Prediction of PCDDs/PCDFs emissions from municipal incinerators by geneticprogramming and neural network modeling

The potential emissions of PCDDs/PCDFs from municipal incinerators have rec eived wide attention in the last decade. Concerns were frequently addressed in the scientific community with regard to the aspects of health risk asse ssment, combustion criteria, and the public regulations. Without accurate p rediction of PCDD/PCDF emissions, however, reasonable assessment of the hea lth risk and essential appraisal of the combustion criteria or public regul ations cannot be achieved. Previous prediction techniques for PCDD/PCDF emi ssions were limited by the linear models based on a least-square-based anal ytical framework, such that the inherent non-linear features cannot be expl ored via advanced system identification techniques. Recent development of g enetic algorithms and neural network models has resulted in a dramatic grow th of the use of non-linear structure for optimization and prediction analy ses. Such approaches with the inherent thinking of artificial intelligence were found useful in this study for the identification of non-linear struct ure in relation to the PCDD/PCDF emissions from municipal incinerators. Exa mples were drawn from the emission test of PCDDs/PCDFs through the flue gas discharge from several municipal incinerators in both Europe and North Ame rica. Although the neural network model may exhibit better predictive resul ts based on the performance indexes of percentage error and mean square err or, model structure cannot be directly identified and expressed for illustr ating the possible chemical mechanism with respect to the PCDD/PCDF emissio ns. However, the tree structured genetic algorithms, or so-called genetic p rogramming, can rapidly screen out those applicable non-linear models as we ll as identify the optimal system parameters simultaneously in a highly com plex system based on a small set of samples.