A novel methodology to manage uncertainty for multidisciplinary design and optimization in conceptual design

Uncertainty in design and simulation affects the quality of product directly during the process of MDO, which should be considered to help designers to make the design decisions, especially at conceptual design stage. In traditional approaches, this uncertainty is ignored in the hope that it is not significant to the decision making. In this investigation, firstly, three main uncertainties in MDO of conceptual design are pointed out and carefully analyzed. Then, an innovative methodology integrating extract knowledge and probabilistic method to manage these uncertainties is presented. Considering practical application and eliminating the uncertainty in configuration, we propose a promising method combining Fuzzy c-means algorithm (FCM) and Rough Sets theory (RST) to deduce the configuration rule. Furthermore, probabilistic kriging model is utilized as an approximation model to reduce global computational expense of complex product. Sensitivity analysis (SA) is used to reduce uncertainty of inputs of simulation, and mean square error (MSE) is employed to assess model error to reduce model uncertainty. Finally, the validity and necessity of this methodology are demonstrated through the conceptual design of bulk carrier.