The Model Driven Multivariate Control Chart (MDMVCC) platform has two primary functions: monitoring and diagnosing.
• Use multivariate control charts to monitor a multivariate process.
• You can interactively drill down to investigate the contributions of individual variables to the overall signal to diagnosis the process.
For more information about multivariate control charts, see Kourti and MacGregor (1996).
You can construct a model driven multivariate control chart using current or historical data. A control chart is considered to be a Phase I chart if the limits are based on current data; a control chart is considered to be a Phase II chart if the limits are based on target statistics from a historical data set. In a Phase I chart, you check that the process is stable, and if so, establish target statistics for the process. In a Phase II chart, you use the target statistics from Phase I in order to monitor new process observations.
To construct a Phase II model driven multivariate control chart, first identify a period of time during which the process is stable and capable. Then, perform the following steps:
1. Develop a Phase I control chart to verify that the process is stable over this period. The data used in Phase I provides a historical data set.
2. Save the target statistics for this historical data set.
3. Monitor the on-going process using the Phase II control chart.
See Example of a Model Driven Multivariate Control Chart with Historical Data.