Power Facilities' Exception Event Monitoring Based On Image

With the continuous developments of economic technology and other various aspects in our country,we demand a bigger amount of electricity from all aspects in life.Therefore,it needs to detect accurately and timely for the electric power facilities'nomal operation.But because of the influencing factors such as the natural conditions and subiective conditions in recent years,electric power accidents of power facilities become abnormal events,which lead to the huge loss in the economic.As for this situation,this paper presents an abnormal event detection method based on image used to detect abnormal events occurred in power facilities.This article first analyzes the electric power facilities abnormal events(such as transmission lines ice cover and transformer or insulater man-made destruction,etc.)happening reasons,influence factors,etc.,which then mainly introduce the image preprocessing methods based on these abnormal events,which are used for edge feature extraction methods.At last the article introduces three aspects of the abnormal event detections:transmission lines ice detection and thickness measurement,transformer and insulater's edge detection,tower and transmission line next to the pedestrian detection.The power facilities'edge detection and number of statistics methods are mainly introduced in this paper.In the aspect of electric power facilities'edge detection,it introduces the traditional methods of edge detection operators,image threshold segmentations and modern intelligent control algorithms:including particle swarm optimization(PSO)and structured forest algorithm,etc.,which have carried on the experimental comparison and analysis to these methods.According to the edge detection results,the structured forest algorithm's features are conclused and improved to satisfy the demands in the complex image simulation.In the pedestrian statistics,it introduces the combined method of background modeling,Adaboost classifier and kalman improved filter,which is proposed at the same time to shoot video of the pedestrians statistics.According to experiment,the algorithm is practical which meets the needs of the actual system.