HYBRID PARTICLE SWARM OPTIMIZATION AND RECURSIVE LEAST SQUARE ESTIMATION BASED ANFIS MULTIOUTPUT FOR BLDC MOTOR SPEED CONTROLLER

Brushless Direct Current (BLDC) motor speed control has been widely developed to obtain high performance in its operation. However, most of the controllers still used conventional controllers that have some drawbacks whenever operated for the different BLDC motor. This paper proposes BLDC speed controller by implementing multioutput Adaptive Neuro Fuzzy Inference System (ANFIS). ANFIS algorithm is able to control the speed of the BLDC motor according to the desired reference value. The average of steady state error achieved using ANFIS is 0.1% and the rise time is 2.7437 s when the reference speed is 4000 rpm. ANFIS learning process uses hybrid Particle Swarm Optimization (PSO) and Recursive Least Square Estimation (RLSE) methods supervised by Fuzzy-PID. PSO and RLSE can train the multi-output ANFIS data very well. The best training data is achieved when the value of lambda is 1 with RMSE error of 0.05364. The execution time of ANFIS algorithm on microcontroller is 96 mu s.