The control of an inverted pendulum system based on simple and adaptive fuzzy regulators

8th International Conference on Advanced Concepts in Mechanical Engineering

Gilc, G.^1 ; Ionescu, M.^1 ; Bizdoac, N.G.^2 ; Diaconu, I.^2

Industrial and Automation Engineering Department, Faculty of Technical, Medical and Behavioral Sciences, Constantin Brancui University of Targu-Jiu, Targu-Jiu, Romania^1

Mechatronics and Robotics Department, Faculty of Automation, Computers and Electronics, University of Craiova, Craiova, Romania^2

关键词: Adaptive fuzzy regulator;    Adaptive technique;    Center of gravity method;    Inverted pendulum;    Inverted pendulum system;    Linguistic values;    Linguistic variable;    Programming environment;   

Others  :  https://iopscience.iop.org/article/10.1088/1757-899X/444/4/042019/pdf
DOI  :  10.1088/1757-899X/444/4/042019

来源: IOP

【 摘 要 】

This paper aims to implement 2 applications for the regulation of the inverted pendulum in the Matlab/Simulink programming environment, the first with the simple fuzzy regulator and the second with the adaptive fuzzy regulator. Both applications use at the input 2 sizes: the regulation error and the variation of the regulation error, resulting in a quasi-PD fuzzy regulator with dynamics. For the design of fuzzy regulators, 3 linguistic variables were considered: error, error variation (error derivation) and force (command). The linguistic values attached to the three linguistic variables are: high negative, small negative, zero, positive low, high positive. For the theta angles that are made by the inverted pendulum in its motion, we will have positive for the movement to the right of the vertical axis and negative for the movement to the left of the vertical axis. The rules base for fuzzy regulators contains 25 rules made up of combinations of the five linguistic values. The fuzzy regulators are Mamdani type, and defuzzification is done by the center of gravity method. The analysis of the two control systems reveals the superiority of the adaptive techniques, since they not only compensate for the nonlinearities present in the process, but also have the extra capacity of learning from the unfolded events.