A two-stage power system frequency security multi-level early warning model with DS evidence theory as a combination strategy. (December 2022)

• Record Type: Journal Article
• Authors: Wu, Junyong; Li, Lusu; Shi, Fashun; Zhao, Pengjie; Li, Baoqin
• Is Part Of: International journal of electrical power & energy systems. Volume 143(2022)
• Publication Date: 2022-12
• Subjects: Frequency safety early warning -- 1D-CNN -- Integrated learning -- DS evidence theory -- Safety margin and risk degree; Electrical engineering -- Periodicals; Electric power systems -- Periodicals; Électrotechnique -- Périodiques; Réseaux électriques (Énergie) -- Périodiques; Electric power systems; Electrical engineering; Periodicals; 621.3
• DOI: 10.1016/j.ijepes.2022.108372
• Languages: English
• ISSNs: 0142-0615

Abstract

Abstract: With the goals of carbon emission reduction and carbon neutralization put forward by countries all over the world, renewable energy clusters are connected to the grid on a large scale, which leads to the problem of frequency security of the power system again. Therefore, this paper proposes a two-stage power system frequency security multi-level early warning model (FSMEWM) with DS evidence theory as a new combination strategy. The model consists of two stages: frequency security multi-level early warning and frequency security margin and risk degree prediction. In the first stage, three 1D-CNN with different structures are selected as sub-classifiers, and DS evidence theory is used as the combined strategy to integrate the results of sub-classifiers, which can evaluate whether the frequency of the system after disturbance will exceed the safety early warning limit. The second stage consists of three regression predictors, which can further give the safety margin and risk degree of early warning samples according to the early warning results of the first stage, to provide a reference basis for whether to start emergency control and what control strategy to choose. Finally, this paper takes the improved IEEE 10 machine 39 bus system as a simulation example to verify the effectiveness and computational efficiency of the model. It also shows that when taking DS evidence theory as the combination strategy, ensemble learning has better performance in early warning accuracy, early warning stability, robustness, and anti-noise ability. Highlights: This paper proposes and constructs a two-stage frequency security multi-level early warning model (FSMEWM) for power systems. The model can use time-series data sampled from the power system to give more refined multi-level frequency security early warning results. The two-stage FSMEWM can give the frequency safety multi-level early warning level in the first stage, and the safety margin and risk degree in the second stage. It can provide a comprehensive reference for the subsequent emergency control. Based on the ensemble learning, this paper introduces DS evidence theory as a new combination strategy for the first time. Compared with the common method, DS evidence theory has better performance in accuracy, robustness, and anti-noise ability.