Institution profile

ONERA

Academic institutioneurope · fr
Official website
Research library75linked papers
Opportunities0open roles
Selected work

Representative Papers

Bayesian optimization for mixed variables using an adaptive dimension reduction process: applications to aircraft design

Apr 11, 2025

In multidisciplinary design optimization—particularly aircraft design—the presence of high-dimensional mixed variables (continuous, integer, and categorical) causes combinatorial explosion in the hyperparameter space of Bayesian optimization surrogate models. To address this, we propose a Partial Least Squares (PLS)-based adaptive dimensionality reduction framework. It dynamically learns variable coupling structures to significantly compress the surrogate hyperparameter space, while integrating mixed-variable encoding with an adaptive hyperparameter selection mechanism to balance modeling accuracy and tuning efficiency. Evaluated on analytical benchmarks and two real-world aircraft design cases, our method achieves >30% faster convergence and improves optimal solution quality by 12–18% compared to genetic algorithms, while reducing hyperparameter count by ~60%. Our core contribution is the first application of PLS for hyperparameter pruning in Bayesian optimization, enabling efficient and robust optimization under high-dimensional mixed-variable settings.

8 citations3 influentialRead paper

Constrained Multi-Objective Bayesian Optimization with Application to Aircraft Design

Jun 20, 2022AIAA AVIATION 2022 Forum

Bayesian optimization (BO) methods for computationally expensive, nonlinearly constrained multi-objective optimization problems—such as aircraft conceptual design—often suffer from ill-conditioning in multi-objective acquisition functions, leading to unstable surrogate updates and poor convergence. Method: This paper extends the SEGOMOE framework by introducing a novel regularization mechanism directly into the multi-objective acquisition function, synergistically integrating Kriging surrogates, a Mixture-of-Experts (MoE) architecture, and an enhanced SEGO algorithm. Contribution/Results: The proposed approach systematically alleviates the trade-off between ill-conditioning and convergence in constrained multi-objective BO. Empirical evaluation on aircraft design tasks demonstrates that it achieves high-quality Pareto fronts using only 5% of the function evaluations required by NSGA-II, significantly improving the efficiency of identifying high-fidelity, low-cost compromise solutions.

8 citationsRead paper

Surrogate-Based Optimization of System Architectures Subject to Hidden Constraints

Jul 27, 2024AIAA AVIATION FORUM AND ASCEND 2024

This work addresses the challenge of implicit constraints—manifested as evaluation failures—arising from unreliable physics-based simulations in system architecture optimization. To tackle this, we propose a surrogate modeling framework that integrates probabilistic feasibility prediction with Bayesian optimization. Methodologically, we introduce a novel hybrid discrete Gaussian process to model the Probability of Validity (PoV), coupled with an interior-point selection strategy based on a minimum PoV threshold; the framework natively supports hierarchical design variables and multi-objective optimization. Our approach achieves the first successful solution for a jet engine architecture optimization task with a 50% simulation failure rate. Across multiple synthetic benchmarks and real-world case studies, it significantly improves convergence robustness and optimization success rate. The implementation is publicly available as the SBArchOpt Python library.

2 citationsRead paper

Tensor-based multivariate function approximation: methods benchmarking and comparison

Jun 05, 2025

This study addresses the tensorization and approximation of multivariate functions. We construct the first standardized benchmark suite of multivariate functions—including nonsmooth, symmetric, and irrational types—and systematically convert them into high-dimensional tensors. A comprehensive evaluation framework is proposed to quantitatively assess tensor-based surrogate models—including the multivariate Loewner framework (mLF), rational approximation, and tensor neural networks—across accuracy, computational efficiency, and hyperparameter robustness. We provide a novel in-depth analysis of mLF’s applicability boundaries, accompanied by reproducible implementations. Additionally, we develop a unified evaluation protocol and practical guidelines. The outcomes constitute an interdisciplinary tensor approximation toolchain (integrated into MDSPACK), supporting model selection and algorithmic improvement. This work establishes a reusable, open benchmarking infrastructure for scientific computing and surrogate modeling.

1 citationsRead paper
Recent publications

Latest Papers

FleetScape: A Mixed Reality Sandtable for Spatial Supervision and Control of Scalable Drone Fleets

Jul 28, 2026

Current drone swarm control interfaces rely on single-vehicle interaction paradigms, which struggle to meet the complex supervisory demands of large-scale formations across spatial, temporal, and safety dimensions. This work reframes swarm control as a spatial interaction task and introduces FleetScape, a novel mixed reality (MR) sandbox system that externalizes multidimensional data—such as mission status, environmental context, and safety constraints—into spatialized visualizations. FleetScape enables seamless transitions between manual intervention and autonomous supervision, integrating high-fidelity building inspection simulation, synchronized multi-drone–environment data streams, and spatial interaction mechanisms. User studies demonstrate that FleetScape significantly enhances operators’ situational awareness and clarifies transitions between control modes; however, as swarm size increases, situational awareness degrades, prompting users to adapt their supervisory strategies.

0 citationsRead paper

Hard Guarantees at a Measured Price: Entropy-Stable Learned Finite Volumes for Compressible Flow

Jul 22, 2026

This work addresses the limited physical admissibility and generalization capability of existing learning-based solvers for compressible flows under comparable computational costs. The authors propose a learning-based finite volume scheme for the two-dimensional Euler equations on unstructured grids, which enforces physical consistency through entropy-stable interior fluxes and admissibility constraints. Central to their approach is an “unlearned skeleton” architecture that integrates scale-invariant inputs, a lower bound on specific entropy, and a spatial gating mechanism, enabling significant improvements in Mach number extrapolation and generalization to complex boundaries without retraining. Experiments demonstrate that the method remains free of non-physical failures across 36 rollout steps; the unlearned skeleton achieves optimal performance on periodic problems; and the spatial-gated variant outperforms baseline methods on unseen wall geometries, reducing Mach extrapolation error by up to 33%.

0 citationsRead paper