Physics-Informed Slope Dynamics for Low-Computational Intelligent Diagnosis and Fault-State Evaluation in Spacecraft Propulsion

This study addresses the critical challenge of propulsion health assessment in space missions by proposing a novel physics-grounded framework for fault-state evaluation and valve-opening ratio estimation. Inspired by similarity analysis and physics-based feature design, the method leverages transient pressure–time slopes as a core physical feature, paired with valve-opening parameters, for efficient anomaly classification. Validation used the PHMAP 2023 Data Challenge dataset, replicating a 2 MPa water-propellant spacecraft propulsion system monitored by seven pressure sensors at 1 kHz over three valve open–close cycles. The framework extracts pressure-slope features from transient water-hammer behaviors, establishes a deterministic mapping to predict actual valve openings during solenoid valve faults, and distinguishes normal, solenoid-valve-fault, and bubble-contamination conditions. It achieved 100% classification accuracy on the test dataset, including unseen spacecraft and novel fault modes, with a maximum opening-estimation error of merely 0.92%. Combining strong interpretability with low computational complexity, the physics-driven approach circumvents the heavy resource demands of traditional models and offers high potential for real-time onboard implementation. It bridges physical principles and data-driven execution, providing critical technical support to enhance mission reliability and autonomous maintenance in satellite propulsion and unmanned aerospace vehicles.