Back-Propagation Neural Network Optimized by K-Fold Cross-Validation for Prediction of Torsional Strength of Reinforced Concrete Beam.

Bijan Samali,Masoud Mohammadi,Maria Rashidi,Yang Yu,Zhaoqiu Lyu,Thuc N Nguyen,Andy Nguyen

doi:10.3390/ma15041477

Bijan Samali, Masoud Mohammadi + Show 5 more

Open Access

https://doi.org/10.3390/ma15041477

Copy DOI

Abstract

Due to the limitation of sample size in predicting the torsional strength of Reinforced Concrete (RC) beams, this paper aims to discuss the feasibility of employing a novel machine learning approach with K-fold cross-validation in a small sample range, which combines the advantages of a Genetic Algorithm (GA) and a Neural Network (NN) to predict the torsional strength of RC beams. This research study not only utilizes the application of a Back Propagation (BP) neural network and the Gene Algorithm-Back Propagation (GA-BP) neural network in the prediction of the torsional strength of the RC beam, but it also investigates neural network parameter optimization, including connection weights and thresholds, using K-fold cross-validation. The root mean square error (RMSE), mean absolute error (MAE), mean square error (MSE), mean absolute percentage error (MAPE) and correlation coefficient (R2) are among the evaluation metrics used to assess the performance of the trained model. To elaborate on the superiority of the proposed network models in predicting the torsional strength of RC beams, a parametric study is conducted by comparing the proposed model to three commonly used empirical formulae from existing design codes. The comparative findings of this research study demonstrate that the performance of the BP neural network is highly similar to that of design codes; however, its accuracy is inadequate. After improving the weights and thresholds by k-fold cross-validation and GA, the prediction of the BP neural network shows higher consistency with the actual measured values. The outcome of this study can be used as a theoretical reference for the optimal design of RC beams in practical applications.

Highlights

Reinforced Concrete (RC) is a complex construction material due to the complexity of its properties and high maintenance conditions
The results showed that ANN outperformed design code in predicting the torsional strength and confirmed the potential feasibility of ANN in predicting the torsional strength of RC beams
In the machine learning (ML) modelling approaches, fuzzy logic, random forests and support vector machines have been reported in predicting concrete mechanical properties such as compressive strength and elastic modulus that are largely consistent with the simulation results of neural networks [11,24,25]

Summary

Introduction

Reinforced Concrete (RC) is a complex construction material due to the complexity of its properties and high maintenance conditions. Based on a large number of research studies for RC structures, the form of stress combinations, initial crack angles, dislodgement of concrete, aggregate damage, etc., are continuously incorporated into the calculations and optimized to obtain better results [6,7,8,9] This literature provides more accurate predictions and increases the complexity of calculating the torsional strength of RC beams. In the ML modelling approaches, fuzzy logic, random forests and support vector machines have been reported in predicting concrete mechanical properties such as compressive strength and elastic modulus that are largely consistent with the simulation results of neural networks [11,24,25] These methods, except neural networks, usually require a significant computational effort in finding an optimal solution to a complex. The design codes from different sources such as ACI-318-2014 [1], TBC-5002000 [3] and BS8110 [4] are used to predict the results and compare them with the results predicted by the model of the BP neural network

Data Collection and Analysis

Design Code

K-Fold Cross-Validation

BP Neural Network and Genetic Algorithm

Model Parameter Setting

Results and Discussion