Journal Article

B-spline empowered functional Bayesian optimization for composite fuselage shape control

Li, Yongxiang, Liu, Ziyu

2026

Overview

Composite fuselage shape control is a representative black-box optimization problem, where the fuselage shape is a high-dimensional functional response with strong spatial correlations and each evaluation is costly. Existing functional Bayesian optimization methods for this problem often suffer from prohibitive computational burdens due to high dimensionality and inadequate representations of spatial correlations in fuselage shapes. To address these limitations, this study proposes a B-spline empowered functional Bayesian optimization (BFBO) method, in which fuselage shapes are represented using low-dimensional B-spline control points and their correlations are modeled using a multivariate Gaussian process (MGP). Based on the predictive distribution of control points, a generalized chi-square expected improvement criterion is derived to guide the optimization process. Numerical results show that BFBO achieves improved performance in terms of convergence speed and optimization accuracy, demonstrating its effectiveness for fuselage shape control and its potential as a practical optimization framework for the design and manufacturing of engineering systems involving functional responses.

• Publisher: IOP Publishing
• Subject: Airframes / Bayesian analysis / Correlation / Design optimization / Fuselages / Gaussian process / Optimization / Predictive control / Shape control