Table of Content

15 May 2021 Volume 34 Issue 5

  

Application of Compressive Sensing of CFD in Permeability Calculation

GUO Long,YAO Shuxin,ZHENG Fawei

Electronic Science and Technology. 2021, 34(5):  1-6.  doi:10.16180/j.cnki.issn1007-7820.2021.05.001

Abstract ( 217 )   HTML ( 17 )   PDF (1385KB) ( 40 )  

Figures and Tables | References | Related Articles | Metrics

The petroleum industry needs to obtain the permeability of porous material model by CT scanning, but the number of meshes in the calculation model is usually very large, which is usually larger than 2 000×2 000×2 000 voxels, Therefore, it is impossible to solve the hydrodynamic problems by a microcomputer in a short time. If the low-density grid can be reduced with the amount of simulation decreased accordingly, and then a mathematical fitting recovery algorithm can be implied to restore the simulation results to the high-precision grid, the low-performance equipment may complete the fluid simulation task in a short time. In this study, compressed sensing is introduced into the Euler grid computation to recover the flow field data of the grid with low precision sampling meshes. Three sandstone cores are randomly selected and compared with the results of high-precision grid calculation, and the error is less than 4%. Therefore, this method can optimize the calculation efficiency in the permeability calculation of fluid simulation of porous rock.

An EMC Prediction Method of Vehicular Communication Equipment

SONG Bingxin,LU Hongmin,LI Minyue,MENG Xiaojiao,WAN Jianpeng

Electronic Science and Technology. 2021, 34(5):  7-12.  doi:10.16180/j.cnki.issn1007-7820.2021.05.002

Abstract ( 231 )   HTML ( 6 )   PDF (809KB) ( 26 )  

Figures and Tables | References | Related Articles | Metrics

With the continuous improvement of the level of combat information, it is more and more complicated of vehicle electromagnetic environment, and the prediction of the electromagnetic compatibility of vehicular communication equipment is of great importance. A method for screening and predicting electromagnetic compatibility of vehicular communication equipment is proposed in this study. This method considers the intermodulation of different interference sources and the superposition of the interference quantity when calculating the interference margin. The key indicator is selected to analyze the performance of the communication equipment. The prediction method performs fast screening, amplitude screening, frequency screening, detailed calculation and performance analysis on the vehicle transmitter in order, and gives the interference situation of the vehicle receiver. In the prediction example, the interference margin of the evaluated receiver is 9.4 dB, the speech intelligibility is reduced by 36%, the communication distance is reduced by 43.7%, and the bit error rate is 4.2×10^-3. The disturbed prediction results are in line with the actual situation.

Study of Apple Coloration Detection Algorithm Based on H-S Histogram

YANG Lingxiao,WANG Zhenying,LIU Qunpo,WANG Gaowei,YANG Yanchao

Electronic Science and Technology. 2021, 34(5):  13-17.  doi:10.16180/j.cnki.issn1007-7820.2021.05.003

Abstract ( 263 )   HTML ( 6 )   PDF (1111KB) ( 41 )  

Figures and Tables | References | Related Articles | Metrics

In order to solve the problem that the detection result of apple color by machine vision is easily affected by the brightness of the picture, an algorithm of apple skin color rate detection based on H-S histogram is proposed in this study. The algorithm use a small number of samples of red, green and yellow colors to generate different H-S histograms. When detecting the coloring rate, the hue value and saturation value of each pixel in the image are extracted, and the H-S histogram is used to analyze the proportion of the red area on the surface of the apple to obtain the coloring rate of the apple surface. The test results prove that this method can effectively detect the coloring rate of the apple surface. The detection results obtained are basically consistent with those observed by human eyes. The average error is less than 8.3%, and the detection results are not easily affected by the brightness of the picture.

Research on Main Steam Temperature Control System Based on Fuzzy PID Load Tracking

REN Xinrui,MA Lixin

Electronic Science and Technology. 2021, 34(5):  18-23.  doi:10.16180/j.cnki.issn1007-7820.2021.05.004

Abstract ( 330 )   HTML ( 9 )   PDF (842KB) ( 42 )  

Figures and Tables | References | Related Articles | Metrics

In view of the problem that the traditional PID fails to realize load tracking control in real time. a fuzzy PID controller is designed by combining traditional PID control with fuzzy control. The real-time load of the generator is used as the response and input to construct a load tracking fuzzy PID main steam temperature control system. Correction values of the three parameters of the PID controller are obtained through a fuzzy algorithm, and the three parameters of the PID are adjusted online by real-time tracking load, and the main steam temperature are controlled to be stable near the set value. Load fluctuation, load mutation and smoke disturbance are performed and the experimental results show that the error between real-time system data and target temperature value is 1~2 ℃, the overshoot of the system decreases by 1.3% compared with the traditional method, and the adjustment time shortens by 33.4%. These results indicate that the proposed system has a strong load tracking ability and better robustness to load fluctuation.

An Evaluation Method of Vehicle-Mounted Communication Equipment Performance Based on BP Neural Network

MENG Xiaojiao,ZHANG Shiwei,LI Xiaojian,LI Minyue,SONG Bingxin,LU Hongmin

Electronic Science and Technology. 2021, 34(5):  24-28.  doi:10.16180/j.cnki.issn1007-7820.2021.05.005

Abstract ( 345 )   HTML ( 14 )   PDF (699KB) ( 40 )  

Figures and Tables | References | Related Articles | Metrics

With an increasing number of vehicular communication equipment, the communication system is faced with increasingly serious electromagnetic interference problem. In this study, based on the characteristics of high nonlinear fitting accuracy and strong self-tuning of BP neural network, an evaluation method is proposed to evaluate the performance of vehicular communication equipment under the complex electromagnetic environment. According to the critical technology indicators of vehicular communication equipment, an evaluation system including transmission, reception and interaction is established, and an evaluation model based on BP neural network structure is constructed. With the use of MTALAB software, large amounts of sample data are adopted to train and optimize the BP neural network model structure, and to improve the evaluating model accuracy. The validation results indicate that the normalized mean square error of the model reaches -36 dB, and the evaluation error is small.

Pedestrian Detection Algorithm Based on Multiple Feature Fusion

GU Wei,LI Feifei,CHEN Qiu

Electronic Science and Technology. 2021, 34(5):  29-34.  doi:10.16180/j.cnki.issn1007-7820.2021.05.006

Abstract ( 387 )   HTML ( 15 )   PDF (1580KB) ( 57 )  

Figures and Tables | References | Related Articles | Metrics

As a typical problem of object detection, pedestrian detection has been an active research topic in recent years. Pedestrian detection is widely used in intelligent transportation, autonomous driving, video surveillance, behavior analysis and other fields, but there are many problems to be solved. In this study, a multi-channel feature model based on multiple feature fusion which consists of a non-deep learning branch, a body branch and a limb branch is proposed. A small number of high-quality pedestrian candidate areas are extracted through the non-deep learning branch, thus reducing the computational cost cause by exhaustive search by sliding window and improving the computational efficiency. The body branch and limb branch obtained from the feature of multi-layer convolutional channel are applied to detect pedestrians through the overall human body information and the semantic information of human body parts, respectively. Caltech pedestrian dataset and INRIA pedestrian dataset are adopted to train and test the proposed model. Experimental results show that combine with the output of each branch, the proposed pedestrian detector have a lower miss rate. The miss rates on INRIA pedestrian dataset and Caltech pedestrian dataset are 8.24% and 19.78%, respectively.

Design of Text Title Generation Prototype System Based on Neural Network

ZHANG Shisen,SUN Xiankun,YIN Ling,LI Shixi

Electronic Science and Technology. 2021, 34(5):  35-41.  doi:10.16180/j.cnki.issn1007-7820.2021.05.007

Abstract ( 292 )   HTML ( 8 )   PDF (2211KB) ( 35 )  

Figures and Tables | References | Related Articles | Metrics

In view of the traditional manual methods cost a lot of manpower and time and can not deal with the problem of massive of non-standard texts, a prototype system of generating text titles is designed in the proposed study. In the prototype system, the non-standard text is calculated by the encoder-decoder model which is based on neural network to produce an accurate title. In the encoder part, the bidirectional long short-term memory neural network is adopted to make full use of the semantic connection between contexts. In the decoder part, one-way neural network is used for decoding operation, and attention mechanism is added to alleviate information loss and improve the effect of title generation. The evaluation indexes of ROUGE-1 and ROUGE-L obtained by experiments on LCSTS data set are 29.91 and 24.68, proving the effectiveness of the title generation prototype system.

ECG Classification Based on Bispectrum and Spectral Features

LIU Shu,SHAO Jie,ZHANG Yiting,ZHANG Shanzhang

Electronic Science and Technology. 2021, 34(5):  42-46.  doi:10.16180/j.cnki.issn1007-7820.2021.05.008

Abstract ( 435 )   HTML ( 12 )   PDF (671KB) ( 55 )  

Figures and Tables | References | Related Articles | Metrics

As the lowest high-order spectrum, bispectrum has many excellent characteristics, but it also has the defect of large calculation. Based on bispectrum matrix, a feature extraction method for two-dimensional ECG signal is proposed in the present study. The spectral flatness, spectral brightness and spectral roll off are combined to form the eigenvector, which is further combined with the support vector machine classification method founded on radial basis kernel function to realize the classification and recognition of ECG signals. The ECG signal in MIT-BIH database is applied to verify the method. Experimental result show that the two-dimensional spectrum feature extraction method based on bispectrum proposed in this study has a small amount of computation and an accuracy of 93.4%, indicating that the proposed method can effectively realize the diagnosis of arrhythmia and the classification of ECG signals.

Accelerated Spectral Clustering Based on Improved Landmark Selection

XU Hangfan,LIU Cong,TANG Jiangang,PENG Dunlu

Electronic Science and Technology. 2021, 34(5):  47-53.  doi:10.16180/j.cnki.issn1007-7820.2021.05.009

Abstract ( 212 )   HTML ( 7 )   PDF (828KB) ( 31 )  

Figures and Tables | References | Related Articles | Metrics

In order to solve the problems that the traditional landmark-based spectral clustering algorithm is susceptible to unevenly distributed landmark points and outlier landmark points, and its sampling methods such as K-means consume a large time and space when face large-scale data. This study proposes an accelerated spectral clustering based on improved landmark selection. The algorithm uses the standard deviation of the pairwise similarity matrix between landmark points to measure the uniformity of the distribution of landmark points. It selectes the landmark points set uniformly distributed from landmark points sets generated randomly, and then removes outlier landmark points with low local density. The sparse similarity matrix is constructed by the obtained landmark points set and the original data set. K-means clustering is performed on data points generated by the first k right singular feature vectors of the landmark points set to obtain the final clustering result. This study theoretically analyzes the time complexity and space complexity of the algorithm and performed experimental verification. Experimental results show that the algorithm is 3%~10% higher than that of the random sampling method, and the time-consuming is 50%~60% less than that of the K-means sampling method.

A Matrix Multiplication Mapping Technology Based on NOC Multi-Core System

WANG Yang,WANG Xiaolei,YUAN Ziang,YUAN Ruming

Electronic Science and Technology. 2021, 34(5):  54-60.  doi:10.16180/j.cnki.issn1007-7820.2021.05.010

Abstract ( 283 )   HTML ( 5 )   PDF (897KB) ( 37 )  

Figures and Tables | References | Related Articles | Metrics

Matrix multiplication is the basic operation of modern signal processing. Improving the parallel processing capacity of data has important practical significance for improving the operation performance of matrix multiplication. In this study, task scheduling and resource allocation are carried out for the intensive computing of matrix multiplication in different dimensions based on NOC multi-core system, and a variety of mapping algorithms suitable for different matrix multiplication are implemented, and the peak performance can reach 5078 MFLOPS. The designed operation unit is relatively independent and reconfigurable, which has good expansibility and generality for matrix multiplication of any dimension. It overcomes the limitation of I/O bandwidth and computing resources in fixed structure, which leads to low efficiency and poor expansibility. Through the analysis of the experimental results of matrix multiplication of different dimensions, the correctness and high performance of the design are verified.

An AlGaN/GaN High-Electron Mobility Transistor with N-Buried Layer

ZHANG Fei,LIN Mao,MAO Hongkai,SU Fangwen,SUI Jinchi

Electronic Science and Technology. 2021, 34(5):  61-65.  doi:10.16180/j.cnki.issn1007-7820.2021.05.011

Abstract ( 320 )   HTML ( 8 )   PDF (776KB) ( 39 )  

Figures and Tables | References | Related Articles | Metrics

In order to further improve the breakdown voltage of GaN HEMT and keep low on-resistance, an AlGaN/GaN HEMT with an N-type GaN buried layer is proposed. The buried layer reduces the peak value of the electric field in the high field region by adjusting the electric field distribution of the device, thereby reducing the leakage current when the device is off-state.The horizontal electric field distribution in channel between the gate and drain is more uniform, which improves the breakdown voltage of device. The Sentaurus TCAD simulation demonstrates that the N-type GaN buried layer can significantly improve the breakdown voltage of the device. The breakdown voltage of the new structure reaches 892 V,which is 68% higher than the 530 V breakdown voltage of the traditional structure. The buried GaN layer has no effect on the on-state characteristics of the device, so that the device retains low on-resistance. These results indicate that the proposed structure has a good application prospect in the field of power devices.

A Review of Research on Decision-Making Method of Autonomous Vehicle Based on Reinforcement Learning

ZHANG Jiapeng,LI Lin,ZHU Ye

Electronic Science and Technology. 2021, 34(5):  66-71.  doi:10.16180/j.cnki.issn1007-7820.2021.05.012

Abstract ( 2071 )   HTML ( 63 )   PDF (734KB) ( 371 )  

References | Related Articles | Metrics

The decision-making system can integrate environment and ego vehicle information, so that the autonomous vehicle produces safe and reasonable driving behavior, which is the core technology to realize the autonomous driving. Reinforcement learning algorithm adopts a self-supervised learning method, so that the decision-making system of autonomous vehicles can autonomously learn the optimal decision model through continuous improvement of its strategy during the interaction with the environment, which provides a direction for building an effective decision-making system.This study summarizes the research progress in recent years of the decision-making method based on reinforcement learning in terms of improving decision accuracy, improving decision-making breadth, and dealing with uncertain factors. The improvement of decision-making accuracy mainly depends on the introduction of deep learning algorithm with strong representation ability and the hierarchical abstraction technology that can decompose complex tasks to alleviate the dimension disaster. The uncertainty is considered by partially observable Markov decision process to improve driving safety.

Study on On-Line Detection of Surface Defects of Flat Enameled Wire

SONG Zhangming,HE Huiyong,HUANG Yuejun

Electronic Science and Technology. 2021, 34(5):  72-78.  doi:10.16180/j.cnki.issn1007-7820.2021.05.013

Abstract ( 337 )   HTML ( 12 )   PDF (1556KB) ( 49 )  

Figures and Tables | References | Related Articles | Metrics

In order to realize the on-line detection of flat enameled wire surface defects, a detection system based on image processing is designed in this study. The system uses a frame-to-frame difference method to separate the background of the flat enameled wire. In view of the jitter interference in the background-separated image, a debounce process is proposed. For images with defects, features such as the location, area, and perimeter of the defects are extracted using connected domain analysis and boundary tracking algorithms. The experimental results show that the accuracy of the system for detecting defects reaches 95%, which reduces the processing time of defect feature extraction, indicating that the system can meet the requirements of online detection and has good application value.