A stochastic unsupervised competitive learning algorithm for the design of cellular manufacturing systems

Cellular manufacturing systems (CMS), an application of group technology (GT) derives the benefit of cost advantage of mass production effect for multi-product small-lot-sized production. There are many applications to this concept of group technology in engineering manufacture. In CMS, machine tools are grouped into cells and each one of them is generally dedicated to the manufacture of a part-family. In this paper, a stochastic unsupervised learning algorithm (SUCLA) has been developed. This is a two-layer feedforward network trained with competitive learning. The proposed algorithm is used for the simultaneous formation of machine cells and part families. This model is tested with a considerable size of data sets available in the open literature. This algorithm has been found to be a flexible and powerful tool for the design of CMS.<
>