A BELIEF NETWORK APPROACH TO OPTIMIZATION AND PARAMETER-ESTIMATION - APPLICATION TO RESOURCE AND ENVIRONMENTAL-MANAGEMENT

An approach to use Bayesian belief networks in optimization is present ed, with an illustration on resource and environmental management. A b elief network is constructed to work parallel to a deterministic model , and it is used to update conditional probabilities associated with d ifferent components of that model. The divergence between prior and po sterior probability distributions at the model components is used as a n indication on the inconsistency between model structure, parameter v alues, and other information used. An iteration scheme was developed t o force prior and posterior distributions to become equal. This remove s inconsistencies between different sources of information. The scheme can be used in different optimization tasks including parameter estim ation and optimization between various policy options. Also multiobjec tive optimization is possible. The approach is illustrated with an exa mple on cost-effective management of river water quality. (C) 1998 Els evier Science B.V. All rights reserved.