基于BP人工神经网络的绝缘子泄漏电流预测  

Prediction of Leakage Current of Outdoor Insulators Based on BP Artificial Neural Network

作　　者：毛颖科[1] 关志成[2] 王黎明[2] 乐波[2] 

机构地区：[1]清华大学电机系,北京市海淀区100084 [2]清华大学深圳研究生院,广东省深圳市518055

出　　处：《中国电机工程学报》2007年第27期7-12,共6页Proceedings of the CSEE

基　　金：国家自然科学基金项目(50377020)。~~

摘　　要：外绝缘泄漏电流和绝缘子的污闪过程存在密切的对应关系,相对湿度是影响泄漏电流产生和发展的关键因素之一。首先对两种常用悬式绝缘子进行人工污秽试验,利用泄漏电流测量系统记录其在运行电压作用下、不同相对湿度时的泄漏电流波形并对此进行分析;采用BP人工神经网络的方法,建立了不同湿度下的泄漏电流最大值之间的对应关系;选择Levenberg-Marquardt快速学习算法对建立的BP神经网络进行训练。利用部分试验数据进行的验证以及采用不同方法获得的泄漏电流随相对湿度变化曲线均表明,使用该神经网络来建立不同表面受潮状态时泄漏电流最大值之间的关系是准确有效的。Leakage current of outdoor insulators is strongly relative to the contamination flashover process. Relative humidity is one of the key factors that influence the generation and development of the leakage current. Artificial pollution tests were carried out on two kinds of typical suspension insulators. Waveforms of Leakage current under operating voltage and different relative humidity were recorded by a leakage current monitoring system, and the data was analyzed. A BP Artificial Neural Network （ANN） was used to establish the relationships of maximum leakage currents with different relative humidity. The ANN was trained by the Levenberg- Marquardt fast training algorithm. Both verification of experimental data and comparisons of relationships between maximum leakage current and relative humidity obtained by different methods both show that the ANN is effective to relate maximum leakage currents with different surface moist condition.

关 键 词：绝缘子 泄漏电流 人工神经网络 相对湿度 等 值附盐密度 

分 类 号：TM216[一般工业技术—材料科学与工程]