Paper Title
Neural Network Modeling for Optimization of Electron Beam Welding Process Parameters in C45 Steel

Abstract
This paper considers an experimental investigation of geometry characteristics of cross-sections of the heat affected zone at electron beam welding (EBW) of carbon (0.45 wt.%) steel 45. The influence of the EBW process parameters - the welding speed, the beam current and distance from the main surface of the magnetic lens toward the surface of the welded sample is investigated by development and deployment of a two-layer feed-forward neural network. Requirements and constraints for geometrical characteristics the heat-affected zone (HAZ) are set for the cross-sectional area, depth, surface (top) width and mean width at the middle part of the weld. Multicriterial optimization solutions can be considered for EBW process parameter setting at tests of electron beam optical systems and for production of quality welds with the desired properties. Keywords - Electron Beam Welding, Neural Networks, Optimizations, Heat-Affected Zone, C45 Steel.