Hits:

Date of Publication:2022-10-10

Journal:Journal of Dalian University of Technology

Volume:61

Issue:3

Page Number:272-279

ISSN No.:1000-8608

Key Words:"ANN; ANN; BP neural network model; FRP reinforced concrete; bond strength; prediction"

CN No.:21-1117/N

Abstract:The interface bond strength of FRP(fiber reinforced polymer)reinforced concrete is one of the important indicators to evaluate its mechanical properties.Based on the experimental data available in the literature,a database including 292 groups of pull-out test results of FRP reinforced concrete is established,and thus the bond strength between FRP bars and concrete is predicted by using artificial neural network.The database is randomly divided into two data sets,of which 242 groups of data are used for training and 50 groups of data are used for simulation prediction.A threelayer artificial neural network model is trained by using the back propagation algorithm,where seven parameters are considered in the input layer,namely FRP type,surface form,FRP rebar diameter, anchorage length,failure mode,concrete compressive strength and normalized concrete cover thickness.The output layer is specified as the interface bond strength between FRP bars and concrete.The results indicate that the BP neural network model has strong capability of prediction and generalization,and the predicting error is minor.This method can integrate many considerations those influence the interface bond strength between FRP bars and concrete,and give out accurate predicting results.

Note:新增回溯数据