Comparative Research on Prediction Model for Strength Activity Index of Iron Tailings Replacing Cement Based on Ensemble Learning
In order to accurately predict the strength activity index of iron tailings replacing cement(referred to as iron tailings strength activity index),based on the experimental data,the prediction performance of iron tailings strength activity in-dex prediction models established by different ensemble learning methods was evaluated,and compared with other single ma-chine learning prediction models.The parameters considered in the prediction model are water-solid ratio,particle size,silica content,iron oxide content,magnesium oxide content,alumina content,calcium oxide content,sulfur trioxide content,other chemical composition content and iron tailings content.The results show that in the ensemble learning method,the extreme gra-dient boosting model has the best prediction performance and accuracy,followed by the histogram gradient boosting model and the gradient boosting model,which are better than the single machine learning model(support vector machine model and linear regression model).The prediction accuracy of the random forest model is better than that of the linear regression model,but slightly inferior to the support vector machine model.In the practical application process,iron tailings containing sulfur triox-ide,alumina,magnesium oxide and iron oxide can be selected as much as possible,because compared with other substances,i-ron tailings containing such substances are beneficial to replace cement,thereby increasing the strength activity index of iron tailings.The research can lay a foundation for the application of iron tailings in the field of cement-based materials.
iron ore tailingsstrength activity indexensemble learningiron tailings dosagegradient boosting model
NETL
NSTL
万方数据
