J. Jarmulak, E.J.H. Kerckhoffs, L.J.M. Rothkrantz, “A software environment for neural control of simulated plants”, in: Neural Network World (International Journal on Neural and Mass-Parallel Computing and Information Systems), Volume 5, Number 6, 1995, ISSN 1210-0552; pp. 873-892.
- pdf – the layout is messed up a bit by WordPerfect because the file was from a very old WP version
Abstract: Several control architectures making use of neural networks and capable of controlling nonlinear systems are presented. In order to facilitate research into the properties and possibilities of neural controllers, a special software tool has been developed. The design of the tool based on object-oriented paradigm is described. Results of experiments with model-based predictive control of a simulated nonlinear water column system are presented and compared to classical PID control. The availability of proper software makes construction of a neural controller not much more difficult than construction of a conventional PID controller.
Key words: adaptive control, identifiers, model-based control, neural networks, nonlinear systems, predictive control, software tools