| Author: | Zhang, Pin |
| Title: | Data-driven modelling of soil properties and behaviours with geotechnical applications |
| Advisors: | Yin, Zhen-yu (CEE) Yin, Jian-hua (CEE) |
| Degree: | Ph.D. |
| Year: | 2022 |
| Award: | FCE Awards for Outstanding PhD Theses (2022/23) PolyU PhD Thesis Award - Merit Award (2023) |
| Subject: | Soil mechanics Soil mechanics -- Data processing Hong Kong Polytechnic University -- Dissertations |
| Department: | Department of Civil and Environmental Engineering |
| Pages: | xvii, 218 pages : color illustrations |
| Language: | English |
| Abstract: | Understanding soil properties and behaviours are fundamental to geotechnical design. Myriad empirical and analytical models have been proposed for prediction accordingly but they tend to be site-specific and increasing parameters need to be calibrated for constitutive models. With the increasing data in the geotechnical domain, machine learning (ML) has emerged as a new methodology to directly learn from raw data to identify soil properties and behaviours. Its applicability has been proved to be promising because of its versatility and strong fitting capability. Nevertheless, the current ML-based data-driven models still exhibited limitations including lack of interpretability, dependency on numerous high-quality data and poor generalization ability, thus they are still far away from application to engineering practice. To this end, this study aims to elaborate data-driven models for predicting soil properties and mechanical behaviours merely based on their micro computed-tomography (µCT) images, as well as facilitate their applications in geotechnical engineering. First, a set of ML-assisted algorithms is developed for automatically reconstructing three-dimensional real particles from µCT images and subsequently identifying their particle size and morphology. Bayesian inference is incorporated into the ML algorithms for enhancing the interpretability of the data-driven model. Then, a multi-fidelity residual neural network incorporating Bayesian uncertainty is proposed to leverage existing knowledge and limited high-quality data for modelling mechanical behaviours of soils. In this context, a multi-scale data-driven model is proposed from the identification of particle size and morphology to the prediction of their mechanical responses together with fabric evolution. Finally, the developed data-driven models are integrated with finite element code for modelling boundary value problems and the results are compared with conventional numerical modelling methods and measurements for the validation. The proposed data-driven modelling methods are successfully used to predict various soil properties such as compressibility, creep, strength and permeability, behaviours such as anisotropy and dilatancy and boundary value problems. |
| Rights: | All rights reserved |
| Access: | open access |
Copyright Undertaking
As a bona fide Library user, I declare that:
- I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
- I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
- I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.
By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.
Please use this identifier to cite or link to this item:
https://theses.lib.polyu.edu.hk/handle/200/11992