Research on Virtual Prototype and Digital Test Method of Pump-Fed Propulsion System

Regarding the difficulty in designing the complex propulsion system, research on the virtual prototype and digital testing methods of the pump-fed propulsion system was carried out. A set of generalized mathematical models of the propulsion system was built based on the Modelica language. The continuous-discrete hybrid simulation was adopted to address the long-term on-orbit mission profile. A digital testing system for the propulsion system virtual prototype was developed, including model calibration, surrogate models based on deep learning, sensitivity analysis, and uncertainty quantification. Next, the above tools were used to analysis a propulsion system under development. First, the unmeasurable parameters were identified to improve the accuracy of the model. Then, the identified model was used to study the pressurization performance as well as the impact of mixing ratio control on engine performance. Further, a surrogate model of 18 parameters and the remaining pressure of the gas bottle was constructed, and the global sensitivity and uncertainty quantification analysis was conducted. The results showed that a 90% confidence interval of the residual pressure of the gas bottle is [4.8 MPa, 7.5 MPa], which meets the requirements of the pressure regulator. The virtual prototype and digital testing methods have great application potential in assisting system design, predicting system failures and identifying system risks.