On the afternoon of January 21, 2021, at the invitation of Professor Zhang Wengang, Associate Chair of SCE,CQU, research fellow, Dr Wang Zezhou from School of Civil and Environmental Engineering, National University of Singapore, presented the "Deep-learning techniques to geotechnical reliability analysis in spatially variable soils " online.
Dr Wang’s main research topics include: Inversion statistical analysis of deep foundation pit engineering inspection data; Deep learning and engineering reliability analysis of soil spatial variability; Large-scale variable finite element analysis. He conducted an effective and comprehensive analysis of geotechnical engineering structures based on the finite element analysis method of soil spatial variability. Considering the large amount of calculation of the current geotechnical engineering spatial variability analysis, the proposed method is practical by combining the convolutional neural network (CNN) with the spatial variability of the soil. The feasibility of using convolutional neural network for spatial variability analysis in geotechnical engineering is verified. At the same time, it is compared with the existing alternative model method. Compared with the existing methods, the convolutional neural network has improved from the dual perspectives of accuracy and computational efficiency, and it is found that the convolutional neural network also has certain advantages in the wide application compared with other methods.
Finally, the faculty and students actively discussed relevant issues with Dr. Wang, and Dr. Wang answered them patiently.
This academic presentation was chaired by Dr Wang Lin. It attracted faculty and students from different universities and institutes. The number of online attendees was up to 341.