Adaptive Tracking Controller for Induction Motor Drives Using Online Training of Neural Networks

This paper explores means of controlling the dynamics of the stator currents of an induction motor. A neural-network-based identification and control scheme is presented. A single artificial neural network is trained to capture the nonlinear dynamics of the motor. A control law is derived using the dynamics captured by the network, and employed to force the stator currents to follow prescribed trajectories. The proposed architecture adapts and generalizes its learning to a wide variety of loads and, in addition provides the necessary abstraction when measurements are contaminated with noise. Extensive simulations reveal that neural designs are effective means of system identification and control for time-varying nonlinear systems, in the presence of uncertainty. The effects of parameter changes on the performance of the network is addressed. Particular emphasis is placed on the effects of sudden, random load torque changes. The difficulties addressed by this paper include incomplete system knowledge, nonlinearity, noise, and delays.