Prediction of mechanical properties of Q345qE steel welded joints based on machine learning

The mechanical properties of welded joints are the key indicators to evaluate the welding quality, which are mainly subject to the rationality of the welding process and the technical level of the welders. Traditionally, enterprises screen out the optimal welding scheme by implementing a large number of welding procedure qualification tests, but this method is costly and inefficient. In this study, the prediction model of tensile strength and yield strength of Q345 qE steel welded joints was established based on the machine learning algorithm of BP neural network. The results show that the prediction accuracy of the model is high, the R2 value is more than 0.96, and the relative error is controlled within\(\pm\) 3MPa (tensile strength) and \(\pm\) 5MPa (yield strength). This prediction method is expected to reduce the dependence on laboratory tensile tests, reduce experimental costs, and has significant application value for actual production.