Fuzzy compromise: An effective way to solve hierarchical design problems

The importance of being able to decompose largescale engineering problems cannot be overemphasized. Many decomposition algorithms require that all variables, constraints and objectives be specified. In practice this is often unattainable, hence we propose to formulate design problems using fuzzy set theory. Modelling design problems in a fuzzy environment has three advantages. First, some problems may be modelled more accurately using fuzzy relationships rather than using crisp (not fuzzy) ones. Second, fuzziness effectively extends the solution space. Third, the extent of fuzziness may be decreased as the design evolves, thus one model may be used throughout the design process. A fuzzy decomposition method using the fuzzy compromise Decision Support Problem (DSP) is presented in this paper. Using this approach, although the system modelled may be fuzzy, the solution is crisp. The method is demonstrated using a portal frame problem, results from the fuzzy formulation are compared with results from a crisp formulation.