An ideal seed non-hierarchical clustering algorithm for cellular manufacturing

Abstract

This paper describes the development of a non-heuristic algorithm for solving group techology problems. The problem is first formulated as a bipartite graph, and then an expression for the upper limit to the number of groups is derived. Using this limit, a non-hierarchical clustering method is adopted for grouping components into families and machines into cells. After diagonally correlating the groups, an ideal-seed method is used to improve the initial grouping. A quantitative criterion called grouping efficiency is then developed for comparing alternative solutions. The algorithm and the criterion are demonstrated through an example.