Interval Estimates Of Process Capability For Non-normal Distribution And The Calculation Of Defect Rate

Process capability, is known as the actual processing capacity of the production process which is under the control status. Usually, process capability is indicated by the volatility of the product quality. Traditional process capability indices are built under the premise of normality assumption. Actually, in most situations, the production process does not follow a normal distribution. Thus, the discussion of the process capability index for non-normal distribution has become a popular topic in recent years.Bootstrap method does not need the original distribution of the sample. It is a new methodology of re-sampling which can be used to refer the characteristics of the population's distribution. This is suitable to the discussion of the process capability index for unknown distribution. With the Bootstrap method, this paper gives the standard, percentile and t-percentile bootstrap confidence intervals of several process capability indices for non-normal distribution and compare the advantages and disadvantages of several bootstrap confidence intervals.Although process capability index can be used as a criterion of to evaluate a production process, the most important criterion will still be the detect rate. However, when sample size is not big enough or the actual detect rate is too small, it is possible that the defect rate from the sample is0, which would impact our evaluation on the process capability. Therefore, this paper gives the defect rate based on non-parametric kernel estimation method which can avoid the0defect rate effectively.