Improved air-breathing propulsion system modelling: Physics-Data integrated driven adaptive correction of input and output parameters for onboard applications

Supersonic air-breathing propulsion systems have been widely adopted as the power for reusable supersonic vehicles, marking a major advancement in aerospace technology. The fidelity and computational efficiency of propulsion system models have become critical research priorities in aerospace engineering, particularly as increasingly expanded flight envelopes make the inlet's significant influence on performance models impossible to ignore. Moreover, the precision of traditional physics-based models is often limited by unmodelled factors and measurement uncertainties. In this study, the Improved Integrated Modelling with Adaptive Correction (IIM-AC), driven by physics-data integration, is proposed for propulsion systems to enhance modelling accuracy. This method comprises an experimentally verified component-level model, an Input Parameter Correction Module (IPCM), and an Output Parameter Correction Module (OPCM). The IPCM is to facilitate real-time adjustments to deviations in inlet air-flow and uncertainties in input fuel-flow measurements through its specially designed algorithm. Specifically, within the framework of IPCM, a filtering mechanism is applied to the feedback error based on physically derived methods to enhance the stability of air-flow correction, while a fuel-flow self-adjustment scheme is designed using the rotor speed measurement as the reference. The OPCM consists of a selectively chosen residual learning model for output parameters, which is employed to correct the measurements obtained from output sensors. In the simulated envelopes simulation, the IIM-AC achieved a better correction effect on the concerned parameters, with mean-absolute-percentage errors around 0.02 to 0.23. Its real-time performance on an actual controller fulfils operational requirements (IPCM: 0.0047 ms, OPCM: 0.0974 ms), thereby proving its onboard potential.