Abstract:In order to realize analog-to-digital conversion with low distortion and high dynamic range, a new high-precision Sigma Delta modulator system is proposed. The Sigma-Delta modulator firstly adopts a new type of second-order single-loop one-bit quantization structure, where adds two feedforward paths and adjusts the logical relationship between the core integrator and the adder module. At the same time, in order to realize the second-order shaping of quantization noise and the delay-free transmission of input signals, the transmission function of the integrator is further adapted and improved. Matlab based system-level simulation results show that, the signal-to-noise ratio (SNR) of the proposed modulator is 106.6dB, the effective number of bit (ENOB) is 17.41bit, the second harmonic distortion is -82.7dB and the dynamic range is 104.76dB under the conditions of sampling signal frequency of 1024KHz and input signal bandwidth of 1KHz, which provides a new direction for the research and development of Sigma-Delta modulator with high-order MASH structure.

Abstract:Prediction of dissolved gas concentration in oil is very important for transformer early fault detection. a prediction model based on singular spectrum analysis (SSA) combined with long and short-term memory network (LSTM) is proposed to improve the prediction accuracy. First, To solve the problem of data leakage in traditional sequence decomposition, a sampling strategy based on SSA stepwise decomposition is proposed. Then based on this strategy, the concentration sequence of dissolved gas in original oil with complex characteristics is decomposed into relatively single trend component and fluctuation component. Finally using LSTM network for each component for single step and multi-step prediction respectively. The predicted values of each component are accumulated to obtain the prediction result of original gas concentration. The example shows that compared with the single LSTM model, the overall prediction accuracy of the proposed method is higher in the experimental days.

Abstract:In order to reduce the resistance caused by the adhesion of Marine organisms to the ship sailing and improve the transport efficiency of the ship, a underwater ship hull cleaning ROV based on six thrusters is designed. Firstly, the corresponding software and hardware scheme of the robot was designed, and the driving and measuring components of the robot were selected according to the actual application requirements. Then, the structure of the ROV was designed using Solidworks 3D mechanical drawing software, and the structure assembly and system test of the ROV prototype were carried out. At the same time, in order to improve the performance index of ROV attitude control process of ship cleaning, an improved variable integral PID control algorithm is proposed. According to the change of the amount of deviation, the whole attitude control process is planned into different stages, and the switching coefficient of the integral term is adjusted to meet the needs of different stages. Finally, the improved ROV attitude control system was simulated and tested by MATLAB simulation software. The results show that the algorithm has shorter transition time and better stability than the traditional single closed loop and double closed loop PID control algorithm, and it is a stable and efficient ROV attitude control mode.

Abstract:Ultrasonic defect detection is a mainstream means of defect recognition, and two-dimensional convolution neural network has always been the main technology in this field. Generally, attribute features are extracted from two-dimensional C-scan, D-scan and other images for recognition and classification. These studies mainly use two-dimensional convolution layer, which will consume a lot of resources. Among all kinds of defect identification methods, ultrasonic echo signal analysis is one of the most important and useful tools. In this study, features are extracted from the original time domain ultrasonic signals, firstly, the data are collected using the JPR-600C air-coupled ultrasonic nondestructive testing system from the laboratory, and then the one-dimensional t-sne network model is constructed and optimized by using different hyperparameters, t-sne visualization and other means. Finally, ultrasonic signal defect recognition and classification is realized. The experimental results show that the performance of the CNN model proposed in this paper is better, and the correct rate of defect identification is 97.57%, which is higher than other machine learning methods, which provides some assistance for the automation of defect recognition.

Abstract:The occurrence of waterlogging disasters can seriously affect the safety of people's lives and property around rivers, and it is important to keep track of the water level changes in rivers in a timely and accurate manner. In response to the shortcomings of the current traditional water level measurement and monitoring methods in terms of timeliness and safety, a water level remote monitoring system based on a pulsed laser and a cloud platform is designed. The measurement device is based on an STM32 microcontroller, which emits a pulsed laser to a floating board in a measurement barrel, detects the distance from the device to the floating board on the water surface, and finally calculates the depth of the water level of the river to be monitored. Laboratory simulation results show that the relative error is less than 1.7% and the coefficient of variation is less than 1.2% when the water level depth is between 200mm and 800mm when the measuring device is fixed at a height of 1m. The wireless transmission module enables real-time uploading of water level data, and the cloud platform can accurately display the water level and measurement time, indicating that the remote water level monitoring system is feasible and can achieve real-time remote monitoring of river water level, and has certain application value.

Abstract:For variable sampling period networked control systems with time-varying delay and random packet-dropouts, a collaborative design method of robustcontrol and sampling cycle scheduling is proposed. Using the maximum membership degree defuzzification method in fuzzy control, the sampling period is switched between several fixed values according to the size of system error, and the state feedback matrix K satisfying theperformance index under each sampling period is calculated respectively. The controller selects the corresponding feedback matrix according to the current sampling period to calculate the control quantity，which can avoid the strong conservatism caused by considering both time-varying sampling period and time-varying delay. When robust controller is designed, Markov chain is used to describe the random packet-dropouts of the system, which makes the modeling closer to the actual situation. Using the nominal point method to determine the norm bound of the system uncertainty composed of time delay can further reduce the conservatism. Finally, numerical simulation results verify the feasibility and effectiveness of the cooperative design method.

Abstract:In order to overcome the problem that traditional control equipment such as remote controller and ground station can not control UAV flexibly and conveniently in complex environment, a UAV end-side control system based on data glove is proposed, which can control UAV through gesture. Firstly, a wireless data glove based on STM32, which integrates flexible sensors and inertial sensors, is used to collect training and test data. According to the data obtained from the data glove, BP neural network deployed to STM32 embedded processor is used for end-side gesture recognition. At last, the gesture is converted into matched UAV control command and sent to the UAV to realize the control. A total of 400 recognition verifications were carried out for 8 kinds of gestures, and the gesture recognition rate was 97%. The UAV simulation experiment is carried out through the Airsim simulation platform. The recognition accuracy of the basic control commands of UAV corresponding to eight gestures is 100%, which shows that the gesture recognition effect of the system is ideal. Finally, the test flight is carried out in the real scene, and multiple participants can successfully complete the complex flight with a total route length of 35m specified in advance within 1 minute. The experiment shows that the UAV can respond quickly to the gesture, and the gesture control method provided by the system is simple and convenient, which can realize the real-time and stable control of the UAV at the end side.

Abstract:For the safe and stable operation of electric vehicle DC charging piles, this paper proposes a charging pile fault prediction algorithm based on improved support vector machine. The algorithm first performs preprocessing such as missing value filling and normalization in the operating parameters of the charging pile; then the preprocessed data is input into the support vector machine model for training, and then the firefly algorithm is introduced for improving the sparrow algorithm to search for the parameters for the support vector machine model. The optimal model is obtained; finally, the obtained optimal model is using to predict and diagnose the operation state for the charging pile to do judge whether the charging pile is faulty. The experimental results show that the prediction accuracy of the prediction algorithm in this paper could reach 94.68%, which is much higher than 72.34% of the traditional support vector machine model.

Abstract:In order to obtain the relative position information of thunder source generated by lightning, a three-dimensional cross array composed of seven microphones is established by using spatial geometric azimuth estimation algorithm. On this account, a lightning localization method is proposed to realize array-based omnidirectional lightning detection. Firstly, according to the distances from the thunder source to each microphone, the coordinate calculation formula of the thunder source is deduced. Among them, a pair of microphones along the Z axis are adopted to determine the positive and negative of the parameters in the Z direction of the coordinate. Then, based on the theory of indirect measurement error, the relationship between element spacing, angle and distance of thunder source and localization accuracy is studied and analyzed. Simulation results show that the coordinate error rate of thunder source measured by this method is about 0.01%, and the angle error rate is about 0.005%. This not only achieves good localization effects, but also effectively solves the problems of large amount of calculation and low accuracy of the existing works.

Abstract:The technology of release source location is of great significance in the prevention and control of harmful gas leakage and diffusion. The traditional location algorithm relies on the odor concentration gradient or wind direction to search, but when dealing with the turbulent environment with sparse odor, it is easy to lose the target and lead to the search failure, while the infotaxis algorithm takes the information obtained in the search process as a clue to maximize the entropy reduction when choosing the direction, and the search performance is better. First of all, the commonly used gas diffusion model is studied and analyzed, and then according to its characteristics, the infotaxis algorithm is simulated based on the gas turbulent diffusion model, and the feasibility of the algorithm is verified. After that, the influence of the search distance length on the algorithm is studied through comparative experiments, and the robustness of the infotaxis algorithm is proved. In order to further improve the performance of the algorithm, the search characteristics of basic quadrilateral path unit, hexagonal path unit and eight-point path unit are analyzed, and an improved quadrilateral search path unit is proposed. A large number of comparative experiments are carried out to verify that the proposed search path unit enhances the adaptability of the algorithm, reduces the search time and improves the search efficiency.

Abstract:With the increasing demand for electricity, the reliability and safety of electric energy metering devices have attracted much attention. Aiming at the problem of low accuracy of fault diagnosis of electric energy metering device, an improved seagull algorithm optimized support vector machine (ISOA-SVM) model is studied and designed. In order to make up for the deficiency of seagull optimization algorithm (SOA), an improved seagull optimization algorithm (ISOA) with better optimization performance is proposed. The internal parameters of SVM are optimized by ISOA, and the fault diagnosis model of electric energy metering device based on ISOA-SVM algorithm is constructed. The experimental results show that under the same evaluation index, the average value of 50 fault diagnosis of ISOA-SVM model is as high as 96.575%, which is 6.681%, 5.63%, 11.95% and 12.79% higher than that of PSO-SVM, SOA-SVM, SVM and ELM model. It shows that the designed ISOA-SVM algorithm has strong robustness and good fault diagnosis performance.

Abstract:When the aircraft needs to change the path temporarily in case of emergencies during navigation, the efficiency and reliability of the route planning algorithm are urgently required. An improved Dijkstra algorithm with pre search is proposed to solve this problem. A more objective track evaluation function is established by using the normalized entropy weight method to simplify the multi-objective track optimization model. The backtracking function of D algorithm is realized by adding the pre search process with depth of one, which solves the problem of high failure rate of path search under complex constraints due to insufficient relaxation of classical D algorithm. In addition, a break mechanism is added in the pre search traversal process to further reduce the operation time. Simulation results show that the operation time of the proposed algorithm is reduced by 46% compared with ordinary backtracking D algorithm. And the algorithm is close to intelligent algorithm in path search success rate and the accuracy of the shortest path decision under complex constraints, which can meet the requirements of fast path planning under complex conditions.

Abstract:Aiming at the requirement of flat bandpass filter in modern wireless communication system, a hairpin resonator with parallel load is proposed, and a fourth-order flat bandpass filter is constructed by using the proposed resonator. Firstly, a new parallel resonator is designed by setting a resistor at the strongest magnetic field (central position) of the hairpin resonator, so that the inherent quality factor can be flexibly adjusted by changing the resistance value; Secondly, the fourth-order flat bandpass filter is constructed by using the coupling topology with high and low intrinsic quality factors. Finally, a fourth-order bandpass filter with transmission zero and flat bandpass is fabricated on the printed circuit board. Experimental results show that the measured results of the fourth-order bandpass filter are in good agreement with the simulation results. Compared with other multi-order bandpass filters, the fourth-order bandpass filter has the inherent quality factor that can be flexibly controlled and the passband flatness is less than 0.2dB.

Abstract:In the shore-air bistatic HF/VHF radar system,the sending and receiving station have high requirements on the synchronization accuracy of time, frequency and phase, and it is difficult to realize, and the receiving station has high requirements on the size and weight of the synchro controller.In this paper，the synchro controller of shore-air bistatic HF/VHF radar system is designed based on digital signal processing, programmable logic gate array and GPS (DSP + FPGA + GPS).The UTC information and PPS signal in GPS signal is used to ensure the time synchronization of the two stations, and the independent high stable oscillator is calibrated by FPGA combined with PPS signal to ensure the frequency and phase synchronization of the two stations. The hardware structure is simplified to facilitate the integration of the whole receiving system, and the synchronization control parameters can be flexibly selected, and the startup time is flexible and controllable.In sending and receiving two stations set up the same synchro controller performance testing, the results show that the synchro controller realizes the synchronization of time, frequency and phase of the two stations, and the timing control signals generated are accurate, the synchronization accuracy of two stations is under 5ns, the coherence of bistatic radar system can be guaranteed in timing , and small in size, light in weight, easy to use, which fully meets the requirements of air-shore bibase HF/VHF radar system.

Abstract:In order to reduce a high operation cost of maintenance and recondition, as well as improving the security capability, we employed an improved YoloX-s detection method for the fault of closure detectors. By elevating the PANet path fusion network of the proposed model, a fusion with shallow feature layer is further strengthened; In addition, we added the CA(Coordinate Attention) attention mechanism to the model for the more detailed information in the target area. Moreover, the CIoU loss function is selected to enhance a positioning accuracy, which is aimed at the overlapping area, the center point distance and the aspect ratio between a target frame and a detection frame. After various tests, the experimental results showed that the presented model has a better comprehensive performance compared with the existing YoloX-s model. Furthermore, an average accuracy of moving contact reached 97.73%, an average accuracy of static contact reached 98.83%, and an average accuracy reached 98.28%.

Abstract:Ball valve door is a kind of use of ball as a closing valve, with its easy operation, easy maintenance and versatility is widely used in the industrial field. In order to solve the problem that it is difficult to detect the safety of ball valve doors during operation, a torque measuring platform is designed in this paper, which aims to monitor and identify the safety state of ball valve doors by collecting, processing and analyzing the bending moment of ball valve door skid bar. The valve lever torque measuring platform is composed of signal amplification, active low-pass filter, data acquisition board and other modules to measure the bending deformation caused by the valve lever rotation. The software of measuring platform is designed and developed based on MATLAB to process and store the torque data of ball valve door lever and draw the data graph. The experimental results show that the support vector machine algorithm is used to classify 8 kinds of state data of ball valve door collected by torque measuring platform, and the accuracy rate is 91.67%. It provides powerful data support for automatic identification and early warning of valve door faults.

Abstract:To solve the problem of Insufficient matching accuracy of the weak texture and disparity discontinuity regions in the image, A stereo matching algorithm combining superpixel segmentation and cross-scale PatchMatch is proposed. Firstly, multi-scale images are obtained by the gaussian under-sampling, and superpixel segmentation of each scale image. Then, based on the four-color theorem, corroding superpixel boundaries makes the iterative propagation of 3D labels on superpixels sub-modular and independent, and the generated sub-modular energy is optimized by the Graph Cut (GC) algorithm. Finally, aim to make 3D label iterative propagation can be cross-scale GC optimization to obtain the optimal disparity map, a cross-scale energy function model is proposed to constrain the consistent energy of 3D labels of the same pixel at different scales. Experimental results on Middlebury data set show that the average mismatch rate of the proposed algorithm for 21 groups of weak texture and complex texture images is 2.20%. Compared with other improved PatchMatch stereo matching algorithm, the false matching rate is reduced by 10.1%. Visualization of disparity map mismatched regions shows the proposed algorithm is better than other improved PatchMatch stereo matching for weak texture and disparity discontinuity regions algorithm.

Abstract:Style variations among different cameras is an important challenge in the field of person re-identification. To smooth the camera style disparities and enrich the diversity of pedestrian samples, this paper explicitly learns invariant features among cameras through a style transfer approach. Specifically, a cycle consistent adversarial networks (CycleGAN) is used to generate transformed images with other camera styles for each pedestrian, and along with the original samples, form the augmented training set. In addition, this paper uses an attention mechanism to reweight the feature channels to extract more discriminative pedestrian appearance features, and finally, the multi-task loss is used to supervise the training process of the re-identification network. The experimental results show that the mAP and top-1 metrics of the method in this paper achieve 86.5%, 95.1% and 77.1%, 87.2% on the public datasets Market1501 and DukeMTMC-reID, respectively, which are better than the existing algorithms. Camera style transfer as a data augmentation approach effectively expands the dataset and reduces the human labeling cost, while improving the identification accuracy in multi-camera scenarios.

Abstract:Cloud detection is an important step in the preprocessing of remote sensing images. The accuracy of cloud detection directly affects the accuracy of subsequent remote sensing image applications. Aiming at the problems of poor generalization ability and serious misdetection and missed detection in the existing cloud and cloud shadow segmentation tasks, this paper designs an improved convolutional neural network model, which uses U-Net as the backbone network and adds efficient channel attention mechanism and modifies activation function. The remote sensing image is used as input and put into the U-shaped cloud image segmentation model based on efficient channel attention for training. After obtaining the optimal weight, the remote sensing image segmentation result including cloud area, cloud shadow area and background area is output. The experimental results show that, compared with the existing segmentation models, this model has lowest parameters, highest accuracy, and best generalization effects in the segmentation task of clouds and cloud shadows.

Abstract:Aiming at the problems existing in common measurement methods of fragment velocity and scatter peculiarity, the integrated measurement method of fragment velocity and scatter peculiarity is studied, and an array-type integrated measurement system is designed in this paper. The integrated measurement system takes the comb target as the array element to construct the measurement target. Based on the multi-channel synchronous timing method, the flight time and target location of fragment are obtained. Combined with the correlation measurement principle, the fragment velocity and scatter peculiarity are obtained. The validation test result shows that the system can realize the integrated measurement of fragment velocity and scatter peculiarity. In the measurement of fragment scatter peculiarity, compared with the results of manual statistics, the fragment capture rate and the measurement accuracy of distribution density is improved by at least 3%, and the measurement error of scatter Angle is less than 5%. It verifies the feasibility of the integrated measurement system and the reliability of the measurement results. The system provides an effective measurement method for the integrated measurement of fragment velocity and scatter peculiarity.

Abstract:Partial discharge is an important cause of insulation breakdown of high-voltage electrical equipment, but also an important indicator of insulation deterioration, in view of the current switchgear partial discharge conventional detection methods have a small amount of detection information, poor timeliness, low diagnostic accuracy and other issues, this paper proposes a convolutional neural network detection method that can be integrated in mobile devices, and for the actual situation there is a problem of uneven discharge class samples, a fault sample generation method is proposed. The collected ultrasonic signal is de-denoised and pre-processed into a two-dimensional temporal spectrogram by short-term Fourier transform, and the partial discharge pattern is identified in the input convolutional neural network, and the adversarial network is used to generate the fault sample for the problem of uneven fault sample in the actual scene. The example experiments show that the accuracy rate of the proposed method in this paper reaches more than 97%, the computing power reaches 0.27 seconds under the condition of t710 computing power of the mobile terminal, and the error of the MSE generated data sample is lower than 0.067.

Abstract:In network communication system, image processing system, and so on, multiple subsystems would access external memory simultaneously. Multi-channel memory controller can solve this problem effectively. With the increasement of data and the improvement of processing unit performance, traditional multi-channel controllers can not meet the requirement of high-speed memory access for system because of low bandwidth utilization. To solve the aforementioned problems, we propose a new kind of DDR4 multi-channel controller on the FPGA in this paper. The controller is defined by simplified user interface and supports ring buffer in network communication, which reduces the use complexity and improves the universality. Multi-channel access conflicts can be solved efficiently by adopting circular priority arbiter. Meanwhile, the bandwidth utilization of the system has been improved. Besides, the rewinding access to the ring buffer is realized by a sharding mechanism. The simulation results are consistent with that on Xilinx KCU116 FPGA. When testing 4096 MB records, the maximum effective bandwidth of the system is 78.3Gbps, and the bandwidth utilization rate reaches 94.0%.

Abstract:In the noisy environment, the highly complete extraction of cable partial discharge signal is the key to the research of cable insulation on-line monitoring. In this paper, a partial discharge signal denoising method based on improved EEMD is proposed. Firstly, EEMD decomposition is carried out on the original partial discharge signal, Fourier transform is carried out on each order modal component (IMF) decomposed, and its corresponding amplitude variance is calculated as the threshold. Combined with the local weighted regression scatter smoothing algorithm, the IMF component mixed with noise and effective signal is filtered. Finally, it is superimposed and reconstructed with the IMF component of pure PD signal to obtain the filtered partial discharge signal. Through simulation and laboratory data analysis, the signal-to-noise ratio and waveform similarity coefficient of partial discharge signal after noise reduction are improved, which proves the effectiveness of the noise reduction method proposed in this paper.

Abstract:In order to reduce the loss of data due to rainfall attenuation and to provide the "best" data for users, the master and diversity data in the path diversity reception mode are optimally synthesised to reduce the impact of rainfall on the Ka-band data transmission link of the Fengyun-4 geostationary meteorological satellite. In this paper, the rainfall fading budget model provided by the International Telecommunication Union's Radiocommunication Sector (ITU-R) is used to fully calculate the rainfall fading situation, and a data merit synthesis system is designed to discriminate the data quality by the fill and error codes of the Advanced Orbiting Systems (AOS) frame data, and to select high-quality data frame by frame for synthesis and distribution to users. Because of the different attenuation effects of rainfall on different frequency bands, the system was tested in a simulated operational environment using Ka-band and X-band data. The test showed that the data merit synthesis system was able to effectively judge the quality of the input data and generate distribution data that met the service requirements, effectively reducing the impact of rainfall on the Ka-band data. The results show that the data merit synthesis system can fully play its role in the path subset, ensuring the high-quality reception of meteorological satellite data and the use of Ka frequency resources.

Abstract:Acoustic logging data plays an important role in horizon calibration and reservoir inversion. However, due to the limitations of equipment and geological environment, the actual acoustic logging curve is often distorted. To provide reliable data support for oil and gas exploration and improve the accuracy of reservoir prediction, a well-to-seismic joint inversion method based on parameter optimization residual network is proposed to reconstruct the distorted acoustic logging curve. Considering that the traditional artificial neural network cannot express the strong nonlinear relationship between well and seismic, this method uses the residual network (ResNet) in deep learning to build an intelligent inversion model. Through network design, parameter selection and model training, a better mapping expression between well and seismic can be found. Considering the characteristics of logging curve and the deficiency of MSE loss, a cost-sensitive loss function Fusion is designed to further improve the overall inversion accuracy of the model. Experiments are carried out on real seismic data and logging data, compared with the inversion results of Fully Connected Neural Network (FCNN) and Multiple Linear Regression (MLR), it shows that the accuracy of the acoustic logging curves inverted by the proposed method is higher, the correlation coefficient reaches 0.912, and the root mean square error is reduced to 13.399. Using the proposed Fusion loss to invert the acoustic logging curve, the correlation coefficient increases by 2.5%, and the root mean square error decreases by 17.4%.