Determination of optimum operation cases in electric arc welding machine using neural network

Abstract

With arc welding machines, welding is only performed at optimum operating points. Determination of optimum operating points is important so as for welding machines which will be produced in future to be developed in a manner to operate in such parts. In this study, an Artificial Neutral Networks method was used in order to determine the optimum operating points of Electric Arc welding machine. For this purpose, a measurement system used to get the current measurements during the welding operation. A welding process includes some stages like initial case; transient case and operation case respectively. So as to use ANN model, a data set was established via time series. ANN is trained with 90% of data set and tested with 10% thereof. At the end of the test, a prediction of 97.49% was made according to the regression value. And according to the MSE value, it was understood that a successful prediction was made with an error of 0.00353075 values.

Energy approach in analysis of microplasma powder surfacing modes

“…Measurement, recording and mathematical treatment of electric parameters of the welding arc have been applied for a long time to study the features of arcing, regularities of the process of electrode metal transfer, as well as welding equipment testing [1][2][3][4][5][6][7][8][9][10].…”

“…Data acquisition systems [5][6][7][8][9][10] have become widely applied recently for recording, processing and analysis of process parameters of the welding arc. These systems allow:…”

“…• using specialized programs to perform digital processing and analysis of electric signals, corresponding to process technological parameters: welding current, arc voltage, electrode wire feed rate, welding head displacement, shielding gas flow rate, etc. Basic methodological support of data acquisition systems [1-4] was complemented by plotting arc dynamic volt-ampere characteristics [5], application of Fourier transforms for frequency analysis of electrode metal mass transfer [5,6], neural networks for optimization of welding parameters during arcing [7].…”