Abstract：Seal recognition is an indispensable part of the intelligent office. The current stage of seal recognition method is to directly input the scanned electronic documents into the neural network model for identification, facing the problems of unable to accurately locate the position of the seal, low accuracy of bending text recognition. Aiming at the above problems, this paper proposes an efficient stamp text recognition method, which uses the diffuse water filling algorithm to process the grayscale image for seal image feature enhancement, which ensures the accuracy of Chinese seal detection, and introduces the polar coordinate conversion operation to ensure the integrity of text features. In order to evaluate the effectiveness of the proposed method, multiple sets of comparative experiments were carried out in the existing and other network models. Experimental results show that the existing network model fused the text features extracted by the method shows excellent recognition results.

Abstract：In order to improve the screening and diagnosis efficiency of hepatic hydatid disease, and make up for the shortage of medical resources in some areas, this paper proposes an intelligent typing method of hepatic hydatid disease based on Swin Transformer, which combines the convolution attention mechanism model, and realizes the automatic classification of five types of cystic hydatid disease by learning the whole and local details of images. In order to verify the superiority of the model, the prediction model proposed in this paper is compared with common classification models. The results show that the classification accuracy based on the improved Swin Transformer model can reach 92.6% on the test set. The experimental results show that compared with other algorithms, the improved Swin Transformer network can better classify the ultrasonic images of hepatic cystic echinococcosis, and this method can be extended to other medical applications.

Abstract：Aiming at the problem of poor recognition accuracy of license plate recognition system in haze scene, an improved license plate recognition model is proposed. The model uses the improved dark channel apriori defogging algorithm for defogging. Considering the color distortion and other problems when the original defogging algorithm processes the haze image with bright areas, firstly, the atmospheric light value is limited by the threshold value. Secondly, the introduction factor is optimized. And finally, the tolerance mechanism is introduced to correct the transmittance, and the image brightness is adjusted to improve the image visualization effect. The simulation results show that the performance of PSNR, SSIM, enterprise and e improved by 1.934 dB, 0.082, 0.235 and 38.995 respectively. The recognition test of the license plate image before and after defogging shows that the recognition accuracy of the license plate is improved by 22%, which proves the superiority of the proposed model.

Abstract：Aiming at the problems of false positive detection and low recall in the detection of small targets by the general target detection algorithm, a small target detection algorithm IPH(involution prediction head) is proposed, which is applied to the detection head of YOLOv4 and YOLOv5. The experimental results on the VOC2007 data set show that the detection accuracy APs(AP for small objects) of YOLOv4 after using IPH is improved by 1.1% compared with the original algorithm, and the APs on YOLOv5 is improved by 5.9%. Through further verification of the intelligent traffic detection data set, IPH algorithm and desampling can effectively improve the accuracy of small object detection and reduce false positive detection and missed detection.

Abstract：Crude oil, as an important strategic material, plays an important role in many fields such as my country′s economy and military. This paper proposes an algorithm MSA-YOLO(MultiScale Adaptive YOLO), which is optimized on the basis of the YOLOv4 algorithm, and is experimented based on the remote sensing image dataset mainly based on Jilin-1 optical remote sensing satellite images，to make identification and classification of oil storage tanks. The algorithm optimization contents include: in order to simplify the oil storage tank monitoring model and ensure the efficiency of the model, prune the multi-scale identification module in the network structure of YOLOv4; use the k-means++ clustering algorithm to select the initial anchor frame to accelerate the convergence of the model;use CIoU-NMS-based optimization to further improve inference speed and accuracy. The experimental results show that compared with YOLOv4, the number of parameters of MSA-YOLO model is reduced by 25.84%; the model size is reduced by 62.13%; in the GPU environment of Tesla V100, the training speed of the model is increased by 6 s/epoch, and the inference speed is increased by 15.76 F/s; the average accuracy is 95.65%. At the same time, the MSA-YOLO algorithm shows more efficient characteristics in the comparative experiments of various general target recognition algorithms. The MSA-YOLO algorithm has universal feasibility for accurate and real-time identification of oil storage tanks, and can provide technical reference for remote sensing data in the field of energy futures.

Abstract：Intelligent computing chips are hardware carriers that process massive amounts of data and reflect computing capabilities, and are an important support for productivity in the digital economy era. With the large-scale application of new-generation information technologies such as artificial intelligence, 5G, big data, and blockchain, the demand for computing resources in the digital transformation of the industry and the upgrading of industrial intelligence has increased exponentially. In addition, the development of integrated circuits has entered the post-Moore era, and the bottleneck caused by the defects of the current computing architecture(Von Neumann architecture) has become more and more prominent, and the development has been seriously challenged. The integrated computing architecture, new computing paradigms, and new computing devices based on new materials and new processes that meet diverse computing needs have also entered a stage of active innovation, and intelligent computing chips have ushered in a golden age of rapid development. This paper comprehensively analyzes the strategic importance of intelligent computing chips in the digital economy era, the development paths and trends of domestic and foreign technologies, and sorts out the main industrial patterns and development opportunities of intelligent computing at home and abroad.

Abstract：Firstly, this paper introduces the research background of smart barracks and its importance in the field of national defense. Secondly, combined with the problems and challenges existing in the initial stage of the research and construction of smart barracks, this paper discusses and evaluates the system architecture of smart barracks and the research status and achievements at all levels. Finally, combined with the technical development trend and project construction experience, this paper discusses the future research direction and development trend of smart military camp.

Abstract：With the rapid development of various industrial fields, the market demand for thin specifications, high strength strip products increases rapidly. The cross-section shape of hot rolled strip is the main evaluation index of hot rolled strip product quality. Based on data mining technology, the data in the mill database are analyzed and processed. The data mining technology combines deep belief neural network(DBN) and back propagation(BP) neural network algorithms to construct a prediction model of strip thickness distribution. The DBN-BP algorithm is composed of several restricted Botlzmann machines(RBM) stacked layer by layer, and the weight matrix and bias of the network are obtained by unsupervised layer-by-layer training method for the BP network, while the BP neural network fine-tunes the whole network by means of error back propagation. This method overcomes the disadvantages of BP network falling into local optimum due to random initialization of weight parameters and long training time. Compared with the BP algorithm, the probability of predicting the midpoint thickness error is within ±5.6 μm by the DBN-BP method is 95%, while the prediction error of BP algorithm is within ±11 μm. Through the analysis of the prediction results of the cross-section shape of the strip, it can be seen that the DBN-BP deep learning method has more advantages than the BP algorithm in predicting the edge thickness of the strip.

Abstract：Entity alignment is an important technical method to realize the fusion of knowledge bases from different sources. It is widely used in the fields of knowledge graph and knowledge completion. The existing entity alignment models based on graph attention mostly use static graph attention network and ignore the semantic information in entity attributes, resulting in the problems of limited attention, difficult fitting and insufficient expression ability of the model. To solve these problems, this paper studies the entity alignment method based on the structure modeling of dynamic graph attention. Firstly, the single hop node representation of the target entity is modeled by GCN. Secondly, the multi hop node attention coefficient is obtained and entity modeled by using the dynamic graph attention network, and then the single hop and multi hop node information output by GCN and dynamic graph attention layer is aggregated by layer-wise gating network. Finally, the entity attribute semantic extracted by external knowledge pre training natural language model is embedded and concatenated to calculate similarity. This method has been improved in three types of cross language datasets of DBP15K, which proves the effectiveness of applying dynamic graph attention network and integrating entity attribute semantics in improving entity representation ability.

Abstract：In the complex scene of shooting range test, the test site often involves the changeable natural environment including dust, strong light, occlusion, etc. A single target tracking algorithm associated with dynamic features is proposed to track fast moving targets in this case. Firstly, the gated recurrent unit is used to extract the time series dynamic characteristics of the target which need to be tracked, so as to obtain a set of candidate processing target frames. Then，convolutional network is adopted to extract the depth convolution features of the candidate target frame and determine target position, as well as separating the background convolution features. In the tracking process, the separated background convolution feature map is applied to update network parameters to enhance the robustness and adaptability of network. Experimental results show that the proposed algorithm can adaptively track moving target in the shooting range image acquisition system, which can still maintain excellent robustness and adaptability in the context of complex environment.

Abstract：In the process of multi-lane data transmission in PCIe, when the arrival time of data in each lane is inconsistent, the issue of skew will be introduced. In order to ensure that the receiver of each lane can process the received data simultaneously and correctly, it is necessary to preprocess the transmitted data. This paper presents a De-skew logic circuit, which explains how to use synchronous FIFO to realize multi-lane De-skew and complete the corresponding logic design. UVM and VIP technology are used to build a verification platform, the test results verify the correctness and feasibility of the design.Compared with other common solutions, the logic design has comprehensiveness, advantages and reusability.

Abstract：Based on the domestic 40 V bipolar process, a fully differential operational amplifier is designed, which is applied to the driver module of the signal conditioning chip. The amplitude of the output signal of the operational amplifier can be adjusted by an external resistor. The overall circuit structure includes input stage, intermediate stage, output stage, bias circuit, feedback circuit and reference circuit. The input stage adopts a common base amplifier circuit, which has a wide operating frequency band; the intermediate stage adopts a common collector-common emitter circuit structure to improve the voltage gain; the power output stage constitutes a loop integrator and a resistor network that provides common mode feedback, so that the output common mode voltage is concentrated between the positive and negative power supplies, and the output signal amplitude can be adjusted by an external amplitude control resistor. When the power supply voltage is ±15 V, the test results are：the output voltage amplitude range is 1.488 V~18.57 V, the DC offset voltage is -169 mV, the output short-circuit current is 65 mA，and the total harmonic distortion is 41.2 dB.

Abstract：With the increasingly close communication between power enterprises and users, the scale of user power information collection is increasing, the types of data collection are increasing, and the requirements for data collection frequency are higher. At present, the research on user power information acquisition system is mostly to optimize the system architecture, and there is little research on the acquisition method. This paper proposes a fast meter reading method based on edge computing, which sinks the computing power and data storage resources to the location close to the data source, and provides low delay and efficient acquisition for the terminal equipment nearby. Compared with the time cost and reliability of traditional meter reading methods, the experimental results show that marginalizing the aggregated data improves the acquisition rate and success rate of intensive user power information.

Abstract：This paper presented a fast custom made method based on multi-dimensional matrix type software library for the multi-model software construction of space data process center. Firstly, the architecture is introduced for the space data process center system, that is multi-layer and multi-state complex structure satisfied with cloud computation. Secondly, the deployment mode of micro module room is introduced adapt to the computer hardware of the space data process center. And the structure of the system is presented to divide the operation with the operation place, to divide the software management with hardware management. At last, a matrix type dynamic custom software library is presented, which could construct the new application flexibly and customizably, load and unload the satellite management system mirror file quickly. And the process flow of the operation is also introduced for quick generation and management. This method satisfied with the multi-series and the multi-satellite mission requirement of space data process center. The system development cost is reduced effectively and efficiently. And the system operation and maintenance are reduced obviously.

Abstract：Massive multiple-input multiple-output(MIMO) systems can effectively improve the spectrum efficiency. When the antenna scale gradually tends to infinity, the minimum mean square error(MMSE) detection algorithm can achieve near-optimal detection performance. However, due to the matrix inversion required in the algorithm, which brings extremely high computational complexity, it is difficult to implement in a massive MIMO system. The Richardson algorithm can achieve the detection performance of the MMSE algorithm in an iterative form without matrix inversion, but the algorithm is greatly affected by its relaxation parameters. In the Richardson algorithm combined with the steepest gradient descent algorithm (SDNSR), the error of the relaxation parameter can be compensated by the gradient descent algorithm, but the computational complexity is increased. This paper firstly uses the idea of deep expansion to map the iterative process of SDNSR to a deep detection network (SDNSR-Net); then, by modifying the network structure and adding trainable parameters，the computational complexity is reduced and the detection accuracy is improved. The experimental results show that SDNSR-Net is superior to other typical detection algorithms in the case of different signal-to-noise ratios and antenna configurations in the uplink massive MIMO system and can be used as an effective detection scheme in practice.

Abstract：In order to solve the problem of low recognition rate of phase shift keying and quadrature amplitude modulation signals when using time-frequency image classification under the condition of low signal-to-noise ratio, this paper proposes a method of signal feature fusion. Firstly, the method calculates the high-order cumulant of the received signal and obtains the one-dimensional numerical eigenvector. Then, the time-frequency diagram of the received signal is obtained by time-frequency analysis, and the one-dimensional image feature vector is extracted by convolution neural network. The two kinds of feature vectors are connected to obtain one-dimensional fusion feature vector. Finally, the fused feature vector is input into the full connection layer and the classification results are output. The simulation results show that under the condition of about 1 dB, the recognition rate of phase shift keying and quadrature amplitude modulation signals can be improved by about 10%～30% compared with the method of single image feature.

Abstract：To address the problems of message scheduling and link planning in TTE networks, which rely on manual configuration and are difficult to afford large network structures, a simulation study of the communication link planning method for TTE networks is conducted, and a link planning model is constructed to find the solution to the problem by a new heuristic algorithm, the brainstorming algorithm. Through experimental simulation comparison, the effectiveness of the brainstorming optimization algorithm and its more stable execution than genetic algorithm in the late planning stage are verified, which improves the network transmission efficiency and has a very wide application prospect.

Abstract：In order to solve the problem that the gray level difference between hidden objects and human body in millimeter wave image is small and the shape is changeable, a hidden object detection algorithm based on visual saliency is proposed. After bilateral filtering, combined with OTSU and morphological operation，the algorithm completes the pretreatment to obtain the human body region. And then the foreground is located based on the significance calculation in the frequency domain. After background suppression，the significance map is generated，so the detection is completed. Experimental results show that compared with the typical active millimeter wave imaging detection algorithm, the detection rate of the proposed algorithm increases by 5.87% and 9.08%, respectively, and the proposed algorithm has better detection performance.

Abstract：In order to accurately design control system of the uninterruptible power supply(UPS) inverter, impact of the delay induced by digital control on the inverter performances is analyzed, and the design method of the inverter system controller considering the delay link is proposed. The mathematical model of the inverter system is established based on the state space averaging method, and the influence of the control delay link on the system stability is analyzed. Aiming at the single-phase UPS inverter power supply, the digital PID double-loop control technology is applied, and a single-phase UPS inverter control algorithm with easy parameter setting is obtained. The generalized Z-transformation theory is introduced to describe the delay link of digital control, and the model of the control system is improved. The modified design method of the digital PID controller parameters considering the delay link is given. Finally, experimental verification is carried out by MATLAB/Simulink simulation and experiment. The experimental results show that the output voltage regulation accuracy of the UPS inverter control system based on the digital PID is less than 1%, and the THD(total harmonic distortion) is less than 2%. When the load end changes suddenly, the time for the system to return to normal is about 1.2 ms. The steady-state performance and dynamic response performance of the inverter system output are effectively improved.

Abstract：As an important part of aerospace mission engineering, the tracking, telemetry and command(TT&C) center system currently adopts technologies such as duplexing, server clusters, disk redundancy, data backup and recovery to ensure the uninterrupted operation, stability and reliability of the aerospace mission system. But there are also such problems such as software and hardware tight-coupling, low resource sharing efficiency, and poor system expansion capabilities, all of which limit the development of the entire aerospace TT&C system. Regarding these problems, in this paper a TT&C center system based on integrated cloud architecture is studied and designed. The system is implemented on-demand customization of services, resource sharing and interoperability, and flexible expansion. Moreover the system is verified by actual aerospace missions.

Abstract：Wavelet transform is widely used in the research of logging signal denoising, and the selection of wavelet parameters directly affects the final denoising effect, so it is necessary to design an algorithm to obtain the best wavelet transform parameters of logging signal. In this paper, the random inertia weight strategy is innovatively proposed to change the weight parameters of particle swarm optimization algorithm, which improves the convergence speed of particle swarm optimization algorithm, enhances the ability of searching for optimization, and obtains the optimal wavelet transform parameters. The optimal wavelet transform parameters are applied to the wavelet denoising of soft threshold method, which can effectively separate the useful signal and useless noise. This algorithm can effectively improve the signal-to-noise ratio of logging signal, reduce the root mean square difference, and realize the effective removal of noise in logging signal.

Abstract：With the increasing frequency of space launch missions, constant application of new spacecraft and tracking and control equipment, and the urgent demand of multi-level integrated command, it puts forward higher universality and adaptability requirements for space launch command and monitoring system. In order to solve these problems, this paper proposes a novel flexibility data processing method of command and monitoring system. By defining protocol model and topic model in off-line mode, this method realizes the map of the internal and external communication protocol. On the other hand, the component methodology is adopted to accomplish the receiving, parsing, storing, and transmitting operation of external data in online mode. The experimental results show that this method successfully decouples the processing logic and data protocol by using the definition of various models, achieves the flexible information exchange for command and monitoring data without revising the software architecture, and sharply improves the preparation efficiency of the command and monitoring system for space launch missions.

Abstract：An intelligent storage and measurement system of multiple continuous shock waves based on SRAM and Flash is proposed for the complex environment of continuous experiment of combat. The system uses FPGA as the core controller and the multi-channel intelligent storage of continuous shock wave is realized by using acquisition and storage technology combined with signal trigger identification and marking. The signal conditioning of 4 piezoresistive sensors realizes the automatic adjustment of the baseline, which ensures the flexibility and applicability of the system. Partition storage management of SRAM and pipeline storage of Flash realize the fast storage of data under multi-channel and multi-trigger times.The experimental result indicates that the proposed distributed measurement system can realize the intelligent storage of signals under 4 channels, 1 MS/s sampling rate and 32 consecutive triggers, and can be applied to the continuous explosion shock wave measurement in experiment of combat.