Identification of faulty insulators by using corona discharge analysis based on artificial neural network

It was found from the tests in our laboratory that statistic analysis of insulator corona current pulses could identify the existence of faulty insulators on strings or transmission line towers. Some basic researches on corona discharge characteristics of insulators and insulator strings were investigated further in terms of polarity, amplitude, and pulse count of corona current pulses. Insulator corona characteristics are observed to be different from piece to piece. Therefore, a BP network was developed to classify voltage across an insulator and further identify the existence of faulty insulators on towers. It is confirmed that the developed BP network method yields satisfactory performance in terms of training time and quality of classification