Distribution-Free Process Monitoring with Conformal Prediction
Christopher Burger
TL;DR
This work addresses the fragility of traditional Statistical Process Control (SPC) under non-normal and high-dimensional data by introducing a distribution-free, model-agnostic framework based on Conformal Prediction (CP). It develops Conformal-Enhanced Control Charts that replace parametric limits with conformal thresholds, enabling uncertainty visualization and leading indicators such as uncertainty spikes, and Conformal-Enhanced Process Monitoring that reframes MSPC as online anomaly detection using conformal p-values. The approach provides marginal coverage guarantees and a simple, interpretable monitoring signal via a p-value chart, while preserving the intuitive, visual nature of conventional charts. This framework promises more robust, proactive quality control suitable for complex modern manufacturing environments, with practical relevance supported by calibration-based procedures and clear future directions for adaptive non-conformity scores and real-world validation.
Abstract
Traditional Statistical Process Control (SPC) is essential for quality management but is limited by its reliance on often violated statistical assumptions, leading to unreliable monitoring in modern, complex manufacturing environments. This paper introduces a hybrid framework that enhances SPC by integrating the distribution free, model agnostic guarantees of Conformal Prediction. We propose two novel applications: Conformal-Enhanced Control Charts, which visualize process uncertainty and enable proactive signals like 'uncertainty spikes', and Conformal-Enhanced Process Monitoring, which reframes multivariate control as a formal anomaly detection problem using an intuitive p-value chart. Our framework provides a more robust and statistically rigorous approach to quality control while maintaining the interpretability and ease of use of classic methods.