Defect classification of laser metal deposition using logistic regression and artificial neural networks for pattern recognition

Detecting laser metal deposition (LMD) defects is a key element of evaluating the probability of failure of the produced part. Acoustic emission (AE) is an effective technique in LMD defect detection. This work presents a systematic experimental investigation of using AE technique for detecting and classifying different defects in LMD. The defects generated during LMD simulate AE sources on deposited material while the AE sensor was mounted on the substrate to capture AE signals. An experiment was conducted to investigate the ability of AE to detect and identify defects generated during LMD using a logistic regression (LM) model and an artificial neural network (ANN). AE features, such as peak amplitude, rise time, duration, energy, and number of counts along with statistical features were extracted and analyzed. Additionally, frequency analysis using fast Fourier transformation was conducted on the AE signal. The results show that AE has considerable potential in LMD monitoring for assessing the overall deposition quality and identifying defects that can significantly reduce the strength and reliability of deposited material, and consequently, increase the risk of a component's failure.