Use of fuzzy logic to describe constraints derived from engineering judgment in genetic algorithms

Engineering optimization frequently runs into difficulties due to constraints. If an application is insufficiently constrained, it will not yield a unique optimal solution. Traditionally, underconstrained problems may be solved manually, applying engineering judgment to select the most plausible of a range of solutions which meet the initial set of constraints. The engineering judgment used to select good solutions is often qualitative and may even include contradictory elements. This paper addresses the issue of incorporating engineering judgment into the objective function of an optimizer in order to restrict the range of solutions. Fuzzy logic is used to represent engineering judgment since it is able to represent the qualitative nature of the knowledge and give an assessment of the overall quality of each solution. The approach has been used successfully to match a physically based gas turbine blade cooling design model to experimental results