CAJ | 학술논문

研究基于Box-Cox变换与Johnson变换方法计算PCIs的有效性，比较Box-Cox变换与Johnson变换计算PCIs的性能优劣，仿真结果显示，Box-Cox变换与Johnson变换均能将常见概率分布转换为正态分布，成功率超过97%，而Johnson变换的转换能力更强，成功率超过99.5%。在计算PCIs的有效性方面，Box-Cox变换与Johnson变换的性能差异不大，利用PCIs估计的产品不合率与理论不合格率相比，相对误差均在2%以内。
연구기우Box-Cox변환여Johnson변환방법계산PCIs적유효성，비교Box-Cox변환여Johnson변환계산PCIs적성능우렬，방진결과현시，Box-Cox변환여Johnson변환균능장상견개솔분포전환위정태분포，성공솔초과97%，이Johnson변환적전환능력경강，성공솔초과99.5%。재계산PCIs적유효성방면，Box-Cox변환여Johnson변환적성능차이불대，이용PCIs고계적산품불합솔여이론불합격솔상비，상대오차균재2%이내。
The Box-Cox transformation and Johnson transformation methods are studied, their calculation performance for PCIs are compared. The simulation results show that the Box-Cox transformation and Johnson transformation can both convert common probability distributions to normal distribution, with success rate over 97%, while Johnson transformation conversion ability is stronger, with success rate more than 99.5%. In terms of the effectiveness of the PCIs calculation, the performance difference between Box-Cox transformation and Johnson transformation is not big, the substandard products rate calculated form PCIs is compared to theoretical substandard products rate and the relative error is within 2%.