Early and accurate fault detection and diagnosis for chemical production plants minimizes downtime, increases the safety of plant operations, and reduces manufacturing costs. Univariate charts have a limited ability to detect and diagnose faults in multivariable processes. This has led industry to develop more effective process monitoring methods leveraging the advances in artificial intelligence and the increase in computing power.
This video series will illustrate Multivariate Statistical Process Control (MSPC) - an industrially proven method to efficiently and clearly distinguish normal from abnormal process behavior. This involves the use of key statistical machine learning techniques. Learn how to apply this to a simulated methanol-ethanol distillation column for predicting process upsets. The app is free to download from MATLAB File Exchange.
Part 1: Multivariate Analysis for Process Monitoring Learn about the industrial applications of multivariate analysis and success stories.
Part 2: Phase I: Modeling Steady State Conditions Create a PCA model from steady state conditions using a GUI.