Need for AI in Tribology

Introduction

Artificial intelligence (AI) has impressive learning capabilities and rapid processing speeds which offers significant support to researchers by quickly identifying valuable patterns, trends, and associations within complex data. Tribology is the study of friction, wear, and lubrication, and is inherently multidisciplinary, involving interactions across mechanics, thermology, electricity, optics, magnetics, and more. These tribological behaviors are system-dependent which evolve over time and occur at multiple levels and scales. Therefore, the integration of AI in tribology is not only advantageous but also extensively important for enhancing our understanding and innovate within this diverse field. For example, AI’s robust capabilities in data and pattern recognition can identify correlations between various signals (such as vibration, acoustics, electrical, and sound pressure) and wear, enabling the early detection of mechanical wear. Additionally, AI can simulate and predict tribological behaviors under a range of operational conditions, which is essential for designing more durable and reliable mechanical systems.

Early Research

As early as 1986, Tallian employed computer-aided approaches in tribological design. By 1997, researchers began exploring the use of neural networks to predict tribological properties, though this area but saw limited enthusiasm. In 2017, Wu et al. utilized random forest algorithms for wear prediction which sparked significant interest among researchers. This development led to the gradual adoption of AI across various tribology research domains. However, the complexity of these machine learning methods and their varying applicability to specific research needs in tribology posed challenges. To address these issues, the concept of “tribo-informatics” has been proposed, focusing on the systematic integration of AI and tribology. This new direction has quickly gained significant attention in the field.

Figure-1 AI in different tribological fields [1]

Application of AI in different Tribology research

Basic Tribological theories

The application of AI technology in tribology can be divided into three main areas: explaining tribological behavior mechanisms, advancing micro- and nano-tribology, and analyzing behavior patterns based on standard tribological experiments. It is important to note that AI’s primary role is to establish data associations among different categories of information, rather than directly explaining mechanisms. To enhance the effectiveness of AI in the fundamental theory of tribology, integrating physical models can improve the interpretability of AI’s computational results. Additionally, methods such as regression, classification, clustering, and dimensionality reduction can be employed to derive mechanisms and understand data patterns after processing.

Intelligent Tribology

Intelligent tribology represents the importance of AI application in tribology research and is the field most impacted by AI advancements. Intelligent tribology focuses on evaluating operational reliability and predicting the lifespan of tribological systems in critical engineering sectors such as transportation equipment, energy systems, and mechanical processing. As technology has progressed, this field has expanded to include intelligent lubrication and friction material design. Thus, intelligent tribology can be categorized into two main areas: status monitoring, fault diagnosis, and life prediction of tribological systems, and intelligent lubrication and friction material design.

Component Tribology

The fundamental components of tribology include elements that contain the core elements of a tribological system and perform key tribological functions within mechanical systems. These components primarily consist of bearings, gears, tires, fasteners, and seals. Bearings, in particular, are among the most complex and widely used components in tribology, and AI technology is extensively applied in bearing research.

Extreme field Tribology

The advent of extreme service environments such as deep sea, polar regions, deep space, and deep underground, tribological systems are facing challenges. This includes in operating conditions like high speed, heavy load, extreme temperatures, and special environments such as strong radiation and high vacuum. These conditions often lead to severe friction and wear, generating various intense derivative signals, which complicates friction and wear testing, online monitoring, and fault diagnosis. However, the integration of AI technology offers new and more effective solutions to these challenges. AI’s role in extreme tribology is substantial, though the reliability of the data relies heavily on its quantity and authenticity. Obtaining accurate tribology information is a significant challenge in this field. By utilizing AI, extreme tribology can establish correlations between simulated environmental data and real working conditions, helping to bridge the gap between theoretical and practical data.