An approach for module decomposition based on fuzzy pattern recognition

Abstract

Modular layout is the basis of modular design for complex product that makes up of multi-domain systems. Furthermore, module decomposition is the precondition for the layout. So the way of information expression, processing and system dividing becomes critical. In this paper, a methodology based on fuzzy pattern recognition is proposed to resolve the module decomposition problem. Firstly, the correlative information among elements is given in the form of constraint network model. In this way, correlative information of product is organized into a tri-view model, the functional, the structural and the behavioral. Secondly, fuzzy pattern models for the functional and the structural are obtained by transforming correlative information into pattern vector. Meanwhile, the ideal patterns for both of them are suggested according to the objectives of planning. Then, fuzzy distance between random pattern and the ideal is passed to GA to evaluate and optimize the layout until the appropriate one. After that, consistent multi-view is built up through correlation among the three views. Finally, a case study about engine is presented to illustrate the application of the proposed approach.