Intelligent identification method of power grid monitoring alarm events based on knowledge base and large language model

Power grid monitoring alarm event refers to the abnormal state notification triggered by monitoring signals during the operation of various devices, networks or business systems in the power system. These alarm events are not only huge in number, but also diverse and complex in correlation. It is difficult to effectively deal with complex and changeable alarm scenarios when evaluating the risk of alarm signals, resulting in low stability and low accuracy of intelligent identification of alarm events. Therefore, this paper proposes an intelligent identification method of power grid monitoring alarm events that integrates knowledge base and large language model. By dividing multi-attribute labels, transforming monitoring alarm data into binary network and calculating feature set, the construction process of knowledge base is designed. Using the natural language processing ability of large-scale language model combined with structured knowledge base data, the structured information retrieved by combining knowledge base is deeply understood and reasoned. On this basis, the fusion process of knowledge base and large language model is designed, and finally the intelligent identification of power grid monitoring alarm events is realized. The experimental results show that the defect risk level of this method is low, and the effect is good when evaluating the defect risk of alarm signals, and the stability keeps rising, and the alarm accuracy is high. In practical application, it can realize the abnormal and accurate alarm of power grid monitoring alarm event information.