No abstract

No abstract

In order to perform systems analysis or synthesis, it is compulsory to deduce a model of the process. Articial Neural Networks ANN have shown their suitability to identify nonlinear dynamic processes without modelling them theoretically. Since no modelling is performed, the important issue for the Neural Network approach is to determine the required time delays. In this paper, di erent methods are presented that make it possible to reach this goal. First, some pruning methods are presented to detect nonrequired input neurons belonging to certain time delays. In order to avoid the high computational e orts of these methods, a new approach is presented which is based on the estimation of the gradient v ector of the system nonlinearity. All methods are applied to a continuous-stirred tank reactor.

This paper deals with the design of iterative learning controllers (ILC) based on extended state space models for non-linear cyclic process control. In order to design a suitable learning operator, knowledge about the plant's dynamical behaviour is needed which implies that a system model has to be set up. It is expedient to acquire a state space model of the plant using identification methods. Here we deal especially with the case, that a linear model represents system dynamics inadequately. We start with a non-linear model and linearize the system along the current trajectory, thus obtaining a linear time variant model. Using this as basis, we develop methods for identification and control of the nonlinear process. Experimental results show that a good system model is also useful to perform a pre-training for the ILC; this is especially interesting in case large deviations from a desired system output trajectory must be avoided. The presented algorithms have been implemented and tested experimentally with a real-life nonlinear processing plant. Iterative learning Controllers with reduced sampling rateWe use the techniques developed in [2] to design stable linear ILC by choosing an appropriate sampling time (for the ILC). In principle the generic time invariant state space model is represented by:In the course of this paper modifications to this model are added to take the nonlinear dynamics into account.

No abstract