8 Industry 4.0 in Welding
267
ability to detect most of the welding defects [22]. In order to overcome the limitation
of being applicable as an offline inspection method, these NDT techniques have been
clubbed with intelligent algorithms such as neural networks and fuzzy reasoning to
implement them in real time. A review article [23] in this regard can be referred which
discusses on the application of radiography, ultrasonic and eddy current inspection
techniques. The use of NDT methods has also been explored for quality monitoring
of FSW technique. While the X-ray technique has been found to be able to detect
the voids, it failed to detect the other types of defect such as “kissing-bond” defect
[24] and tiny root flaws. The tiny voids were captured by using the ultrasonic c-scan
technique. This suggests that different NDT methods would be required for fulfilling
the need of quality control. The ultrasonic phased array technique has also been
utilized for inspection of the welds in FSW [25].
8.3.1.2 Vision-Based System
Other than these, the vision-based systems also come in the category of direct monitoring. These systems have been utilized to extract information about the molten pool
in welding. The system makes use of light which is projected onto the workpiece
to be welded, and the distance within the vision system and the workpiece forms
one of the reference measurements. During welding, whenever a variation occurs,
this distance varies leading to change in the position of the reflected beam which is
captured by using a camera. The captured data is then processed digitally to infer
useful information about the welding technique. The inferences include weld seam
tracking, edge detection, groove, tracking of beginning and end positions of the weld,
and information about the weld pool [26]. The vision-based systems can either be
referred as active or passive, depending upon the nature of illumination.
The active vision-based system refers to the use of a source of light to illuminate
the workpiece area, and the passive vision-based use the arc light for imaging [27].
A number of experiments have utilized the vision-based systems for tracking of the
seam during arc and laser welding techniques [28–32]. These works have utilized
active vision sensor, such as laser for detection. Beside the impressive results obtained
in these works, they have been criticized for the high capital investment, inability in
tracking complex seams and error in look-ahead detection [27]. This is so because the
successful implementation requires proper illumination of the targeted areas, which
limits them to be explored in laboratories rather than in industry. An alternative to this
problem is the passive-based vision sensor which also has been explored for tracking
the weld seam [27]. Using this sensor, the cost has been reported to be reduced by
15 times as compared to that of the active laser system [27]. However, this increases
the pre-processing exercise, as the images suffer from the intensity of the arc light
and have random noises.
Seam tracking is a major issue in the robotic welding [33]. As mentioned earlier,
the robots have been identified as one of the core technologies for implementation of
automation and control in the manufacturing sector. The robots which are utilized for
arc welding have laser system attached to the torch. Robots track the seam and control
267
ability to detect most of the welding defects [22]. In order to overcome the limitation
of being applicable as an offline inspection method, these NDT techniques have been
clubbed with intelligent algorithms such as neural networks and fuzzy reasoning to
implement them in real time. A review article [23] in this regard can be referred which
discusses on the application of radiography, ultrasonic and eddy current inspection
techniques. The use of NDT methods has also been explored for quality monitoring
of FSW technique. While the X-ray technique has been found to be able to detect
the voids, it failed to detect the other types of defect such as “kissing-bond” defect
[24] and tiny root flaws. The tiny voids were captured by using the ultrasonic c-scan
technique. This suggests that different NDT methods would be required for fulfilling
the need of quality control. The ultrasonic phased array technique has also been
utilized for inspection of the welds in FSW [25].
8.3.1.2 Vision-Based System
Other than these, the vision-based systems also come in the category of direct monitoring. These systems have been utilized to extract information about the molten pool
in welding. The system makes use of light which is projected onto the workpiece
to be welded, and the distance within the vision system and the workpiece forms
one of the reference measurements. During welding, whenever a variation occurs,
this distance varies leading to change in the position of the reflected beam which is
captured by using a camera. The captured data is then processed digitally to infer
useful information about the welding technique. The inferences include weld seam
tracking, edge detection, groove, tracking of beginning and end positions of the weld,
and information about the weld pool [26]. The vision-based systems can either be
referred as active or passive, depending upon the nature of illumination.
The active vision-based system refers to the use of a source of light to illuminate
the workpiece area, and the passive vision-based use the arc light for imaging [27].
A number of experiments have utilized the vision-based systems for tracking of the
seam during arc and laser welding techniques [28–32]. These works have utilized
active vision sensor, such as laser for detection. Beside the impressive results obtained
in these works, they have been criticized for the high capital investment, inability in
tracking complex seams and error in look-ahead detection [27]. This is so because the
successful implementation requires proper illumination of the targeted areas, which
limits them to be explored in laboratories rather than in industry. An alternative to this
problem is the passive-based vision sensor which also has been explored for tracking
the weld seam [27]. Using this sensor, the cost has been reported to be reduced by
15 times as compared to that of the active laser system [27]. However, this increases
the pre-processing exercise, as the images suffer from the intensity of the arc light
and have random noises.
Seam tracking is a major issue in the robotic welding [33]. As mentioned earlier,
the robots have been identified as one of the core technologies for implementation of
automation and control in the manufacturing sector. The robots which are utilized for
arc welding have laser system attached to the torch. Robots track the seam and control