局部放电灰度图象分维数的研究

STUDY ON FRACTAL DIMENSION OF PD GRAY INTENSITY IMAGE

局部放电模式识别被普遍认为是一种预测高电压设备绝缘状况的有效手段。本文提出一种适用于局部放电模式识别的局部放电分形特征提取方法。该方法在估计分维数的改进差盒计维数(MDBC)算法的基础上,提取局部放电灰度图象分维数和二阶广义分维数以及局部放电高值灰度图象分维数,共同构成局部放电模式识别特征。针对高电压设备内部局部放电和外部放电干扰,设计了五种放电模型,通过放电模型实验获得的大量放电样本数据,构造出相应的局部放电特征提取图象,计算出分形特征参数,输入人工神经网络进行识别的结果表明,采用该方法具有良好的识别效果。

Partial discharge(PD)pattern recognition is widely considered as an effective method to evaluate insulation condition of high voltage(HV)apparatuses.This paper brings forward a method to extract PD fractal features used for PD pattern recognition.On the base of fractal dimension evaluation with modified differential box-counting(MDBC) method,the fractal dimension (FD) and the 2nd order generalized dimension of PD gray intensity image and FD of the relevant high gray intensity image are extracted and then used as PD pattern features.Five discharge models are designed according to internal PD and external discharge interference of HV apparatuses.Large quantities of discharge samples are acquired with discharge model test and then the relevant PD pattern images are constructed.From those images the fractal features are computed and inputted into artificial neural networks for PD pattern recognition.The computation results show that the method proposed in this paper is effective for PD pattern recognition.