一种可用于绝缘子沿面放电状态识别的算法  

Novel Algorithm for Discerning Surface Discharge State of Contaminated Insulators

作　　者：贺博[1] 高乃奎[1] 万军[1] 张爱德[1] 孙健[1] 彭宗仁[1] 

机构地区：[1]西安交通大学电力设备电气绝缘国家重点实验室,西安710049

出　　处：《西安交通大学学报》2008年第10期1269-1274,共6页Journal of Xi'an Jiaotong University

基　　金：国家自然科学基金资助项目(90210006)

摘　　要：在深入分析污秽绝缘子沿面放电特性的基础上,发展建立了一种新型的模式识别算法.该算法借鉴Bayes分类原理来实现电流的判别,借鉴模糊数学理论来实现湿度的判别,然后通过仲裁机制在综合电流、湿度信息的基础上确定出状态类属.检验结果表明,该算法具有好的通用性,利用一次试验数据训练出的算法在完成另外次试验数据的识别时具有很高的准确率,并且误判均发生在相邻状态,所以非常适合基于泄漏电流和湿度信号的放电状态识别.It is important for accomplishing online monitoring and risk prediction of flashover from contaminated insulators to develop efficient algorithm for identifying surface charge state of insulators. Following the analysis of surface discharge character, a novel discernment algorithm is proposed, where the discernment process contains three steps： the discharge states are discerned based on leakage current via Bayes discernment algorithm; the stages are discerned based on humidity through fuzzy theory; the results from the first and two steps are combined by an arbkration operation and the final result about the discharge state is achieved. The validity testing results show that the novel algorithm is endowed with a good universality and the algorithm trained out by one group data enables to identify the discharge state of the other artificial test with good accuracy and the mistake happens within the adjacent states. So the algorithm is suitable for discerning surface discharge state of HV insulators through LC and humidity signal.

关 键 词：沿面放电 模式识别 模糊化 交互验证 

分 类 号：TM216.1[一般工业技术—材料科学与工程]