Moments-based clustering techniques for manufacturing cell formation

Abstract

Clustering and data organization algorithms and array sorting methods are most widely used in solving machine-part grouping problems in Group Technology (GT). One of the most recent search algorithms for part-family identification is the close-neighbour search algorithm. However the method does not suggest a solution that tries to eliminate exceptional elements after the block diagonal structure is produced. This paper presents a Method of Moments approach such as Weighted Moment, End Load Ratio and Differential Moment Methods to solve this problem. The effectiveness of these algorithms is tested against sixteen problems from the literature. The results show that the algorithms are very efficient and reliable.