ADAPTIVE SEARCH MANAGER: An Engineering Design Decision Support Approach

Engineering design often involves several objectives. True engineering design solutions are not necessarily the global optimum as described by some mathematical simulation with respect to one criterion [Parmee and Denham (1)]. Often designer interaction is required to take into account many different criteria. In the case of multimodal design problems there may be quite different design solutions that perform similarly quantitatively, but have large differences in their degree of multi-criteria satisfaction. Criteria may include manufacturability, choice of materials, maintainability, reliability, specific customer requirement etc., many aspects of which can be qualitative in nature. Integrating all of these criteria into one comprehensive evaluation function is difficult and at times can be misleading. If the criteria are quantitative in nature a multiobjective genetic algorithm [Goldberg (2)] can be utilised. An attempt has been made at the Plymouth Engineering Design Centre to obtain multiple “good” design solutions based on the most important quantitative criteria. The “good” design solutions are qualitatively evaluated for other less well-defined criteria and then sensitivities of each of these design solutions are calculated. The final decisions are left to the designer.