A review of the multi-dimensional application of machine learning to improve the integrated intelligence of laser powder bed fusion

Abstract

Laser powder bed fusion (LPBF) as one of the most promising additive manufacturing (AM) technologies, has been widely used to produce metal parts and applied in fields such as medical and aerospace. However, the full development of LPBF is still limited by the existence of a limited variety of materials suitable for LPBF, a single structure of printed components, defects in the processing, and complex post-processing. Machine learning (ML), as the core of artificial intelligence (AI), is expected to be an effective tool for LPBF related researches due to its ability to find latent relationships among numerous research issues. This paper provides a newly comprehensive review of ML applications to LPBF. The ML-assisted three stages of LPBF, including the pre-processing, in-situ processing and post-processing, are systematically reviewed. In the pre-processing phase, the application of ML in designing lightweight structures as well as high-performance materials are analyzed in detail. The optimization on the identification of the process and real-time monitoring during the in-situ processing stage are thoroughly discussed. In the post-processing stage, the material-structure-performance (MSP) relationship and their optimization are summarized. Based on the comprehensive review, challenges and perspectives for ML multiscale application development are eventually envisaged. This work contributes a great help to utilize ML to AM and suggests meaningful guidelines for future ML applications to manufacturing work.