A data-driven method for operation pattern analysis of the integrated energy microgrid

Abstract The variability of renewable energy generation and diverse load demands have led to diverse operation patterns of the integrated energy microgrid (IEM). However, there is a lack of systematic analysis of operation patterns from massive operational scenarios. Considering the uncertainty of the load and renewable energy, this paper proposes a data-driven method to identify the normal and extreme operation scenarios, then extracts all potential operation patterns of the IEM. Furthermore, the evaluation indices for the extracted operation patterns are given to quantify the economy, security, and energy-saving rate. The proposed method involves the normalization, kernel principal component analysis (KPCA), the enhanced K-means, and uniform manifold approximation and projection (UMAP) techniques. The effectiveness and superiority of the proposed method are verified by comparison with a conventional method using principal component analysis (PCA) and K-means algorithms under a modified industrial park testbed. The testbed combined with a 14-bus modified distribution power system and an 11-node heat system employs the simulation under isolated and grid-connected operating circumstances. The results show that the accuracy and effectiveness of the proposed method to identify extreme scenarios are better than that of the conventional method. In addition, the extracted operation patterns with no duplication in energy allocation and the performance of these patterns in economy, security, and energy-saving rate are demonstrated.