Purpose:To propose a new nonparametric control chart (called the kLINK chart) based on a k-linkage ranking algorithm that calculates the ranking of a new observation relative to the in-control training data. Summary:An individual control chart can be used to monitor to each individual quality characteristic. This may lead to unsatisfactory results when a process involves a number of correlated quality characteristics. While using multivariate control charts when assumed normality is not present, the calculated probabilities of Type I and Type II error rates derived from the control mechanisms become unreliable where as nonparametric techniques control the probabilities of false alarms. A nonparametric procedure utilizes the historical in-control data to represent the underlying distribution. Multivariate nonparametric control charts perform reasonably well in the situations for which they were designed. The present study proposes an alternative nonparametric control chart that is based on a ranking algorithm, the k-Iinkage ranking (kLINK). The article demonstrates that the kLINK chart is a logical, efficient, and robust control chart method with flexible control boundaries that can effectively monitor multivariate processes in a variety of nonnormal situations. (27 refs.)
展开▼
