| Summary: | 碩士 === 國立暨南國際大學 === 資訊管理學系 === 92 === This thesis deals with the priority quota assignment problem for integrated production control of a semiconductor manufacturing chain (SMC). For each lot in the SMC, a production priority is tagged to indicate its relative importance. The performance of a lot is evaluated according to its priority, rather than sophisticated production targets. To help healthy fab management, lot priorities are classified into several levels. The higher priority of a lot is, the shorter cycle time it takes. As there are only limited production resources at a time, a quota for each priority level is needed.
We develop a multi-class, tandem queueing model to capture the dynamics of prioritized lots in a SMC. The nonpreemptive, head-of-the-line priority queueing discipline is adopted. Given the processing rates of stations in the SMC, our priority quota assignment problem is to determine for each priority level the total number of lots such that the overall throughputs are maximized and the specified cycle times of each priority level are satisfied.
We formulate the priority quota assignment problem into a nonlinear programming problem. We use the penalty method to relax the nonlinear constraints, and add a penalty term to the objective function. The penalized programming problem is then solved by the gradient method. We develope a priority quota assignment (PQA) algorithm to solve the relaxed programming problem.
Numerical results demonstrate the feasibility and effectiveness of the developed PQA algorithm. We prove its feasibility with a toy example where there are only two stages in there. Examples of realistic scale are conducted to test the effectiveness of the PQA algorithm. All the numerical results indicate that the developed PQA algorithm is found and effective in solving the priority quota assignment problem in a semiconductor manufacturing chain.
|
|---|
