染污玻璃绝缘子泄漏电流特性及其闪络电压预测

发布时间：2018-03-07 13:04

本文选题：泄漏电流　切入点：污秽度　出处：《电网技术》2014年02期 　论文类型：期刊论文

【摘要】：有效预测绝缘子闪络电压是防止发生污闪事故的重要手段,泄漏电流是分析检测绝缘子闪络电压的重要方法。在人工污秽实验室进行大量的试验,模拟运行电压下污秽度与相对湿度对泄漏电流的影响,从不同角度提取了能够反映绝缘子表面污秽度及相对湿度的4个泄漏电流特征量:泄漏电流脉冲幅值熵S、脉冲幅值Ih、能量比K及能量E,并得到它们之间的变化规律。提出基于广义回归神经网络(generalized regression neural network,GRNN)的绝缘子闪络电压预测模型,将4个特征量S、Ih、K、E及相对湿度作为GRNN模型的输入量,闪络电压作为GRNN模型的输出量。预测结果与试验结果对比分析可知,相对误差小于7.33%,表明提出的绝缘子闪络电压预测GRNN模型的预测结果与试验结果基本一致,能够有效地对绝缘子闪络电压进行预测。
[Abstract]:Effective prediction of insulator flashover voltage is an important means to prevent pollution flashover. Leakage current is an important method to analyze and detect insulator flashover voltage. The influence of pollution degree and relative humidity on leakage current under running voltage is simulated. From different angles, four characteristic quantities of leakage current which can reflect the contamination degree and relative humidity of insulator surface are extracted: leakage current pulse amplitude entropy S, pulse amplitude value Ih, energy ratio K and energy E, and the variation gauge between them is obtained. The prediction model of insulator flashover voltage based on generalized regression neural network (GRNN) is proposed. In this paper, the four characteristic variables, Schi-Ihke E and relative humidity, are taken as the input of GRNN model, and the flashover voltage is taken as the output of GRNN model. The results of prediction and test are compared and analyzed. The relative error is less than 7.33, which indicates that the predicted results of the GRNN model for insulator flashover voltage prediction are in good agreement with the experimental results, and can effectively predict the insulator flashover voltage.
【作者单位】：国网湖南省电力公司电力科学研究院;输配电装备及系统安全与新技术国家重点实验室(重庆大学);
【基金】：国家重点基础研究发展计划项目(973项目)(2009CB72-4501/502/503)~~
【分类号】：TM216

【参考文献】