24 June 2025 in Data Science and Engineering
Industrial processes often involve the generation—and the analysis—of multivariate time series data, which poses several challenges from the anomaly detection perspective. In addition to the need to detect previously unseen anomalies, the high dimensionality of industrial datasets introduces the complexity of simultaneously analyzing multiple features and their interactions. Finally, industrial datasets are typically highly imbalanced, with minimal information on anomalous processes. To address these issues, we propose a novel anomaly detection framework that introduces two embedding models, based on Time2Vec and Discrete Wavelet Transforms, leveraging their capabilities to represent multivariate time series as vectors while capturing and preserving temporal dependencies and combining them with several classifiers to enhance the overall performance of anomaly detection. We tested our solution using a publicly available benchmark dataset and a real industrial use case, particularly data collected from a Bonfiglioli gear manufacturing plant. The results demonstrate that, unlike traditional reconstruction-based autoencoders, which often struggle with sporadic noise, our embedding-based solutions maintain high performance across various noise conditions.
26 November 2025 in Internet of Things
13 July 2026 in SSRN Electronic Journal
Predictive Maintenance (PdM) has become a key component of modern industrial systems, enabling data-driven maintenance strategies that improve reliability, reduce downtime, and optimize operational costs. Among PdM tasks, Remaining Useful Life (RUL) estimation plays a central role by forecasting the time before a component reaches the end of its operational life, thereby supporting proactive maintenance planning and informed decision-making. However, accurate RUL estimation remains challenging due to complex degradation dynamics, varying operating conditions, and limited failure data. This paper presents an industrial RUL estimation platform developed for MEP Saws to monitor the health of band saw blades and support human-centered predictive maintenance. The proposed solution integrates a scalable big-data architecture for industrial data acquisition, storage, and processing with a machine-learning-based prognostic service. At the core of the platform is a two-phase prognostic ML system that continuously assesses blade health through sensor-derived time-series features. The first phase identifies early degradation states, while the second phase predicts imminent failure through a staged warning mechanism designed to balance timely detection and false-alarm mitigation. Presenting therefore an end-to-end, RUL-based prognostic system that combines scalable industrial data management, machine learning, and human-centered decision support, providing actionable insights for more effective maintenance strategies in real-world manufacturing environments.