A Security Monitoring Method for Distributed Photovoltaic Systems Based on Log Association and Fusion Analysis Technology

Traditional log analysis relies heavily on predefined rules, failing to detect unknown attacks promptly. Integrating log analysis with machine learning improves accuracy and the ability to detect complex attacks. This paper focuses on the log data of distributed photovoltaic (PV) systems, combining clustering algorithms and association analysis to enhance security monitoring techniques. Key contributions include: Proposing a parallel improved Apriori algorithm based on hash storage and transaction weighting to enhance association rule mining performance. This algorithm compresses the dataset by deduplication, reduces the number of database scans, and utilizes MapReduce for parallel computation, significantly improving efficiency. Developing a complete solution for anomaly detection in distributed PV systems using association rule mining. This solution addresses the low efficiency of handling massive log data through data collection, feature selection, and model training and detection, establishing a normal behavior rule base for anomaly detection.The proposed methods effectively improve the accuracy and efficiency of anomaly detection in distributed PV systems using log association fusion analysis technology.