基于神经网络理论,采用典型的BP神经网络模型,以橡胶粒径、橡胶掺量、水胶比、水泥用量以及试块的冻融循环次数作为输出,分别以橡胶集料混凝土冻融循环后的相对动弹性模量和立方体相对抗压强度作为输出,建立两个5-5-1的三层BP神经网络模型。结果表明:基于BP神经网络模型的预测结果与试验结果较好吻合。 Based on the neural network theory, a typical BP neural network model was used to output the rubber particle size, rubber content, water / rubber ratio, the amount of cement used and the freezing and thawing cycles of the test pieces. After the rubber aggregate concrete was frozen and thawed Relative dynamic elastic modulus and relative compressive strength of cube as output, two 5-5-1 BP neural network models are established. The results show that the prediction results based on BP neural network model are in good agreement with the experimental results.