Models for predicting the compressive strength of bagasse ash concrete based on emsemble learning
To obtain an efficient and accurate ensemble learning model for predicting the compressive strength of sugarcane bagasse ash concrete,four ensemble learning models,namely eXtreme Gradient Boosting(XGBoost),Random Forest(RF),Light Gradient Boosting Machine(LightGBM),and Adaptive Boosting(AdaBoost)were established.The predictive capabilities of these models were compared,and the ensemble learning model with the optimal predictive performance was identified.The Shapley Additive Explanation(SHAP)value method was employed to quantitatively study the impact of each input variable on the compressive strength of sugarcane bagasse ash concrete.Firstly,compressive strength experiments were conducted for sugarcane bagasse ash concrete.Based on experimental data and literature information,an ensemble learning database consisting of five input variables,namely cement content,water-to-cement ratio,sugarcane bagasse ash admixture content,fine aggregate content,and coarse aggregate content,was built.Subsequently,four evaluation metrics,namely determination coefficient,mean absolute error,root mean square error,and reliability index,were used to assess the predictive capabilities of the models.In the performance comparison,it was observed that the XGBoost model exhibited the highest predictive accuracy.The evaluation metrics for the training set of the XGBoost model were determined as follows:determination coefficient of 0.976,mean absolute error of 1.811,root mean square error of 2.344,and reliability index of 0.875.The impact of each input variable on the compressive strength of sugarcane bagasse ash concrete was ranked from highest to lowest as follows:cement content,fine aggregate,coarse aggregate,sugarcane bagasse ash admixture content,and water-to-cement ratio.Cement content had a positive effect on concrete compressive strength,and the compressive strength of concrete was not significantly reduced when the sugarcane bagasse ash admixture content was below 10％.This study provides useful reference for predicting the compressive strength of sugarcane bagasse ash concrete and explaining influencing factors.It holds value in advancing research and applications of environmentally friendly materials such as sugarcane bagasse ash concrete.
ensemble learningsugarcane bagasse ash concretecompressive strengthSHAP value methodpredictive model
NETL
NSTL
万方数据
