基于统计参数的自适应网络推理系统在局部放电缺陷识别中的应用

Application of ANFIS to Partial Discharge Defect Recognition Based on Statistical Parameters

高压电力设备在发生绝缘劣化的早期,内部会出现局部放电现象,笔者依据检测得到的局放信号,提出了采用基于统计参数的自适应网络推理系统进行绝缘缺陷模式识别的方法。自适应网络推理系统是神经网络和模糊逻辑的结合,通过模糊逻辑进行识别系统建模,利用神经网络训练系统参数。设计并实验了4种绝缘缺陷模型,对多周期的局放信号进行相位分布及幅值分布统计,提取表征局放特性的统计参数,总结了不同缺陷模型局放特征的区别。实际的检测结果表明,经过训练后的局放缺陷识别系统,能够有效地对各种缺陷的样本数据进行分类,达到良好的识别效果。

Partial discharge（PD） phenomenon happens at the early stage of insulation degradation in power apparatus.To classify the insulation defect according to the acquired PD signals,an adaptive neuro-fuzzy inference system（ANFIS） based on statistical parameters are introduced to the PD pattern recognition.The structure of ANFIS is modeled by fuzzy logic and trained by neural network.Four types of insulation defect models are designed and tested,and statistical parameters are extracted from the tested PD signals.The difference of statistical parameters among defect models is also summarized.The verification results show that the PD pattern recognition system can effectively classify different kinds of insulation defects and reach high recognition rate.