Research On Insulator Fault Diagnosis Method Based On Leakage Current Characteristic Analysis

Although domestic and foreign scholars have studied the problem of denoising and feature extraction of leakage current,a method for online monitoring the state of transmission line insulators,and used pattern recognition method to achieve classification of leakage current and insulator fault diagnosis,but these methods are all flawed,and there are still many problems.In order to make up the deficiency of existing methods,aiming at the problem of the leakage current characteristics of insulator and fault diagnosis based on leakage current,the research of leakage current denoising and feature extraction is carried out in this paper.The current insulator fault diagnosis method is analyzed and studied,and deep learning RBF neural network is further studied.Finally,a fault diagnosis model based on RBF neural network is established.By analyzing the features of the leakage current signal,it is found that the leakage current is easily affected by many factors of the external environment.This paper mainly study the influence of relative humidity and equivalent salt concentration on the maximum leakage current so as to obtain the corresponding pollution degree of leakage current.In order to reduce the impact of signal noise on the measurement results,this paper firstly introduces the denoising method of leakage current,analyzes several common denoising method,then gets a wavelet transform method.At the same time of removing transmission line insulator leakage current noise,the wavelet transform method can not only determine the two values of the optimal wavelet basis and the best threshold,but also give the scientific calculation method of the decomposition level.To a certain extent,it can better denoise the leakage current signal.Then,the characteristics of insulator leakage current are studied in this paper,and the influence of relative humidity and equivalent salt deposit density on the maximum leakage current is analyzed respectively.Through the characteristics of leakage current feature analysis,the maximum leakage current of 3 pieces of XP-70 insulator string was obtained.With the increase of the equivalent salt deposit density,the maximum value of leakage current showed a trend of increasing linear relationship.However,with the increase of relative humidity,it showed a non-linear increase.The third harmonics and fundamental amplitudes of leakage currents at different humidities were also analyzed.With the change of RH,the ratio increased.The obtained relative humidity and equivalent salt deposit density are used as input variables of the RBF model.Finally,an insulator fault diagnosis model based on two-input and one-output of RBF neural network is established to predict the degree of contamination on the surface of the insulator.And compared with the simulation test results to analyze and test the model accuracy.It is of great significance to study insulator diagnosis for leakage current feature analysis.