基于TPE-XGBoost算法的再生粗骨料混凝土抗压强度预测模型

Prediction model of compressive strength of recycled coarse aggregate concrete based on TPE-XGBoost algorithm

为了更好地预测再生粗骨料混凝土的抗压强度,提出了基于极限提升树(XGBoost)算法的再生粗骨料混凝土抗压强度预测模型;利用再生粗骨料混凝土数据库,对数据进行预处理,利用树结构概率密度估计贝叶斯优化(TPE-BO)方法优化模型参数;通过实例对再生粗骨料混凝土抗压强度预测模型进行对比验证。结果表明:数据预处理和TPE-BO超参数优化方法均能在一定程度提升模型性能;与随机森林、K邻近回归、支持向量机回归、梯度提升决策树模型相比,提出的模型有更高的预测精度和泛化能力;高性能抗压强度预测模型可为再生粗骨料混凝土的研究和实践提供依据,同时也为再生混凝土性能预测提供新途径。

In order to better predict the compressive strength of recycled coarse aggregate concrete,a compressive strength prediction model for recycled coarse aggregate concrete based on extreme gradient boosting(XGBoost)algorithm was proposed.Taking the recycled coarse aggregate concrete database as the research data set,the data set was preprocessed,and the Bayesian optimization(BO)method was used to estimate the tree-structured parzen estimator(TPE)to optimize the model parameters.The comparative verification of compressive strength prediction models for recycled coarse aggregate concrete was carried out through examples.The results show that data preprocessing and TPE-BO hyperparameter optimization methods can both improve model performance to a certain extent.Compared with random forest,K-nearest neighbor regression,support vector machine regression,and gradient boosting decision tree models,the proposed model has higher prediction accuracy and generalization ability.The high performance compressive strength prediction model provides a basis for the research and practice of recycled coarse aggregate concrete,and also provides a new approach for predicting the performance of recycled concrete.