Improved Fault Detection Approach for Injection Molding

Statistical process control (SPC) for injection molding typically consists of setting alarm limits on a few important molding variables for real-time part containment. A new multivariate statistical process control (MSPC) method is presented that utilizes the measurement of process variable relationships to achieve a more robust quality control system. This approach not only detects process outliers outside of an allowable alarm range but can detect changes in correlation structure between multiple dependent variables, which often is the cause for out of spec parts.