EXPERIMENTAL COMPARISON OF CONTROL STRATEGIES

Advanced conventional, internal model, model predictive, process-model -based generic model, neural-network-based, and fuzzy-logic control st rategies have been implemented and evaluated on a nonlinear fluid flow and heat exchange pilot plant. Control models, which included both st eady-state and dynamic versions, were generated using empirical, pheno menological and neural-network approaches. Evaluation criteria include d manipulated variable action, ease of implementation and tuning, as w ell as measures of controlled variable performance. Copyright (C) 1997 Elsevier Science Ltd.