摘要
采用正交实验设计与BP神经网络相结合,建立了油页岩渣砖配比与抗压强度之间的非线性关系模型,以正交实验数据为基础进行神经网络训练,得到了油页岩渣砖的BP神经网络预测模型.并对模型进行验证,预测结果基本满意.同时用训练好的BP神经网络预测模型建立了配方中单因素分析,对寻找最优配方有一定的指导意义.最终达到缩短实验周期,节省人力、物力的目的.
The BP neural network based on orthogonal design is used to establish a nonlinear rela prediction model between oil shale waste brick and compressive strength. Based on orthogonal test, tion the BP neural network model was trained. The prediction results show the trained model is good. The model can predict compressive strength with any brick formula. In addition, the model can do factors analysis, which is sandand cement have influence on compressive strength. It has guiding significance for finding optimal formula. It also can shorten the period of research and save the manpower and material resource.
出处
《暨南大学学报（自然科学与医学版）》
CAS
CSCD
北大核心
2010年第1期39-42,共4页
Journal of Jinan University(Natural Science & Medicine Edition)
基金
广东省科技计划项目(2007B03100017)
关键词
油页岩渣砖
配方
BP神经网络
正交设计
oil shale waste brick
formula
BP neural network
orthogonal design
