Model-Free Unsupervised Anomaly Detection Framework in Multivariate Time-Series of Industrial Dynamical Systems

TL;DR

This paper introduces a model-free, explainable anomaly detection framework for multivariate industrial time-series, emphasizing reduced data requirements and incremental learning, validated through two case studies.

Contribution

It proposes a novel, physically-inspired feature engineering approach for anomaly detection that does not rely on explicit models, enhancing interpretability and adaptability in industrial settings.

Findings

Effective in detecting anomalies with limited training data

Compatible with incremental learning and operator feedback

Validated on two industrial case studies

Abstract

In this paper, a new model-free anomaly detection framework is proposed for time-series induced by industrial dynamical systems.The framework lies in the category of conventional approaches which enable appealing features such as a learning with reduced amount of training data, a high potential for explainability as well as a compatibility with incremental learning mechanism to incorporate operator feedback after an alarm is raised and analyzed. Although these are crucial features towards acceptance of data-driven solutions by industry, they are rarely considered in the comparisons that generally almost exclusively focus on performance metrics. Moreover, the features engineering step involved in the proposed framework is inspired by the time-series being implicitly governed by physical laws as it is generally the case in industrial time-series. Two examples are given to assess the efficiency of the proposed approach.