Graph-neural network approach in cellular manufacturing on the basis of a binary system

Abstract

In the past several years, many studies have been carried out on cellular manufacturing. Group technology is a manufacturing philosophy in which similar parts are identified and grouped together to take advantage of their similarities in manufacturing and design. The main problem in the development of cellular manufacturing is that of cell formation. In this paper, a graph-neural network approach is given for cell formation problems in group technology. Effort has been made to develop an algorithm that is more reliable than conventional methods. A graph-neural network has the advantages of fast computation and the ability to handle large scale industrial problems without the assumption of any parameter and the least exceptional elements in the presence of bottleneck machines and/or bottleneck parts. Two examples from the literature have been solved to demonstrate the advantages of the algorithm.