Article ID: | iaor19982157 |
Country: | United Kingdom |
Volume: | 35 |
Issue: | 10 |
Start Page Number: | 2773 |
End Page Number: | 2792 |
Publication Date: | Oct 1997 |
Journal: | International Journal of Production Research |
Authors: | Zolfaghari S., Liang M. |
Keywords: | neural networks |
This paper reports an ortho-synapse Hopfield network (OSHN) for solving machine grouping problems. An objective-guided search scheme is proposed to lead the search process. The OSHN structure and the objective-guided search scheme have been implemented in an algorithm. Unlike the back propagation or ART1 neural networks, the proposed approach does not require training process and is not affected by the initial arrangement of the machine-part incidence matrix. As compared with the original Hopfield neural network, the computational efficiency and the solution quality can be considerably improved due to the reduced synapses in the OSHN and the objective-guided search process. The performance of the proposed algorithm has been tested using 28 notable problems from the literature and compared favourably with the solutions obtained in the literature.