A fuzzy control charting method for individuals

Abstract

Shewhart Statistical Quality Control (SQC) charts provide a graphical depiction and record of sample data points, to enable immediate recognition of special causes affecting the process output quality. These charts can detect changes in the mean level, and various other types of unnatural behaviour by the process. Control limits, together with heuristic run and zone rules, are used for the identification of these process behaviours, resulting in a complex situation in which charts must be examined for a number of features. This paper describes a simple method for identifying a criterion representing the state of control of manufacturing processes having a single critical variable characteristic. A fuzzy logic approach has been used to provide this single 'in-control' value. Two fuzzy sets are used, the Centred set and the Random set. The paper explains how the degree of membership of these sets is determined for each sample and how a single crisp in-control value is determined. The behaviour of the fuzzy control chart for three typical unnatural patterns (shift, trend and cyclical) is examined in detail. Some experimental results are provided, comparing the approach with the traditional methods used for control charting of individuals, and the potential of the approach is discussed.