Tuning fuzzy PD and PI controllers using reinforcement learning

• Published in: ISA transactions Vol. 49; no. 4; pp. 543 - 551
• Main Authors: Boubertakh, Hamid, Tadjine, Mohamed, Glorennec, Pierre-Yves, Labiod, Salim
• Format: Journal Article
• Language: English
• Published: Kidlington Elsevier Ltd 01.10.2010; Elsevier
• Subjects: Aircraft; Algorithms; Applied sciences; Artificial Intelligence; Classical PID controllers; Computer science; control theory; systems; Computer Simulation; Control system synthesis; Control theory. Systems; Controllers; Equipment Design; Exact sciences and technology; Fuzzy; Fuzzy Logic; Fuzzy PID controllers; Fuzzy set theory; Industry - instrumentation; Mathematical analysis; Polyimide resins; Reinforcement; Reinforcement (Psychology); Reinforcement learning; Takagi–Sugeno fuzzy systems; Tuning; Tuning Fuzzy PID controllers; Fuzzy PID controllers; Tuning Fuzzy PID controllers; Takagi–Sugeno fuzzy systems; Classical PID controllers; Reinforcement learning; Integral proportional control; Membership function; Takagi-Sugeno fuzzy systems; Control synthesis; PD control; Two input two output system; Fuzzy logic; Fuzzy control; Self regulation; SISO system; Learning algorithm; Differential integral proportional control; Intelligent control
• ISSN: 0019-0578; 1879-2022; 1879-2022
• DOI: 10.1016/j.isatra.2010.05.005

In this paper, we propose a new auto-tuning fuzzy PD and PI controllers using reinforcement Q -learning (QL) algorithm for SISO (single-input single-output) and TITO (two-input two-output) systems. We first, investigate the design parameters and settings of a typical class of Fuzzy PD (FPD) and Fuzzy PI (FPI) controllers: zero-order Takagi–Sugeno controllers with equidistant triangular membership functions for inputs, equidistant singleton membership functions for output, Larsen’s implication method, and average sum defuzzification method. Secondly, the analytical structures of these typical fuzzy PD and PI controllers are compared to their classical counterpart PD and PI controllers. Finally, the effectiveness of the proposed method is proven through simulation examples.