Online Distribution Network Reconfiguration Using Incremental Reinforcement Learning

View at publisher Summarise with VIRAT
Authors: Quan Zhang, Jiajie Ling, Zhichao Pan, Yuzhong Gong, Guangchao Geng, Jiang Quanyuan
Cited by: 0
References: 14

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Online network reconfiguration (NR) is critical to improve the operational quality of active distribution networks (ADN). This paper proposes an online NR method based on incremental reinforcement learning (IRL) to overcome the bottlenecks of RL in exploration efficiency and generalization for long-term deployment. Firstly, as the ADN operates, a mixed-integer second-order cone programming (MISOCP) solver solves the NR model in a near-online manner, continuously providing IRL with feasible candidate topologies as the state and action spaces. Subsequently, when new topologies are detected, IRL expands its state and action spaces while performing incremental training within recent power flow scenarios to adapt to changes in ADN operating conditions. Case studies in modified IEEE 33-bus system demonstrate that the proposed method outperforms benchmarks in terms of timeliness, adaptability and optimality.

Fields of Study & Subjects

Optimal Power Flow Distribution Microgrid Control and Optimization Smart Grid and Power Systems

Publication details

Journal / source
IEEE PES International Meeting (PES IM)
Pages
1–5
Publication date
18 January 2026
Language
EN