A Research Of Ultraviolet Pulse Detection Of Faulty Insulators

Currently, porcelain insulators are still the most commonly used insulators in overhead transmission lines. It will threat the safety of power grid if insulator string contains faulty insulator or even zero resistance insulator. In order to detect faulty insulators in overhead transmission lines, and to improve work efficiency of maintenance crew, this paper investigates UV pulse method in detecting faulty insulators. With the UV Corona Detector this paper presents, maintenance personals will be able to detect faulty insulators conveniently and precisely in overhead transmission lines.The UV Corona Detector employs UVtron sensor, which works within the solar blind zone, so it can avoid the interference of sunlight. After discussing how the UV sensor works, this paper presents the method of improving the driving circuit. Following testing shows that it can improve sensor’s sensitivity. With sapphire lens, the UV Corona Detector can improve detection distance greatly. Based on embedded microprocessor core development board, this paper implements the peripheral circuits, and integrates them with other circuits into the UV Corona Detector. Laboratory detection system is set up to collect UV pulse signal of normal insulators and faulty insulators. Leakage current detection circuits and others are also designed.Based on embedded Linux, this paper develops the data collecting software for the UV Corona Detector. Each function is divided into a module. And the working state switching is managed by FSM(finite state machine), which can improve the software reliability. The structure of Linux character driver and the thought of data collection are explained. And a Linux character driver specialized for data collection is developed. So the data collection promptness is improved.This paper analyses the statistic characteristics of UV pulse data, and the results are confirmed by fellows. From the fluctuation of the UV pulse, this paper guesses that the frequency characteristics of normal insulators and faulty insulators are different. This paper compares the FFT of normal insulator and faulty insulator, and uses FFT results as feature vector to carry out binary classification with SVM. Then, this paper tries to do SVM binary classification using periodogram method PSD and Welch method PSD as feature extraction method. After comparing of all the classification results, this paper concludes that using Welch method as feature extraction method has the best result.