The Forecasting Method For Surface Filthy Of The Suspension-type Insulator Based On Insulator Leakage Current Measurement

Insulators are important insulation parts and frequent occurrences of pollution flashover cause enormous losses in the power system. In order to prevent the occurrence of flashover, insulator surface contamination conditions is particularly important to predict. Among the prediction methods, leakage current online monitoring method can be achieved, and some of its characteristic quantities can be extracted, and has been widely used in the experimental studies. When conduct tests for porcelain insulators and composite insulators in the artificial climate chamber, first select appropriate equipments to built experiments platform in the laboratory, then the measurement system of leakage current is developed based on Lab VIEW software. The system can display real-time leakage current waveform, extract and save some of leakage current characteristic quantities, and make analysis in time domain and frequency domain.Under the different humidity conditions, test the different contamination levels of insulators, mainly by the following conclusions:The discharge pulse number of insulator leakage current and leakage current RMS and the ratio between low harmonic wave such as three and five and the base wave as well as cumulative charges increase with ESDD increscent, this trend is more obvious especially in high humidity conditions. Each feature value has the same trend except in the case of low humidity, therefore, effects caused to the leakage current are different with two sides above, ESDD and NSDD should be distinguished when pollution is forecast. Through the comparison of composite insulator leakage current’s various stages of development in the time domain and frequency domain characteristics, we find that the high frequency components has an important guiding role in early warning of change in flashover. The intuitive understanding to changes of high-frequency components could be obtained through the analysis in the frequency domain. It has a significant inhibitory effect for the leakage current development because of surface hydrophobicity of the composite insulators.The prediction model of insulator surface ESDD and NSDD is established based on rough sets and BP neural network, the neural network hidden layer nodes can be simplified, then simplify the structure of BP neural network, and Improve the training speed. At last, verify the prediction model by artificial test samples. The results show that it is feasible for BP neural network model based on rough sets to predict the surface contamination of insulators.