基于神经网络的S 弯喷管-涡扇发动机多维度仿真

The internal flow of the serpentine nozzle is complex, with numerous parameters affecting its aerodynamic performance. Traditional component-based zero-dimensional engine models cannot accurately assess the aerody-namic performance impact of the serpentine nozzle on the overall engine. In this paper, a high-fidelity performance prediction model for serpentine nozzles is established using the back propagation (BP) neural network, and is coupled with a zero-dimensional turbofan engine model. This integrated approach is employed to investigate the influence of the baseline serpentine nozzle on engine speed, altitude characteristics, and component behaviors, as well as the dif-ferences in nozzle and engine performance with various geometric parameters. The results indicate that compared to axisymmetric nozzles, the engine equipped with the serpentine nozzle experiences a decline in performance. Specifi-cally, at sea-level static conditions, the engine’s thrust decreases by 4. 50%, while the fuel consumption increases by 4. 75%, the fan bypass ratio decreases by a maximum of 0. 33% at sea level, accompanied by a reduction in surge margin. Conversely, at an altitude of 12 km, the fan bypass ratio increases by a maximum of 0. 28%, and the surge margin increases. These variation trends in fan operating characteristics can be attributed to the differences in throttling effects of the serpentine nozzle on the core and bypass of mixing chamber at different altitudes. Additionally, increas-ing the length-to-diameter ratio of the serpentine nozzle from 2. 2 to 3. 0 enhances its performance, resulting in an 8. 0% increase in thrust coefficient and a 4. 8% increase in discharge coefficient. The multi-dimensional coupling model between serpentine nozzle and engine established in this study can effectively evaluate the changes in engine perfor-mance and component characteristics following the installation of serpentine nozzles of varying geometric parameters.