基于贝叶斯优化高斯过程回归法的再生混凝土力学性能预测

为了更准确地预测再生骨料混凝土的抗压强度与弹性模量，建立了一个比以往研究大的数据库，具有730组数据，为建立可靠的预测模型奠定了基础。提出了贝叶斯优化的高斯过程回归方法，选取再生粗骨料体积分数、水灰比、混合粗骨料吸水率、细骨料与总骨料比、粗骨料与水泥比、混合粗骨料饱和表面干密度等6个参数作为影响因素，同时建立了再生骨料混凝土抗压强度和弹性模量预测模型。通过比较抗压强度、弹性模量的预测值与实验值，发现二者较为接近，说明该方法具有一定的可靠性。将贝叶斯优化的高斯过程回归与高斯过程回归、支持向量机回归、随机森林回归、人工神经网络进行比较，并选取4个统计指标对模型进行评价，结果表明贝叶斯优化的高斯过程回归预测抗压强度和弹性模量精度较高，相关系数分别达到了0.91和0.93。这说明贝叶斯优化的高斯过程回归方法对预测再生骨料混凝土的抗压强度和弹性模量同时适用。

Prediction of mechanical properties of recycled concrete using Bayesian optimization-based gaussian process regression method

In order to more accurately predict the compressive strength and elastic modulus of recycled aggregate concrete, a large database with 730 sets of data is established compared with previous studies, which lays a foundation for the establishment of a reliable prediction model. A Bayesian optimized Gaussian process regression method was proposed, and six parameters, including volume fraction of recycled coarse aggregate, water-cement ratio, water absorption rate of mixed coarse aggregate, ratio of fine aggregate to total aggregate, ratio of coarse aggregate to cement, and dry density of saturated surface of mixed coarse aggregate, were selected as influencing factors, at the same time, a prediction model of compressive strength and elastic modulus of recycled aggregate concrete was established By comparing the predicted value of compressive strength and elastic modulus with the experimental value, it is found that the two values are close, which indicates that the method has certain reliability. Gaussian process regression of Bayesian optimization with Gaussian process regression, support vector machine regression, random forests regression, artificial neural network are compared, and selected four statistical indicators to evaluate the model, the results show that the Gaussian process regression of Bayesian optimization to predict the compressive strength and elastic modulus of high precision, the correlation coefficient reached 0.91 and 0.93 respectively. This shows that Bayesian-optimized Gaussian process regression method is suitable for predicting the compressive strength and elastic modulus of recycled aggregate concrete.