Control of Complex Systems Using Bayesian Networks and Genetic Algorithms

      Tshilidzi Marwala
University of the Witwatersand, Department of Electrical and Information Engineering, P/Bag 3, Wits, 2050, South Africa
E-mail: t.marwala@ee.wits.ac.za
 
 

Abstract

A method based on Bayesian neural networks and genetic algorithm is proposed to optimally control complex systems.  The proposed methodology is applied to control fermentation process where sugar is converted into alcohol using yeast, a living organism, as a catalyst.  The relationship between input variables and output variables is modelled using Bayesian neural networks trained using hybrid Monte Carlo method.  A feedback loop based on genetic algorithm is used to change input variables so that the output variables are as close to the desired target as possible without the loss of confidence level on the prediction the network gives.  The objective function constructed in this regard, is a weighted sum of square of errors between the target and the neural networks output as well as the confidence levels given by the Bayesian networks.  The proposed procedure is found to reduce the distance between the desired target and measured outputs significantly.

To read the full article,
please log in:
If you haven't registered already,
you can do so for free:
 
© Copyright 2000-2004 Professor F.R. Hall & Dr I. Oraifige, University of Wolverhampton.
 
close this window