which deviate both sideways and in height。 In Figure 3。8, a work-piece (T-pipe) is shown as an example of a complex weld path that is difficult to generate a program for and thus, a sensor with a capability to both track and generate the weld path is a suitable technique。

3。3Monitoring

The ability to monitor the weld quality automatically is important in order to reduce production costs and to assure and improve weld quality。 An automatic detection system should be able to classify different weld defects such as porosity, metal spatter, irregular bead shape, excessive root reinforcement, incomplete penetration and burn-through。

Monitoring systems for weld parameters such as ADM III, Arc guard, Analysator Hannover 10。1 and Weldcheck are commercially available [2],[6]。 They all work in a similar way: voltage, current and other process signals are measured, presented and compared with preset nominal values。 An alarm is triggered when any difference from the preset values exceeds a given threshold。

Thus, an important feature of monitoring is that it is done during welding and using data that exist during the welding process。 To be able to make any judgment about the quality, reference data must be available including models or algorithms that describes and evaluate measured parameter。

An important task of any monitoring system which is used for quality assurance or quality control purposes is to be able to present the data with respect to quality measures as consistently as possible。 This means that alarm thresholds defined must be correlated with real weld defects or relate to specifications defined in the WPS。 An important aspect in this context is to understand that the welding process displays a more unstable situation when the data frequency of the readings are increased, and consequently, measurements of process parameters at lower frequencies, providing they display mean values, will display a more stable process。

The information within the WPS does not normally account for this but includes nominal operating data for different controllable parameters。 Thus, part of the monitoring system for control purposes is to define alarm thresholds with  respect to the WPS to maintain the process within nominal parameter limits and at the same time produce a weld at the defined quality and productivity levels。

The examples given here are limited to the detection of changes in the weld quality both automatically and on-line in spray GMAW when using signal processing methods。 However, the method as such can in principle be applied to any welding method providing that knowledge exists about the stability criteria of the process and how to measure significant parameters related to the stability。

Gas Metal Arc Welding (GMAW) is widely used in welding applications because of the specific advantages it offers such as reduced spatter and smoother bead appearance as compared with submerged arc welding。 There are two stable metal transfer modes in GMAW: short-circuit metal transfer at low arc voltage; and spray metal transfer at high voltage and high current。

In the field of GMAW of steel, both physical analysis of the welding process [7]- [11] and statistical analysis of real welding signals have been performed [12]-[21]。 However, the problem of classifying the weld with respect to quality is still in focus for research and is an important area to produce efficient control systems which include, if not all, the most important parameters and how these affect the quality and productivity, and the proper definition of the corresponding WPS for control purposes。

Short-circuiting transfer Globular transfer Spray transfer

10。45    10。46    10。47    10。48    10。49     10。5     10。51    10。52    10。53    10。54      10。55