Neural networks for non-linear control

This paper describes how a neural network, structured as a multi layer perceptron, is trained to predict, simulate and control a non-linear process. The identified model is the well-known known innovation state space model, and the identification is based only on input/output measurements, so in fact the extended Kalman filter problem is solved. The training method is the recursive prediction error method using a Gauss-Newton search direction, known from linear system identification theory. Finally, the model and training methods are tested on a noisy, strongly non-linear, dynamic process, showing excellent results for the trained net to act as an actual system identifier, predictor and simulator. Further, the trained net allows actual on-line extraction of the parameter matrices of the model giving a basis for better control of the non-linear process.<<ETX>>