CLASSIFICATION OF SHORT CIRCUIT GMA WELDING USING ARTIFICIAL NEURAL NETWORK
Victor Bicalho Civinelli de Almeida1; Karla Aparecida Perine Lagares1; Eduardo Pestana de Aguiar1; Moisés Luiz Lagares Júnior1
1 Universidade Federal de Juiz de Fora
doi:10.20906/CPS/COB-2015-1860
Resumo
Short circuit gas metal arc welding has an important role in manufacturing processes and is widely used in an industrial base. In order to achieve high quality control levels in weldments, investigations are carried out with the aim to find a way to monitor the quality of weldments in real time. The Laprosolda research group has created a criterion to quantify the metal transfer stability in short circuit process that is based on calculating the IVcc index and metal drop diameter - the two parameters used by Laprosolda criterion. The proposal in question makes use of measured data obtained from three different gas flow rate weld beads performed in laboratory. Observing such samples, this work introduces a classification for monitoring the gas flow rate in short circuit gas metal arc welding using a Multilayer Perceptron Artificial Neural Network fed by parameters of Laprosolda criterion and statistics from the electrical signal, for the purpose of classifying satisfactorily events in the abovementioned process. The results indicate that this proposal is suitable to be used in welding quality monitoring, generating benefits that will be discussed and thus providing solutions for users to achieve the top of operational excellence.
Palavras-chave: short-circuiting GMA welding ; artificial neural network; gas flow rate