Prediction of molten pool temperature and processing quality in laser metal deposition based on back propagation neural network algorithm

Stability of molten pool temperature directly affects the dimensional accuracy, metallurgical defects, solidification microstructure and mechanical properties of formed parts manufactured by laser metal deposition technology. Therefore, the optimization of the macroscopic morphology and microstructure of the formed parts through the stable control of the molten pool temperature has been intensively studied. In this study, three models, based on the back propagation neural network (BPNN), random forest (RF) algorithms and response surface methodology (RSM) approach, respectively, were used to establish prediction relationships between the deposition input parameters and processing status parameters, geometric morphology and mechanical property parameters. Then, the processing temperature and molten pool characteristics of thirty groups of 316L stainless steel single-track cladding layers were analyzed. The prediction results show that the average prediction error (APE) of the prediction of molten pool temperature, track width, track height and micro-hardness of the deposited layer based on BPNN model are 0.5%, 1.3%, 2.9% and 0.1%, respectively, which are better than the prediction results of RF and RSM models. Then, BPNN model was further used to predict the molten pool temperature and processing quality of the deposited layer under the combination of five new process parameters groups. Objective of this study was to lay the foundation for the subsequent design of the molten pool temperature control system to improve the morphology accuracy and mechanical properties of formed parts.