Robust and predictive control using neural networks

Project members: T.J.J. van den Boom, M. Ayala Botto

The aim of the project is to investigate the possibilities of the application of neural networks to the predictive control of dynamical systems. It contains two sub-projects.

In the first sub-project we aim to establish a link between accuracy of function approximation with a neural network and the stability of the system, leading to a robust model-based control scheme using a nonlinear (neural network) model. Of particular importance is that when bounds can be given on the modeling error, robust control schemes for such systems must be developed which lead to a guaranteed stable control system. The project specifically investigates the relationship between bounds on the network error and stability of the system. If this is achieved, neural control can be applied to real-world applications with guaranteed robustness properties.

In the second sub-project investigate the use of neural networks in the design of analytic constrained predictive controllers for linear systems that combines constraint handling with speed and is applicable to control problems with many constraints. The solution to the model predictive control problem is a continuous function of the state, the reference signal, the noise and the disturbances and hence can be approximated arbitrarily close by a feed-forward neural network. This leads to an analytic constrained predictive controller that combines constraint handling with speed and is applicable to fast systems and complex control problems with many constraints [26,1].

This project is done in co-operation with the IST group of the University of Lisbon.