SA-CoSaMP算法在电能质量扰动信号识别中的应用  

Application of SA-CoSaMP algorithm in power quality disturbance signal identification

作　　者：余林 肖儿良[1] 简献忠[1] Yu Lin;Xiao Erliang;Jian Xianzhong(School of Optical-Electrical and Computer Engineering,University of Shanghai for Science and Technology,Shanghai 20009a,China)

机构地区：[1]上海理工大学光电信息与计算机工程学院

出　　处：《电子测量技术》2018年第21期1-7,共7页Electronic Measurement Technology

基　　金：国家自然科学基金(41075019);上海市高校教师创新基金(ISI0302020)项目资助

摘　　要：针对电能质量扰动信号具有实时随机性和稀疏度未知导致识别率低的问题,提出了基于稀疏自适应的压缩采样匹配追踪算法(SA-CoSaMP)进行电能质量扰动信号分类识别的研究,可有效减少数据采集、降低特征值处理量,并提高识别率。首先,针对电能质量扰动信号,利用压缩感知理论(CS)获取降维的测量信号,基于SA-CoSaMP算法获取稀疏向量;然后,针对7类典型电能质量扰动信号的稀疏向量,经过实验对比提出5种特征值,采用BP神经网络作为分类器。实验结果表明,选取需要处理的原始信号长度2048,所提算法特征提取时处理长度平均为40,保存信号全部原始特征,具有重构精度高、抗噪能力强的优点,识别率为98.87%,高于稀疏自适应匹配追踪算法(SAMP)的94.33%和压缩采样匹配追踪算法(CoSaMP)的95.75%,为电能质量扰动信号识别提供了一种新的思路。In allusion to the problem of power quality disturbance signal with real-time randomness and sparseness unknown lead to low recognition rate,this paper proposes a SA-CoSaMP algorithm based on sparse adaptive coding to research on the recognition of power quality disturbance signal,which can effectively reduce the data acquisition and the eigenvalue processing capacity,and improve the recognition rate.Firstly,according to the power quality disturbance signals,using compressed sensing (CS)theory to obtain the dimensional measurement signal,and the sparse vector is obtained by SA-CoSaMP algorithm.Aiming at seven typical power quality disturbance signals based on the sparse vector,five eigenvalues proposed through experimental comparison,BP neural network is used as a classifier.The experimental results show that the average size of the proposed method is only 40 instead of 2048 for the original signal.It has the advantages of high precision and high anti-noise ability because of containing all the features of the signal.The recognition rate is 98.87%,which is higher than that of sparse adaptive matching pursuit (SAMP) algorithm of 94.33% and compression sampling matching pursuit (CoSaMP)algorithm of 95.75%,therefore,the power quality disturbance signal identification provides a new way of thinking.

关 键 词：电能质量 压缩感知 稀疏向量 BP神经网络 SA-CoSaMP算法 

分 类 号：TM71[电气工程—电力系统及自动化] TH183.3[机械工程—机械制造及自动化]