关键词: 聚类算法, 产品配置, 模块划分, 演化细胞学习自动机, 模糊C均值

Abstract: By analysing some limitations of current clustering algorithms in the process of product module partition, along with the features of product configuration, fuzzy C means, which was criticized because of its unilateral clustering objective function and climb-hill method for searching solutions, was improved to plan product modular architecture. The intelligent clustering algorithm was proposed by combining the improved fuzzy C means clustering algorithm with evolutionary cellular learning automata, which was used for product granularity division. Based on analysis of main influencing factors for configuration design, three quantified indices including satisfaction degree, assembly complexity and variant design complexity, were proposed to evaluate validity for granularity division results and to decide optimal module numbers. Finally, a practical case was provided to demonstrate the feasibility and validity of this method.

Key words: clustering algorithm, product configuration, module division, evolutionary cellular learning automata, fuzzy C means, granularity evaluation