Adaptive Modeling and Anti-error Correction Algorithm for Abnormal Behavior of Power Load Based on Reinforcement Learning

Authors

  • Liang Xiao China Southern Power Dispatching and Control Center, Guangzhou 510700, Guangdong, China
  • Guanghu Xu China Southern Power Dispatching and Control Center, Guangzhou 510700, Guangdong, China
  • Shijin Xu China Southern Power Dispatching and Control Center, Guangzhou 510700, Guangdong, China
  • Quan Yuan China Southern Power Dispatching and Control Center, Guangzhou 510700, Guangdong, China
  • Xiaobo Zhang China Southern Power Grid Artificial Intelligence Technology Co., Ltd, Guangzhou 510700, Guangdong, China

Abstract

In regard to the high uncertainty and error correction risk problems in the identification of abnormal behavior of power load, this paper proposes an adaptive modeling and anti-error correction algorithm based on reinforcement learning. In the adaptive modeling stage, a Markov decision process with random perturbations is constructed; multi-scale state representation is integrated; and the load fluctuation characteristics are dynamically captured. In the anti-error correction stage, a hierarchical safety constraint action space and an interpretable reward function are designed, and the attention-driven strategy network is combined to focus on key features. A dynamic safety barrier module based on load fluctuation entropy is applied
to achieve risk perception and real-time interception of high-risk correction. Experiments show that the retention rate of the proposed model reaches 0.87, and the correction success rate is 92% under a noise intensity of 20db, which is up to 22% higher than the baseline model.

Keywords: reinforcement learning; power load anomaly identification; adaptive modeling; anti-error correction; multi-scale state representation

Cite As

L. Xiao, G. Xu, S. Xu, Q. Yuan, X. Zhang, "Adaptive Modeling and Anti-error Correction Algorithm for Abnormal Behavior of Power Load Based on Reinforcement Learning", Engineering Intelligent Systems, vol. 34 no. 4, pp. 455-468, 2026.

Downloads

Published

2026-07-01

Issue

Vol. 34 No. 4 (2026): International Journal of Engineering Intelligent Systems

Section

Article

License

The submission of a paper implies that, if accepted for publication, it will not be published elsewhere in the same form, in any language, without the prior consent of the publisher. Before publication, authors are requested to assign copyright to CRL Publishing Ltd. This allows CRL to sanction photocopying, and to authorize the reprinting of issues or volumes according to demand. Authors' traditional rights will not be jeopardized by assigning Copyright in this way, as they retain the right to reuse the material following publication, and to veto third-party publication.