Genetic Algorithm and Simulation for Selecting PID-Controllers Parameters in the Walking Robot

Workshop on Materials and Engineering in Aeronautics

材料科学;航空航天工程

Ryadchikov, I.V.^1 ; Gusev, A.A.^1 ; Sechenev, S.I.^1 ; Prutskiy, A.S.^1 ; Nikulchev, E.^2

Kuban State University, Krasnodar

350040, Russia^1

MIREA - Russian Technological University, Moscow

119454, Russia^2

关键词: Computational experiment;    Dynamic stabilization;    Genetic algorithm approach;    Inverted pendulum model;    Mechanical modelling;    Optimization problems;    SIMULINK environment;    Stabilization systems;   

Others  :  https://iopscience.iop.org/article/10.1088/1757-899X/476/1/012023/pdf
DOI  :  10.1088/1757-899X/476/1/012023

来源: IOP

【 摘 要 】

Walking robot dynamic stabilization is a growing field of research as such robots are being constructed for various applications in human environment. The dynamic stabilization systems for such robots are often highly non-linear with the additional demand for rapid change of model mass parameters to meet the design requirements. The article centres round the problem of mechanical modelling of a walking robot and identifying the robot dynamic stabilization system parameters using the genetic algorithm approach. The stabilization system supposed in the paper is based on inverted pendulum model with paired gyroscopic units with limited kinetic momenta, that shapes out the optimization problem. The proposed control system includes a cascade of two PID regulators, the optimal parameters of which are identified using the genetic algorithm approach. The authors give the short description of the approach and suggest the fitness function suitable for this application. The computational experiment is then conducted, and the optimal parameters set is obtained. The disadvantages of the genetic algorithm solver in Simulink environment are pointed out with the approaches to their partial elimination.