Multistage Enhanced Intelligent Aerodynamic Design Method for Hypersonic Forebody/Inlet Integration

Abstract

The aerodynamic design of hypersonic forebody/inlet integration systems represents a challenging multiobjective optimization problem characterized by high-dimensionality, strong coupling, and expensive evaluations. To move beyond conventional empirical approaches, this paper introduces a knowledge-driven intelligent design framework that systematically integrates feature sampling, surrogate modeling, and multistage optimization. At its core, a multistage enhanced multiobjective particle swarm optimization (MSE-MOPSO) algorithm is developed. This novel optimizer incorporates strategies such as chaotic map, random walk, and opposition-based learning to dynamically balance global exploration and local exploitation across different search phases. The framework is supported by an adaptive distribution sampling optimization for improved data efficiency in high-dimensional design spaces. Validation results demonstrate that the proposed MSE-MOPSO achieves a significant improvement in convergence performance (approximately 20% over standard multiobjective particle swarm optimization) while yielding an optimized configuration with enhanced total pressure recovery and lift-to-drag ratio. Further flowfield analysis provides physical insights into the aerodynamic gains, linking the optimization outcomes to underlying flow mechanisms. The study establishes an effective strategy-framework-pattern paradigm for intelligent aerodynamic design, offering a scalable and adaptive solution for complex engineering system optimization.