Digital Twin-Based Real-Time Blockage Detection for Liquid-Cooled Power Electronic Systems

Power electronic systems in autonomous vehicles and mission-critical applications are highly susceptible to overheating due to operational inefficiencies. These systems often rely on liquid cooling loops to maintain safe temperatures. However, blockages in cooling loops can lead to temperature rises, component degradation, and eventual system failure. This paper presents a digital twin (DT) approach that enables early detection of coolant-loop blockages and provides a framework for developing autonomous rectification of cooling loop blockages using only temperature sensors. The proposed method integrates transient and steady-state thermal models to detect anomalies and predict system behavior in real-time. A robust detection logic, comprising five independent threshold-based switches, identifies deviations from normal behavior while avoiding false positives. Experimental validation is conducted using a physical testbed to evaluate performance under varying blockage conditions. The system detects blockages in less than 1 min, with temperatures remaining roughly 25 °C below the maximum operating temperature of 80 °C, well within safe operating limits. This approach eliminates the need for additional sensing hardware and is well suited to embedded and remote applications.