【摘要】 以实验室模拟雾霾环境下污闪得到的大量电流数据为基础,运用ARMA算法建立故障检测模型。研究发现,雾霾天气下的绝缘子污闪与普通闪络有很大差别,且雾霾所含的主要盐分不同,故障特征也不一样。为进一步提取故障特征,提出了两种基于ARMA算法的故障诊断模型:基于ARMA预测误差的雾霾天绝缘子故障诊断模型和基于ARMA预测平均绝对误差的小波能量参数区间模型。前者是套用不同故障模型,通过静态预测误差的大小来判别故障类型;后者结合小波分析方法,克服了ARMA算法在故障特征提取过程中出现低区分度问题,以能量区间的方式区分和表示不同故障,以达到故障识别检测的目的。实验发现,两种模型均具有良好的雾霾天污闪检测识别性能。更多还原

【Abstract】 The most important part of fault model is to extract fault features. This paper establishes an ARMA fault detection model based on a large number of current data obtained from laboratory simulation of haze. In this paper,we found that there are huge differences between common flashover and contamination flashover caused by haze. The different salt type that haze contains will make the characteristics of contamination flashover different. In order to extract fault features further,two fault diagnosis models based on ARMA algorithm are proposed in this paper. One is based on ARMA prediction error in the haze fault diagnosis model,and the other is the wavelet energy parameter interval model based on ARMA prediction mean absolute error. The basic method of the first model is to apply different fault models,and identify the fault types by the size of the static prediction error. The second model analysis method combining wavelet and ARMA algorithm overcomes low extraction discrimination problem which appears in the process in the fault feature,to achieve the purpose of detecting fault recognition. Through verification,it is found that the two models have good recognition performance of haze contamination flashover detection.更多还原