Zusammenfassung
For the automation of multipass welding processes few optical sensors are used because of the heavy distortions caused by the welding arc and the strongly varying geometry of the welding groove. A new intelligent sensor head with an integrated processing unit is able to reduce the distortions and reduce the data in real time. Fast programmable hardware units are able to do the pre-processing. The reduced data showing the height profile of the welding groove are transmitted to a PC. By the use of further distortion filters and fuzzy algorithms characteristic features of the groove are extracted form the height profile. These features are necessary for calculating geometry attributes of the welding groove. These attributes are the input of a model for the adaptation of the welding parameters. To build this model several sources have to be used. There are many sources to build this model and the generation is very time-consuming, therefore an adapting of the model during welding will be feasible. Correcting the automatic welding process by an experienced welder the model can be adapted. After the putting into operation phase the model adapts itself by scanning the seam after welding.
Nutzer
