Neural Inverse Optimal Control of Single-Phase Induction Motors

One of the most used electrical machines in the industry and domestic applications are the Single-Phase Induction Motor (SPIM), due to its low cost and low-price regarding maintenance. In this paper the Neural Inverse Optimal Control (NIOC) based Recurrent High Order Neural Network (RHONN) identifier is developed to control the SPIM flux and mechanical speed. The proposed neural identifier is on-line trained using the Extended Kalman Filter (EKF) based algorithm, which helps to obtain adequate SPIM model even in the presence of disturbances. To synthesize the NIOC, a Control Lyapunov Function (CLF) is selected as a cost function to be optimized. To illustrate the effectiveness of the proposed control scheme, simulations results considering time-varying references tracking and robustness in presence of parameter variations are presented and compared with conventional controllers.