基于KELM-VPMCD方法的未知局部放电类型的模式识别  

Pattern recognition of unknown PD types based on KELM-VPMCD

作　　者：高佳程 曹雁庆 朱永利[1] 贾亚飞[1] GAO Jiacheng;CAO Yanqing;ZHU Yongli;JIA Yafei(State Key Laboratory of Ahernate Electrical Power System with Renewable Energy Sources, North China Electric Power University, Baoding 071003, China;GD Power Development Co., Ltd., Beijing 100101, China)

机构地区：[1]华北电力大学新能源电力系统国家重点实验室,河北保定071003 [2]国电电力发展股份有限公司,北京100101

出　　处：《电力自动化设备》2018年第5期141-147,共7页Electric Power Automation Equipment

基　　金：国家自然科学基金资助项目(51677072)~~

摘　　要：为了解决局部放电类型未知的样本无法被正确识别的问题,提出了一种基于核极限学习机变量预测模型(KELM-VPMCD)的未知局部放电类型的识别方法。通过KELM对已知局部放电类型的训练样本进行训练,然后对各局部放电类型已知的样本建立相应的变量预测模型。利用这些模型对测试样本进行回归预测。根据各样本的预测误差平方和,利用Otsu算法设置误差阈值,通过阈值识别各样本的局部放电类型。识别结果表明,所提方法对于未知的局部放电类型具有较高的正确识别率。In order to solve the problem that the unknown PD （ Partial Discharge） types cannot be recognized con＇ect- ly,a method based on KELM-VPMCD （Kernel Extreme Learning Machine-Variable Predictive Model based Class Discriminate） is proposed to recognize the unknown PD types. The samples with the known PD types are trained by the KELM,and the corresponding VPMs （Variable Predictive Models） are constructed and used for the regression prediction of testing samples. According to the quadratic sum of regression prediction errors, the thresholds are set by Otsu algorithm to recognize the PD types of samples. The recognition results show that the proposed method can re- cognize the unknown PD types with high accuracy.

关 键 词：局部放电 模式识别 核极限学习机 变量预测模型 

分 类 号：TM835[电气工程—高电压与绝缘技术]