Multi-strategy progressive intelligent design method of inward turning inlet

The inward turning inlet, as a key component of air-breathing engines, has its aerodynamic performance directly determined by the basic flowfield. With the diversification of design requirements and the complexity of the design space, intelligent technologies provide new approaches for inlet research. However, current basic algorithms and sequential modes result in limited optimization outcomes. This paper creates a physics-inspired virtual image learning strategy to enhance population diversity and algorithm convergence, and integrates chaos mapping to construct a multi-stage progressive optimization framework. Subsequently, a Multi-Strategy Progressive MOPSO (MSP-MOPSO) is established by combining it with the Multi-Objective Particle Swarm Optimization (MOPSO) algorithm. With the assistance of small-sample intelligent sampling and the surrogate model, optimization of the basic flowfield for inviscid drag and total pressure recovery coefficient is conducted. Comparison with various optimization algorithms verifies that MSP-MOPSO achieves the fastest convergence speed and the best convergence position. Compared to the MOPSO results, the inviscid drag is reduced by 30.83%, and the total pressure recovery coefficient is increased by 7.49%. Furthermore, the effects of the multi-strategy approach and the physical flow of the configuration scheme are analyzed. Finally, the inward-turning inlet is designed through streamline tracing.