This study utilizes background monitoring and alarm signals from Emergency Medical Services (EMS) as data sources, focusing on equipment such as switch knife switches, line endpoints, transformers, and other key components. The research targets periodic monitoring data for power grid regulation systems, with a focus on detecting abnormal data that exceeds predefined limits. Such abnormalities often occur within specific periods for the same type of equipment. To monitor the operational status of the grid, data is collected every minute. Given the unique characteristics of power data, relevant features are extracted and mathematical models are developed using clustering, classification, and other techniques. The performance of these models in predicting abnormalities is then evaluated and compared. By periodically analyzing the collected statistical data, potential issues can be identified in advance. Based on this analysis, early warnings are generated for equipment that is likely to experience abnormal behavior in the future. These warnings include automated summaries detailing the equipment’s name, time of occurrence, affected components, descriptions of the anomalies, and the probability of future abnormalities.