摘要
对电能质量进行监测是用电信息采集系统的重要任务之一。针对电能质量扰动的识别和分类问题,提出一种基于小波支持向量机(support vector machine,SVM)的分类方法。对正常电压信号和6种电能质量扰动信号波形进行仿真,首先对各种扰动信号进行小波分解,提取各层小波重构系数的能量熵作为特征向量;然后改进的粒子群(particle swarm optimization,PSO)算法对SVM核函数参数进行优化;最后,利用优化参数的SVM对扰动测试集进行分类识别。仿真结果表明,与BP神经网络分类方法相比,该方法对扰动识别和分类的准确率达到97.28%,且训练时间和测试时间都有所减小。
The monitoring of power quality is one of the important tasks of power information acquisition system.As for the disturbance identification and classification of power quality,a kind of classification method based on wavelet support vector machine(SVM)is proposed.For the simulation of normal voltage signal waveform and of the disturbance signal waveform of 6 kinds of power quality,the wavelet decomposition for various disturbance signal is made firstly so to extract the energy entropy of each level of the wavelet packet coefficients as the feature vector;then the particle swarm optimization(PSO)algorithm is adopted to optimize the SVM kernel function parameters.Finally,the SVM with optimized parameters is used to perform the classification and identification the disturbance test set.It is shown by the simulation result that the accuracy rate of the method in the disturbance identification and classification is up to 97.28 compared with BP neural network classification method,also the training time and test time are reduced.
作者
何行
夏水斌
张芹
董重重
冉艳春
王汪兵
王先培
HE Xing;XIA Shuibin;ZHANG Qin;DONG Chongchong;RAN Yanchun;WANG Wangbing;WANG Xianpei(State Grid Hubei Electric Power Company Metering Center,Wuhan 430080,China;School of Electronic Information,Wuhan University,Wuhan 430072,China)
出处
《电力电容器与无功补偿》
北大核心
2019年第2期119-124,共6页
Power Capacitor & Reactive Power Compensation
基金
国家自然科学基金(50677047)
国家电网总部科技项目(52153217000A)
