Abstract:In order to efficiently and accurately identify potential risk points in a large and complex network of power communication networks, analyze the impact of power communication from the aspects of network structure and bearer services, and combine the assessment criteria of power accident incident investigation procedures, objectively and scientifically evaluate the power safety incidents caused by power communication and provides decision support for risk management and control of power communication. Based on the content and characteristics of risk management and control, analyzes the causes of accidents in power communication networks, infers the process of accident evolution, proposes solutions and solutions, verify model with high computational efficiency and accuracy through actual verification, that provide scientific basis for accurately locating the causes of accidents and evaluating risk levels. The model that combines the invulnerability, survivability and effectiveness of reliability factors of complex network theory calculate the risk factor values of related characteristics. The path coefficient analysis technique is used to verify the completeness of risk factors; The CNN model of deep convolution neural network is constructed based on the labeled sample data of risk level, it assess risk Level of accidents in a real-time manner.