Distribution-Free Process Monitoring with Conformal Prediction

This paper addresses a key limitation of traditional Statistical Process Control (SPC) – its reliance on statistical assumptions that are often violated in complex manufacturing environments. By integrating Conformal Prediction, the authors propose a more robust and statistically rigorous approach to quality control. The novelty lies in the application of Conformal Prediction to enhance SPC, offering both visualization of process uncertainty and a reframing of multivariate control as anomaly detection. This is significant because it promises to improve the reliability of process monitoring in real-world scenarios.

• •Integrates Conformal Prediction to overcome limitations of traditional SPC.
• •Proposes 'Conformal-Enhanced Control Charts' for visualizing process uncertainty.
• •Reframes multivariate control as anomaly detection using a p-value chart.
• •Aims to provide a more robust and statistically rigorous approach to quality control.