Design of an adaptive output recurrent cerebellar model articulation controller for direct torque control system

Abstract

This study aims to design an adaptive output recurrent cerebellar model articulation controller (AORCMAC), which is embedded into the direct torque control (DTC) system of an induction motor as the speed controller. Similar to the conventional cerebellar model articulation controller (CMAC), the designed AORCMAC also has the advantages of rapid learning, simple architecture, online training, and nonlinear learning abilities. In addition, by incorporating the Gaussian function and recursion, the AORCMAC provides satisfactory dynamic response. This study compares the AORCMAC with the adaptive fuzzy CMAC (AFCMAC) and uses the root mean square error as the indicator for performance assessment. The experiment results verify that the proposed AORCMAC has rapid speed response, and its performance is superior to that of the AFCMAC. The AORCMAC maintains excellent robustness despite changes to the motor parameters and the addition of external load disturbances.