การปรับปรุงวิธีการทางสถิติเพื่อการควบคุมคุณภาพ : กรณีศึกษาโรงงานผลิตหม้อแปลงไฟฟ้า / มาลีญา ฉิมพาลี = Improvement of statistical quality control : a case study of the transformer manufacturing industry / Maleeya Chimpalee
The objective of this research is to determine single acceptance sampling plans for specified consumers and producer risks. Empirical Bayes single sampling plan procedures are derived in which past lots history is used in a supplementary manner to estimate hyperparameters of beta prior distribution. Study in the sample departments found that outgoing inspection plans used by Quality Assurance Department was not based on specified producer and consumers risks. The procedures are developed here for obtaining Shilling-Johnson single sampling plan by both classical method and Empirical Bayes method. Computer programs were written to conduct the calculation and search. Performance of AQL obtained from process average is better than those obtained from BEQ. The sample sizes obtained by Empirical Bayes sampling plans for ERL and SBT visual inspection are zero according to good past lot history. MIL-STD-105E sampling plan is recommended here in order to keep quality record for future Empirical Bayes analysis and be feedback to production unit.