An Axis Fault Detection Method of Hydraulic Turbine Unit Based on Lightgbm

The axis trajectory of hydraulic turbine units is a key parameter to characterize whether hydraulic turbine operates normally. Abnormal axis trajectory affects the running state of hydraulic turbines directly, which affects the quality of power generation adversely. In this paper, we proposed an Axis fault detection method of hydraulic turbine unit based on Light Gradient Boosting Machine (Lightgbm). First, the key attributes are extracted from the historical jigger data, and built a dataset to characterize the state of axis. Then the kernel principal component analysis method is adopted to reduce dimension and redundancy of the dataset, which improves detection precision and efficiency obviously. Finally, the Lightgbm algorithm is optimized by regular loss function, which is used to diagnose the abnormal axial state of hydraulic turbine. The experimental results show that the detection accuracy of the method proposed in this paper is 97.18%, the method has high accuracy and real-time performance, which is of great significance to ensure the normal operation of the turbine unit.