Predictive control using neural networks

The predictive control of nonlinear systems has recently been the subject of several research works and several algorithms, in particular those using fuzzy logic and neural networks. In this paper, we present a method for unconstrained predictive control of nonlinear systems. This method, uses a static neural network as a prediction model and is based on the idea of dividing the predicted output into it´s free and forced parts. Such division of the predicted output allows obtaining analytically the sequence of control signals. We use this technique for the predictive control of a Continuous Stirred Tank Reactor (CSTR).