A Study on Anomaly Detection in Massive Multivariate New Energy Data Based on Isolation Forest

Abnormal detection in massive new energy data is an important research area in new energy power generation. Accurate detection of abnormal values can significantly improve the reliability and maintenance efficiency of equipment operation. This study proposes an anomaly detection method based on the Isolation Forest algorithm, which combines adjacent equipment, associated equipment, and related data within the power station to detect data validity and abnormal points. Through experimental verification of equipment operation data, the Isolation Forest algorithm shows excellent accuracy and efficiency in large-scale data, and can correctly detect abnormal points in massive and diverse data. In addition, this article also studies the anomaly detection system. In response to the problem of real-time detection of massive and diverse new energy data, it proposes a model optimization combined with cost function improvement and an anomaly feature recognition strategy based on CUDA, which improves the real-time and accuracy of anomaly detection in massive and diverse data scenarios.