Article ID: | iaor19952006 |
Country: | United Kingdom |
Volume: | 22 |
Issue: | 8 |
Start Page Number: | 841 |
End Page Number: | 856 |
Publication Date: | Oct 1995 |
Journal: | Computers and Operations Research |
Authors: | Hawaleshka O., Awadh B., Sepehri N. |
Keywords: | heuristics, networks: flow |
A process planning model is developed in conjunction with a genetic algorithm. The model first receives the flow network description of possible process plans. The goal is to find the best plan according to some prescribed objective criteria. A novel stage binary matrix format is then applied to describe each stage in a flow network. This formulation allows proper representation of possible routes as binary strings to be used by a genetic algorithm. Each string is directly mapped onto a zero/one stage matrix. A ‘zero’ bit in these strings represents no connection between two nodes in the flow network, whereas a ‘one’ bit denotes a connection between the corresponding nodes. The binary stage matrix representation also facilitates the employment of a modification procedure to ensure that a single distinct path (indication of a process plan) exists in each individual network. The model, developed in the paper, is first illustrated using a simple network. The performance of the computer program based on this model is then evaluated for different sizes of networks.