A novel fault diagnosis technique for wind turbine gearbox

Abstract

The gearbox is one of the most important parts of a mechanical equipment. The importance of fault diagnosis in rotating machineries for preventing catastrophic accidents and ensuring adequate maintenance has received considerable attention. In this study, a fault diagnosis method based on gearbox vibration signal monitoring is used to differentiate the signal characteristics of different working conditions and improve the accuracy of diagnosis. The time-domain sequence approximate entropy (ApEn) adaptive strategy is used to propose a wind turbine intelligent fault diagnosis algorithm based on a wavelet packet transform (WPT) filter and a cross-validated particle swarm optimized (CPSO) kernel extreme learning machine (KELM). First, the correlation between the parameter requirements of the intelligent diagnosis system and the system complexity analysis is analyzed. Then, the parameters related to the wavelet filter is determined by calculating the ApEn of the time-domain sequence. Finally, a compact wind turbine gearbox test bench is constructed and tested to validate the proposed ApEn-WPT+CPSO-KELM to identify gearbox-related faults for verification. Results show that the proposed ApEn-WPT+CPSO-KELM method can accurately identify four states of the wind turbine gearbox.