Multi-Sensor Fault Diagnosis of Wind Turbine Gearbox Via Ensemble Filter Network

Abstract

The wind turbine gearbox (WTG) is regarded as a pivotal component in modern power generation systems, primarily due to its capability to streamline structural complexity and optimize energy conversion efficiency. Nevertheless, the transmission systems are frequently subjected to premature degradation, which stems from prolonged exposure to stochastic loading patterns and extreme environmental stressors. To address this reliability challenge, a set of symptom parameters is systematically extracted from different aspects. Then, multiple symptom parameters are fed into the proposed ensemble filter network to achieve the precious and robust fault diagnosis of wind turbine gearbox. Experimental validation was conducted on the WTG fault simulation test platform of Beijing Jiaotong University. The results show that the proposed method is competitive with other existing methods in terms of diagnosis accuracy and performance stability.