INTELLIGENT CONTROL OF INDUCTION MOTOR USING RECURRENT FUZZY NEURAL NETWORKS

Abstract

Induction motors play an important and indispensable role in electro-mechanical transmission in the industry. However, current common controllers with fixed parameters have proved less flexible to adapt to harsh industrial conditions. This study proposes a solution using recurrent fuzzy neural networks (RFNNs) to overcome that limitation. Accordingly, a PID controller is combined with a supervisory controller using the RFNN to adjust system responses. Simulation results show that, with the same parameters, when the PID controller runs independently, it has given a high overshoot response. However, when combined with the RFNN – based supervisory controller, the overshoot of system response is eliminated. The experimental results show that by the online training algorithm, the RFNN-based system identifier and the RFNN-based supervisory controller have monitored and rapidly adapted to the changes of the system such as noise affecting or load changing, thereby, they can adjust the control signal is more suitable, overcoming the limitation of fixed parameters of the traditional PID controller.