Anomaly Detection in the Task of Remote Monitoring of Large Rotary Machines Based on the Signal-Based Digital Twin

Anomaly detection plays a crucial role in remote monitoring systems by identifying deviations from normal operating conditions, which may indicate potential faults or performance degradation. Traditional anomaly detection methods often rely on statistical techniques or rule-based algorithms, which may struggle to capture complex relationships in data and adapt to dynamic operational conditions. In this context, the emerging paradigm of signal-based digital twins offers a promising solution. A signal-based digital twin is a virtual model that mirrors the dynamics of a physical machine based on real-time sensor information. Using advanced data analytics, machine learning algorithms, and physics-based models, signal-based digital twins provide a powerful framework for anomaly detection. Integrating such models into remote monitoring systems for decision-making purposes is one of the current challenges. This approach enables proactive maintenance strategies, reduces unplanned downtime, and increases overall equipment reliability. This paper aims to explore the application of signal-based digital twin for anomaly detection in remote monitoring of large rotating machines. It describes the concept of signal-based digital twin, its representation, clarifies the role in anomaly detection, and presents a case study illustrating the practical use in remote monitoring system. By bridging the gap between data analysis, machine learning, and domain knowledge, signal-based digital twins offer a significant framework for advancing techniques in proactive maintenance, ensuring the reliability, safety, and efficiency of industrial operations.