Article ID: | iaor20105553 |
Volume: | 20 |
Issue: | 7 |
Start Page Number: | 636 |
End Page Number: | 648 |
Publication Date: | Oct 2009 |
Journal: | Production Planning & Control |
Authors: | Silva Cristvo |
Keywords: | production, artificial intelligence: decision support, decision: studies |
This article describes the development of a decision support system, called ‘PHIL’, for a fibre production scheduling problem. An algorithm to obtain production plans for the spinning department of a case study company is presented. This article shows how the algorithm has been implemented in ‘PHIL’ to allow interaction between the decision it proposes and the human planner. This interaction is possible because of the way in which ‘PHIL’ deals with the problem constraints that differ in terms of ‘hardness’. The methodology uses formal rules but keeps some decision room for the planner to use informal decision practices that are almost impossible to encode. ‘PHIL’ is evaluated comparing the plans obtained with plans generated informally by the human planner. The same data sets are also used to compare plans produced using only formal practices with plans produced by the proposed methodology. The results of this evaluation show that the proposed methodology, which allows a decision maker to use his/her personal knowledge of the production process when generating production plans, produces the best schedules.