@article{Wang_2023, title={3D-dimensional Effective Stress Analysis of Wetting and Wetting Trapping Process in Wet-submerged Loess Tunnel Surrounding Rock Based on BP Neural Network}, volume={10}, url={https://publications.eai.eu/index.php/ew/article/view/3988}, DOI={10.4108/ew.3988}, abstractNote={<p>INTRODUCTION: China’s loess is vast. Loess has apparent high strength and resistance to deformation once encountered with water immersion and humidification, fusible salts precipitated on the surface of soil particles, the soil’s carry alkalization strength is relatively reduced, while the vertical tubular pores in the soil accelerate the infiltration of water, the earth will be in the self-weight or the overlying loads of the additional action of the soil body will produce a significant settlement deformation, which results in the structural damage of the upper building, which is the loss of the wetting of subsidence.</p><p>OBJECTIVES: From China’s practical point of view, the humidification and wetting process of wetted loess tunnel peripheral rock is deeply discussed and analyzed, and the water content distribution characteristics of wetted loess tunnel peripheral rock are sought.</p><p>METHODS: Using the particle swarm algorithm, four neural optimization network models, namely, radial basis neural network (RBFNN), generalized regression neural network (GRNN), wavelet neural network (WNN), and fuzzy neural network (FNN), are simulated and created for the analysis of three-dimensional effective stresses in the process of humidity and wetness subsidence in the surrounding rock of loess tunnels of a northwestern city in China and a central city in China.</p><p>RESULTS: By analyzing the comparison graphs between the predicted and actual values of these four models on the test data of two sets of experimental data, the distribution of the proportion of the expected difference to the true value, and the results of the calculation of the three error indexes, it can be found that when using the four neural networks, namely, RBFNN, GRNN, WNN, and FNN, for the analysis of the three-dimensional effective stresses during the process of increasing wetting and wetting of the surrounding rock of the tunnel in the soil-wetted loess, the prediction performance of the WNN is the best.</p><p>CONCLUSION: The soil’s unsaturated settlement characteristics differ for different water contents and humidification times. The shorter the period, the more the soil column water content difference. With the continuous increase of water content change in the soil layer, the distribution of water content change in the loess soil column tends to be relatively uniform, and the difference in damage rate between the upper and lower layers tends to be reduced—the amount, time, and pressure of humidification controls wet subsidence.</p>}, journal={EAI Endorsed Transactions on Energy Web}, author={Wang, Wen}, year={2023}, month={Sep.} }