Publication
Real-time condition monitoring is crucial for the reliable and efficient operation of complex systems. However, relying solely on physical sensors can be limited due to their cost, placement constraints, or inability to directly measure certain critical parameters. Virtual sensing addresses these limitations by leveraging readily available sensor data and system knowledge to estimate inaccessible parameters or infer system states. The increasing complexity of industrial systems necessitates deployments of sensors with diverse modalities to provide a comprehensive understanding of system states. These sensors capture data at varying frequencies to monitor both rapid and slowly varying system dynamics, as well as local and global state evolutions of the systems. However, this leads to heterogeneous temporal dynamics, which, particularly under varying operating conditions, pose a significant challenge for accurate virtual sensing. To address these challenges, we propose a novel Heterogeneous Temporal Graph Neural Network (HTGNN) framework for virtual sensing. HTGNN explicitly models signals from diverse sensors as distinct node types within a graph structure, enabling the capture of complex relationships between sensors. Additionally, HTGNN integrates context from operating conditions, derived from exogenous variables such as control settings and external environmental factors into the model architecture. This integration allows HTGNN to adapt to diverse operating and environmental conditions, ensuring accurate and robust virtual sensing. We evaluate the effectiveness of HTGNN using two newly released, publicly available datasets: a test-rig bearing dataset with diverse load conditions for bearing load prediction and a comprehensive year-long simulated dataset for train–bridge–track interaction, aimed at predicting bridge live loads. Our extensive experiments demonstrate that HTGNN significantly outperforms established baseline methods in both bearing and bridge load prediction tasks, particularly under highly varying operating conditions. These results highlight HTGNN's potential as a robust and accurate virtual sensing approach for complex systems, paving the way for improved monitoring, predictive maintenance, and enhanced system performance. Our code and data are available under https://github.com/EPFL-IMOS/htgnn.