Power system abnormal data identification method

• Main Authors: QIAO LIN, LIU SHUJI, HU NAN, WU HE, LI ZHAO, XU ZHIYUAN, XU LIBO, LIU BIQI, CHEN SHUO, LI LIGANG, QU RUITING, LYU XUMING, LU BIN, SONG CHUNHE, WANG ZHONGFENG, FU YATONG, SHEN LI, CUI SHIJIE, ZHOU QIAONI, RAN RAN
• Format: Patent
• Language: Chinese; English
• Published: 07.07.2020
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; DATA PROCESSING SYSTEMS OR METHODS, SPECIALLY ADAPTED FORADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL, SUPERVISORYOR FORECASTING PURPOSES; HANDLING RECORD CARRIERS; PHYSICS; PRESENTATION OF DATA; RECOGNITION OF DATA; RECORD CARRIERS; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE,COMMERCIAL, FINANCIAL, MANAGERIAL, SUPERVISORY OR FORECASTINGPURPOSES, NOT OTHERWISE PROVIDED FOR

The invention relates to a power system abnormal data identification method. The method comprises the steps of: training a neural network by taking the normal data of a power system as a training sample; inputting to-be-detected data into the trained neural network to obtain a residual error sequence; clustering the residual training based on an affine propagation clustering algorithm; and performing abnormal data judgment according to the features of each category and intra-category distances. According to the method, neural network training is carried out by using a chaotic particle swarm algorithm, and data clustering is realized by using the affine propagation clustering algorithm, so that a calculation amount can be remarkably reduced. The method does not depend on sampling distribution, and effectively improves the accuracy of abnormal data identification of the power system. 本发明涉及一种电力系统异常数据辨识方法，包括将电力系统正常数据作为训练样本，训练神经网络；将待检测数据输入训练后的神经网络，获得残差序列；基于仿射传播聚类算法对残差训练进行聚类；根据各个类别的特征和类内距离进行异常数据判断。本发明利用混沌粒子群算法来进行神经