Application of Artificial Neural Network Model for Optimization in Main Deck Cargo Ship Welding

Journal article

Authors/Editors

Strategic Research Themes

Metal (Materials Processing )

Publication Details

Author list: ปรัชญา เพียสุระ;Pasapitch Chujai

Publisher: คณะวิศวกรรมศาสตร์ มหาวิทยาลัยเทคโนโลยีราชมงคลธัญบุรี

Publication year: 2020

Volume number: 18

Issue number: 1

Start page: 57

End page: 68

Number of pages: 12

ISSN: 27302148

eISSN: 26974339

URL: https://journal.engineer.rmutt.ac.th/enjournal/index.php/enjournal/article/view/475

Abstract

This research proposes the optimization of main deck cargo ship welding with submerge arc welding process (SAW) in high strength steel ASTM A131 EH36 grade. The mathematic modeling for tensile strength predicting was based on the artificial neural networks (ANN) with back-propagation learning algorithm and supervised learning. The SAW process parameters were studied the welding current, voltage and travel speed. The resulting SAW welding specimens were examined using tensile strength tests, bending tests which were observed microstructure with scanning electron microscopy (SEM) and determine a suitable mathematic model. The Levenberg-Marquart training algorithm was also train for weight and bias network. The two learning function, including learning gradient descent (Learngd) and learning gradient descent with momentum (Learngdm) were used in ANN model. The activation function of log-sigmoid for input layer, tan-sigmoid for hidden layer of 1 and 2, purelin for output layer was assigned. The research results reveal that using a ANN model with the proposed mathematical model, which represents 3 neurons for the input 8 neurons for layer 1 layer 2 for 10 neurons and 1 neuron for output layer (3-8-10-1) with learning function of Learngd. The mean square error (MSE) of ANN model is 0.000106 and the coefficient of determination (R2) is 0.99947. The optimum from ANN model were welding current of 340 amperes, 26 volts, and 20 centimeter

Keywords

กระบวนการเชื่อมใต้ฟลักซ์, แบบจำลองโครงข่ายประสาทเทียม, พื้นดาดฟ้าเรือ