Reinforcement Learning for Accelerated Aerodynamic Shape Optimisation

2025-07-23

AI Summary

To address the high computational cost of CFD simulations, low efficiency of dimensionality reduction, and poor interpretability of results in aerodynamic shape optimization, this paper proposes an adaptive optimization framework integrating reinforcement learning with surrogate modeling. Methodologically, it introduces an Actor-Critic–guided MCMC sampling strategy, coupled with local parameter neighborhood temporal freezing and manifold-aware dimensionality reduction, to accelerate global search while minimizing simulation calls. Additionally, a differentiable surrogate model combined with gradient-guided extremum localization enables quantitative importance assessment and physical interpretation of critical geometric features. Experimental results on canonical fluid optimization tasks demonstrate over 60% reduction in CFD evaluations, significantly improved convergence speed, and support for attribution analysis of flow-field extremum origins. The framework establishes a new, efficient, and physically interpretable paradigm for data-driven aerodynamic design.

Technology Category

• Search and Optimization: Sampling/Simulation-based Search
• Machine Learning: Optimization
• Reasoning under Uncertainty: Stochastic Optimization

Application Category

• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

We introduce a reinforcement learning (RL) based adaptive optimization algorithm for aerodynamic shape optimization focused on dimensionality reduction. The form in which RL is applied here is that of a surrogate-based, actor-critic policy evaluation MCMC approach allowing for temporal 'freezing' of some of the parameters to be optimized. The goals are to minimize computational effort, and to use the observed optimization results for interpretation of the discovered extrema in terms of their role in achieving the desired flow-field. By a sequence of local optimized parameter changes around intermediate CFD simulations acting as ground truth, it is possible to speed up the global optimization if (a) the local neighbourhoods of the parameters in which the changed parameters must reside are sufficiently large to compete with the grid-sized steps and its large number of simulations, and (b) the estimates of the rewards and costs on these neighbourhoods necessary for a good step-wise parameter adaption are sufficiently accurate. We give an example of a simple fluid-dynamical problem on which the method allows interpretation in the sense of a feature importance scoring.

Problem

• Develop RL-based optimization for aerodynamic shape design
• Reduce computational effort in CFD simulations
• Interpret optimization results for flow-field improvement

Innovation

• RL-based adaptive optimization for aerodynamic shapes
• Surrogate actor-critic MCMC with parameter freezing
• Local parameter changes speed up global optimization