Design Of Substation Intelligent Monitoring System Based On Human Behavior Recognition

With the rapid development of China’s electric power industry,the demand for the safety of substations is getting higher and higher.In the actual production and operation,the safety protection of the substation is the top priority of the power grid company’s management measures.The substation undertakes the mission of power supply and transmission in the operation of the power grid.When the substation is out of operation due to safety problems,it will bring unpredictable risks to the safety of people’s lives and property.A large part of the safety problems of the substation are caused by the non-standard actions of human beings,such as the irrelevant personnel crossing the fence of the substation and the staff not operating in accordance with the safety regulations on the site.The existing substation monitoring systems all transmit the video from the camera to the security personnel,and the human brain can judge whether the human movement will bring security risks.This way wastes manpower and material resources,so a substation intelligent monitoring system that can automatically analyze the human body’s behavior is very necessary.The principle of substation intelligent monitoring system is based on human behavior recognition algorithm.Human behavior recognition algorithm is an important subject in the field of computer vision.Human behavior recognition algorithm has been widely used in rescue and disaster relief,man-machine interconnection,medical and sports research and criminal behavior early warning.Human behavior recognition algorithms can be roughly divided into two categories: one is the classical recognition algorithm based on spatial feature extraction and SVM classifier,and the other is the dual-flow network algorithm based on deep neural network.The human behavior recognition of the classical recognition algorithm is divided into three stages,which are: the target segmentation stage,the feature extraction stage and the classification recognition stage.At the stage of object segmentation,the basic principles of interframe difference method,background subtraction method and background modeling method are described,and the image is processed by morphological filtering.In the feature extraction stage,the feature extracted in this paper is HOG feature,which is a common method to describe the target feature in computer vision.Finally,the HOG feature is input into the SVM classifier.The two-stream network algorithm can extract spatiotemporal features from complex scenes.In this paper,the combined network model of PCCNN and bi-lstm is used to identify the human body in the substation.In this paper,two kinds of different human body recognition algorithms are implemented on the software,and the recognition accuracy and calculation efficiency of the two kinds of algorithms in the identification of the substation human body movement target are compared,so as to select a better calculation scheme.After the realization of the human behavior recognition algorithm,the warning of dangerous human movements is carried out.After realizing the basic algorithm of substation intelligent monitoring,the substation intelligent monitoring system is integrated into the secondary equipment of the substation to realize the intelligent monitoring function of the substation on the hardware.