Study on Correction of Continuous Compaction Quality Test Results Considering Influence of Rolling Parameters
To reduce the detection errors caused by rolling parameters such as driving speed,direction,and vibration frequency of the roller,two correction models of CMV were developed using the extreme gradient boosting(XGBoost)al-gorithm and multiple linear regression method,with Evd(dynamic modulus of deformation)as the target for correction to compare the applicability of the correction models and analyze the influence of rolling parameters on CMV errors.The re-search indicates that the XGBoost-corrected results have a high correlation of 0.8 with Evd and an error of only 1.7％,while the regression-corrected results show a lower correlation of 0.65 with Evd and a higher error of 9.2％.Thus,the XGBoost correction model is more suitable for reducing the detection errors of CMV caused by rolling parameters.High-speed compaction leads to significant negative errors in CMV.Different driving directions cause errors of roughly equal magnitude but with opposite positive and negative values.The influence of vibration frequency on CMV errors is non-line-ar and depends on the specific frequency values.Finally,based on the corrected results from field compaction test data,the reliability of the XGBoost correction model in reducing CMV detection errors was validated.
万方数据
维普
