Durability evaluation of modified basalt fiber reinforced concrete in high altitude cold regions: Based on AE, DIC and deep learning prediction

The complex climate and environmental characteristics in high altitude cold regions, such as low air pressure, strong radiation, and freeze-thaw cycles, have a significant impact on the durability of concrete. In order to improve the durability of concrete in high altitude cold regions, this paper uses nano-SiO 2 -KH560 modified basalt fiber (BF) and nano-SiO 2 to synergistically enhance the durability of concrete, which provides a new idea for improving the durability of concrete in complex environments in alpine regions. The uniaxial compression test (0.3 MPa/s) of concrete was carried out by acoustic emission (AE) and digital image correlation technology (DIC). The characteristics of acoustic emission parameters and crack propagation law of concrete under different curing systems and environments were explored, and the damage degree of concrete was predicted by deep learning method. The results show that with the addition of KHSi-BF, the concentration range of ringing count and peak frequency is further narrowed than that of N2B2. The relationship between the proportion of shear cracks corresponding to K S of each specimen is: KHSi-N2B2 < N2B2 < OPC < LOPC, indicating that the low-pressure curing environment, nanoparticles and modified fibers have a greater impact on concrete, and the KHBF-SiO 2 cementitious system significantly improves the durability of concrete. The results of DIC analysis show that the damage of LOPC is more serious than that of OPC. In addition, the concrete of KHBF-SiO 2 cementitious system shows excellent performance under the action of low pressure curing and freeze-thaw, mainly due to the synergistic effect of KHSi-BF and nano-SiO 2 , which not only enhances the interfacial bonding between BF and cement matrix, but also optimizes the pore structure, and significantly improves the mechanical properties and durability of concrete. The RMSE (0.47 MPa ~ 0.85 MPa) and MAPE (1.59% ~ 2.02%) of CNN-LSTM-Attention are significantly smaller than the index values of CNN-GRU-Attention and LSTM prediction models, which proves that the prediction results of CNN-LSTM-Attention model are more appropriate to the actual values. The research results provide an important reference for the innovation and application development of concrete materials in alpine regions, and also provide a theoretical basis for the safety assessment and analysis of concrete in complex environments.