Problem-dependent search for optimal parameter instantiation order in design spaces

Abstract

The paper refers to the domain of engineering design and introduces a method for obtaining an optimal instantiation process for the design parameters in a design problem by using extended, problem- and path-dependent search within the digraphs that these parameters form. Design parameters are considered as entities being interrelated via deterministic associative relationships and forming digraphs that accommodate declarative and procedural design knowledge. They participate in establishing formal metrics for the estimation of importance and design effort. Factors with values assigned by the designer(s) define the required effort for the instantiation of primary parameters. A method is presented and an algorithm is proposed for obtaining optimal instantiation lists. The algorithm uses extended problem- and path-dependent search within the parameter digraphs. Iteratively, in each step, the algorithm performs extended search in order to: (a) locate the optimal path, (b) instantiate its primary parameter and (c) propagate its value throughout design space. Thus information and knowledge of major importance is produced early in the design process with minimum instantiation effort. The design of a belt conveyor carrying unit loads is used as a case study for the suggested method. The proposed method can optimally guide the instantiation process. Violations of requirements, specifications and constraints for important parameters are promptly realized, modifiable primary parameters are easily located and, as a result, the necessity for changes to parameter interrelationships can be subsequently justified.