路基工程Issue(1)：8-13,6.DOI:10.13379/j.issn.1003-8825.202404052
基于机器学习的频域法快速检测含水率可靠性研究
Research on Reliability of Frequency Domain Method for Rapid Detection of Moisture Content Based on Machine Learning
摘要
Abstract
Based on the experimental section of the new Shanghai—Chongqing—Chengdu high-speed railway,the performance of FDS-100 moisture sensor in the detection of filler moisture content by frequency domain method was studied,and its applicability and reliability in practical field application were evaluated.Through univariate regression,multivariate regression model and XGBoost algorithm,the field test data and instrument detection data were analyzed to explore the factors of filler moisture content.The result shows that,there is a significant correlation between particle size and instrument detection results and the moisture content of filler,especially the particle content below 0.075 mm;the content of particles with a particle size of 40.000 mm and 5.000 mm has a weak correlation with the test results of filler moisture content;when XGBoost model is used to predict,the predicted value of more than 75%coarse particles(particle size>2.000 mm)in the sample exceeds the upper limit of error,and the predicted value of samples with more than 10%fine particles(particle size＜0.250 mm)is relatively accurate.关键词
路基填料/含水率/频域法传感器/数据修正/机器学习Key words
subgrade filler/moisture content/frequency domain sensor/data correction/machine learning分类
建筑与水利引用本文复制引用
王学朋,罗京,周晋筑..基于机器学习的频域法快速检测含水率可靠性研究[J].路基工程,2026,(1):8-13,6.基金项目
国家自然科学基金项目(51478481) （51478481）
