A Genetic Algorithm‐Optimized LSTM Framework for Mechanics‐Conditioned Reconstruction of Stress–Strain Curves in High‐Performance Cementitious Materials

Mechanics‐conditioned reconstruction of nonlinear stress–strain responses in high‐performance cementitious materials is essential for interpreting experimentally characterized material behavior and supporting the safe design of underground, geotechnical, and structural systems. Owing to the heterogeneous nature and complex damage evolution of these materials, conventional constitutive models often fail to capture their full mechanical behavior under varying loading conditions. This study introduces a Genetic Algorithm Optimized Long Short‐Term Memory (GA‐LSTM) framework that combines evolutionary optimization and deep learning to reconstruct complete stress–strain curves of high‐performance cementitious materials conditioned on known or experimentally derived mechanical descriptors, including elastic modulus, peak stress, peak strain, and toughness. Experimental data from uniaxial compression tests were systematically preprocessed using cleaning, normalization, interpolation, and Savitzky–Golay filtering to preserve physical integrity. The genetic algorithm optimized key hyperparameters, including network depth, hidden units, dropout rate, and sequence length under a multi‐objective, physics‐informed strategy. The proposed model achieved superior curve‐reconstruction accuracy (, RMSE MPa) with minimal deviation in peak stress and post‐peak toughness. Comparative statistical analyses (ANOVA, Tukey's HSD, and Pearson correlation) verified the model's reliability and robustness. The developed GA‐LSTM framework offers a scalable and interpretable tool for mechanics‐conditioned constitutive reconstruction, mechanical‐response characterization, and geomechanical simulation within experimentally characterized high‐performance cementitious material domains.