Journal of Materials Informatics, Journal Year: 2025, Volume and Issue: 5(3)
Published: April 16, 2025
Ablative materials, a special type of thermal protection material, are widely used in extremely high-temperature environments such as hypersonic vehicles and re-entry capsules. They effectively mitigate heat conduction to the interior through ablation at material surface. Based on traditional physical models machine learning techniques, we systematically investigated mapping relationship between multiple parameters responses within carbonized layer pyrolysis ablative materials. By employing high-throughput modeling sure independence screening sparsity operator (SISSO) method for feature selection, first revealed that different layers dominated by distinct properties (e.g., density, conductivity, capacity, etc. ). The explicit relationships functioning features response curves associated with single-/double-layer structures well established. After key parameter based SISSO, further developed deep neural network surrogate model, capable accurately predicting entire process layers.
Language: Английский