Principal component analysis for evaluating the feasibility of cellular manufacturing without initial machine-part matrix clustering

Abstract

We present a set of diagnostic tools for machine-part matrix clustering. Machine-part matrix clustering is traditionally viewed as the first step in designing a Cellular Manufacturing System. Clustering is achieved by permuting the rows and columns of the matrix to get a Block Diagonal Form (BDF). Without actual matrix clustering being done, most methods for clustering fail to predict objectively the adaptability of the given matrix for Cellular Manufacturing. This paper presents an approach which predicts clusters accurately without initial machine-part matrix clustering. In addition, it also provides quantitative information on whether the matrix encourages Cellular Manufacturing without obtaining or partitioning the BDF. Lastly, the method also identifies the part families corresponding to each machine group in the same sequence for both dimensions of the matrix. This yields a BDF rather than a Block Checkerboard Form for the clustered matrix which happens when the machine groups and part families are not correctly sequenced in both dimensions of the matrix.