Modeling and Prediction of Mechanical Properties of Mild Steel Weldments Using Artificial Neural Network (ANN) Technique Based on Simulation Software

الملخص

 Tungsten Inert Gas welding, also known as Gas Tungsten Arc Welding (GTAW), is an advanced arc welding process that becomes a popular choice when a high level of weld quality or considerable precision welding is required. However, the major problems of the TIG welding process are its slow welding speed and limited ability to produce material with lower thickness in a single pass. In this work, TIG welding has been performed on a 10 mm thick EN-5A mild steel plate without using any filler material. They developed a model using the artificial neural network (ANN) technique based on the Japan Welding Engineering Society (JWES) software to find the significant effect between the welding inputs process parameters, namely current, velocity, and voltage, on the mechanical properties namely: HAZ maximum hardness Hv-5, and weld metal tensile strength. After the simulation process, the maximum hardness of HAZ, Hv-5, and the weld metal tensile strength of the weld were investigated by finding predicted and optimum values for each model. The implemented validation for each model was done using the mean absolute percentage error MAPE and Nash Sutcliffe efficiency (NSE), respectively. It was observed that the ANN technique gave a mean absolute percentage error MAPE low, and Nash Sutcliffe efficiency (NSE) high, indicating that each model is accurate and excellent.  The ANN technique is an accurate prediction model of the HAZ maximum hardness Hv-5, and weld metal tensile strength of the weld. Therefore, they are recommended for predicting of the weld of the arc welding process.