Power Information System Alarm Noise Reduction Technology Based on Graph Theory and Information Entropy

In order to deal with the redundancy, irrelevance and repetition of massive alarm information in power information system, this paper proposes an alarm noise reduction technology based on graph theory clustering and information entropy. Based on the topological structure of the device and the correlation between the alarms, the node-edge model is established. The CH index (Calinski-Harabasz) method is used to find the appropriate number of clusters to cluster the alarm information. The information entropy method is used to optimize and redundancy. Finally, the average output alarm of each system accounts for 36.63% of the original alarm. The proposed method can effectively mine the similarity and difference of alarm information, automatically and dynamically classify alarms, reduce misjudgment, help network administrators of power grid enterprises to screen out core alarm information, and improve management efficiency and quality.