Future research directions and applications of artificial intelligence in tribology

Artificial intelligence (AI) is transforming tribology by enabling the prediction and understanding of friction, wear, and lubrication at a higher level than conventional methods. This is important because 20–23% of global energy is wasted due to friction and wear, and 30–40% of machine failures are due to these issues. This review analyzes recent developments, existing problems, uses, and prospective research on AI in tribology. It also holds tribo-informatics that is a combination of tribology and data science. The review deals with multiple scales of tribology, including small particles up to large components and long-term wear experiments, as well as existing machine-learning methods. These methods include physics-informed machine learning, which uses major equations, such as lubrication and wear equations. Studies have shown that traditional machine learning (ML) can predict friction and wear with 90–95% accuracy, whereas deep learning improves wear prediction by 20–40%. CNN-based methods can identify wear states with 95–98% accuracy, and hybrid methods predict the remaining life of parts with less than 5–10% error. Challenges include limited data, complex interactions, hard-to-understand models, high resource needs, and privacy issues, making most industrial data unavailable. Despite this, AI improves production efficiency by 15–30%, reduces unexpected failures by 20–40%, and cuts maintenance costs by 10–25%. These have enormous effects on aerospace, renewable energy and manufacturing industries. Others are that it needs open databases, legal explainable AI to show why it made its decisions and digital twins and autonomous labs to improve diagnostics, lubricant design, and energy-saving systems.