基于BP神经网络的再生混凝土抗压强度预测

Prediction of Compressive Strength of Recycled Concrete Based on BP Neural Network

再生混凝土抗压强度影响因素多、力学性能易劣化,为提高再生混凝土品质,必须对其强度特性进行深入研究。针对常规回归分析方法预测再生混凝土存在的问题,利用非线性映射能力良好的BP神经网络算法进行再生混凝土抗压强度预测。该预测模型以粗骨料吸水率、水灰比和水泥掺入比作为输入层,以再生混凝土28 d抗压强度作为输出层,中间隐含层节点数为10。仿真结果表明,该模型平均相对误差仅为3.04%,线性相关系数大于0.94,该方法具有简单高效的特点。

The compressive strength of recycled concrete is influenced by many factors and its mechanical properties are easy to deteriorate.In order to improve the quality of recycled concrete,it is necessary to conduct in-depth research on its strength characteristics.In view of the problems existing in the prediction of recycled concrete by conventional regression analysis method,BP neural network algorithm with good nonlinear mapping ability is used to predict the compressive strength of recycled concrete.The prediction model takes water absorption rate of coarse aggregate,water-cement ratio and cement incorporation ratio as the input layer,takes28d compressive strength of recycled concrete as the output layer,and the node number of hider layer in the middle is 10.The simulation result shows that the MRE of the compressive strength prediction model is only3.04%,and the R is greater than 0.94,indicating that the method is simple,efficient and economical.