Optimum Design of Carbon Fiber-Reinforced Polymer (CFRP) Beams for Shear Capacity via Machine Learning Methods: Optimum Prediction Methods on Advance Ensemble Algorithms – Bagging Combinations

Optimum Design of Carbon Fiber-Reinforced Polymer (CFRP) Beams for Shear Capacity via Machine Learning Methods: Optimum Prediction Methods on Advance Ensemble Algorithms – Bagging Combinations

Melda Yucel (Istanbul University-Cerrahpaşa, Turkey), Aylin Ece Kayabekir (Istanbul University-Cerrahpaşa, Turkey), Sinan Melih Nigdeli (Istanbul University-Cerrahpaşa, Turkey) and Gebrail Bekdaş (Istanbul University-Cerrahpaşa, Turkey)
DOI: 10.4018/978-1-7998-0301-0.ch005

Abstract

In this chapter, an application for demonstrating prediction success and error performance of ensemble methods combined via various machine learning and artificial intelligence algorithms and techniques was performed. For this reason, two single method was selected and combination models with Bagging ensemble was constructed and operated in the direction optimum design of concrete beams covering with carbon fiber reinforced polymers (CFRP) by ensuring the determination of design variables. The first part was optimization problem and method composing from an advanced bio-inspired metaheuristic called Jaya algorithm. Machine learning prediction methods and their operation logics were detailed. Performance evaluations and error indicators were represented for prediction models. In the last part, performed prediction applications and created models were introduced. Also, obtained prediction success for main model generated with optimization results was utilized to determine the optimum predictions about test models.
Chapter Preview
Top

Introduction

Artificial Intelligence (AI) methods are effective in solving multidisciplinary engineering problems. Also, AI methods can be trained with optimization methodologies to provide the prediction of optimization results. In this chapter, the authors present a study showing the application of the predictive success and error performance of ensemble methods employing various machine learning and artificial intelligence algorithms. Two single methods were selected, and combination models with a Bagging ensemble were constructed. The optimal design is that of using concrete beams with a covering of carbon-fiber-reinforced polymers (CFRP) by ensuring the determination of design variables for the minimization of CFRP material in order to increase the shear capacity of the beam. For an RC beam using CFRP, the width, spacing, and application angle of the CFRP strip are the design variables. Their optimization has previously been done (Kayabekir, Sayin, Bekdas, & Nigdeli, 2017; Kayabekir, Sayin, Nigdeli, & Bekdas, 2017; Kayabekir, Sayin, Bekdas, & Nigdeli, 2018; Kayabekir, Bekdaş, Nigdeli, & Temür, 2018) by using several metaheuristic algorithms—namely, Flower Pollination Algorithm (FPA) (Yang, 2012), Teaching-Learning-Based Optimization (TLBO) (Rao, Savsani, & Vakharia, 2011), and Jaya Algorithm (JA) (Rao, 2016).

Complete Chapter List

Search this Book:
Reset