Research On Video Intelligent Analysis Algoritihm For Insulator Flashover

Insulator flashover refers to an electrical discharge along the surface of the solid insulator, when the air or liquid around the solid insulator is breakdown, which is an important factor affecting the safety of power transmission equipment. In order to analyze the characteristics of insulators flashover,manual recognition of flash images based on the video of insulator flashover experiment is widely used. Manual recognition of flash images is a huge workload, inefficient and unreliable.The application of the intelligent analysis algorithm based on machine vision technology to the detection and classification of insulator flashover is of great significance to the study of insulator flashover characteristics.Aiming at the characteristics of the flashover image, the intelligent analysis algorithm based on machine vision is applied to detect and classify the insulator flashover image and the software for detecting and analyzing flashover image is developed. The main work of this thesis is as follows:(1) The segmentation algorithm suitable for insulator flashover is studied. According to the characteristics that the difference between the flashover and background is obvious, the Otsu threshold segmentation algorithm is adopted, and the scheme based on problems in segmentation is improved. Image enhancement based on visual saliency is adopted,Canny edge detection operator for image edge and the internal filling is used, and the filling result and Otsu segmentation are fused. Finally, the fusion results is morphological processed.(2) The parameter optimization of SVM classifier is studied.Optimize the parameters by combining the grid search method and genetic algorithm, reduce the training time reduce, and avoid the situation when blind selection of initial population leads to local extremum.(3) The insulator flashover images are classified, and the classification results are evaluated by confusion matrix. The insulator flashover video analysis system is developed to detect and to classify different flashover types.