Prediction of tensile behaviour of hybrid laser arc welded Inconel 617 alloy using machine learning models

The current study deals with hybrid laser arc welding (HLAW) of Inconel 617 alloys under different heat inputs (0.516 kJ/mm and 0.615 kJ/mm) whereas the room and high temperature tensile properties are considered as an output parameter. The high temperature tensile tests are conducted at 650 °C, 700 °C, and 750 °C respectively. The effect of heat input on the tensile properties of welded joints was comparatively investigated using tensile tests. The base metal shows superior tensile strength (814 MPa) than the weld joints. The maximum high-temperature tensile strength of weld joints is 538 MPa was achieved at the heat input of 0.516 kJ/mm and temperature of 750 °C. Temperature has a significant effect on the tensile properties of weld joints. A significant drop in tensile strength was observed at higher temperatures and high heat input (0.615 m/min). In addition, a significant improvement of tensile strength at room temperature and low heat input (0.516 kJ/mm). This works deals with the application of various regression analysis methods like Linear Regression (LR), Decision Tree Regression (DTR), Random Forest Regression (RFR), Polynomial Regression (PR), Gaussian Process Regression (GPR) and Support Vector Regression (SVR) were used for predicting the room and high temperature tensile behaviour of hybrid laser arc welded Inconel 617 alloy. The effectiveness of each model was evaluated in the training and validation stages and the results were compared to experimental data. To assess the effectiveness of the designed models, the two quantitative standard statistical measures of root mean squared error (RMSE) and R² were applied. The statistical performance factors of the models reveal the superior performance of PR model was found better than LR, DTR, RFR, SVR, and GPR in predicting the tensile strength of joints.