Statistical methods for multivariate process control do not provide engineering tools for the diagnosis of manufacturing processes. Rather, they characterize and monitor the randomness inherent in manufactured products to identify time periods or production batches which show irregular or atypical characteristics. It is up to the production engineer to diagnose the cause or causes of these irregularities and to define the appropriate action. This article proposes a new diagnostic approach for multivariate process measurement vectors using a process-oriented basis. Many potential production problems have characteristic signatures that can be detected in the multivariate quality vector. Each signature can be used as a basis element in the process-oriented basis. The representation of the multivariate bias or variance using this basis will identify a small set of potential causes, those associated with basis elements (signatures) with the largest coefficients.
All Science Journal Classification (ASJC) codes
- Safety, Risk, Reliability and Quality
- Industrial and Manufacturing Engineering