In-process penetration monitoring of the pulsed laser welding process remains a great challenge for achieving uniform and reproducible products due to the highly complex nature of the keyhole dynamics within the intense laser-metal interactions. The main purpose of this study is to investigate the feasibility of acoustic emission (AE) measurement for penetration monitoring based on acoustic wave characteristics and deep learning. Firstly, a series of laser welding experiments on aluminum alloys were conducted using high-speed photography and AE techniques. This allowed us to in-situ visualize the complete keyhole dynamics and elucidate the generation mechanism of acoustic waves originating from pressure fluctuations at the keyhole wall. Then, an adaptive time-frequency technique namely VMD (Variational Mode Decomposition) was proposed to characterize the acoustic energy distribution among the nine subsignals with low-frequency and high-frequency components under different welding penetrations. Lastly, a novel hybrid model combing CNN (Convolutional Neural Network) and LSTM (Long Short Term Memory) was designed to deeply mine the spatial and temporal acoustic features from the extracted frequency components. Extensive experiments demonstrate that our proposed approach yields a remarkable classification performance with a test accuracy of 99.8% and a standard deviation of 0.21, which obtains a high recognition rate. This work is a new paradigm in the digitization and intelligence of the laser welding process and contributes to an alternative way of developing an efficient end-to-end penetration monitoring system.

Plasma arc deposition as an additive manufacturing technology has unique advantages for producing parts with complex shapes through layer-by-layer deposition. It is critical to predict and control the temperature field during the production process due to the temperature distribution and gradients determining the properties and performance of the part. Numerical simulation approaches, such as the finite element method, which provides a large amount of data for machine learning modeling, thus reducing the overhead of experimental measurements, are widely used in machine learning. In this paper, we propose a neural network combined finite element method and process prediction workflow. A one-dimensional convolutional neural network model for predicting 2D temperature distribution is developed by training the collected data on the planar temperature field of titanium–aluminum twin-wire plasma arc additive manufacturing and the finite element method. The results show that the predicted temperature mean square error is only 0.5, with less than a 20 °C error in peak temperature and a relative error below 1%. The proposed transfer learning method achieves the same training loss and is 500 iterations faster than basic training, which improves the training speed by 25%. The current study confirms the accurate performance of the ML model and the effectiveness of the optimization method.