ARMS — application of AI and OR methods to resource management.pdfVIP

A Comparison of Competitive Neural Network with Other AI Techniques in Manufacturing Cell Formation 1 2 1 Gurkan Ozturk , Zehra Kamisli Ozturk , and A. Attila Islier 1 Eskisehir Osmangazi University, Industrial Engineering Department, Bademlik 26030, Eskisehir, Turkey {gurkano, aislier}@ogu.edu.tr 2 Anadolu University, Open Education Faculty No:312, 26470, Eskisehir, Turkey zkamisli@anadolu.edu.tr Abstract. The cell formation (CF) problem aims to transform the in- cidence matrix into block diagonal form. Numerous techniques are de- veloped for this purpose ranging from mathematical programming to heuristic and AI techniques. In this study a simple but effective com- petitive neural network algorithm (CNN) is applied and compared with genetic algorithms, tabu search, simulated annealing and ant systems by making use of some well known data sets from literature. As a result at 14 out of 15 cases, better results are obtained by CNN. Keywords: Cell formation, cellular manufacturing, competitive neural network, AI techniques. 1 Introduction Wemmerlov and Hyer [1] assert that cellular manufacturing is an application of group technology in which all or a portion of firm’s manufacturing system has been converted into cells. A manufacturing cell is a cluster of machines or processes located in close proximity and dedicated to the manufacture of a family of parts. The parts are similar in their processing requirements, such as operations, tolerances, and machine tool capacities.