Statistical methods for process control do not provide engineering tools for diagnosis. Instead, they characterize and monitor randomness inherent in the manufactured products to identify time periods which show atypical characteristics. The production engineer must identify the cause or causes for the irregular process behavior. Many potential production problems have characteristic signatures (patterns) that can be detected in the multivariate quality vector. Each pattern can be used as a basis element in a process-oriented basis. The representation of the multivariate quality vector using this process-oriented basis allows useful diagnostic information from the multivariate SPC data. The set of potential causes is identified by those patterns associated with the largest coefficients in the representation.
展开▼