A PATTERN-RECOGNITION ALGORITHM FOR AN (X)OVER-BAR CONTROL CHART

Control charts are very useful in identifying process problems. With t he implementation of automated data collection methods and computerize d plotting on control charts, a need exists for identification procedu res that can alert the operator to unnatural patterns and special caus es. This proposed pattern-recognition algorithm is based on traditiona l recognition methods modified for recognizing nonrandom patterns on a n (x) over bar control chart. Extensive simulations tested the capabil ity of the algorithm to detect cycles and systematic patterns on an (x ) over bar control chart. Average run lengths are computed and compare d with other relevant studies. Extensions and enhancements are also ad dressed.