Control of dilution rate of Inconel 625 alloy surfacing based on elman network algorithm

This paper established the feedback Elman neural network model of a 5-8-3 structure, in order to analysis the correlation between the input parameters of surfacing welding process including the arc length, welding current, welding speed, wire feeding speed and shielding gas flow rate and these output variables such as the height,the widththe dilution rate of the surfacing weld. The result showed that the Elman model prediction results were more accurate than BP and GRNN neural network. A four dimensional image to determine surfacing welding process window with error bound of δ≤5% was established on the basis of simulation calculation results of the Elman model, in which arc length (X), welding current (Y) and wire feed speed (Z) as the space coordinates and surfacing welding dilution rate function δ=f(X, Y, Z) as the objective function. The dilution rate δ were 2.55% and 3.32% respectively, the relative error of dilution rate of the simulation was about 0.8%, which confirmed the feasibility and reliability of the Elman model predictions.