Evaluation of MIG/MAG Welding with Short-Circuit Metal Transfer Using Computational Tools and Statistical Modeling

Abstract

Abstract  This work analyzed data from MIG/MAG welding experiments performed in short-circuit operation mode using three gases: 100% argon, argon plus 25% carbon dioxide and 100% carbon dioxide, enabling comprehensive analyses of the following parameters: short-circuit time; open arc time; short-circuit voltage; open arc voltage; current rise rate; number of shorts and reference voltage. Correlation tests (Pearson, Spearman, Kendall) were performed to analyze the quality of the models, including the application of Shapiro-Wilk tests to verify the normality of the residues, making it possible to create a correlation matrix with the parameters and gases tested. The correlation between the open arc time and the number of shorts, and between the open arc time and the current rise rate showed strong correlation for all gases. The correlations between the current rise rate and the number of shorts, the short-circuit time and the short-circuit voltage, as well as the correlation between the short-circuit voltage and the reference voltage, demonstrated lower significance for all gases. Regarding the time series analysis, the ARIMA and ETS models were adjusted and evaluated with specific metrics, indicating the need for improvements in the model to increase the accuracy of the predictions.

Key-words:
Correlation tests; Correlation matrix; Time series