Machine Component Clustering with Mixing Technique of DSM, Jaccard Distance Coefficient and k-Means Algorithm

Abstract

This study aims to introduce a design method of Jaccard Distance Coefficient with k-Means algorithm for machine component clustering into independent modules so that a machine can be easily modified to achieve its requirement functions. In this study, Jaccard Distance Coefficient is firstly applied to generate relation matrices. After that, six results of distance coefficients and their clusters (2.02 with 2 clusters, 1.95 with 3 clusters, 1.89 with 4 clusters, 1.83 with 5 clusters, 1.75 with 6 clusters and 1.66 with 9 clusters) are calculated corresponding to relation matrices from the first step. Nextly, k-Means algorithm is used to take care of these six results for analyzing the most proper level coefficient. The result shows that the modules of 3 clusters with distance coefficient of 1.95 is the best outcome at the 0.7074 natural value.