Recent advances in machine learning for defects detection and prediction in laser cladding process

As a fundamental component of artificial intelligence, machine learning has gained considerable prominence within the domain of laser cladding in recent years. By employing algorithms to analyze data, discern patterns and regularities, rendering predictions and decisions, machine learning has significantly influenced various aspects of laser cladding processes. The emergence of defects during the cladding procedure poses substantial challenges for the quality and performance of the cladding layers. Addressing the reliability and reproducibility of cladding quality is a paramount concern within laser cladding technology. Leveraging data-driven machine learning algorithms enables the monitoring and detection of defects throughout the laser cladding process. Moreover, these algorithms offer avenues for feedback regulation of the cladding process, optimizing parameters, and mitigating cladding defects, thereby establishing this as a research frontier. This paper presents an overview of the typologies and formation mechanisms of defects encountered during laser cladding, elucidates the signal characteristics and expounds upon monitoring principles and methodologies employed within the laser cladding process. Additionally, it synthesizes advancements in machine learning methodologies for signal feature extraction, defect classification, and predictive modeling within the laser cladding process. Furthermore, it encapsulates prevalent machine learning models and algorithms employed for defect detection. The findings highlight the efficacy of machine learning algorithms in detecting defects within laser cladding coatings, while concurrently establishing correlations between feature signals, coating defects, and cladding processes. Presently, supervised learning algorithms dominate the landscape of research, yet the potential of unsupervised and semi-supervised learning algorithms, with their diminished data annotation requirements, garners increasing attention within the realm of laser cladding process monitoring. Collectively, the research findings delineate key focal points and avenues for future exploration within the realm of machine learning methodologies applied to laser cladding processes. • Typical defects and their formation mechanisms were summarized. • Typical laser cladding monitoring technologies were reviewed. • Typical machine learning algorithms for defects classification and recognition were reviewed.