Optimal transport detects anomalies in industrial sensor data
Optimal Transport for Efficient, Unsupervised Anomaly Detection on Industrial Data
Machine Learning
Summary
Finding unexpected problems in machines is important but hard when lots of data has no labels. The authors developed a method using Optimal Transport that learns what normal sensor readings look like and spots unusual behavior without needing labelled examples. Their approach adjusts to changes in data over time and handles noise and missing information well. They tested it on data from ships and heating systems, where it found issues accurately with fewer false alarms and worked efficiently.
What this means in practice
- •For industrial operations teams: Monitor machine sensor data in real-time to detect faults without needing labelled training data or heavy configuration.
- •For building maintenance engineers: Identify unusual patterns in HVAC sensor readings to prevent system failures by using adaptive anomaly detection that handles noisy and incomplete data.
Authors
Abigail Langbridge, Fearghal O'Donncha, James T Rayfield, Bradley Eck
Abstract
Effective anomaly detection frameworks are a central pillar of the Industry 4.0 paradigm. In this paper, we introduce an Optimal Transport (OT)-based framework for anomaly detection, designed to detect deviations from normal behaviour in time-series sensor data. The OT-based method requires minimal user input and adapts to real-time data without the need for labelled training data. Our method effectively addresses existing limitations related to data labelling, generalisability, and scalability, demonstrating resilience against short-term fluctuations, noise, and data gaps - common challenges in industrial environments. Additionally, our method provides counterfactual explanations improving the auditability of the approach when deployed in industrial settings. The proposed method learns the mapping between normal and observed operating conditions through a sliding reference window that adapts to the dynamicity of the data. We evaluate our approach on three industrial datasets, from shipping, industrial HVAC systems, and publicly available benchmark data. The method was highly effective in identifying anomalies and reducing false positives, outperforming traditional methods, while maintaining computational efficiency and ease of configuration.