Abstract:

Control charts are tools used to detect an incorrect course of production processes. In classic terms, estimating process parameters (location and variability) is based on the estimation of these parameters using historical data assuming that it runs correctly and that the data is free of contamination. In a situation when outliers and/or other contaminations appear, classic estimators can lead to a significant extension and/or narrowing of the range of control limits. As a consequence, this type of chart proves to be useless. The solution may be that the use of robust estimators or distribution-free control charts. The use of robust estimators provides good estimates in both the absence and presence of outliers. The article presents a review of literature in the field of robust control charts, as well as a definition of directions for further research in this area.