摘要
采用BP人工神经网络模型,以砌体高厚比、块材强度及粘结剂强度作为输入变量,以砌体抗压强度作为输出变量,建立网络模型模拟输入变量和输出变量间的非线性关系,提出了用于生土基砌体抗压强度计算的简化公式,并将模型预测结果与试验值、计算值进行了对比分析.结果表明:在样本空间内,本研究所建立的10隐含层神经元BP神经网络模型对生土基砌体抗压强度具有较好的预测性能,且简化公式的计算精度及稳定性均较好,计算值与试验值的比值均值为0.92,方差为0.28,可用于对生土基砌体单轴抗压强度的计算.
This study used back propagation(BP)artificial neural networks to predict the compressive strength of masonry prisms formed by earth blocks and cement mortar or mud,by using three parameters:the height-to-thickness ratio of prisms and the compressive strength of the mortar and that of the blocks.A simplified formula for calculating the compressive strength of the earth masonry prism was proposed.The predicted results were compared with the experimental values and the calculated values of the existing formulas.The results show that the BP neural network model with 10 hidden layer neurons has a good predictive performance for the compressive strength of earth masonry.The accuracy and stability of the simplified formula are superior with the mean value of the ratio between the calculated value and the experimental value is 0.92,and the standard deviation is 0.28,which can be used to determine the compressive strength of earth masonry prisms.
作者
兰官奇
王毅红
张建雄
董飞
LAN Guanqi;WANG Yihong;ZHANG Jianxiong;DONG Fei(School of Civil Engineering,Chang’an University,Xi’an 710061,China)
出处
《华中科技大学学报(自然科学版)》
EI
CAS
CSCD
北大核心
2019年第8期50-54,共5页
Journal of Huazhong University of Science and Technology(Natural Science Edition)
基金
国家自然科学基金资助项目(51478043)
关键词
生土材料
砌体
抗压强度
人工神经网络
计算方法
earth materials
masonry
compressive strength
artificial neural networks
computational methods