摘要
利用RBF神经网络对水泥水化热进行预测,根据水泥水化热的影响因素,建立了12个输入节点、1个输出节点的RBF神经网络模型。通过27组试验数据,验证了模型的可靠性,并与BP神经网络进行了比较。结果表明,RBF神经网络预测效果明显优于BP神经网络,前者不仅预测速度快,而且预测精度高,相对误差小于4%,在水泥水化热预测中具有广阔的应用前景。
An RBF neutral network model is proposed and used for predicting the hydration heat of cement in the paper.According to the influential factors of the hydration heat of cement,a prediction model of RBF network with 12 input vectors and 1 output vectors is established,whose reliability is proved with 27 groups of test values.And a comparison between the RBF model and the BP neutral network model is conducted,too.The results show that the predicted values taken from the RBF neutral network model are even closer to the real experimental ones than the BP neutral network model,and the relative errors of the former are less than 4%,both of which show that the RBF model provides a new method of quickly predicting the hydration heat of cement and has a wider application in the future.
出处
《国防交通工程与技术》
2011年第3期31-33,37,共4页
Traffic Engineering and Technology for National Defence
关键词
水化热
直接法
RBF神经网络
预测模型
hydration heat
direct method
RBF neural network
prediction model