1993 Fiscal Year Final Research Report Summary
A Study on the Analysis Method of Large Scale Production Systems Using Queueing Network Model
| Project/Area Number |
03832038
|
|---|
| Research Category |
Grant-in-Aid for General Scientific Research (C)
|
| Allocation Type | Single-year Grants |
|---|
| Research Field |
社会システム工学
|
|---|
| Research Institution | Aoyama Gakuin University |
|---|
Principal Investigator |
KURODA Mituru Aoyama Gakuin University, Depart of Industrial and Systems Engineering, Professor, 理工学部, 教授 (80082791)
|
|---|
| Project Period (FY) |
1991 – 1993
|
| Keywords | queueing network model / simulated annealing / tabu search / inverse queueing network analysis / semiconductor manufacturing lines / quasi-equilibrium / compulsory simulation |
|---|
| Research Abstract |
It is actually impossible from the viewpoint of computation time to optimize a large scale production system using simulation. In this research, we presented an optimization technique coupling a queueing network model and any local search algorithm, which is called "inverse queueing network analysis(IQNA)". It was applied to the models of hypothetical production systems which possess the properties of semiconductor manufacturing lines. As the results, the following conclusions were obtained.
(1)IQNA provides the analytical results which are close to the simulation results carried out under the same conditions in the case of constant demands.
(2)Assuming the existence of quaisi-equilibrium of production systems, IQNA can analyze them in the case that the demands vary from period to period. The errors between the analytical and simulation results depend on the ratio of the length of a period and the average residence time of the work-in-process.
(3)The errors are caused by the delay between the change of inputs and the resultant outputs. The fact was confirmed by carrying out compulsory simulation.
(4)The computation time of IQNA can be reduced almost by half using both simulated annealing and tabu search in comparison with using only simulated annealing as a local search algorithm.
|
