Intelligent manipulation of calibration parameters in numerical modeling

Abstract

Numerical models are often employed to simulate flow and water quality processes in coastal areas. Model manipulation is always required, particularly during the set-up, since a slight change of model parameters may lead to quite different results. Nowadays, with the maturity of numerical modeling paradigms, there is abundant heuristic knowledge relating design parameters to outcomes. Techniques are also available to improve modeling accuracy and/or speed. It is desirable to incorporate the existing heuristic knowledge about model manipulation in the form of a rule base and to furnish intelligent manipulation of calibration parameters. In this paper, a prototype expert system for model manipulation for flow and water quality is developed and implemented by employing an expert system shell. The architecture and main components of the system are delineated. Expertise in model manipulation, including knowledge as to what is the correct manipulation direction and how to enhance manipulation effectiveness, are incorporated. It is shown that, through the successful development of this prototype system, expert system technology can be integrated into numerical modeling systems.