Neural network for optimization of existing control systems

The purpose of this paper is to develop methods to use neural network based controllers (NNC) as an optimization tool for existing control systems. Two different methods are suggested. One uses the NNC as a feedforward controller and the other uses the NNC as a feedback controller. The main advantage of these methods is that they make it possible to retain an existing traditional control system. In these methods the NNC is doing the optimization and not the stabilization of the process. A thermal mixing process is used as a test system, which is a multivariable and nonlinear process. The method based on feedforward has been tested and shows very good performance.