*Stability on the GMAW Process DOI: http://dx.doi.org/10.5772/intechopen.90386*

**Figures 13** and **14** show a summary of the signals and methods used to measure or estimate the indexes. Consequently, the current and voltage signals are widely used, as well as the camera in the image processing and the microphone for the

**Figure 15** summarizes the parameters and variables used in the studies showing that among the most influential in the stability of the process, current, voltage, wire feed speed, short-circuit time, arcing time, and short-circuit frequency can be

analysis of acoustic signals.

*Welding - Modern Topics*

mentioned.

**Figure 12.**

**Figure 13.**

**18**

*Sensors used to measure.*

*Cluster of terms (two as a frequency of occurrence of the term).*

Also, it is possible to classify the indexes into groups according to their purpose, those that are oriented to the monitoring of the metallic transfer, and the analysis of the stability of the arc and the process in general. **Figure 16** shows the percentage by group; **Figure 17** shows the technique used to develop the indexes for those groups. It is important to emphasize that these concepts are widely correlated.

**Figure 14.** *Methods used to estimate the indexes.*

**Figure 15.** *Parameters and variables used in the studies.*

**Figure 16.** *Percentage by group.*

**3.1 Highlights of the works of the last 5 years**

*DOI: http://dx.doi.org/10.5772/intechopen.90386*

used as demonstrated in the present study.

bead in real time is now a reality.

**3.2 Innovative techniques**

**Figure 19.**

**21**

gas arc welding (GMAW).

*Stability on the GMAW Process*

An analysis of the works in the field of stability in the last 5 years was made and allowed to find the following trends. There is a considerable increase in the study and application of works in pulsed GMAW (**Figure 19**). This increase is caused by the known improvements in quality and productivity with respect to regular metal

Another trend that could be identified is the increase in research that integrates

classical statistics techniques and novel machine learning algorithms. It is well known that with the increase of the computing processing capacities, the data analysis, big data, and machine learning have had a significant boom since 2009. The welding area has not been oblivious to the use of such techniques, although it should be noted that in the area of stability, classical statistics is more commonly

Already in recent years, some interesting solutions have been presented. Alizadeh and Omrani [41] integrate successfully the Taguchi method with back-propagation neural network (BPNN) technique for controlling quality in offline mode. Gyasi et al. [42] are employing an artificial neural network (ANN) to predict geometric characteristics of the welded cord. Wan et al. [43] integrate multiple linear regression analysis and back-propagation neural network to estimate the weld quality. Yue-zhou et al. [44] use sound monitoring and develop a classification algorithm with SVM (support vector machine). Sumesh et al. [45] use machine learning algorithms for weld quality monitoring, acoustic signature,

In addition, there has been an increase in the use of artificial intelligence algorithms

Another area that has been highly developed in recent years and future perspectives is image processing. A great number of algorithms have been created for high performance in this subject. Thanks to these advances, the monitoring of the weld

It is known that metal transfer has a direct influence on the stability of the process and on the final quality of the welding. Consequently, it has been widely

welding (GMAW), a modification of GMAW, used to control the metallic

But innovative techniques continue to appear in this field with future prospects of great interest. In this case, they were identified as laser-enhanced gas metal arc

studied as demonstrated in the present review of the literature.

*Number of publications for years in the area of pulsed GMAW.*

and sensorial fusion. Two powerful techniques have enabled the monitoring and control of welding processes in real time. Also, and as expected, we already find in the literature novel proposals for applications of artificial intelligence and robotics.

and the perform classification use J48 and random forest algorithms.

**Figure 17.** *Techniques used to develop the indexes classified by groups.*

#### **Figure 18.**

*Classification of the main methods used for monitoring.*

Note that the highest percentage of investigation is aimed at the study of metal transfer stability. It is also evident that the main processing techniques to develop the indexes were the mathematical formulation and statistical methods. Although in the case of metal transfer, image processing is widely used, mainly to define the transfer mode and drop size.

**Figure 18** shows a taxonomy that details the methods used to measure the stability of the welding process and the techniques associated with them. The techniques used were divided according to Weglowski [40] into traditional and nontraditional.

Finally, to find a trend and a possible vision of the direction of future studies, the following was analyzed:
