摘要
为了预测复合盐侵蚀后混凝土的相对动弹性模量,在分析BP神经网络原理的基础上,提出用BP神经网络模拟混凝土相对动弹性模量变化率与复合盐溶液质量分数、侵蚀时间之间关系的方法。根据侵蚀试验的实际工况,分别建立了三维输入向量,一维输出向量的BP神经网络模型,通过39组试验,验证了模型的可靠性与精确性。结果表明:实测结果与预测结果相吻合,并且平均误差百分比为2.08%,该BP神经网络模型能较准确地快速预测侵蚀后混凝土的相对动弹性模量变化率。
Based on the basic theory of BP neutral network,the method which intend to simulate the relationship between relative dynamic elastic modulus change rate of concrete and the mass fraction of composite salt solution and erosion time is presented.According to the practical engineering,a three-dimensional input vector BP and a one-dimensional output vector BP are set respectively.This model is proved to be a rather reliable and accurate one via 39 sets of experiment.With a average error of 2.08%,the experimental results and the predicting data fits well.It can be concluded that the BP neutral network model is able to predict the relative dynamic elastic modulus of concrete after erosion in a relatively precise and quick way.
作者
李扬
王伯昕
陈冬昕
王清
张中琼
LI Yang;WANG Boxin;CHEN Dongxin;WANG Qing;ZHANG Zhongqiong(College of Construction Engineering,Jilin University,State Key Laboratory of Frozen Soil Engineering,Institute of Economics and Technology,Changchun Electric Power Supply Compan)
出处
《混凝土》
CAS
北大核心
2018年第7期21-23,共3页
Concrete
基金
国家自然科学基金重点项目(41430642)
中国博士后科学基金项目(2015M581403)
冻土工程国家重点实验室开放基金(SKLFSE201514)
关键词
BP神经网络
混凝土
相对动弹性模量变化率
盐蚀
BP neutral network
concrete
relative dynamic elastic modulus change rate
salt erosion
