Abstract:Currently, power grid enterprises have accumulated a large amount of user behavioral data in their operations, and how to effectively monitor and analyze the user''s operational behavior in complex business systems is a crucial issue for guaranteeing the safety and high-quality operation of power grids. In this paper, we propose a user abnormal behavior detection method based on graph vector computation. The method constructs a normal business process operation flow chart by analyzing a general business operation manual. The behavioral links are constructed according to the time sequence for the user temporal behavior data, and the complex user temporal behavior is converted into a content semantic graph. The flowchart and timing diagram are then converted to numerical vectors using graph vectorization techniques. Then, the traditional support vector machine (SVM) model is used to implement the graph vector matching computation of user behaviors to discover the similarities and differences between the content semantic graph of user behaviors and the operational flowchart of business processes. Finally, in-depth analysis and mining of business anomalous behaviors in different domains are carried out to optimize the detection results and improve the accuracy and efficiency of detection. The experimental results show that the method can effectively identify and analyze the abnormal behaviors of grid users in different business domains, which is of great significance for enhancing the security and stability of the grid system..