Dynamic multi-objective optimization of scramjet inlet based on small-sample Kriging model

Abstract

The generic inlet is depicted based on a smooth Bézier curve, and the results and insights from high-dimensional dynamic multi-objective optimization of small-sample high Mach number axisymmetric scramjet inlets are discussed in detail. The optimization is performed by integrating a Kriging surrogate model-assisted improved congestion distance multi-objective particle swarm optimization algorithm and computational fluid dynamics simulation. The steady-state flow field is derived by solving the Euler equation using self-developed hypersonic internal and external flow coupling numerical simulation software, which is designed to minimize inlet surface area and drag while improving the total pressure recovery factor. The results revealed that the generic inlet can achieve a total pressure recovery capability exceeding 95%, with minimal surface area and drag. The prediction error, mean absolute percentage error, of the performance dynamic surrogate model based on Kriging is less than 1%, and the performance parameter optimization shows an improvement greater than 8% compared to static multi-objective optimization results. Ultimately, the obtained Pareto solution set is grouped by K-means feature recognition, contributing to a comprehensive understanding of the flow physics knowledge related to optimal geometric local shape control. Finally, an inward-turning inlet is designed by streamline tracking technology based on the optimized axisymmetric scramjet inlet primary flow field.