Clustering and decision tree based analysis of typical operation modes of power systems

Abstract

The increasing penetration of renewable energy resources has greatly changed the pattern of the modern power system. In this case, the operation modes of power systems are becoming much more complex and the traditional experience-based method is no longer practical in typical operation modes analysis. In this paper, a clustering and decision tree-based scheme is proposed for the analysis of the typical operation modes of power systems. Specifically, the k-means++ clustering algorithm is adopted to classify the operation data into different groups, which represent the typical operation modes. The group labels and several important operation features are used to construct a decision tree for the quantitative description of different kinds of typical operation modes. In addition, feature importance is analyzed and the decision tree is pruned with a balance between complexity and classification accuracy. Case studies are conducted based on the actual system planning data of the Jiangsu power grid to verify the effectiveness of the proposed scheme. From the results, system operators could get deep insights into system planning with a high share of renewable energy resources.

Keywords

• Data-driven methods
• Operation features extraction
• Operation modes analysis
• Power system