Sangjin Jung, Timothy W. Simpson
Modularity indices based on Design Structure Matrices (DSMs) have been utilised to help measure modularity and cluster a product’s architecture into independent or coordinated modules, but many metricshavedifficulty(a)measuringthemodularityofdifferenttypes ofarchitecturesinreal-worldproductssuchasbus-typearchitectures and (b) optimising module boundaries in architectures. After review- ing existing modularity indices and clustering algorithms, we intro- duce new modularity indices that can capture the degrees of (1) con- nection strengths within each independent module and between different modules, (2) density of connections within each module and between modules, (3) proximity of interactions to the diagonal of the DSM, and (4) density of connections between buses and other components. Moreover, the proposed metrics can serve as objective functions to obtain optimal DSMs to maximise modularity. A com- parative analysis of the proposed modularity index shows that the proposed metric can measure the modularity for 28 different types of DSMs unlike other metrics. Also, the module definition results for an aircraft engine with complex connections between components indicate that clustering using the new modularity indices can help obtain optimal modular product architectures with higher modular- ity compared to existing clustering results in the literature.
© 2001-2024 Fundación Dialnet · Todos los derechos reservados