A Matrix-based Clustering Approach for Developing Modular Architectures

摘要

The purpose of this paper is to employ resemblance coefficient adopted form hierarchical cluster analysis for identifying product architecture. Defining product architecture is a challenging task which is a key activity in the engineering design of modular products. However, no solid matrix-based clustering method can be found in the literature of engineering design. Design Structure Matrix (DSM) is used to visualize the product architecture in view of matrix and to develop the required building blocks for identifying product architecture. The DSM architectural representations are then used to develop a new matrix-based clustering approach based on hierarchical cluster analysis and minimum description length (MDL) principles. The core technique is to investigate the coupling value between the components of the product in order to identify the modular structure of product architecture. The new clustering approach is demonstrated using two real complex products and the results are compared with clustering arrangement proposed by Genetic Algorithm (GA) and human experts. The proposed method is capable of providing competitive clustering arrangements and yield comparable results with results of GA in faster time while allowing the designers to postpone the decisions about crucial parameters like number of modules and size of modules after sorting analysis.