On-line optimisation of a fuzzy drive controller using genetic algorithm

This paper describes the application of genetic algorithms to the tuning of a fuzzy controller for a brushless dc motor drive. The fuzzy controller has two inputs, speed error and estimated load torque, and generates two current demand signals. These current demands are summed before being input to a proportional-integral current controller. The fuzzy controller membership function parameters are tuned on-line using the genetic algorithm to optimise the drive´s performance in the presence of changes in speed demand and load torque. Experimental results illustrate the efficiency of the technique and the impact of genetic algorithm parameters on the tuning process.