摘要
利用水泥对风化千枚岩进行改良处理,依据正交试验理论进行多水平、多因素的无侧限抗压强度室内试验,将试验数据作为训练样本,通过BP神经网络建立非线性预测模型,对不同水泥掺量、压实度、龄期的改良风化千枚岩的无侧限抗压强度进行预测分析。预测结果虽然与试验数值存在误差,但是基本可以满足工程精度要求。建立预测模型后,可以大大减少试样数量,节省大量的人力、物力和时间,并为路基填筑提供理论根据。
The cement was used to improve the weathered phyllite.Based on the theory of orthogonal test,the laboratory experiments of unconfined compressive strength with multi levels and multi factors were carried out,and the experiment data was used as training sample.The nonlinear prediction model was established by back-propagation(BP)neural network to predict and analyze the unconfined compressive strength of modified weathered phyllite with different cement contents,compaction degrees and ages.Although there are some errors between the predicted results and the experimental values,they can basically meet the requirements of engineering accuracy.After the establishment of the prediction model,the number of test samples can be greatly reduced,a lot of manpower,material resources and time can be saved,and the theoretical basis for subgrade filling can be provided.
作者
王鹏
邱洲
WANG Peng;QIU Zhou(CCCC Second Highway Consultants Co.,Ltd.,Wuhan 430056,Hubei,China;CCCC Northeast Regional Headquarters,Shenyang 110100,Liaoning,China)
出处
《筑路机械与施工机械化》
2020年第9期6-9,共4页
Road Machinery & Construction Mechanization
关键词
改良千枚岩
BP神经网络
无侧限抗压强度
预测模型
modified phyllite
back-propagation neural network
unconfined compressive strength
prediction model