Adaptive Neuro-Fuzzy Control with Fuzzy Supervisory Learning Algorithm for Speed Regulation of 4-Switch Inverter Brushless DC Machines

Principle of a new adaptive neuro-fuzzy inference system (ANFIS) with supervisory learning algorithm is introduced and is used to regulate the speed of a four-switch, three-phase inverter (FSTPI) brushless DC (BLDC) drive. The proposed algorithm has advantages of neural and fuzzy networks. To enhance of drive´s performance, instead of well-known back propagation learning method, a fuzzy based supervisory learning algorithm is used. This newly developed design leads to a controller with minimum structure and improved dynamic performance. System implementation is relatively easy since it has minimum fuzzy rules and membership functions as compared with the conventional fuzzy and/or neural networks, used for electrical drive applications. In order to demonstrate the proposed ANFIS controller abilities to follow the reference speed and to reject disturbances, its performance is simulated and compared with that of a conventional PI controller