摘要
针对水泥熟料28天抗压强度预测存在影响因素多,变量之间相互交叉大,预测精度低的情况。运用MATLAB的神经网络工具箱,分别建立起熟料化学成分、熟料三率值与28天抗压强度之间的BP神经网络预测模型(模型1和模型2)。将模型运用于四川某水泥厂水泥熟料28天抗压强度预测。预测值与实测值的均方差分别为0.8450和0.7577,模型2预测结果优于模型1,说明通过熟料化学成分计算得出的熟料三率值作为输入变量建立的预测模型预测结果更好,从而为水泥熟料强度预测提供了一种有效的方法。
Pointed to the statue of The Recasting of traditional 28-day compressive strength of cement clinker with many influence factors,large cross-cutting between variables and Low accuracy of prediction.It is established that distinct chemical composition of clinker,clinker three rate values between the 28-day compressive strength of BP neural network prediction model（model 1 and model 2） by using the MATLAB neural network toolbox which used in prediction of cement clinker 28 days compressive strength in a cement plant in Sichuan province.The mean values of relative errors of model 1 and model 2 are 0.8450 and 0.7577,Model 2 better than the model 1,respectively,and the results of the prediction are proved close to the experimental ones 。
出处
《微计算机信息》
2010年第28期227-228,129,共3页
Control & Automation
关键词
BP神经网络
水泥熟料
抗压强度
预测
BP network
cement clinker
compression strength
prediction
