基于神经网络的绝缘子故障诊断

Insulator Fault Diagnosis Based on Neural Network

绝缘子是高压架空输电线路的关键部件之一,受到导线自重、雷电、风力、冰雪、粉尘污染等因素的影响,易造成绝缘子掉串、断裂、污秽等故障。上述故障容易导致电网跳闸等停电事故,进而造成巨大的经济损失,因此,对输电线路中绝缘子故障诊断具有重要意义。通过对故障绝缘子图像的特征分析,研究了基于神经网络技术实现绝缘子多种故障的自动分析与识别方法。主要根据图片的特征找出特征向量,采用附加动量与自适应速率相结合的改进型BP算法建立神经网络,实验证明改进方法有较高的识别率。

The insulator is one of the key components of the high-voltage overhead transmission lines,lead weight,lightning,wind,snow and ice,dust pollution and other factors,which can easily lead to the insulator string out,broken,filthy failure.Through the analysis of the characteristics of the image of faulty insulators,this paper studied the automatic analysis and recognition method of a variety of faults of insulators based on neural network technology.This paper identified the feature vector using additional momentum based primarily on the characteristic of the picture and established a neural network by adopting improved BP algorithm of the combination of additional momentum and adaptive speed.Experiments show that the method has a higher recognition rate.