Case reuse based product fuzzy configuration

This paper proposes a new probabilistic approach to deciding product fuzzy configuration by using the cases in the case library. The main idea relies on condition probability calculation for deriving the probability of selecting an undecided subassembly with given the known selection of a group of decided subassemblies. The predefined product structure is used as the predetermined condition, and the up consistency rule is proposed as the preprocessing step. The conditional probability is extended as the core algorithm. The evaluation method of judgment threshold is designed based on the maximum likelihood estimation. A practical example is applied for proving the validity of this method. By comparing with case-based reasoning (CBR), this new method avoids the properties information blind spots in product configuration procedure and avoids the redundant constraints post-checking procedure. As the result, this method can solve the huge complex configuration data problem, and truly make the fuzzy configuration possible.