03.1 - Aerodynamics – CFD Methods and Validation
RESEARCH ON AERODYNAMIC-STEALTH INTEGRATED DESIGN METHODOLOGY BASED ON MPPO ALGORITHM
K. Gan¹, H. Zhao¹, S. Wang¹; ¹Northwestern Polytechnical University, China
Traditional design optimization methods often suffer from low efficiency and a tendency to converge to local optima when applied to the multidisciplinary, high-dimensional, and nonlinear problem of aircraft aerodynamic- stealth design. To address these issues, a Modified Proximal Policy Optimization (MPPO) algorithm is proposed. By optimizing the extraction of the surrogate objective function from the traditional PPO algorithm, the MPPO algorithm effectively enhances the quality of samples selected by the model. This mechanism reduces unnecessary perturbations to the environment while simultaneously improving the algorithm's global search capability. Integrating MPPO with surrogate optimization techniques, this study develops an MPPO-based integrated aerodynamic-stealth optimization design framework.To validate the effectiveness of the proposed MPPO algorithm, aerodynamic-stealth optimization is conducted on both the NACA 65,3-014 airfoil and an X- 47B-like configuration. The results are compared against those obtained using the traditional Particle Swarm Optimization (PSO) algorithm. The optimization case studies demonstrate that, compared to the PSO algorithm, the MPPO algorithm significantly accelerates the convergence speed of the aerodynamic-stealth shape optimization. Furthermore, the final optimized designs obtained by MPPO outperform those from PSO in both aerodynamic and stealth performance metrics. These results indicate that the proposed MPPO-based integrated aerodynamic-stealth design methodology possesses higher optimization efficiency and stronger global optimization capability.
DOI: 10.71945/icas2026_0462