Takefumi Kubota

Controlling of torch attitude and seam tracking using neuro arc sensor

It is important to realize intelligent welding robots to obtain a good quality of weld. For this purpose, it is required to detect the deviation from the center of the groove, the torch height, and the torch attitude. In order to simultaneously detect these, the authors propose a neuro arc sensor as the sensor fusion by using a neural network. First, the authors deal with the welding phenomena as the melting phenomena in the electrode wire of the MIG welding. Next, the training data of the neural networks are made from the numerical simulations. A neuro arc sensor is trained so as to get the desired performance by the backpropagation method. The welding experiments are carried out to examine the performance of the neuro arc sensor. A good performance of the neuro arc sensor is obtained. By using it, the seam tracking can be performed in the T-joint welding.

Sensor fusion using neural network in the robotic welding

It is important to realize intelligent welding robots to obtain a good quality of the welding results. For this purpose, it is required to detect the torch height, the torch attitude, the deviation from the center of the gap. In order to simultaneously detect those, the authors propose sensor fusion by using the neural network, i.e., the information concerning the welding torch is detected by using both the welding current and the welding voltage. First, the authors deal with the welding phenomena as the melting phenomena in the electrode wire of the MIG welding and the CO/sub 2/ short circuiting welding. Next, the training data of the neural networks are made from the numerical simulations. The neuro arc sensor is trained so as to get the desired performance of the sensor. By using it, the seam tracking is carried out in the T-joint.

Back bead control of the one-side robotic welding with visual sensor-cooperative control of current-waveform and torch motion for change of gap and welding position

Control of the weld pool in a first layer of the one-side multi-layer welding is important to obtain a good quality of the weld. For this purpose, the authors propose a new method, the switch back welding method, to get the stable back bead. In the forward stroke, the visual robot observes the groove shape and the root gap. In the backward process, it observes the surface of the weld pool every half cycle of the weaving to investigate whether the pulsed droplet is deposited to the root edge of the base metals and whether the deposited metal bridges the gap between the both base metals or not. The welding speed, the current waveform and the distance of the backward and forward stroke are obtained by using the fuzzy inference to get the stable and wide back bead.

Application of neural network to detection of arc length, extension length and root gap in robotic welding

Full penetration control of the weld pool in a first layer of the one-side multilayer welding is important to obtain a good quality of welding. For this purpose, the authors propose a new method, the switch back welding method, to obtain a stable back bead. A welding torch is not only oscillated in the groove, but also moved backward and forward. Both voltage and current are entered to neutral networks to estimate the wire extension and the arc length. Moreover, the gap and the deviation of the oscillation center from gap center are estimated. Training data are made from experimental results. Performance of the arc sensor is examined by giving testing data to the neural networks.