Automatically Updated Multivariate Control Charts with JMP Live (2025-US-PO-2526)

This presentation examines a workflow created using a multivariate control chart for an industrial process system. First, a PLS model is created to predict a result from the process, such as yield or the protein content of a product, just to name a couple examples. From there, a model-driven multivariate control chart (MDMCC) is created, representing the variables critical to making that prediction. 

The MDMCC is then uploaded to JMP Live. In addition, the data used to create the model and MDMCC is pulled via an add-in that was created to pull data from a PI data historian. Within JMP Live, scripting is done to automate the data pull so that it refreshes on its own, creating a live model-driven multivariate control chart. It is an excellent tool for process engineers, since it can indicate T2 and squared prediction error deviations. We like these charts because they provide more information than typical control charts, since they not only show deviations in the univariate control of variables but also deviations in the relationship between the x variables, which is not seen by typical control charts.