Integrated cooling and power generation system addressing thermal protection and long-term power supply demands for scramjets: A novel design approach

Hypersonic vehicles represent a new strategic frontier in aerospace technology, with the scramjet as their mainstream propulsion system. However, the scramjet presents significant challenges in thermal protection and electricity supply due to the long-term high-Mach-number operation and the lack of power drive shaft. Designing integrated cooling and power generation systems (ICPGSs) based on the supercritical CO₂ Brayton cycle can effectively solve these problems, with the advantages of high efficiency, compactness and non-coking. Nonetheless, its long-term operation under off-design conditions deviates from the rated-design condition. A novel integrated design approach is proposed by combining ensemble learning with heuristic algorithms to ensure a truly optimal all-condition performance. Different from existing studies that treat off-design analysis as a post-hoc validation, this design approach embeds a feedback mechanism into the rated-design optimization through ensemble-learning-based surrogate models and a synchronous iterative optimization of design and operation through heuristic algorithms, resulting in an improvement of up to 21.59%, which yields both a lower heat dissipation load and a larger electricity supply capacity. To reduce the unacceptable time consumption of numerical solutions in the optimization process, six surrogate models are considered and compared. A rapid design method based on a Bayesian-optimized XGBoost model is proposed, which enables a rapid and accurate prediction of system performance under diverse conditions, with the reduction in time consumption reaching up to 96.88%, and the mean absolute percentage error being only 1.59%. The findings provide theoretical support for the optimal design of ICPGSs and the performance enhancement of scramjets.