Regularization-based Continual Learning for Anomaly Detection in Discrete Manufacturing

TL;DR

This paper explores regularization-based continual learning methods to improve anomaly detection in discrete manufacturing, enabling models to adapt to process changes without forgetting previous knowledge.

Contribution

It evaluates and compares various regularization strategies for continual learning applied to industrial anomaly detection using real-world metal forming data.

Findings

Regularization strategies improve adaptability to new manufacturing processes.

Certain methods outperform others in maintaining detection accuracy over time.

The approach enhances robustness of anomaly detection in dynamic industrial environments.

Abstract

The early and robust detection of anomalies occurring in discrete manufacturing processes allows operators to prevent harm, e.g. defects in production machinery or products. While current approaches for data-driven anomaly detection provide good results on the exact processes they were trained on, they often lack the ability to flexibly adapt to changes, e.g. in products. Continual learning promises such flexibility, allowing for an automatic adaption of previously learnt knowledge to new tasks. Therefore, this article discusses different continual learning approaches from the group of regularization strategies, which are implemented, evaluated and compared based on a real industrial metal forming dataset.