This paper presents a framework for applying AI techniques to mechanical engineering design. In particular, we focus on the determination of the limiting factors, i.e., the bottlenecks, of system design. We refer to the bottlenecks as active constraints and propose a method called Active Constraint Deduction (ACD.) A Knowledge-base is constructed from the activity-data, each of which includes a constraint predicted to be active, the validity intex that indicates the confidence of the claim, and the justification for the claim. Given a set of design constraints, ACD deduces the candidate active constraints from the system specification. The deduced information is then used in conjunction with monotonicity analysis to determine the actual active constraints on which the design decisions should be based. The use of ACD allows designers to rapidly obtain the optimal solution. We illustrate the proposed method by an example: the system design of coal-fired power generation plants.

This content is only available via PDF.
You do not currently have access to this content.