A two-timescale neurodynamic algorithm for optimal power flow problem of radial networks

作     者：Li, Jin Lu, Jianquan Tang, Tianyu Lou, Jungang 

作者机构：Southeast Univ Sch Math Nanjing 210096 Peoples R China Huzhou Univ Yangtze Delta Reg Huzhou Inst Intelligent Transpor Huzhou 313000 Peoples R China 

出 版 物：《INFORMATION SCIENCES》 (Inf Sci)

年 卷 期：2025年第700卷

核心收录：

学科分类：12[管理学] 1201[管理学-管理科学与工程(可授管理学、工学学位)] 08[工学] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

基　　金：Natural Science Foundation of Jiangsu Province [BK20240009] National Natural Science Foundation of China Jiangsu Provincial Scientific Research Center of Applied Mathematics [BK20233002] 

主　　题：Neurodynamic optimization Optimal power flow Nonlinear equations Recurrent neural network Meta-heuristic algorithm 

摘      要：The issue of optimal power flow (OPF) is a basic and a critical challenge within the analysis of electrical power systems. In radial networks, the OPF problem is formulated as a branch flow model with nonlinear equations, which makes conventional approaches difficult to be applied for solving such OPF problem, including one-layer neurodynamic approach. In this article, we propose a two-timescale neurodynamic approach based on recurrent neural networks (RNNs) to deal with the OPF optimization problem. The optimality and convergence of this two-timescale neurodynamic model are rigorously proved, indicating that it can converge to a local optimum. Subsequently, a collaborative neurodynamic approach with meta-heuristic algorithm, which has been proven to almost certainly converge to global optimum, is used to perform the global optimization and seek a global optimum. Finally, the two-timescale neurodynamic approach based on RNNs is applied to a 6 bus power system for validation and comparison with other methods. Simulation results demonstrate that the proposed approach can effectively achieve both local and global optimization, thereby verifying its correctness and effectiveness.