Abstract
This paper deals with the problem of selecting efficient solution procedures for solving lot-sizing scheduling problems in production planning environments based on a number of factors underlying production demands, which were identified in the literature to have influence on the performance of the solution procedures. The factors identified thus far are the coefficient of variation in demand (CVD), inventory ratio (s/h), average time between order (TBO) and the length of the demand horizon.The conventional selection procedures of lot-sizing techniques is to feed Master Production Schedule (MPS), which is based on customer delivery dates, as an input to the Material Requirement Planning (MRP) which explodes all dependent demands to produce a series of due dates along the product structures. The common practice by the production industry today is to test the available MRP lot-sizing techniques with the MPS input and the one giving the least total production cost is selected to be employed in the long run. However, the information concerning the underlying production demand factors are seldom used to evaluate the performance of the lot-sizing techniques prior to choosing the most appropriate techniques to be employed in the production planning environment. Evidently the whole process necessitates laborious computing time and power.The central theme of this paper is the experiment in using the Extended-Delta-Bar-Delta learning rule in identifying the proper lot-sizing techniques to be used given a production demand with a particular underlying structure. The learning rule is found to be effective for decreasing the learning time of artificial neural networks (ANN) in predicting one criterion, namely the appropriate lot-sizing technique, from several predictors, namely the factors underlying the demand structures. Thus an immediate result of this experiment is that we can reduce the time that is normally required to complete a whole production planning cycle by reducing the amount of time required to translate an MPS requirement into a realizable MRP schedule.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have