| Summary: | 碩士 === 國立屏東科技大學 === 工業管理系 === 90 === Scheduling has been one of the major tasks in operational management. It has direct impact on the efficiency and effectiveness of production operations. In order to obtain a better production schedule to improve the performance of daily production operations and to increase the competition of the business organization, numerous academic researchers and industrial participants have devoted tremendous amount of effort in developing scheduling algorithms. Due to the combinatorial characteristics, the scheduling problems are classified as NP-hard. Therefore, various heuristics, such as genetic algorithm, simulated annealing, and others, have been developed and tested to show the strength of obtaining closed-to-optimal solutions. Besides, it is quite common that new orders and/or rush orders may arrive after the master production schedule has been determined. The need of scheduling new arrivals creates the problem of ‘re-scheduling’. Scheduling and re-scheduling become typical hurdles for production planning personnel. It is very common, when rescheduling is needed, to reschedule all the pending orders. Such rescheduling method results in building a new mater production schedule every time when a new order is received. It is very typical that the production planning department holds a large number of pending orders. To reschedule all the pending orders increases the responding time to confirm a new order and to alter scheduled activities, including material planning, human resource planning, and etc. In the modern industrial environment, longer responding time means less profit. Therefore, how to shorten the responding time of rush orders, i.e., how to reschedule rush order in less time, has become one important issue in production planning domain.
This thesis develops a two-stage methodology to perform scheduling and rescheduling activities. The first stage determines the master production schedule. At the second stage, a partial rescheduling algorithm is developed to shorten the responding time to newly arrived rush orders. The partial rescheduling algorithm only reschedule the first several orders and the rush order and leaves the sequence of the rest orders unchanged. The problem is formulated as a nonlinear problem, which encompasses production costs, material and finished products storage costs, communication cost for changing delivery time, emergent procurement cost and penalty cost for late delivery. At both stages, genetic algorithm is used to solve scheduling and re-scheduling problems. The results show that the partial rescheduling methodology has advantages in terms of responding times and costs. It is also found that the numbers of orders to be rescheduled are between 5 and 7. Statistical tests show that capacity utilization, material storage cost, and communication cost have significant effect on rescheduling.
|
|---|
