Cellular manufacturing is the cornerstone of many modern flexible manufacturing techniques, taking advantage of the similarities between parts in order to decrease the complexity of the design and manufacturing life cycle. Part-Machine Grouping (PMG) problem is the key step in cellular manufacturing aiming at grouping parts with similar processing requirements or similar design features into part families and by grouping machines into cells associated to these families. The PMG problem is NP-complete and the different proposed techniques for solving it are based on heuristics. In this paper, a new approach for solving the PMG problem is proposed which is based on biclustering. Biclustering is a methodology where rows and columns of an input data matrix are clustered simultaneously. A bicluster is defined as a submatrix spanned by both a subset of rows and a subset of columns. Although biclustering has been almost exclusively applied to DNA microarray analysis, we present that biclustering can be successfully applied to the PMG problem. We also present empirical results to demonstrate the efficiency and accuracy of the proposed technique with respect to related ones for various formations of the problem.
