基于多元回归统计下的BFRP筋-混凝土粘结性能研究

Study on multivariate regression statistical analysis of BFRP-concrete bonding properties

玄武岩纤维复合(BFRP)筋替代混凝土结构中的钢筋是解决传统混凝土结构中钢筋锈蚀问题的重要方法。本文通过对拉拔试验的统计分析发现,粘结强度的影响因素不能用单因素准确表达。随着混凝土抗压强度的提高,BFRP筋与混凝土的粘结强度会有一定的提升,且平均粘结强度与相对保护层厚度成正相关,随着相对保护层厚度的增加,粘结强度的增幅放缓。基于多元统计下的平均粘结强度与诸因素之间的统计规律,发现了非线性统计方法与试验结果更为吻合,增加的拟合参量越多,模型的预测精度越高,敏感因素为混凝土抗压强度、水灰比以及BFRP筋的拉伸强度。并利用BP神经网络模型对粘结强度进行了预测,平均误差为8.9%,满足工程使用的要求,为BFRP筋-混凝土粘结强度的预测提供了方法与依据。

The replacement of basalt fiber reinforced ploymer(BFRP)bar in concrete structures is an important method to solve the problem of steel corrosion in traditional concrete structures.Through the statistics of the relevant scholars pull-out tests,it is found that the single factor can inaccurately express the influencing factors of bond strength.With the increase of concrete compressive strength,the bond strength of BFRP bars and concrete will be improved.The average bond strength will be improved.It is positively correlated with the thickness of the opposite protective layer.As the relative thickness increases,the increase in bond strength slows down.Based on the statistical law of average bond strength and various factors,it is found that the nonlinear statistical method is more suitable with the experimental results.As the fitting parameters increase,the prediction accuracy of the model will be higher.It was obviously shown that the influencing factors are concrete compressive strength,water-cement ratio and tensile strength of BFRP bars.The BP neural network model is used to predict the bond strength,and the average error is 8.9%.It can meet the requirements of engineering use,it also can provide a method and basis for the prediction of BFRP-concrete bond strength.