Domangue, Rickie; Patch, Steven C. (1988, ASQC) James Madison University, Harrisonburg, VA; University of North Carolina-Asheville, Asheville, NC
To control a normally distributed quality characteristic, a series of observations or sample means is taken. These values are standardized by the target mean and standard deviation and then squared. The proposed paper considers a quality control scheme based on the exponentially weighted moving average of these squared deviations (EWMASD). The process is stopped when this EWMASD exceeds the asymptotic expected value of the in-control EWMASD plus a constant, called the control limit, times the asymptotic standard deviation of the in-control EWMASD. This procedure is capable of detecting a change in location or an increase in dispersion.
The distribution of run length depends on the weighting constant, starting value, and control limit. A discussion on the selection of these parameters is given. For given values of the weighting constant and starting value, the control limit is selected to give the desired in-control Acceptable Reliability Level (ARL). Nomograms are provided for this purpose for certain starting values and control limits.
A case study is presented illustrating the design and implementation of the proposed procedures. For a fixed in-control ARL of 250, the out-of-control ARL's of some of the proposed schemes are compared with those of some established quality control schemes. The out-of-control processes considered in this comparison include changes in location, increases in dispersion, and combinations of the two.