A study on the estimate of weld bead shape and the compensation of welding parameters by considering weld defects in horizontal fillet welding

Generally, though we use a vision sensor or arc sensor in the welding process, it is difficult to define the welding parameters which can be applied to weld quality control. The important parameters in arc welding are arc voltage, welding current, welding speed and they affect the weld bead shape, the stability of the welding process and weld quality. Therefore, it is difficult to determine the unique relationship between the weld bead geometry and the combination of various welding conditions. Due to the various difficulties, we intend to use fuzzy logic and neural networks to solve these problems. Therefore, the combination of fuzzy logic and neural networks has an effect on removing the weld defects, improving the weld quality and finding the desired weld bead shape. Finally, this system can help us make an estimate of the weld bead shape and remove the weld defects.