Multivariate Time Series Anomaly Detection in Industry 5.0

2025-03-20

AI Summary

Addressing the challenge of anomaly detection in high-noise, unlabeled, and multi-source heterogeneous industrial multivariate time-series data under Industry 5.0, this paper proposes an embedded robust representation learning framework. Departing from conventional reconstruction-based autoencoders, it integrates TCN/Informer variants with a lightweight downstream classification/scoring module, enabling joint unsupervised and weakly supervised training. The method explicitly models noise robustness via deep temporal embedding. To our knowledge, it is the first to validate cross-noise-condition stability on real-world Bonfiglioli production-line data. Experiments on authentic factory datasets demonstrate that the framework achieves a 12.6% higher F1-score than state-of-the-art autoencoders and maintains >91% detection accuracy under strong noise. This significantly enhances equipment fault early-warning and process safety monitoring capabilities.

Technology Category

• Machine Learning: Representation Learning
• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Data Mining & Knowledge Management: Anomaly/Outlier Detection

Application Category

• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

Industry5.0 environments present a critical need for effective anomaly detection methods that can indicate equipment malfunctions, process inefficiencies, or potential safety hazards. The ever-increasing sensorization of manufacturing lines makes processes more observable, but also poses the challenge of continuously analyzing vast amounts of multivariate time series data. These challenges include data quality since data may contain noise, be unlabeled or even mislabeled. A promising approach consists of combining an embedding model with other Machine Learning algorithms to enhance the overall performance in detecting anomalies. Moreover, representing time series as vectors brings many advantages like higher flexibility and improved ability to capture complex temporal dependencies. We tested our solution in a real industrial use case, using data collected from a Bonfiglioli plant. The results demonstrate that, unlike traditional reconstruction-based autoencoders, which often struggle in the presence of sporadic noise, our embedding-based framework maintains high performance across various noise conditions.

Problem

• Detect anomalies in Industry 5.0 environments
• Analyze noisy multivariate time series data
• Improve anomaly detection with embedding models

Innovation

• Embedding model combined with Machine Learning
• Time series represented as vectors
• High performance in noisy conditions