A data-driven synergistic optimization framework for thermo-mechanical properties of oriented fiber-reinforced aerogel composites

In thermal protection systems particularly for aerospace and energy applications, the development of thermal insulation materials that simultaneously maintain mechanical robustness and dimensional stability under extreme conditions remains a challenge. Although oriented fiber-reinforced silica aerogel composites exhibit superior thermal and mechanical performances, their further application is hindered by the intrinsic trade-offs between microstructure and macroscopic properties. To address this limitation, this work proposes a data-driven synergistic optimization framework for oriented fiber-reinforced silica aerogel composites, facilitated by multiscale structure optimization to achieve a multifunctional integration. A nanoscale-informed thermo-mechanical theoretical model based on the real nanostructure of silica aerogels was developed to quantitatively correlate the density with thermal conductivity, elastic modulus, and thermal expansion coefficient. Furthermore, a hierarchically coupled multiscale modeling strategy for fiber-reinforced aerogel composites was proposed to achieve nano-micro-macro matched thermo-mechanical numerical predictions and experimentally validated using samples prepared in-house. We developed a multi-objective optimization approach that combines a finite-element (FE) database with an artificial neural network (ANN) surrogate and the Non-dominated Sorting Genetic Algorithm II (NSGA-II). The integrated optimization delivers exceptional properties: ultralow thermal conductivity (0.0295 W m⁻¹ K⁻¹), high elastic modulus (24.06 MPa), and low thermal expansion coefficient (4.87 × 10⁻⁶ K⁻¹), at a fiber volume fraction of 12.2% and an orientation angle of 18.5°. This work can resolve the thermo-mechanical performance–microstructure trade-offs, advancing the collaborative optimization of oriented fiber-reinforced aerogel composites. The present data-driven optimization framework could be straightforward for more general multifunctional thermal insulation composites.