This paper provides the first comprehensive evaluation and analysis of modern (deep-learning) unsupervised anomaly detection methods for chemical process data. We focus on the Tennessee Eastman process dataset, which has been a standard litmus test to benchmark anomaly detection methods for nearly three decades. Our extensive study will facilitate choosing appropriate anomaly detection methods in industrial applications.
@article{arxiv.2303.05904,
title = {Deep Anomaly Detection on Tennessee Eastman Process Data},
author = {Fabian Hartung and Billy Joe Franks and Tobias Michels and Dennis Wagner and Philipp Liznerski and Steffen Reithermann and Sophie Fellenz and Fabian Jirasek and Maja Rudolph and Daniel Neider and Heike Leitte and Chen Song and Benjamin Kloepper and Stephan Mandt and Michael Bortz and Jakob Burger and Hans Hasse and Marius Kloft},
journal= {arXiv preprint arXiv:2303.05904},
year = {2023}
}