(1 西安建筑科技大学土木工程学院， 西安 710055；2 西安建筑科技大学结构工程与抗震教育部重点实验室， 西安 710055；3 陕西省结构与抗震重点实验室， 西安 710055；4 西部装配式建筑工业化协同创新中心， 西安 710055)

［摘要］为研究型钢再生混凝土结构的粘结破坏规律及利用Elman神经网络算法预测其粘结强度的方法，选取再生混凝土取代率、再生混凝土强度、再生混凝土埋置长度、型钢保护层厚度、箍筋直径及箍筋间距作为影响因素，设计并制作了36个型钢再生混凝土推出试件。通过推出试验获得了型钢再生混凝土结构的粘结破坏规律并定义了3个平均特征粘结强度。其次，基于试验结果将取代率为0，50%和100%的30个试件作为训练样本建立了型钢再生混凝土构件粘结强度的Elman神经网络模型，最后利用该模型对取代率为30%的9个试件的粘结强度进行了预测。与试验结果对比表明，建立的Elman神经网络模型能够准确地预测型钢再生混凝土结构的粘结强度，神经网络在结构工程领域具有较大的应用前景。 

［关键词］型钢再生混凝土；粘结强度；Elman神经网络；粘结强度预测 

中图分类号：TU398-9 文献标识码：A文章编号：1002-848X(2021)16-0035-07

Study on the prediction method of bond strength of steel reinforced regenerated concrete structure based on Elman neural network algorithm 

BAI  Guoliang1,2,3,4,  LIU  Biao1,  XU  Zhenhua1,  YIN  Yuguang1 

(1 School of Civil Engineering, Xi’an University of Architecture & Technology, Xi’an 710055，China；2 Key Lab of Structural Engineering and Earthquake Resistance, Ministry of Education (XAUAT), Xi’an 710055，China；3 Shaanxi Key Lab of Structure and Earthquake Resistance (XAUAT), Xi’an 710055，China；4 Collaborative Innovation Center for Assembled Buildings in Western China (XAUAT), Xi’an 710055，China) 

Abstract: In order to study the bond failure law of steel reinforced recycled concrete structure and the method of predicting the bond strength by Elman neural network algorithm, 36 push out specimens were designed and made with the replacement ratio of recycled concrete, the strength of recycled concrete, the embedded length of recycled concrete, the thickness of protective layer of steel, the diameter and spacing of stirrup as influencing factors. The bond failure law of SRRC structure was obtained by push out test, and three average characteristic bond strengths were defined. Secondly, based on the test results, 30 specimens with replacement ratio of 0, 50% and 100% were used as training samples to establish Elman neural network model of the bond strength of SRRC members. The Elman neural network model of the bond strength was finally used to predict the bond strength of 9 specimens with the replacement ratio of 30%. The comparison with the experimental results show that the Elman neural network model established can accurately predict the bond strength of SRRC structure. The neural network has great application prospects in the field of structural engineering. 

Keywords:steel reinforced recycled concrete structure； bond strength； Elman neural network； bond strength prediction

*国家自然科学基金（51878544），陕西省自然科学基础研究计划（2019JM-597），国家自然科学基金（51478381）。

 作者简介：白国良，博士，教授，博士生导师，Email:guoliangbai@126.com； 通信作者：刘彪，博士，Email:src_lb@126.com。