Fault Detection Based On Corona Discharge In Ultraviolet Image

Under the long-term electric and thermal losses,surface cracks and pollution damage,the breakdown voltage of high-voltage equipment in power grids and railways generates surface corona discharge,which leads to a significant decline in insulation performance and a huge power loss.The ultraviolet light signal generated by corona discharge is an important parameter for detecting insulation degradation and corona discharge in time.The analysis of ultraviolet signal is of great significance for preventing accidents and maintaining equipment.This paper first analyzes the working principle of the dual optical path of the ultraviolet imager,and gives a detection method for solving the discharge detection by using the relevant parameters in the output image of the ultraviolet imager,aiming at the current two-channel ultraviolet imager based on the full blind ultraviolet imaging technology.In the discharge detection method,only one parameter of the photon is used to quantify the discharge intensity,and the problem of the outer insulation and the occurrence of the flashover margin cannot be effectively judged.Therefore,an ultraviolet imaging method that uses ultraviolet imaging technology to select available characteristic parameters and assists in quantifying discharge intensity is given.In this paper,the grayscale histogram is obtained by graying out the conventional image,and the threshold of the gray histogram is selected based on the bimodal method to obtain the ideal threshold,and then the ideal binarized spot image is obtained.In this paper,the direct opening operation is used to mathematically filter out the noise caused by the sky,the white background of the device,and the framing frame and time.This method can filter the noise points far smaller than the construction elements,thus making the image smoother.Finally,the genetic algorithm-based back propagation propagation(Genetic Algorithm-Back Propagatio)neural network model is used to study the relevant parameters in the ultraviolet image of high-voltage equipment.The size of the spot area under different gains at different distances is calculated,and the predicted spot area of the model is solved.The problem of large error with the actual spot area provides a basis for fault diagnosis.The experiment proves that the condition of the model equipment is consistent with the actual equipment,which proves that the gray histogram analysis is used to determine the threshold and segment the image.Then use the open operation to perform digital morphological filtering,and finally calculate the area of the corona discharge spot and perform neural network prediction.It has an important role in the detection of corona discharge faults.