A Novel AI-Based Method for Spur Gear Early Fault Diagnosis in Railway Gearboxes

Abstract

Artificial intelligence (AI) applications have started to take place in our lives due to increasing data collection and processing capabilities with developing technology. In this regard, AI-based early fault diagnosis technologies, which have started to gain reliability in automotive, aviation, and wind turbine fields, have begun to use for railway gearboxes in terms of defect detection and predictive maintenance. Gears are one of the most significant components of powertrain systems. The AIbased fault diagnosis has become more prominent in recent years to predict the remaining useful life of gearbox systems. The gearbox early fault diagnosis plays an important role in both security and reducing high maintenance costs. This issue is of great importance in terms of rail vehicle safety and reliability in a medium to long term perspective. This paper deals with an approach of transferability to railway gearboxes of AI-based gear early fault diagnosis methods from other industries. A vibration-based early fault diagnosis approach and test setup are proposed for railway gearboxes. Early gear crack diagnosis is performed using MATLAB with machine learning algorithms. The proposed test setup allows different degrees of tooth cracks in railway gearboxes to be detected at different operating speeds. As a result, it is observed that the proposed AI-based approach is suitable to identify railway gearbox faults and can be adaptable in rail-based transportation systems.