{"title":"Comparative analysis of machine learning techniques for accurate prediction of unfrozen water content in frozen soils","authors":"Jiaxian Li, Pengcheng Zhou, Yiqing Pu, Junping Ren, Fanyu Zhang, Chong Wang","doi":"10.1016/j.coldregions.2024.104304","DOIUrl":null,"url":null,"abstract":"<div><p>Unfrozen water content (UWC) plays a critical role in determining the thermal, hydraulic, and mechanical properties of frozen soils. Existing empirical, semi-empirical, and theoretical models for UWC estimation have limitations in terms of accuracy as well as generalizability. To address these challenges, the present study explored the application of six machine learning techniques to predict UWC in frozen soils: Random Forest (RF), eXtreme Gradient Boosting (XGBoost), Light Gradient Boosting Machine (LightGBM), K-Nearest Neighbors (KNN), Support Vector Regression (SVR), and Backpropagation Neural Network (BPNN). Considering the UWC hysteresis phenomenon between the freezing and thawing processes, experimental UWC data collected from the literature were separated into two sub-datasets: freezing branch dataset (FBD) and thawing branch dataset (TBD). Based on that, a comprehensive framework integrating Bayesian optimization and 10-fold cross-validation was established to optimize the six models' hyperparameters and to evaluate their performance. The results highlighted significant variations in the predictive capability among the six machine learning models, with ensemble methods (i.e., RF, XGBoost, LightGBM) generally demonstrating superior accuracy. Feature importance analysis, robustness checks, and uncertainty quantification further elucidated the strengths and limitations of each model. The present study provides profound insights into the selection and application of machine learning models for accurately modeling the properties of frozen soils for cold regions science and engineering.</p></div>","PeriodicalId":10522,"journal":{"name":"Cold Regions Science and Technology","volume":"227 ","pages":"Article 104304"},"PeriodicalIF":3.8000,"publicationDate":"2024-08-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Cold Regions Science and Technology","FirstCategoryId":"5","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S0165232X2400185X","RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"ENGINEERING, CIVIL","Score":null,"Total":0}
引用次数: 0
Abstract
Unfrozen water content (UWC) plays a critical role in determining the thermal, hydraulic, and mechanical properties of frozen soils. Existing empirical, semi-empirical, and theoretical models for UWC estimation have limitations in terms of accuracy as well as generalizability. To address these challenges, the present study explored the application of six machine learning techniques to predict UWC in frozen soils: Random Forest (RF), eXtreme Gradient Boosting (XGBoost), Light Gradient Boosting Machine (LightGBM), K-Nearest Neighbors (KNN), Support Vector Regression (SVR), and Backpropagation Neural Network (BPNN). Considering the UWC hysteresis phenomenon between the freezing and thawing processes, experimental UWC data collected from the literature were separated into two sub-datasets: freezing branch dataset (FBD) and thawing branch dataset (TBD). Based on that, a comprehensive framework integrating Bayesian optimization and 10-fold cross-validation was established to optimize the six models' hyperparameters and to evaluate their performance. The results highlighted significant variations in the predictive capability among the six machine learning models, with ensemble methods (i.e., RF, XGBoost, LightGBM) generally demonstrating superior accuracy. Feature importance analysis, robustness checks, and uncertainty quantification further elucidated the strengths and limitations of each model. The present study provides profound insights into the selection and application of machine learning models for accurately modeling the properties of frozen soils for cold regions science and engineering.
期刊介绍:
Cold Regions Science and Technology is an international journal dealing with the science and technical problems of cold environments in both the polar regions and more temperate locations. It includes fundamental aspects of cryospheric sciences which have applications for cold regions problems as well as engineering topics which relate to the cryosphere.
Emphasis is given to applied science with broad coverage of the physical and mechanical aspects of ice (including glaciers and sea ice), snow and snow avalanches, ice-water systems, ice-bonded soils and permafrost.
Relevant aspects of Earth science, materials science, offshore and river ice engineering are also of primary interest. These include icing of ships and structures as well as trafficability in cold environments. Technological advances for cold regions in research, development, and engineering practice are relevant to the journal. Theoretical papers must include a detailed discussion of the potential application of the theory to address cold regions problems. The journal serves a wide range of specialists, providing a medium for interdisciplinary communication and a convenient source of reference.