Acoustic emission for in situ process monitoring of selective laser melting additive manufacturing based on machine learning and improved variational modal decomposition

Selective laser melting (SLM) additive manufacturing overcomes the geometric limits of complex components produced with traditional subtractive methods, which has significant advantages in designing and manufacturing special-shaped components. However, due to the lack of adequate and effective process monitoring, it is difficult to ensure the reliability of as-built parts and the stability of the additive manufacturing process. Therefore, it is necessary to monitor the as-built part quality of the SLM process. An in situ quality monitoring method of acoustic emission (AE) based on machine learning and improved variational modal decomposition (VMD) is proposed in the present work. The VMD parameters are adjusted based on the whale optimization algorithm (WOA) and average energy entropy to realize the adaptive decomposition of the AE signals. Each sub-mode is evaluated according to the signal energy, the feature vector used for SLM printing quality prediction is extracted. Finally, the artificial neural network (ANN) and support vector machine (SVM) are employed for quality prediction. The improved VMD method is compared with empirical modal decomposition (EMD), aiming to verify the predictive validity of printing quality in the SLM process. The results show that predicting SLM printing quality based on improved VMD is better than the EMD method. Meanwhile, it is verified that online monitoring of SLM for improving printing quality can be achieved based on the AE technique.