Rotating Machinery Fault Diagnosis using Light Gradient Boosting Machine

Rotating machinery is widely used in modern industrial technology. Timely diagnosis of faults of rotating machinery equipment is of great significance to maintain the reliability and safety of the whole system. Since the development of fault diagnosis technology, there have been many diagnosis methods that can be applied to rotating machinery, and these methods have achieved good results. However, many of these methods cannot balance the relationship between diagnostic accuracy and timeliness very well, and require high computing capabilities of the device, which is not conducive to algorithm deployment on hardware devices, and the long diagnosis time is not conducive to real-time monitoring of the rotating machinery. This paper takes the core component bearing of rotating machinery equipment as the object, and proposes a fault diagnosis method for rotating machinery based on light gradient boosting machine (LightGBM). In this paper, two kinds of bearing data sets are used for ten-fold cross-validation, which can achieve high accuracy and very short training time. The experimental results show that LightGBM has higher diagnostic accuracy and better real-time performance.