Load–frequency control: a GA-based multi-agent reinforcement learning

• Published in: IET generation, transmission & distribution Vol. 4; no. 1; pp. 13 - 26
• Main Authors: Daneshfar, F., Bevrani, H.
• Format: Journal Article
• Language: English
• Published: Stevenage Institution of Engineering and Technology 01.01.2010; The Institution of Engineering & Technology
• Subjects: Applied sciences; Control systems; Controllers; Dynamical systems; Electrical engineering. Electrical power engineering; Electrical machines; Electrical power engineering; Exact sciences and technology; Flexibility; Learning; Miscellaneous; Multiagent systems; Nonlinear dynamics; Power electronics, power supplies; Power networks and lines; Regulation and control; Reinforcement; Performance evaluation; Parallel algorithm; Integral proportional control; Parameter estimation; Control system; Reinforcement learning; Optimization method; Control synthesis; Load frequency control; Power electronics; Frequency estimation; Flexibility; Non linear phenomenon; Linear model; Operating conditions; Multiagent system; Electrical network; Genetic algorithm; Power control; Problem solving; Parallel computation; Non linear effect
• ISSN: 1751-8687; 1751-8695
• DOI: 10.1049/iet-gtd.2009.0168

The load-frequency control (LFC) problem has been one of the major subjects in a power system. In practice, LFC systems use proportional-integral controllers. However, since these controllers are designed using a linear model, the non-linearities of the system are not accounted for and they are incapable of gaining good dynamical performance for a wide range of operating conditions in a multi-area power system. A strategy for solving this problem, because of the distributed nature of a multi-area power system is presented by using a multi-agent reinforcement learning approach. It consists of two agents in each power area; the estimator agent provides the area control error signal, based on the frequency bias, and estimation and the controller agent uses reinforcement learning to control the power system, in which genetic algorithm optimisation is used to tune its parameters. This method does not depend on any knowledge of the system and it admits considerable flexibility in defining the control objective.