Table of Content

15 August 2021 Volume 34 Issue 8

  

Research and Application of Vehicle Face Recognition Algorithm

ZHAN Yinwei,ZHU Baiwan,YANG Zhuo

Electronic Science and Technology. 2021, 34(8):  1-7.  doi:10.16180/j.cnki.issn1007-7820.2021.08.001

Abstract ( 360 )   HTML ( 111 )   PDF (2219KB) ( 64 )  

Figures and Tables | References | Related Articles | Metrics

The parking lot charging system based on the license plate recognition technology has problems such as the automatic deduction of unlicensed cars and the fee evasion of the fake-licensed vehicle. In order to solve this problem, it is necessary to automatically and accurately identify the vehicle based on other features except for license plate. The vehicle face can effectively express the feature information of the vehicle. A vehicle face recognition algorithm based on convolution neural network is proposed in this study. This algorithm automatically and accurately identify the vehicle according to the vehicle face image, and effectively solves the problem of excessive dependence on license plate in vehicle identification. The experimental results show that the accuracy of the algorithm reaches to 94.23%, which is obviously better than that of the traditional algorithm, which proves the feasibility of the algorithm. Finally, integrating the face recognition technology into the intelligent parking fee collection system effectively solves the problem that the unlicensed car cannot automatically deduct the fee and the fee evasion of the license plates.

Batch Mixed Training Traffic Sign Recognition Based on Improved VGG16 Network

LIAO Luming,ZHANG Wei

Electronic Science and Technology. 2021, 34(8):  8-13.  doi:10.16180/j.cnki.issn1007-7820.2021.08.002

Abstract ( 414 )   HTML ( 75 )   PDF (970KB) ( 47 )  

Figures and Tables | References | Related Articles | Metrics

In view of the problem of low recognition rate and long recognition time of the existing traffic sign recognition methods, based on convolution neural network method, an improved convolutional neural network model based on VGG16 is proposed. The number of convolution layers, convolution kernels and pooling layers of the VGG16 network model are modified to enhance the feature extraction ability and simplicity of the network model. The experimental datasets are enhanced by random rotation, scaling, offset and contrast adjustment, and the network model recognition rate is improved by activation function, mixed batch training and early termination regularization methods. In the experiment of the German traffic sign recognition benchmark, the recognition rate of the improved VGG16 network model is 98.98%, and the recognition time of single traffic sign is only 0.24 ms. Compared with other models, the proposed model has obvious advantages in recognition rate and recognition time.

Workflow Recognition Based on Temporal Action Detection

WANG Qingwen,HU Haiyang

Electronic Science and Technology. 2021, 34(8):  14-18.  doi:10.16180/j.cnki.issn1007-7820.2021.08.003

Abstract ( 166 )   HTML ( 98 )   PDF (675KB) ( 33 )  

Figures and Tables | References | Related Articles | Metrics

In a complex manufacturing environment, it is difficult for workflow recognition based on action recognition to localize the start and end times of each activity in a workflow from the video. In order to localize the activities temporally in the workflow from the video, according to R-C3D, a workflow recognition framework based on temporal action detection is proposed. The proposed workflow recognition method introduces a random sparse sampling strategy to reduce the redundancy between adjacent frames, and uses Res3D network to extract spatio-temporal features in the video. In addition, Soft-NMS strategy is used to eliminate the highly overlapping and low confidence proposals. Experiment results show that compared with R-C3D, the proposed method can improve the speed of training and detecting about 80% and 85%, respectively without loss of accuracy, proving the effectiveness of the method for workflow identification.

Fundus Blood Vessel Image Segmentation Combining Resnet50 and U-Net

SI Mingming,CHEN Wei,HU Chunyan,YIN Zhong

Electronic Science and Technology. 2021, 34(8):  19-24.  doi:10.16180/j.cnki.issn1007-7820.2021.08.004

Abstract ( 745 )   HTML ( 125 )   PDF (1448KB) ( 129 )  

Figures and Tables | References | Related Articles | Metrics

The characteristic information of the fundus blood vessel image has high complexity, and the existing algorithms have the problems of insufficient retinal microvascular segmentation and wrong segmentation of the focus area. In view of these problems,a fundus color blood vessel image segmentation model combined RESNET 50 and U-Net is proposed. Through data enhancement, the number of datasets is increased, so as to improve the robustness of the algorithm. The adaptive Gamma correction is used to improve the brightness information and reduce the interference of the image artifacts. In the coding part, the residual network is used to extract the deep-seated features of the color image. After four times up-sampling and skip-connection of the feature map, pixel level segmentation is achieved. In each convolution layer, the ReLU activation function and batch normalization are used to improve the performance of the model. A large number of experiments in the open dataset DRIVE show that the specificity of the proposed algorithm is 0.989 2, the accuracy is 0.967 6, and the sensitivity is 0.812 0.

Research on Processing Methods of False Signals Caused by Channelization

WU Zhenyu,XIA Houpei

Electronic Science and Technology. 2021, 34(8):  25-30.  doi:10.16180/j.cnki.issn1007-7820.2021.08.005

Abstract ( 469 )   HTML ( 103 )   PDF (940KB) ( 57 )  

Figures and Tables | References | Related Articles | Metrics

In view of the problem of false signals such as cross-channel split signals and rabbit ear effect when using channelization to process signals, this study proposes a method for eliminating false signals by using the parameter characteristics of the false signals. According to the serious time-frequency overlap of the cross-channel split signals of adjacent channels in the method, the cross-channel split signals of the channelization detection result are spliced and fused. According to the parameters such as the amplitude and pulse width of the false pulse of the rabbit ear effect, which are much smaller than the amplitude and pulse width of the original signal, and the occurrence and end times are near the rising and falling edges of the original signal, the rabbit ear pulses of the channelization detection result are directly eliminated. The final detection result is obtained through the above processing. Experimental results show that the method proposed in this study can effectively eliminate false signals and obtain correct detection results after processing the channelized output.

Research on Charging Method of Sensor Network Based on Relay Multi-Hop Model

YE Hengchao,CHENG Zongmao

Electronic Science and Technology. 2021, 34(8):  31-36.  doi:10.16180/j.cnki.issn1007-7820.2021.08.006

Abstract ( 174 )   HTML ( 93 )   PDF (1013KB) ( 27 )  

Figures and Tables | References | Related Articles | Metrics

Under the condition of ensuring that the wireless rechargeable network can last for a long time, based on the relay multi-hop charging model, the problem of maximizing the energy utilization of the charging car in the sensor network is studied. To solve this problem, a relay node selection algorithm and path planning algorithm is proposed. The relay node selection algorithm is used to select relay nodes from sensor nodes and serves as a stop for charging cars, and uses the number of nodes covered by the relay as the basis for selection. The path planning algorithm is based on Hamiltonian loop re-optimization to plan the moving path for the charging car. By using the relay multi-hop charging model and the above algorithm, the charging car can reduce the loss of mobile energy, thereby improving energy utilization. Through simulation experiments, the two strategies of NJNP and TADP and the effects of different scale networks are compared. The results show that the proposed scheme can improve energy utilization by 15%~25%, and has good performance in large-scale sensor networks.

Design of Pulse Power Supply for Micro ECM Based on Interface Acidification

TIAN Yuanbo,LIU Guili,KONG Quancun

Electronic Science and Technology. 2021, 34(8):  37-42.  doi:10.16180/j.cnki.issn1007-7820.2021.08.007

Abstract ( 241 )   HTML ( 98 )   PDF (1573KB) ( 32 )  

Figures and Tables | References | Related Articles | Metrics

The high-frequency pulse power supply is one of the important factors to achieve precise micro electrochemical machining. In order to take into account the machining locality and increase the erosion rate of workpiece, the study develops a high-frequency pulse power supply based on interface acidification mechanism, which is able to add acid to the workpiece/electrolyte interface during the pulse interval period. The micro-electrochemical machining experiment is carried out using the developed high-frequency pulse power supply. The pulse power supply is utilized to process array holes on 0.5mm thick 304 stainless with 0.5 mol·L^-1 NaNO₃ electrolyte. The experiment results show that compared with the two-electrode pulse power, the erosion rate increases from 3.21×10^-4 mm³·s^-1 to 6.55×10^-4 mm³·s^-1 by using high-frequency pulse power to process array holes. The workpiece erosion rate is significantly improved, which proves the reliability of the designed pulse power supply.

A On-Demand Multi-Node Priority Charging Scheduling Algorithm

SHUI Jiusheng,WANG Ran

Electronic Science and Technology. 2021, 34(8):  43-49.  doi:10.16180/j.cnki.issn1007-7820.2021.08.008

Abstract ( 315 )   HTML ( 115 )   PDF (1895KB) ( 49 )  

Figures and Tables | References | Related Articles | Metrics

Most of the existing mobile charging scheduling schemes in wireless rechargeable sensor networks focus on periodic charging schemes and single-node charging models, but periodic charging ignores the dynamic changes of the network and often results in low charging performance, and the single-node charging model has low charging efficiency and poor scalability. In view of this problem, based on the multi-node charging model, an on-demand multi-node priority scheduling algorithm is proposed to reduce the number of dead nodes and energy consumption in this study. At the same time, a new linear model related to the distance of the base station is proposed for the energy consumption model of the sensor. In grouping nodes, in order to reduce the number of stopping points and balance the charging time of the nodes, a centroid-based clustering algorithm is developed. In the path planning, the remaining life of the node and the distance between the stopping point and the mobile charger are also considered. Finally, experiments verify that the proposed algorithm can effectively reduce node death and energy consumption.

Time-Varying Formation Control for Heterogeneous Nonlinear Multi-Agents System

WANG Weike,ZHANG Wei,SONG Fang,LONG Lin

Electronic Science and Technology. 2021, 34(8):  50-57.  doi:10.16180/j.cnki.issn1007-7820.2021.08.009

Abstract ( 205 )   HTML ( 109 )   PDF (1344KB) ( 38 )  

Figures and Tables | References | Related Articles | Metrics

In practical applications, it is necessary for multiple agents to work together according to the desired formation. At this time, the multi-agents system is heterogeneous. In view of this problem, this study investigates the time-varying formation control problem of heterogeneous nonlinear multi-agents system based on the theory of consensus. A formation control strategy is designed by using the nonlinear parameter decomposition method to represent the heterogeneous nonlinear terms in a unified form. An adaptive time-varying formation control strategy is designed based on adaptive strategy and state information feedback using the leader-follower method under directed topology, and the stability of the control strategy is proved by using Lyapunov stability theory. The simulation carried out in the case of one leader with three followers shows that the system is basically stable when the time is 40 s, and the system is stable and formed the desired time-varying formation when the time is 60 s, which verifies the effectiveness of the control strategy.

Research on Flexible Load Control Strategy of Air Conditioning in Power Demand Side

WANG Fuzhong,LI Runyu,ZHANG Hongwei,GUO Jiangzhen,ZHANG Li

Electronic Science and Technology. 2021, 34(8):  58-63.  doi:10.16180/j.cnki.issn1007-7820.2021.08.010

Abstract ( 967 )   HTML ( 121 )   PDF (1242KB) ( 137 )  

Figures and Tables | References | Related Articles | Metrics

With the widespread use of air conditioning in various buildings, air conditioning load has become an important factor affecting the stable operation of regional power grids. If air conditioning load is not regulated, it will affect the normal electricity consumption of the entire area. Based on this, this study proposes an adaptive fuzzy PI three-closed-loop control strategy based on the framework of layered regulation of air-conditioning load areas. The method combines air-conditioning load information and fuzzy control, and flexibly regulates air-conditioning loads by forming a three-closed-loop PI control, and establishes a simulation model to verify the feasibility of the strategy. The simulation results show that the control accuracy of the control method can reach nearly 99%, which provides a new method for precise load control. The control strategy has been applied in a university, providing theoretical support and practical reference for the future promotion of flexible control technology.

Analysis of MMC-HVDC DC Single Pole Grounding Fault Considering the Influence of Control Scheme

SHI Xi,CHENG Peng

Electronic Science and Technology. 2021, 34(8):  64-70.  doi:10.16180/j.cnki.issn1007-7820.2021.08.011

Abstract ( 300 )   HTML ( 78 )   PDF (1302KB) ( 46 )  

Figures and Tables | References | Related Articles | Metrics

At present, in order to solve the problem that the converter control is not fully considered in the single-stage ground fault characteristics of the existing flexible HVDC transmission system, this study proposes a transient characteristics analysis method for single-stage ground fault of MMC-HVDC system considering the influence of control strategy. According to the grounding mode of neutral point, the short circuit current path model after the single point grounding fault on DC side is constructed. By analyzing the influence of the nearest level control strategy on the physical parameters in the fault path, the change law of number of sub-modules on the bridge arms that locates on faulted pole is revealed. The formulas of short circuit current and capacitor voltage of sub-modules are derived, which accurately shows the transient characteristic of single point fault on DC side for MMC-HVDC system. The RTDS simulation results show the validity of formulas of short circuit current and capacitor voltage.

The Loss Calculation and Analysis of Hybrid Modular Multilevel Converter Considering Sub-Module Balancing Constraint

ZHANG Jie,ZENG Guohui,HUANG Bo

Electronic Science and Technology. 2021, 34(8):  71-78.  doi:10.16180/j.cnki.issn1007-7820.2021.08.012

Abstract ( 213 )   HTML ( 54 )   PDF (1080KB) ( 30 )  

Figures and Tables | References | Related Articles | Metrics

The voltage level and transmission capacity of flexible DC transmission projects are constantly increasing, and the loss of modular multilevel converter has attracted much attention. At present, the commonly used loss analysis methods are for the analysis and calculation of full-bridge and half-bridge modular multilevel converters. Based on the operating characteristics of the hybrid modular multilevel converter, a hybrid MMC loss analysis method that considers the constraints of submodule voltage sharing is proposed in this study. According to the reference direction of the voltage and current of the hybrid MMC bridge arm in a control cycle, the sub-module switching situation is analyzed.The mathematical model of the hybrid MMC is established according to the switching situation of different types of sub-modules for loss analysis calculation. By extracting the loss data of the insulated gate bipolar transistor to fit and establish a mathematical model, the detailed analysis of the mutual influence between the junction temperature and the loss of the converter is carried out. Finally, experimental simulation analysis is performed to verify the accuracy of the proposed method.

Research on Grid Connection of Photovoltaic Power Generation Systems Based on a New Robust Droop Control Method

YAN Zhenbin,ZHENG Bochao,ZHOU Zhiyong

Electronic Science and Technology. 2021, 34(8):  79-86.  doi:10.16180/j.cnki.issn1007-7820.2021.08.013

Abstract ( 293 )   HTML ( 8 )   PDF (1340KB) ( 43 )  

Figures and Tables | References | Related Articles | Metrics

Photovoltaic power generation system is easy to be affected by the environment and load changes, which will generate frequency and power fluctuations in the process of grid connection. The traditional PI control method is slow in response and poor in stability. To solve this problem, a new robust droop control design method is proposed in this study. An integral sliding mode controller is designed in the double closed-loop current inner loop control structure of bi-directional interconnected converter. The designed controller can not only suppress the voltage and frequency fluctuation of photovoltaic power generation system in the process of grid connection, but also ensure the stability of power system in different environmental conditions. The simulation results in MATLAB/Simulink show that the new robust droop control method is faster than the traditional PI control, and the fluctuation amplitude is reduced by nearly half, which ensures the continuous and stable operation of the grid after the photovoltaic power generation system is connected to the grid.