Process capability analysis in the presence of autocorrelation

摘要

The problem of performing process capability analysis in multivariate environment when autocorrelations are present is discussed. It is shown that when the systematic nonrandom phenomenon induced by autocorrelation is ignored the variance estimate obtained from the original data is no longer an appropriate estimate for use in the process capability analyses. A remadial measure based on an autoregressive integrated moving average (ARIMA) model is proposed. It is also shown that the process varianceestimated from the residual analysis yields remarkable results for the process capability indices.