输电线路绝缘子放电模式的识别

Pattern Recognition of Transmission Line Insulator Discharge

为了正确判断输电线路绝缘子放电的类型以及放电的严重程度,在绝缘子放电声发射试验的基础上,首先针对放电产生的超声波信号(放电声发射),在时域状态下求出反应其波形特征的18个特征指标,利用主成分分析法,在这18个特征指标中提取4个互不相关的主成分的表达式,并解释了每个主成分所代表的含义,以降低指标维数,避免信息重叠为目的,提高放电模式识别的速度和准确性;其次,根据得到的特征指标,应用神经网络对放电模式进行识别。识别结果表明,利用放电声发射信号进行绝缘子放电模式的识别,可以有效地判断绝缘子放电的类型,该方法为电力设备的在线监测与故障诊断提供了一种新思路。

In order to diagnose the type and severity of insulator discharge fault on transmission line, 18 characteristic vectors which can reflect the features of discharge acoustic emission wave are selected and calculated in time domain, then, 4 principal components which are independent each other are extracted from the 18 characteristics through principal component analysis, including the meaning of each principal component is explained. This will reduce the target dimension and avoid duplication of information, and will improve the rate and the accuracy of identification for fault or exception. Secondly, according to the acquired characteristics, neural network is used for the recognition of discharge pattern. The results indicated that using the signals of discharge acoustic emission to recognize the pattern of insulators, could effectively judge the type of insulator discharge mode, this method provides a new thread for the on-line monitoring and fault diagnosing of power equipment.