Model–data hybrid thermal management and health monitoring of the space station application fluid loop

• CPS-based, high-fidelity space station fluid loop model in MWORKS • Digital twin PHM enables state prediction and mission planning • Hierarchical calibration with residual fitting boosts model fidelity • Validated on Mengtian loop; key parameter errors under 5% As China’s space station enters sustained operations, precise thermal management of scientific payloads becomes critical. The application fluid loop that is the core active thermal control subsystem, governs heat removal and temperature uniformity, but traditional telemetry-telecommand workflows cannot meet reliability and efficiency needs under cross-disciplinary payloads, high parallelism, tight resources, and complex space-ground coordination. Building on a cyber-physical systems framework, this study develops model construction and simulation-based application technologies for the fluid loop from a thermal science perspective. A high-fidelity thermohydraulic model integrates first-principles mass-momentum-energy conservation with data-driven components to capture the thermal and hydraulic behavior of drive units, cold plates, piping networks, and scientific payloads. A model-data hybrid-driven PHM framework fuses multi-source telemetry and virtual simulation data, injects typical thermal anomalies (degraded heat transfer, flow maldistribution, pump performance loss), and executes temperature-oriented state prediction to generate enhanced training samples. Case studies show significant gains in thermal state awareness, improved prediction of temperature evolution and safety margins, and more informed mission planning under thermal constraints. Results indicate that model-data hybrid thermal digital twins effectively support safe, efficient, and reliable thermal management of space station experiments, advancing health monitoring, fault diagnosis, and resource scheduling for the application fluid loop.