exist and are visible to the same ground station, through the adjustment of their orbit optimization, the simultaneous visibility of multiple satellites with domestic and foreign ground stations can be avoided, so as to ensure the normal transmission and receiving of data.

Abstract:Simultaneous Wire-less Information and Power Transfer (SWIPT) can reduce energy consumption in the network and extend the lifetime of the energy consuming devices in the network. The object of this study is to transmit beam-forming signals to communicate with users. The receiver uses power divider to separate the signals and collect information and power respectively. A part of the signal flow is used for energy collection, and a part of the signal flow is used for information demodulation. By using beamforming technology, the transmission power of the system is minimized and the sum rate of the system is improved.At the same time, the non-convex problem is transformed into a Semi-Definite Programming (SDP) problem by using the Semi-Definite Relaxation (SDR) technique and S-program.Finally, simulation results show that the beamforming system based on wireless communication and simultaneous interpreting is correct and reasonable.

Abstract:In this paper, a quality control method of acquisition instrument is proposed to solve the problem of long cable and multi-channel of large-scale deep-water seismic streamer, including indoor system operation quality control method and underwater streamer data acquisition quality control method, which is realized by active and passive modes. In the active mode, analog or digital excitation signals can be generated on demand and sent to the acquisition instrument to evaluate the key performance parameters of the instrument; in the passive mode, various state parameters in the operation process of the instrument can be collected in real time to realize the real-time evaluation of the operation quality of the instrument. Based on the complete quality control method independent of acquisition instrument, the quality of acquisition instrument can be effectively controlled, the quality of data acquisition can be guaranteed, and the risk of exploration operation can be greatly reduced. The results of long-term laboratory test and marine operation show that the method is correct and realizable.

Abstract:Aiming at the problem that it is difficult for the Photo-elastic modulation Fourier transform spectrometer to collect a complete period of laser interference signal and cannot perform the real-time restoration of the spectrum, this paper designs a real-time restoration of the Photo-elastic modulation Fourier transform spectrum based on LabVIEW. First, a He-Ne laser with a wavelength of 632.8nm is selected as the light source. After the incident laser passes through the Photo-elastic modulation Fourier transform spectrometer, an interference signal that changes with time is generated, and then the interference signal is converted by a Photo-elastic detector to obtain interference data. Finally, use a high-speed data acquisition card to collect the data [基金项目：1、山西省青年科学基金（201901D211234）；2、山西省自然科学基金（201901D111145） ]and transmit it to the LabVIEW host computer, perform data processing in the spectrum restoration of the interference data, and perform wavelength calibration and restoration analysis. Experiments show that LabVIEW can obtain a complete cycle of the interference signal from the collected monochromatic laser on the host computer, and the spectrum restoration algorithm can be used to restore the spectrogram in real time. The error of the restored laser is less than 1nm, and the half-peak width is 0.01um.

Abstract:In order to meet the needs of leakage monitoring in special environment such as electromagnetic interference, this paper designs an optical fiber lateral coupling leakage monitoring system based on STM32. The system is mainly divided into hardware module and upper computer module. The system hardware consists of optical fiber sensor module and signal converter module. The optical fiber sensor module is composed of POF fiber with defect structure, LED light strip and sensor package shell; STM32 is used as the main control chip in the signal converter module to realize the function of weak light signal acquisition, amplification and transmission. The upper computer of the system uses the filtering algorithm to realize the mean difference comparison of the leakage signal, filter out the occasional pulse, reduce the false alarm rate and system response time, and improve the sensitivity of the sensing band. The experimental results show that the system can work stably, each sensor point can achieve a single point calibration, the leakage monitoring range can reach 8m, and the leakage and false detection rate is less than 2%.

Abstract:To address the high-overload data problem in the bore during the development and testing of small rockets, a wireless missile-borne acquisition system based on MRAM is proposed. The hardware part of the system uses wireless transmission to achieve real-time data reception, uses the high reliability of MRAM to store data, and uses data readback to improve the reliability of the system; the software part uses continuous threshold trigger design to reduce false triggers Probability, using programmable digital conditioning modules to provide better linearity, accuracy and temperature stability for data acquisition. The test results show that the system can collect and store overload data at a sampling rate of 100 kHz under severe conditions in the bore. It can wirelessly receive data within 900 m, and the transmit current is 430 mA, which not only has the advantage of high reliability, It also has the characteristics of miniaturization and low power consumption.

Abstract:In order to solve the problems of low error-tolerance rate and low diagnosis precision of traditional machine learning algorithm in rolling bearing fault diagnosis of Cross-platform under different working conditions, which a rolling bearing fault diagnosis method based on the fusion of Continuous Wavelet Transform (CWT) and Transfer Learning (TL) was proposed in this paper. In this method, the time-domain signals of rolling bearing fault signals under Cross-platform and different working conditions were extracted as source domain samples and target domain samples respectively, and the vibration signals were transformed into two-dimensional signals by CWT algorithm. Then, the fault signals were mapped to the Reproducing kernel Hilbert Space through Kernel function, and the loss function of the Convolutional Neural Network (CNN) was optimized to reduce the distribution difference between the source domain and target domain samples after transfer learning using the Multi-Kernel Maximum Mean Discrepancy (MK-MMD) distance as the metric. Finally, CNN model was used for the pattern recognition of the matched source domain and target domain samples to realize fault transfer diagnosis of Cross-platform rolling bearings under different working conditions. Experimental results show that compared with other methods, the proposed method improves the accuracy and robustness of fault diagnosis of rolling bearings significantly under Cross-platform and different working conditions.