| Summary: | 碩士 === 國立清華大學 === 工業工程與工程管理學系 === 103 === Acceptance sampling plans state the required sample size for inspection and the associated acceptance or rejection criteria for lot sentencing, which have been considered as one of practical tools for quality assurance application. There are several ways to classify acceptance sampling plans. One major classification is by data, i.e., variables and attributes. When a quality characteristic is measurable on continuous scale or very low fraction of defectives, it may be appropriate to use variables sampling plan rather than attributes sampling plan.
Several authors have incorporated Process Capability Indices（PCIs）to develop Single Sampling plan (SSP) and Repetitive Group Sampling Plan (RGSP). Both of them are simple to operate for administration, but it does not considerate the all the available information from the collected samples. The concept of Multiple Dependent State Sampling Plan (MDSSP) has been proposed. It considers the states of the preceding lots and can provide better efficiency of inspection, but it is only developed for attributes sampling plan.
In this thesis, we overcome the disadvantages of SSP and RGSP by considering the concept of MDSSP to develop two sampling plans, called Variables Multiple Dependent State Sampling Plan（VMDSSP）and Variables Multiple Dependent State Sampling Plan Repetitive Group Sampling Plan（VMRGSSP） based on the one-sided capability index. Finally, we will summarize the advantages of the proposed variables plans over than existing variables sampling plans.
|
|---|
