Self-tuning adaptive control of multi-input, multi-output nonlinear systems using multilayer recurrent neural networks with application to synchronous power generators

A multilayer recurrent neural network-based approach for the identification and self-tuning adaptive control of multi-input multi-output nonlinear dynamical systems is developed. An efficient online implementation of the control strategy, by a fast updating of the control actions to track the dynamical variations in the system, is facilitated by the recurrent neural network, which is trained by a supervised training scheme that uses a simple updating rule. An application of this approach for the adaptive control of synchronous power generators under fault conditions is described, and a quantitative performance evaluation is given to bring out certain important characteristic features of the neural network used for control