Tuning of a PI Speed Regulator for Electric Drives by Using Particle Swarm Optimization Algorithm

This paper presents an investigation on the use of the Particle Swarm Optimization (PSO) algorithm on the tuning of a classical PI speed controlled DC motor drive. The motor drive system has two control loops, an inner one for the armature current and an outer one for the speed. The tuning of the current controller is kept unchanged and the PSO is used to tune only the speed regulator. The integral of the absolute speed error is used as the fitness criteria, therefore, the faster the speed response meets the speed demand, the better the particle or individual that represents the tuning. The searching space which represents possible values of the proportional and integral controller gains was limited to keep the system linear. Then, one may know in advance which tuning is the best and the algorithm is put to the test before the changing of some important parameters, such as inertia weight and cognitive and social coefficients. To make the task a little more difficult, the drive system was made non-linear by adding a limit to the controller's output and the PSO was used to do the same job with same parameter variation within the algorithm. Simulation results are presented showing the capability of the PSO algorithm to quickly find the best tuning for the speed controller, representing an important tool for optimization problems.