[2F8] Transfer learning for fault classification in heavy rotating machines
G Greenberg, R Cohen, O Matania, S Shoham, I Dattner and J Bortman
Health monitoring of rotating machines, particularly in the aviation industry, poses a significant challenge: developing machine learning models without direct access to labelled faulty data from the monitored machines. This challenge stems from the stringent safety standards of the aviation industry, which minimise fault occurrences and, consequently, limit the availability of faulty data.
In contrast to traditional fault classification methods, the approach presented here addresses this critical gap by enabling fault detection and classification in scenarios of limited or no faulty data. A novel domain adaptation methodology is proposed, employing transfer function estimation to enable transfer across different sensors (TDS).
In this paper, a case study involving gear wear faults is evaluated, demonstrating the ability of the method to classify fault severity with an increase of at least 20% in accuracy, showcasing its adaptability across domains. This work lays the foundation for advancements in the development of transfer across different machines and digital twins, providing an innovative framework for predictive maintenance and health monitoring in safety-critical industries such as aviation.
In contrast to traditional fault classification methods, the approach presented here addresses this critical gap by enabling fault detection and classification in scenarios of limited or no faulty data. A novel domain adaptation methodology is proposed, employing transfer function estimation to enable transfer across different sensors (TDS).
In this paper, a case study involving gear wear faults is evaluated, demonstrating the ability of the method to classify fault severity with an increase of at least 20% in accuracy, showcasing its adaptability across domains. This work lays the foundation for advancements in the development of transfer across different machines and digital twins, providing an innovative framework for predictive maintenance and health monitoring in safety-critical industries such as aviation.
