Quality prediction of ultrasonically welded joints using a hybrid machine learning model

Ultrasonic metal welding has advantages over other joining technologies due to its low energy consumption, rapid cycle time and the ease of process automation. The ultrasonic welding (USW) process is very sensitive to process parameters, and thus can be difficult to consistently produce strong joints. There is significant interest from the manufacturing community to understand these variable interactions. Machine learning is one such method which can be exploited to better understand the complex interactions of USW input parameters. In this paper, the lap shear strength (LSS) of USW Al 5754 joints is investigated using an off-the-shelf Branson Ultraweld L20. Firstly, a 33 full factorial parametric study using ANOVA is carried out to examine the effects of three USW input parameters (weld energy, vibration amplitude & clamping pressure) on LSS. Following this, a high-fidelity predictive hybrid GA-ANN model is then trained using the input parameters and the addition of process data recorded during welding (peak power). Once trained, the predictive model is tested against seven unseen parameter combinations specimens. Analysing the experimental data shows that the LSS performance envelop is non-linear with respect to the process variables of clamping pressure, vibration amplitude and welding energy. Vibration amplitude is the dominant input parameter affecting the LSS of the joints. At a fixed welding energy, the LSS can be increased by increasing vibration amplitude. However, the effect of clamping pressure on LSS is dependent on the level of welding energy. The resultant GA-ANN model accurately predicts the LSS of unseen test data producing a mean absolute percentage error of 7.51% with a Pearson's correlation coefficient of 0.96 for all data. It is demonstrated that including process data in a closed loop reduces the mean prediction error from 13.17% to 7.51