| Summary: | 碩士 === 國立成功大學 === 工業與資訊管理學系碩士在職專班 === 103 === Summary
Scheduling is an important problem in production planning systems. Determining a method by which to meet customer demand and resource allocation, in addition to factory machine capacity in order maximally reduce the operating cost of the production schedule is an issue. Scheduling needs to be set according to targeted factory scheduling considerations, but in the real world, the targets set at the factory are usually multiple in nature. This study discusses the fact the production schedule planning of TFT-LCD manufacturing company A still relies on staff, and taking in to consideration production diversification constraints and objectives and the fact the this company has its own production database, we proposed an automated scheduling model using the Multi objective Genetic Algorithms(MOGA) in order to provide schedule quality and stability.
Key words: Scheduling , Flexible shop manufacturing system(FMS), Computer integrated manufacturing(CIM), Adaptive planning and Scheduling(APS)
Introduction
Because of global competition, in order to reduce cost to increase return on assets, companies production patterns began to turn pull production and diversification. With changes in production mode, enterprises are facing increasingly shorter production cycle times, increasingly more production conditions, and increased complexity of scheduling problems.
The current production schedule planning of company A is based on the manufacturing plant personnel scheduling, and the scheduling quality performance varies because the individuals doing it are different. Addressing scheduling problems has become a popular research topic. In a recent study, Zhang et al. (2012) proposed the Flexible Manufacturing System (FMS) as a highly automated production systems, the due to the fact that the need for factories to reduce delivery time is considered a very important goal. Ruiz et al. (2008) mentioned that many past studies have been conducted to reduce the differences between theory and reality. In reality, the impacts of the schedule constraints include: available release dates for machines, unrelated parallel machine issues, available machine qualifications, necessity to skip the processing phase, and machine setup time sequence dependency, among other restrictions. Scheduling problems are based on the structural properties of different factories, different consideration limitation to the problem, staff planning and scheduling time constraints, and the inability to guarantee the quality of the schedule, not to mention realistic targets for multiple objects.
Semi-automated environments has real-time automated industrial production data, so automated scheduling has begun to gain increasing importance in the enterprises. We propose the use of real-time factory data through multi-genetic algorithms to provide stable schedule performance and quality.
Materials and Methods
This study considers an automated flexible scheduling production system that orders processing operations in accordance with the path order for each individual product. Each workstation has fixed assumptions for processing time.
In order to address real world scheduling problems and extend the research on this topic, a number of assumptions will be considered as follows:
1. The factory is not only the producing of a single type of product.
2. The batch lot in the machine can’t be interrupted until the process is completed.
3. Production quantities are known.
4. The production processes are not the same.
5. There is no rework.
6. Machine setup time is sequentially dependent.
7. The setup time for each machine is known.
8. The customer demand date is known.
9. The transmission time is not considered.
10. The machine produces only one product at a time.
11. Machine production efficiency varies.
12. Idle time is not included in the setup-time.
This study proposed a methodology for automated scheduling using the Microsoft Visual Basic for Application (VBA) program and Microsoft Access as the database in order to determine feasible solution for scheduling problems through the use of multi-object genetic algorithms.
Results and Discussion
In order to verify the effectiveness of the proposed algorithm, the experimental design made use of Company A as a case for verification and comparison. The system used Excel 2007 VBA to build the program and use Access 2007 as database.
The case assumes the factory has 25 jobs and that the factory process involves 7 operations; each process has to produce a machine group; the number of machine groups includes up to 4 machines and 4 objects respectively: setup-time, maximum output, machine idle time and minimum job delay.
From the results, it was determined that the schedule performance using the proposed algorithm was better, and that the proposed algorithm can provide personnel with decisions and more detailed schedule information, such as dispatch job route and job process start time.
Conclusions
This study considers a flow-shop production scheduling problem occurring in sequence dependent on setup time, taking in to consideration the different production processes and quality problems arising from delays in the production of products resulting from such issues as machine condition so that more realistic and feasible solution can be provided that will lead to more clear schedule information and reduced staff uncertainty.
|
|---|
