Genetic Simulated Annealing Algorithm Used for PID Parameters Optimization

A type of genetic simulated annealing algorithms (GSAAs) is presented, which is used to optimize the parameters of proportional-integral-derivative (PID) controllers. This approach combines the merits of genetic algorithms (GAs) and simulated annealing algorithms (SAAs). By integrating the global search ability of GA with the local search ability of SAA, the search ability of GSAA is much stronger than GA's and SAA's search ability. So, GSAA could find the global optimal solution of the given problem. Furthermore, the adaptive probability for crossover operator and nonuniform mutation operator is used in the GSAA, which can eliminate the phenomena of premature converge. Computer simulation on the speed control system of a kind of mobile robots is relized by Matlab. The results of computer simulation demonstrate that, comparing with the GA and SAA, the response speed of the PID controller can be improved due to the parameters produced from GSAA.

I. Introduction

There are many studies on using genetic algorithms (GAs) or simulated annealing algorithms (SAAs) or both to optimize the parameters of proportional-integral-derivative (PID) controllers. Immune genetic algorithm [1], which was proposed in 2007 by Li, is used to optimize the parameters of a PID controller, and the controller is used in a water level control system. An improved genetic algorithm [2], proposed by Wang, is used to optimize the parameters of a PID Decouping Controller, and this conrtoller is used in a variable flow heating system. Three different PID controllers based on a type of SAAs [3], proposed by Zhang, are used to deal with the constant disturbance of a SMT axial motion control system. Genetic algorithms [4] and simulated annealing algorithms [4] were respectively used to optimize the parameters of robot arm PID controllers by Kwok. An effective and hybrid optimization strategy [5], which integrates the parallel structure of GAs with the controllable jumping property of SAAs, was proposed by Wang, is used to estimate the parameters of models and set parameters for controllers.