基于PSO‑BPNN模型的氯氧镁水泥混凝土耐水性预测

Prediction of Water Resistance of Magnesium Oxychloride Cement Concrete Based on PSO‑BPNN Model

为快速准确地获得具有优异耐水性氯氧镁水泥混凝土(MOCC)的配合比,设计了拓扑结构为4‑10‑2的粒子群优化(PSO)算法-反向传播(BP)神经网络(PSO‑BPNN)模型.该模型的输入层参数为n(MgO)/n(MgCl_(2))、粉煤灰掺量、磷酸掺量和磷肥掺量,输出层参数为MOCC的抗压强度和软化系数;模型数据集为144组,其中训练集数据为100组,验证集数据为22组,测试集数据为22组.结果表明:PSO‑BPNN模型在MOCC抗压强度预测中的评价参数——决定系数R^(2)=0.99、平均绝对误差S_(MAE)=0.52、平均绝对误差百分比S_(MAPE)=1.11、均方根误差S_(RMSE)=0.73;其在软化系数预测中的评价参数——R^(2)=0.99、S_(MAE)=0.44、S_(MAPE)=1.29、S_(RMSE)=0.62;与BP神经网络(BPNN)模型相比,PSO‑BPNN模型具有更强的双参数预测能力,可用于MOCC配合比的正向设计和反向指导.

In order to quickly and accurately obtain magnesium oxychloride cement concrete(MOCC)proportions with excellent water resistance,a particle swarm optimization back propagation neural network(PSO‑BPNN)model with a topology of 4‑10‑2 was designed.The input layer parameters of the above model were n(MgO)/n(MgCl2),fly ash content,phosphoric acid content,and phosphate fertilizer content.The output layer parameters were MOCC compressive strength and softening coefficient.The model establishment data set contained 144 groups,including 100 groups of training set data,22 groups of validation set data,and 22 groups of test set data.The results show that the mean value of each evaluation parameter in the prediction of compressive strength using the PSO‑BPNN model are coefficient of determination R^(2)=0.99,mean absolute error S_(MAE)=0.52,mean absolute percentage error S_(MAPE)=1.11,and root mean square error S_(RMSE)=0.73.The mean value of each evaluation parameters in the prediction of softening coefficient are R^(2)=0.99,S_(MAE)=0.44,S_(MAPE)=1.29,and S_(RMSE)=0.62.This indicates that compared to the BP neural network(BPNN)model,the PSO‑BPNN model has a strong ability to predict dual parameters and can be used for both forward design and reverse guidance of MOCC mix proportions.