Using swarm intelligence algorithm to solve a part clustering model based on weighted directed graph

Mass customisation is an effective mechanism to satisfy the new requirements of changing markets. E-manufacturing can be seen as a good choice to implement mass customisation. As a support technology for mass customisation, part clustering of the product family is very important in the optimal formation of manufacturing cells or systems and it could be more efficient in e-manufacturing mode than in traditional manufacturing mode. This paper is mainly concerned with the part clustering problems for the manufacturing of mass customisation. A mathematical model of part clustering of a product family based on weighted directed graph is put forward. According to the criterion of minimising the manufacturing cost, two hierarchic sub-models and a relevant methodology are presented. As the bottleneck problem of the methodology, process combination is realised by combining the incidence matrixes of the abovementioned weighted directed graph model. Subsequently a swarm intelligence algorithmic model is designed and a software prototype system is developed with Java technology. Finally a clustering case is put forward and studied, which illustrates that the models described in this paper can be used to solve the parts clustering problem in e-manufacturing system for mass customisation effectively.