TECHNIQUES FOR INTEGRATING QUALITATIVE REASONING AND SYMBOLIC COMPUTATION IN ENGINEERING OPTIMIZATION

Abstract

This paper presents techniques that extend theories from the fields of artificial intelligence (AI) and engineering optimization for automating design decision making at three levels of reasoning: qualitative, functional, and numerical. Qualitative reasoning about constraint activity is implemented in SYMON (SYmbolic MONotonicity analyzer) through monotonicity analysis and the maximal activity principle. Functional reasoning is employed in SYMFUNE (SYMbolic FUNclional Evaluator) in the form of algebraic manipulations of the constraint functions and the Karush-Kuhn-Tucker optimality conditions. The techniques are applied to a parametric multiobjective optimal design problem from the literature