Fault Feature Extraction And Diagnosis Of Transmission Components Of Wind Turbines Under Variable Speed Condition

With the strategic goal of achieving "carbon peak and carbon neutrality," wind energy is being widely adopted in power generation to reduce dependence on traditional energy sources,decrease the demand for imported energy,and improve energy security,leading to a cleaner and greener power system.However,during normal wind turbine operation,the turbines are subjected to multiple loads,including random fluctuations in wind speed and power grid,resulting in significant irregular vibrations.Additionally,various noise components are produced by components such as the yaw and hydraulic systems,and flexible support,which cannot be avoided in the vibration data measured by wind turbines.Moreover,as wind turbines operate at variable speeds for extended periods,the frequency spectrum ambiguity caused by wind speed fluctuations affects the efficacy of signal processing methods in the frequency domain during condition monitoring and fault diagnosis of bearings and gears.Furthermore,the complex structure of the transmission system creates multiple transmission paths that modulate the signals from different excitation sources during transmission,making it difficult to accurately identify the fault type.Therefore,it is of great significance to investigate the fault generation mechanisms of gears and bearings under the comprehensive action of complex movement and variable environmental excitation,and explore the potential relationship between fault characteristics,vibration response,load fluctuation,and other variables.This research will enable efficient and accurate feature extraction and fault diagnosis,ensuring the safe and stable operation of wind turbines.In this context,this paper focuses on the key transmission components of wind turbines,namely the gears and bearings,and investigates their fault generation mechanisms,signal transmission paths,and the effects of speed changes on vibration signals.Moreover,this paper explores methods for efficient feature extraction from fault states.The main research contents of this study are as follows:To address the frequency modulation,amplitude modulation,and spectral blurring of vibration signals caused by wind speed,an adaptive iterative instantaneous speed estimation method was proposed in this study.This method involved extracting harmonic components through cepstrum editing and iteratively extracting instantaneous speed information by combining the frequency-domain energy operator and instantaneous frequency error estimation method.The proposed method not only effectively improves the accuracy of instantaneous speed estimation but also enables speed information extraction under conditions of large speed fluctuation through multiple iterations.These developments lay the foundation for subsequent state feature extraction and fault diagnosis using advanced signal processing technology.In compound fault diagnosis of gears and bearings,fault information is often coupled with each other,and bearing fault information may be masked by the stronger vibration energy of gear meshing.To address this issue,this study proposes a composite fault diagnosis method that combines discrete random separation(DRS)technique and improved Autogram method.The method involves instantaneous speed estimation for vibration signals of complex faults,and elimination of periodic component interference caused by gear meshing vibration in fault signals through DRS technology.A new index,consisting of spectral kurtosis and spectral negative entropy,is designed to quantitatively characterize the narrow-band components obtained via maximum overlapping discrete wavelet packet transform and unbiased autocorrelation processing.Moreover,an improved Autogram method is developed to determine the appropriate demodulation frequency band.The effectiveness of the proposed method was demonstrated through testing vibration signals of wind turbine gear/bearing complex faults,indicating that the method can effectively extract complex fault information.The fault characteristic frequency of motor bearing is often masked under strong background noise and electromagnetic interference,posing a challenge for fault diagnosis.In this study,the electromagnetic vibration generation mechanism is analyzed,and a fast spectral coherence method with a double-layer noise reduction fusion is proposed to extract weak fault characteristics of the bearing.Firstly,the instantaneous speed of the fault vibration signal is estimated and the vibration signal is converted from equal time sampling to equal angle sampling.Then,a combined double-layer noise reduction approach using CEEMDAN-SANC is employed to suppress background noise and electromagnetic interference components,effectively overcoming electromagnetic vibration components that might mask the fault signal.Finally,fast spectral coherence is utilized to extract the hidden carrier frequency and cycle(modulation)frequency information for fault diagnosis.The proposed method is verified using vibration data of a wind turbine bearing fault,and the results indicate its strong feature extraction ability for fault information under strong electromagnetic interference.