Method for Anomaly Detection in Distributed Photovoltaic Power Generation Based on Kernel Principal Component-OPTICS Clustering

Addressing the lack of professional monitoring in Distributed Photovoltaic Power Generation Systems (DPV), this study proposes a novel method for detecting anomalies in distributed photovoltaic power output. This method leverages the output correlation of adjacent photovoltaic stations and combines Kernel Principal Component Analysis (KPCA) with the OPTICS clustering algorithm to perform cluster partitioning and dimensionality reduction on the power output data of photovoltaic power stations. The specific steps are as follows: First, the OPTICS algorithm is used to partition the output data of multiple photovoltaic stations into clusters; then, Kernel Principal Component Analysis is applied to reduce the dimensionality of the clustered data. Based on this, an equal-weighted weighted average method is employed to plot the output curve that reflects the overall output condition of the cluster under different weather conditions. Finally, the output range is fitted using quantile regression, which serves as the basis for anomaly detection in DPV clusters.