Research Of Intelligent Video Surveillance System In The Transfomer Substation

Unattended substation has become necessary trend of the electric power network dispatching automation development. So far,substation automation system has achieved 'remote-measure, remote-communicate, remote-control, remote-regulate',the four-remote function,but only rely on the four-remote function, can not fully meet the safety requirements of the unattended substation. This paper presents that intelligent video surveillance system will be used in the substation automation system,the 'remote view' will become of a powerful additional of the traditional four-remote function and a necessary facilities of the safe operating substations,and really reach the unattended substation operating.This paper studies two key technologies of the intelligent video surveillance system,moving objects detection and objects classification. Firstly, in order to achieving objects detection,based on analysis of existing algorithms for moving objects detection,the method on the basis of independent component analysis be adopted in the paper. This method understands monitoring images as mixed signals by background and moving objects,and separates them based on the statistical independence.Because of independent component analysis focuse only on the independence of the signals, and unrelated to the characteristics of signal itself, the algorithm can adapt to the changes in lighting at the scene, and has better robustness and real-time.Then,this paper adopt the algorithm of combination Hu invariant moment and BP neural networks. The method describes the shape characteristics of objects by Hu invariant moment,and classify the objects by them in the BP neural netwok.This algorithm has very good tolerance for the changes of size, location and angle and has a strong ability to learn, fault tolerance and fast computing power,makes the classification is accurate and real-time processing.Finally,this paper simulates the above two algorithms based on the Matlab. Experiment results show that,this method can better achieve the function of moving objects detection and objects classification in the substation and lay the foundation for achieving substation intelligent video surveillance.