Predictive control with neural network model for flexible link robot

This paper proposes the use of predictive control systems coupled with artificial neural networks (ANN) to control the dynamic behavior of a single-link flexible manipulator rotating in a horizontal plane. For systems with unessential nonlinearity and without non-minimum phase properties, it is quite enough to use one-step ahead predictive models to achieve efficient control, although they do not ensure sufficient robustness. For systems with more complicated dynamics, the application of extended predictive control algorithms with neural network model is desirable.