Prediction for emission of environmental burden in GTAW based on combined neural network

The model is established for quantitative predicting the generation of environmental burdens in welding. The key factors for influencing the emissions in GTAW are determined using Taguchi method, including welding current, nozzle height, and welding time. Moreover, the emission model based on RBF-BP neural network was established. It can be predicting the emissions of environmental burden in GTAW when different welding parameters are adapted. The results show that the average error is 6.63% for predicting emissions of environmental burden using RBF-BP combination neural network model. The welding current, nozzle height and welding time are positively correlated with the generation of environmental burdens. That can provide support for reducing the emission of welding environmental burden and formulating reasonable welding process route.