Study On Optimization And S Imulation For Semiconductor Final Testing Scheduling With Two-staged Re-entry

There exist such features as complex processing techniques, numerous equipments with various types, co-existence of parallel and serial processing, as well as the re-entrant feature on semiconductor manufacturing line. Re-entrant feature means products or jobs would visit the same equipment or equipment group more than once at their different stages of processing. As a crucial part of semiconductor manufacturing line, the final testing stage contains all properties mentioned above. Therefore, how to model and optimize the whole or partial semiconductor manufacturing line has become a problem cries for solution academically and industrially.However, the traditional operations research methods and heuristic methods are difficult to tackle such complex scheduling problem. With the development of computer technology, simulation modeling has emerged as an important way out.In this thesis, we utilize ExtendSim platform to model and optimize the scheduling problems of semiconductor final testing with two-stage re-entry (FTMR). We also summarize current relative research theories and methods for scheduling problems of semiconductor final testing stage with re-entry and present a physical and a mathematical model for FTMR, as well as a B-T rule used to solve such problem. And thereafter we construct an ExtendSim simulation model based on the physical and mathematical model for FTMR, and then compare and analyze the advantages and disadvantages of some feeding strategies and dispatching rules commonly used in Semiconductor Final Testing stage. Moreover, the simulation results are optimized by the built-in genetic algorithm block on the software platform. The next part of the dissertation introduces the basic principle of nested partition algorithm, and constructs a block for nested partition algorithm on the ExtendSim software combined with FTMR’s characteristics and the software’s features. Feasibility and effectiveness between the Genetic Algorithm and Nested Partition Algorithm in solving FTMR are compared at the end of this dissertation.