A DATA-DRIVEN METHOD FOR PREDICTING THE OVERALL PERFORMANCE OF A THREE-ROTOR DUAL-VARIABLE CYCLE ENGINE

In order to meet the high maneuverability and long-range design requirements of the new generation of aeroengines, variable cycle engines (VCEs) with variable geometry characteristics and multiple mode adjustments have become a mainstream research direction. A three-rotor dual-variable cycle engine (TDCE) configuration with both turbofan and turbojet modes is proposed for the 0-30 km, 0-5 Ma flight envelope. Based on an integrated aeroengine simulation platform (IASP), a comprehensive performance model of the variable cycle engine is established, and the engine characteristics under four modes, low-subsonic mode (M-Lsub), middle-supersonic mode (M-Msup), high-supersonic mode (M-Hsup), and top-supersonic mode (M-Tsup) are analyzed with various adjustment features. For this configuration, a surrogate model for predicting overall performance deviations is established using a radial basis function neural network to quantify the impact of design performance deviations of rotating components on the overall engine performance parameters. neural network. The overall engine prediction model based on the Radial Basis Function (RBF) neural network approach shows high accuracy with an R-squared value exceeding 0.85, according to the experimental results. This indicates that the model fits the test set well and allows for a rapid evaluation of the effect of component design on overall performance. The outcomes of the investigation can provide technical assistance in the design of the TDCE configuration and enhance the effectiveness of the collaborative design procedure for advanced aeroengines.