Power system dynamic security region algorithm based on LightGBM and improved SVM

• Published in: 2021 International Conference on Power System Technology (POWERCON) pp. 2027 - 2032
• Main Authors: Chen, Tao, Zhao, Ke, Shi, Chenggang, Wan, Zhaohui, Jiang, Zhenyu, Hu, Wei, Ma, Kun
• Format: Conference Proceeding
• Language: English
• Published: IEEE 08.12.2021
• Subjects: big data; Costs; dynamic security region; Feature extraction; Heuristic algorithms; LightGBM; Power system dynamics; Power system stability; Stability analysis; Support vector machines; SVM
• ISSN: 2642-6226
• DOI: 10.1109/POWERCON53785.2021.9697627

Modern power system is featured with high dimensionality of operating data and redundant features. The traditional calculation methods of power system transient stability analysis and dynamic security region face obvious bottlenecks. However, the direct application of artificial intelligence algorithms and big data processing technology will encounter the problems of dimensional explosion and low calculation accuracy. This paper proposes a method for obtaining dynamic security region of power system based on LightGBM algorithm and improved SVM algorithm. The method first uses LightGBM and loop elimination algorithm to achieve key feature subset. Then the SVM algorithm is used to give the analytical expression of the dynamic security region. The improved SVM algorithm considering unbalanced cost in classification is adopted to satisfy the conservative characteristics of power system safety and stability analysis. The CEPRI 36 node system example is used to verify the effectiveness of the model and algorithm.