Analysis On Leakage Current For Polluted Insulator Of Catenary

In China, the railway is an important national infrastructure, the popular traffic, as the backbone role in the comprehensive transportation system. The electrified railway is more and more proportion in the railway business, this is also the trend of the development of railway. Catenary as an important part of electrified railway, is responsible for providing electricity to the locomotive. Generally, catenary are exposed in the air, so the insulator as an important part of a catenary, is very easy to be polluted and lower insulating properties. In recent years, the national environmental pollution is more severe, the fog weather has been persistent in most regions, exacerbating the contamination degree of catenary insulators. It is easy to cause discharge in the insulator surface, polluted flashover occurs, even line tripping, great safety hidden trouble for the normal work of the electrified railway traction power supply system. Therefore, in-depth study of polluted in-depth study of discharge process along the surface of polluted insulator and its influence factors is necessary, through the insulator on-line monitoring system to monitor the running state of the insulator, and know the pollution laws of the insulator surface, in order to find potential factors of pollution flashover, then take appropriate measures of anti-fouling promptly, strengthen insulation management, prevent sudden fault, the maximum to avoid contamination flashover, is very important for electrified railway transport, provide technical support for the sustainable development of electrified railway.In this paper, the artificial contamination tests on composite insulator and porcelain insulator were carried out, using the insulator contamination on-line monitoring device of catenary to collect and stores the leakage current signal and the environment information, from the experiment phenomena and test data can be clearly seen the characteristics of insulator leakage current with the change of the contamination and humidity. Low humidity conditions, the leakage current is very small, and the variation characteristics with the pollution degree are not very obvious; high humidity conditions, the leakage current obviously increased with the increase of pollution degree, the harmonic content also increased. In addition, the anti fouling performance of composite insulator is better than porcelain insulator through the discharge phenomenon and leakage current data.In order to realize the project meaning of the insulator contamination on-line monitoring device of catenary, the locomotive impact was analysed,which is a large influence factor of leakage current in the railway site. When the locomotive passes through, the time domain and frequency domain characteristics of insulator leakage current will be greatly affected, according to the characteristic of the low frequency part (0～10Hz) increased, wavelet energy ratio method is improved, so this effecte on leakage current can be effectively identified, then this section datas were eliminated to eliminate the effect of the locomotive impact. Current sensor in the device can result in offset on the leakage current, this offset was analysised in this paper, by a large amount of data verification, the trend of EMD was used to effectively eliminate the offset. Two denoising methods of leakage current were expounded, namely the singular value decomposition (SVD) denoising and wavelet threshold denoising. From the denoised waveform, SVD denoising filtered texture excessively, smoothed too much noise, so that the signal missed some details; and wavelet threshold denoising is reasonable to filter out noise, the signal is smooth, but also kept the details of the information signal, therefore the wavelet threshold denoising method is more suitable for the leakage current than SVD denoising method.The time domain and frequency domain characteristics of insulator leakage current are analyzed. Three time domain reference volumes are proposed to characterize the surfacecontamination condition of insulator, namely the leakage current effective value, the maximum leakage current and the time domain entropy of leakage current. These five reference volumes and the insulator surface contamination degree showed exponential growth, fitted exponential curve fitting is above 0.95, meet the requirement of engineering application, so the five reference volumes can be used to characterize the surface contamination condition of insulator.Finally, support vector machine (SVM) classification method is proposed to identify the insulator contamination grades based on leakage current waveforms of insulator. The environmental humidity, the leakage current effective value, the maximum leakage current, the three harmonic amplitude and fundamental amplitude ratio of leakage current and the total harmonic distortion rate of leakage as the inputs of SVM model, the dirty four rank as the four kinds output of SVM model, and the Gauss radial basis kernel function is selected, through training sample and testing sample, the results of predictive models are given, it is found that the average error of the establishment of this SVM early-warning model can completely satisfy the practicalrequirements of engineering.