Abstract
Measuring arc length is important to obtain good welding quality regardless of the torch deviation from the base metal and the molten pool. Therefore, it is necessary to detect the arc behavior in the transient state in addition to the steady state. For this purpose, this paper proposes the neural network model which outputs the present wire extension from the previous wire extension, welding current and wire feed rate in every sampling period. The structure of this neural network was determined by considering the differential equation which describes the wire melting phenomena. At the same time, the arc length is calculated by the equation of voltage drop in the extended wire and an approximate expression describing the arc characteristics. To confirm the validity of this system, fundamental experiments were carried out. The arc was directly observed and recorded as image data using a high speed camera. The output data from the neural network were compared with the measured data which were obtained from every captured image. It was found that the system can be applied as an arc sensor because of good responses in the transient state and no steady state error.
