Abstract:Aiming at the security problem of pseudo-random number in cryptographic applications, based on the characteristics of superlattice physical entropy source, this paper analyzes the advantages of stimulated chaos as entropy source, and proposes for the first time that the stimulated chaos oscillation of superlattice is used as entropy source of true random number generator, 12 bit ADC is used to realize digital sampling of stimulated chaos signal of superlattice, and FPGA is used as post-processing unit, A true random, high-performance and high-speed superlattice random number generator is designed and implemented. The real-time generation rate of random number is up to 300mbps. The performance of the superlattice random number generator can pass the standard (NIST sp 800-22) provided by the National Bureau of standards. The results show that the superlattice random number in this paper has good randomness.

Abstract:The battery in the new energy system works in a wilderness environment, and its remote data monitoring is often realized based on wireless communication methods such as GPRS. This transmission method requires the establishment of a dedicated base station, which is costly. Beidou short message communication realizes data transmission through satellite, without the need to set up a base station. Based on the above situation, a battery monitoring system based on Beidou short message communication is designed. The system is composed of a monitoring terminal, a Beidou receiver, and a remote monitoring cloud platform. Aiming at the software and hardware requirements of data collection, data monitoring and Beidou short message communication, corresponding design ideas and implementation schemes are given. At the same time, considering the influence of the wilderness environment on satellite communication, a transmission error control method based on data backup is proposed to improve its transmission reliability. Finally, it is verified by actual tests that the system can effectively transmit monitoring data and locate the battery. Maintenance personnel can monitor the data in real time through the cloud platform, reducing the construction and maintenance costs of remote monitoring.

Abstract:Unmanned aerial vehicles (UAVs) are used in aerial photogrammetry and digital low-altitude remote sensing, which has the characteristics of low operating cost, high efficiency and flexibility, and a wide range of applications. The fuselage and main wing with better drag reduction, weight loss and strength increase were selected, the motor and propeller with the optimal ratio in the ascending, takeoff and cruising segments under the same power condition were calculated, and the metal steering gear with higher power and safety was preferred as the servo. The electronic governor, landing gear, flight control, image transmitter, ground receiver and other components were carefully selected. So the optimized modular low-altitude long-endurance UAV was integrated. In order to reduce the weight of the flight platform, increase the load, extend the flight time and distance, a large number of repeated experiments were carried out to select the type, and various basic functions were tested. Compared with the traditional flight platform, the UAV flight platform has the characteristics of long flight time and long flight distance. It can also be equipped with more equipment such as miniature SLR, downhanging release cabin, weather detector, etc. The taken photos can be easily uploaded with the satellite link, and it can be modeled with one click when connected with the ground station. The UAV can take off by hand and land at various venues, support additional accessories for ejection, taxiing, vertical take-off and landing, and can slide down and parachute. It has broad application prospects in emergency earthquake relief, land and resources survey and monitoring, digital city and major engineering construction, etc.

Abstract:The branch-cut phase unwrapping algorithm is widely used because of its better noise suppression effect. However, the branch-cuts constructed by the branch-cut algorithm may not be the shortest globally, and it is easy to form a closed branch-cuts , resulting in phase unfolding errors. Therefore, an improved branch-cut algorithm based on tabu search is proposed. The nearest neighbor algorithm is used to match the positive and negative residues to obtain the initial solution, and then the tabu search algorithm is used to optimize the initial solution to construct the branch-cuts globally. The algorithm was tested by simulation, and the algorithm reduced the length of the branch-cuts constructed by the branch-cut algorithm by 42% and increased the speed by 28%. And the dynamic three-dimensional shape reconstruction experiment is carried out on the changed facial expressions, and the results show that the improved algorithm has better phase unwrapping accuracy.

Abstract:In order to solve the problems of low precision, poor reduction degree of internal surface morphology and inconvenient measurement of gun barrel inner diameter, an automatic measurement system of gun barrel based on laser triangulation is designed. The system mainly uses laser sensor, single chip microcomputer and stepping motor to design the mechanical measurement structure of propulsion type, and uses the principle of geometric centering to measure the whole body of gun automatically; The minimum measuring inner diameter is 60mm; The sampling data is processed by filtering and cubic spline interpolation, which greatly reduces the measurement error. The measurement system is used to measure the body pipe of different caliber. The measurement results show that the measurement system can work stably and can measure the body pipe with different inner diameter. The measurement resolution is 0.001mm, the measurement accuracy is 0.02mm, the measuring point is accurate, the functions of each part of the upper computer are normal, and the wear defects of the inner wall of the body pipe can be restored well. The measurement system has the advantages of small measurement error, high efficiency, simple operation, high automation and can also be used in the same kind of pipe measurement.

Abstract:The traditional W-band power combiner is difficult to integrate with other planar circuits because of its large size, which limits its application in small platform. To carry out research on high efficiency and small size power combiner. In this paper, we present a novel two-way waveguide power combining structure which uses multiplayer silicon bulk micromachining techniques. A good performance is achieved by utilizing MEMS and three-dimensional wafer bonding stacking technology. Within the operating band from 90GHz to 96GHz, the return loss is better than 22.5dB, the insertion loss is less than 0.22dB and the simulated result demonstrates an overall combining efficiency of 95%.

Abstract:This paper proposes a novel primary double winding coupling topology of the DC transformer (DCT) for DC micro-grid applications. The primary side high-frequency full-bridge inverter is replaced by a novel topology without shoot-through problem. As a result, the reliability of the system is improved, and dead-time is unnecessary to be added to the driving signals thus the dc voltage utilization is higher. Compared to the no dead-time dual-buck inverter, the proposed topology has more advantages in the size and cost, because the primary two windings can be integrated on the same core. Aiming at DC micro-grid system, this paper studies and designs a non-dead zone topology. We carry out equivalent transformation of transformer, the working principle is analyzed, and the calculation and analysis of compensation and efficiency are given. Finally, the simulation and experimental results show that the proposed topology is advanced.

Abstract:In order to effectively reduce the damage of obstructive sleep apnea syndrome (OSAS), a two-level non-invasive home sleep suction machine was designed.Based on STM32, the adaptive inverse control theory is used to control the motor. The difference between the input signal and the output response is taken as the error signal, and the adaptive algorithm is driven by the signal.According to the collected breathing wave shape,Kalman filter is used to predict the timing of inspiratory and exhalatory in the next cycle,and then control the motor to reach the target pressure value.In order to facilitate user experience, real-time display of upper computer waveform is added, so that users can clearly see the adjustment of breathing wave while experiencing ventilator assisted breathing. This design not only controls the time of the motor to reach the set response speed at about 100ms, but also realizes three breathing modes: CPAP mode, S mode and T mode. In CPAP mode, the ventilator continuously outputs positive airway pressure;In the S mode, the next time node of exhalation and inspiration can be estimated by the patient's breath wave shape, and the correct air pressure can be given to assist the patient's breathing.In T mode, the air pressure is controlled by the timer to assist the patient's breathing.This not only improves patient compliance and makes patients feel more comfortable, but also facilitates long-term treatment.

Abstract:For Android malware detection, Most of research proposed multi-type features combined with machine learning to improve the detection rate of malware detection, but rarely considered association between application interface and edge information in call graph. This paper designs a method of Android malware detection based on accessibility feature of application interface. This method extracts accessibility features of application interface based on malicious behaviors, effectively makes feature set contain edge information. Experiments were conducted on 1151 malware collected by VirusShare in 2018 and 1021 benign programs from Google player. Experiments show that random forest is more effective than other four methods in malware detection, and accuracy of model reaches 98.90%. Results show that accessibility features improved recall rate and precision of the model , and is more important than other features in the experiment.

Abstract:This paper analysis the problems of heavy workload, long verification time and requirement of human involvement that exist in verifying the meteorological temperature sensors. By investigating the intelligent measurement technology, this paper proposed an intelligent transportation system. By using a 6-axis robot arm, our proposed algorithm successfully achieved temperature measurement points transition of temperature sensors. Our proposed system realized simultaneously verification of multi-batches sensors with no human involvement. In addition, we also developed a remote console deployed on mobile application. This remote console could achieve functions such as remote operation, remote monitoring and auto-alarm. Using this remote console, the staff could turn on or turn off the device, monitor the working status remotely and take corresponding movement according to the remote alarming system. The experiment result shown that the proposed intelligent verification system efficiently reduced the verification time by making the verification process four times faster. Furthermore, our proposed system released the verification process from the eight working hours restriction and achieves automation in all the verification process.

Abstract:With the continuous development of national defense technology, spread spectrum technology is more and more widely used. In order to meet the communication requirements of gun barrel in flight, a transmitter based on spread spectrum technology is designed. The system consists of RF front-end chip AD9361, digital baseband processing chip FPGA and peripheral circuits; The peripheral circuit includes power module, clock module, balun feeder interface, serial port and debug interface. The software of communication system adopts BPSK modulation technology and spread spectrum technology, and the spread spectrum modulation part adopts gold balanced code and three-level RAM buffer scheme. Finally, based on a ballistic measurement project, the telemetry test is carried out, using the ground receiving station to receive the data sent by the transmitter, recording the modulation waveform and calculating the bit error rate of the received data, which verifies the reliability of the designed missile borne transmitter platform and the correctness of the spread spectrum modulation method.

Abstract:As an important electromagnetic surveying method, Controlled Source Audio Frequency Magnetotelluric (CSAMT) method is operated within a very broad bandwidth, so observational data is vulnerable to varies noise. To solve this problem, real-time data processing method for CSAMT was studied is this paper. The frequency table design method, all-phase analysis, and the judgment and elimination of abnormal value were discussed. Cagniard resistivity and phase measurement under strong disturbances was achieved by integrated utilization of multiple error inspection method as well as anomalous extremum value identification method. Field tests confirmed that the data processing method described in this paper can extract effective Cagnia apparent resistivity and impedance phase from the original data with very low SNR, helping to improve the anti-interference capability and expand the application scope of CSAMT.

Abstract:In this paper, through the research on the longitudinal and transverse depth control of the small amphibious underwater vehicle, through the mechanical structure design of ROV system and the structure distribution of thruster, through the generalized multi degree of freedom equation, under multiple constraints, the bi-directional depth control model is established. Through the UFD fluid analysis in fluent, the hydrodynamic coefficients in different attitude motion control states are obtained Through the research of PID control and fuzzy PID control theory, the PID controller and fuzzy PID controller of depth control is built in Simulink Library of Mtalab. The control effect of PID controller and fuzzy PID controller is compared. The simulation results show that the fuzzy PID control is more effective than PID control Fuzzy PID has good stability and anti-interference ability for depth control and angle control in vertical and horizontal control .

Abstract:AbstractTo summarize, a unified interface service framework of multi-dimensional biometric identification fusion application systems is designed in order to unify the docking mode of various biometric identification subsystems and standardize the multi-dimensional biometric identification data and use environment. The framework follows the standard system of video and image information application system in the public security industry. Based on the REST service architecture, it realizes the standardized definition of the interactive interface between the fusion application system and the biometric identification subsystem by HTTP method, including registration and maintenance, target object comparison, image retrieval target, etc. The framework has changed the traditional biometric identification subsystem independent deployment, integration complexity of higher status. Combined with the REST service architecture of portable, simple, high safety and reliability and so on, the framework implements the standardization of multidimensional biometric identification subsystem integration from the application level, implements the standardization of multidimensional biometric data management, and can effectively improve extensibility of integration application system.

Abstract:Barreled mineral water needs to be tested before leaving the factory to see if there are foreign bodies in it to reduce safety risks. The foreign body detection of mineral water based on computer vision technology is a kind of common method.However, barreled mineral water data is difficult to obtain, and directly deploying models trained on bottled mineral water data to barreled mineral water for detection will result in a sharp performance decline due to domain shift.In order to solve the above problems, a domain adaptive foreign body detection method based on adversarial learning was proposed.To be specific, an automatic device is designed to produce foreign body detection data set of barreled mineral water.Then, a foreign body detection model was trained on a bottled mineral water sample, taking into account its easy availability.Secondly, in order to improve the generalization ability of the model, a domain classifier is designed by introducing the idea of adversarial learning, and the bottled and bottled mineral water are confused and classified by adversarial training to learn domain invariant features.Finally, the effectiveness and superiority of the proposed method are proved by experiments.

Abstract:Indoor positioning based on ZigBee received signal strength index has attracted more and more researchers' attention and use because of its low cost, low hardware power consumption and easy implementation. Due to the influence of multipath effect and shadow effect, the traditional indoor positioning algorithm can not obtain good positioning effect. In order to improve the positioning accuracy of traditional wireless sensor network indoor positioning algorithm, this paper proposes an indoor positioning algorithm based on annealing algorithm (SA) and genetic algorithm (GA) optimized neural network (saga-bp). The initial weight and initial threshold of neural network algorithm are optimized by using annealing algorithm mechanism combined with genetic algorithm. The simulation results show that: the average positioning error of saga-bp algorithm is 0.40 m and the maximum positioning error is 0.83 m by adding a certain amount of random noise in the simulation; compared with the neural network (BP) positioning algorithm and genetic algorithm, the positioning accuracy of improved neural network (GA-BP) positioning algorithm is improved by 56% and 8.6%, which effectively improves the indoor positioning accuracy.

Abstract:Aiming at the problem of object frame positioning error and redundant detection in the top-down human pose estimation algorithm, a top-down human pose estimation algorithm based on deep learning is proposed. The symmetric space transformation network is designed to connect with the single-person pose estimation network to propose high-quality human target frames from inaccurate human body bounding boxes, and parametric pose non-maximum suppression is introduced to eliminate redundant pose estimation , The elimination rule is applied to eliminate similar postures, and the unique human posture estimation result is obtained. Part of the data set is selected for training and testing on the public human pose estimation data set MPII. The experimental results show that the method proposed in this paper can accurately detect the key points of the human body, effectively improve the accuracy of human body pose estimation, and can adapt to crowded people. , Complex scenes with occlusion.

Abstract:Aiming at the problems of insufficient network fusion and low detection efficiency in current target detection using RGB-D images, a feature-level fusion network structure based on attention mechanism is proposed. First, under the backbone network structure based on Yolo v3, the RGB and Depth networks are trained separately with the labeled RGB-D samples, and then the two features are enhanced by the attention module, and finally the final feature weights are obtained by layer-by-layer fusion in the middle of the network. Tested on the challenging NYU Depth v2 data set, the average accuracy of the method in this paper is 77.8%. Through comparative experiments, it is concluded that the fusion network based on the attention mechanism proposed in this paper has significantly improved performance compared with similar algorithms.

Abstract:In order to make up for the shortcomings of the U-Net model and the level set method in segmenting liver CT images, this paper proposes a segmentation algorithm which combines U-Net model with the level set method. The algorithm uses the U-Net model as a prior network to segment the liver in the training phase, and then sends its output segmentation result as a prior feature map to the level set method for further segmentation; by calculating the difference between the two segmentation results, back propagation error, update network parameters, and finally get a complete segmentation algorithm model. Through experimental comparisons on the public data set 3Dircadb, this paper proposes that the sensitivity of the segmentation algorithm and the average Dice coefficient can reach 94.85% and 95.18%, respectively, and the average area overlap error and relative area error are only 9.97% and 7.69%. Compared with other common segmentation algorithms, this algorithm obtains better segmentation results.

Abstract:For the traditional wavelet enhancement algorithm applied to medical CT image processing effect is not ideal.There are many problems, such as the loss of image details while enhancing the image, weakening the edge information of the target in the image and reducing the contrast of the image.In order to maintain good edge detail information, highlight sensitive information of CT image, and better reconstruct enhanced image, a medical CT image enhancement algorithm based on multi-view wavelet transform fusion was proposed.The algorithm using the multiplicative decomposition to get the detail of the original image and clustering of codominant method are used to get the original image is of significant figure and fusion by wavelet transform, and then to weigh the rule of image fusion, respectively for the high frequency weighted average and low frequency signals of the image reconstructed image to handle the big variance.Experimental results show that this method is superior to the existing image enhancement algorithms in terms of image sharpness and highlighting sensitive information, and has a good enhancement effect on CT images.

Abstract:Edge computing has a promising application prospect in the field of Internet of Things, especially cable real-time online monitoring business in a smart grid. However, for the relatively limited resources and capabilities of the edge nodes, such as computing resource and storage resource, it is hard to highly and comprehensively satisfy the high real-time requirements of cable online monitoring tasks. To solve this problem, effectively dynamic task allocation is needed on the basis of efficient utilization and optimization of resource and capabilities of the edge nodes. In this article, a task allocation mechanism for cable real-time online monitoring business based on edge computing is proposed. First, considering the linear distribution characteristics of the cable, the statuses of edge nodes, the processing overhead of tasks, and the scheduling strategy of delay-sensitive tasks, we establish a task allocation model based on edge computing. Second, a task allocation strategy based on improved discrete particles warm optimization is proposed. In our strategy, we focus on the task queuing problem in edge nodes and the optimized task allocation problem among edge nodes. Simulation results show that the task allocation mechanism proposed in this article can effectively reduce the average delay of cable real-time online monitoring businesses, and further improve the security and reliability of the smart grid.

Abstract:With the development of drive systems and sensor technology, the application field of photoelectric tracking system is expanding. The target tracked by the system has also expanded from the traditional target to the new type of target. Traditional targets have the characteristics of high altitude, long distance, fast speed and strong movement rules. The characteristics of the new type of target are low altitude, short distance, slow speed and weak movement law. Faced with the new type of target, the photoelectric tracking system needs to consider further upgrading the tracking technology to improve the tracking ability of the system. Feed-forward control is an effective means to improve the tracking ability of the system. The key of feed-forward control is to obtain real-time and accurate target motion state, but the image sensor that detects the target generally has a time delay that cannot be ignored. Therefore, the research direction of photoelectric tracking technology based on predictive filtering has been formed. This paper reviews the current mainstream four photoelectric tracking technologies based on predictive filtering, and compares the performance of these four methods from the frequency domain and time domain through simulation experiments. Then, the advantages and disadvantages of these four methods are summarized. Finally, the future research direction of this technology is pointed out.

Abstract:Aiming at the fact that the current flow indicators in the wireless sensor network (WSN) cascade model cannot correctly reflect the convergence characteristics of WSN, a wireless sensor network cascade model limited by node capacity and link capacity is proposed. First, the network load is redefined according to the new traffic index direction betweenness. Second, the network cascading survivability model is constructed so that the faulty node can recover from the fault state after a certain time delay. Finally, a route recovery mechanism is proposed, which can effectively improve the survivability of network cascading failures, and a simulation comparison test is carried out.The results show that the increase of the node tolerance coefficient λn cannot reduce the damage caused by the cascading failure, but it can delay the occurrence of the cascading process.The larger the link tolerance coefficient λl, the proportion of overloaded nodes will decrease significantly and the proposed method can help the faulty network to recover in a short time and make the network status more stable.

Abstract:Towards the safety operation of composite insulators, research was conducted for the nondestructive detection of debonding defects in composite insulator using the method of the adjacent capacitance. The feasibility of adjacent capacitive sensing technique for the detection of debonding defects in multilayer dielectric structures and the effect of electrode width and spacing on the detection capability of adjacent capacitive sensors were investigated through numerical simulations and experiments. It is demonstrated that the adjacent capacitive sensing technique can be used for the detection of debonding defects at the interface of multilayer dielectric structures, and the electrode width and electrode spacing have an effect on its detection performance. The detection efficiency and resolution of the adjacent capacitance detection technique can be improved by optimizing the electrode parameters. The research provides a feasible technical solution for debonding detection in composite insulator.

Abstract:Aiming at the problems of less test channels, single type of measured storage media and low degree of automation in the existing memory test system, a universal automatic memory test system was designed. The system consists of upper computer, hardware test circuit and Gigabit Ethernet. The upper computer is developed under the environment of VS2015, and sends test instructions to the system and displays test results. The hardware test circuit completes the reading and return of data from the storage medium. The gigabit Ethernet realizes the high-speed exchange of data between the two. This system can test the performance of the memory with SATA interface and ONFI interface, so as to improve the testing efficiency of the memory and reduce the testing cost. The results of many experiments show that the system has good environmental adaptability and expansibility, and has high engineering application value in aerospace field.

Abstract:To improve the stability of PID controller to the system and reduce the control error, an adaptive neural network PID controller is proposed. Firstly, PID controller is developed in discrete-time model to reduce the problems of controller design in continuous time. Then, an adaptive neural network is defined to adjust the control gain to minimize the tracking error of the UAV during the navigation mission. The important parameters of PID controller are adjusted by gradient descent method. In addition, Kalman filter is used to filter the measured values of sensors to improve the performance of on-line adaptive. The experimental results verify the effectiveness of the proposed controller. The integral of absolute error (IAE) is 2.576×103, and The integral multiplied time of absolute error (ITAE) is 5.152×105. Both indexes are one order of magnitude lower than the classical PID controller.

Abstract:In order to solve the problems of heavy equipment, low automation, and inconvenience in practical application of the regional automatic weather station inspection equipment in Hainan Province, A design plan for meteorological element sensor on-site verification instrument based on the Internet of Things is distributed. By using digital sensor detection technology and wireless networking technology to construct a wireless sensor network(WSN), It can measure real-time meteorological data such as temperature, humidity, air pressure, wind and other data. Though comparing with the site's real-time operating data, It can auto to verificate the results in the field devices and judge whether the running of the equipment is normal. Field test verification shows that the on-site verifier equipment is highly accurate，highly stable，intelligent and portable, It can quickly and accurately provide on-site verification data of regional automatic weather stations, and greatly improve the efficiency of on-site verification,and it’s very suitable for on-site verification of regional weather stations．