A neural network approach for the development of modular product architectures

The clustering of a product's components into modules is an effective means of creating modular architectures. This article initially links the clustering efficiency with the interactions of a product's components, and interesting observations are extracted. A novel clustering method utilising neural network algorithms and design structure matrices (DSMs) is then introduced. The method is capable of reorganising the components of a product in clusters, in order for the interactions to be maximised inside and minimised outside the clusters. In addition, a multi-criteria decision-making approach is used, in order for the efficiency of the different clustering alternatives, derived by the network, to be evaluated. Finally, a case study is presented to demonstrate and assess the application of the method. The derived algorithmic clustering proved to be more efficient compared with the empirical one, and thus, it can be used by design engineers as an effective tool for the derivation of product clustering alternatives.