Condition monitoring of rotating shafts is essential for ensuring the reliability and optimal performance of machinery in diverse industries. In this context, as industrial systems become increasingly complex, the need for efficient data processing techniques is paramount. Deep learning has emerged as a dominant approach due to its capacity to capture intricate data patterns and relationships. However, a prevalent challenge lies in the black-box nature of many deep learning algorithms, which often operate without adhering to the underlying physical characteristics intrinsic to the studied phenomena. To address this limitation and enhance the fusion of data-driven methodologies with the fundamental physics of the system under study, this paper leverages physics-informed neural networks (PINNs). Specifically, a simple but realistic numerical case study of an extended Jeffcott rotor model, encompassing damping effects and anisotropic supports for a more comprehensive modelling, is considered. PINNs are used for the estimation of five parameters that characterize the health state of the system. These parameters encompass the radial and angular position of the static unbalance due to the disk installed on the shaft, the stiffness along the principal axes of elasticity, and the non-rotating damping coefficient. The estimation is conducted solely by exploiting the displacement signals from the centre of the disk and, to showcase the efficacy and precision provided by this novel methodology, various scenarios involving different constant rotational speeds are examined. Additionally, the impact of noisy input data is also taken into account within the analysis and the performance is compared to that of traditional optimization algorithms used for parameters estimation.

Condition monitoring of rotating shafts is essential for ensuring the reliability and optimal performance of machinery in diverse industries. In this context, as industrial systems become increasingly complex, the need for efficient data processing techniques is paramount. Deep learning has emerged as a dominant approach due to its capacity to capture intricate data patterns and relationships. However, a prevalent challenge lies in the black-box nature of many deep learning algorithms, which often operate without adhering to the underlying physical characteristics intrinsic to the studied phenomena. To address this limitation and enhance the fusion of data-driven methodologies with the fundamental physics of the system under study, this paper leverages physics-informed neural networks (PINNs). Specifically, a simple but realistic numerical case study of an extended Jeffcott rotor model, encompassing damping effects and anisotropic supports for a more comprehensive modelling, is considered. PINNs are used for the estimation of five parameters that characterize the health state of the system. These parameters encompass the radial and angular position of the static unbalance due to the disk installed on the shaft, the stiffness along the principal axes of elasticity, and the non-rotating damping coefficient. The estimation is conducted solely by exploiting the displacement signals from the centre of the disk and, to showcase the efficacy and precision provided by this novel methodology, various scenarios involving different constant rotational speeds are examined. Additionally, the impact of noisy input data is also taken into account within the analysis.

Condition monitoring of rotating shafts is a critical task in the maintenance of mechanical systems. Rotating shafts are essential components in many machines, and their failure can result in serious consequences, including system downtime, production loss, equipment damage, and safety issues. Advanced sensor technologies and deep learning algorithms have facilitated data collection and processing, providing vital insights into system health. However, despite the vast availability of sensors, the data used to train these algorithms often consists of single-nature signals, and systems are typically damaged to simulate different faulty scenarios. Additionally, the perceived opacity of deep learning algorithms, often referred to as black-box models, has raised concerns about their credibility in critical domains. Hence, this paper addresses these challenges by (i) proposing various explainable artificial intelligence (XAI) methods in a rotating shaft case study, (ii) using a novel data-fusion approach to combine multiple signals, and (iii) leveraging data from an innovative experimental set-up mimicking real-world industrial machines. The employed experimental set-up is equipped with a diverse array of sensors capable of capturing signals of varying nature, and it streamlines the automated introduction of four distinct fault types in an innovative manner. Applied to convolutional neural networks, the employed XAI methods enhance transparency in deep learning models, providing practical insights for complex systems. This approach not only addresses the limitations associated with single-nature signals and simulated faults but also contributes to the credibility and interpretability of deep learning models in critical applications.