A Fault Diagnosis Method of Insulator String Based on Infrared Image Feature Extraction and Probabilistic Neural Network

In view of the close relationship between insulator fault and surface temperature distribution, a detection method for insulator string is proposed combined infrared image segmentation and artificial neural network, which is based on the analysis of infrared image processing and fault diagnosis of artificial neural network. Firstly, the steel cap and the disk of insulator string are extracted according to their length characteristics, and then the temperature characteristics are calculated, finally, the fault diagnosis model of insulator string is established by probability neural network (PNN) according to the integral temperature distribution rule of a string insulator and the heating law of fault insulator. The K-means clustering method is introduced to eliminate the bad data and improve the accuracy of diagnosis when the temperature characteristics are calculated. The validity and accuracy of the proposed method are verified in the application of 500kV substation insulator string state detection.