Modeling of Penetration Depth in Submerged Arc Welding Using Artificial Neural Network

The penetration depth, which is the distance from the surface of the plate to the bottom of the pool or the bottom edge where melting took place, will have a decisive importance in the strength of the weld metal. Submerged arc welding is a manufacturing process that is directly affected by various input parameters and interactions, and these effects directly affect the penetration depth. This research used an artificial neural network with two hidden layers to find the relationship between process inputs and their effects on weld penetration depth. Arc voltage (V), electric current intensity (I), electrode stick-out (N), welding speed (S), and the thickness of the layer of nanoparticles (F) were selected as input layer neurons and penetration depth as output layer neurons. Also, the investigation of the effect of the input parameters on the penetration depth showed that the increased intensity of the electric current increases the heat input to the welding pool. This, in addition to the rise in the melting of the base metal, also increases the penetration depth. Increasing the arc voltage increases the amount of heat input to the welding pool, but the melting speed of the electrode does not change much, so the penetration depth increases slightly.

Keywords:

Artificial neural networks (ANNs) , Modeling and Optimization , Weld geometry , Nanoparticles , Submerged arc welding (SAW).