Research On Fault Diagnosis Of Wind Turbine Bearing Based On Adaptive Chirp Mode Decomposition

Rolling bearings are an indispensable key component of wind turbines,and their operating status directly affects the efficiency and safety of wind turbines.Conducting research on rolling bearing fault diagnosis for wind turbines is of great significance for ensuring the safe and reliable operation of the entire unit.This article takes the adaptive chirp mode decomposition(ACMD)as the core theory and focuses on the analysis and processing of vibration signals,exploring the diagnosis of rolling bearing faults in wind turbines.Based on the analysis of the signal processing characteristics of ACMD,this research explores key issues such as extracting single weak fault features of bearings,separating compound fault features,fusing multi-channel fault features,and identifying fault features under variable speed.The main work and achievements are as follows:Using numerical simulated experiments of fractional Gaussian noise,the equivalent filtering characteristics of ACMD were studied,and the impact of key parameters on the signal processing characteristics of ACMD was qualitatively analyzed through a simulated model of rolling bearing faults in wind turbines.The results show that ACMD has band-pass filtering characteristics,and the instantaneous frequency and weighted coefficient respectively affect the central frequency position and frequency domain coverage range of the extracted modal components,and reasonable parameter settings can help to extract the target features inherent in the signal.In response to the problem of extracting single weak fault features of rolling bearings in wind turbines under constant speed,a feature extraction strategy that optimizes ACMD and improves the integration of Multi-point optimal minimum entropy deconvolution adjusted(MOMEDA)was proposed.The integrated linear kurtosis index was used as the fitness function of the grassland prairie dog optimization algorithm to jointly optimize the instantaneous frequency and weighted coefficients of ACMD,thus automatically determining the best parameter combination and achieving fast extraction of sensitive modal components of the original signal.The performance of MOMEDA was improved using multi-point kurtosis and autocorrelation energy,and the fault features were further enhanced by deconvolution processing of the obtained modal components.The simulated and measurement signal verification results of weak faults show that the proposed strategy can significantly suppress background noise interference and accurately extract the weak fault features of the bearing.Aiming at the problem of separating complex fault features of wind turbine rolling bearings under constant speed,the iterative modified adaptive chirp mode decomposition(IMACMD)method is proposed.By using the dual constraints of envelope interpolation and Spearman rank correlation analysis,the sensitive frequency bands of different feature information can be located,and the minimum number of modal components and the instantaneous frequency of each modal component can be determined.In addition,by using the aggregated linear kurtosis index to set the weighted coefficient in the IMACMD iteration process and processing the original signal,modal components containing different types of fault features can be decomposed.Simulated and experimental results of complex faults show that the proposed method can effectively decouple and separate the different fault impact components of the bearings,and has advantages in determining compound faults.For the problem of multi-channel fault feature fusion of wind turbine rolling bearings under constant speed,a feature fusion enhancement method based on multi-dimensional data processing of ACMD is proposed by introducing the fast and adaptive multivariate empirical mode decomposition(FAMVEMD)algorithm into the field of mechanical fault diagnosis.The modal overlap ratio index is used to control FAMVEMD to decompose the multi-channel signal,and the modal components in the same frequency range are weighted and reconstructed to achieve a high degree of fusion of correlated features.Subsequent processing using ACMD can effectively improve the signal-to-noise ratio.Simulated and experimental results of multi-channel signals show that the proposed method,with the advantage of multi-channel signal fusion,can enhance the amplification of fault features contained in the signal and help accurately determine the operating status of the bearings.Focusing on the topic of rolling bearing fault identification in variable speed wind turbines,a feature identification scheme based on instantaneous frequency estimation of ACMD is constructed.After envelope analysis of the original signal,local maximum synchronous compression transformation is used to achieve accurate estimation of instantaneous frequency.The obtained instantaneous frequency curve is input to ACMD to extract various order feature components of the envelope signal,and ultimately achieve accurate representation of the timefrequency characteristics of the envelope signal.Simulated and experimental results under variable speed conditions show that the proposed scheme can construct time-frequency characteristics with higher resolution and effectively identify bearing faults under variable speed conditions.The achievements of this paper provide new ideas for further research on the extraction of weak single fault features,separation of complex fault features.fusion of multi-channel fault features,and recognition of fault features under variable speed conditions in wind turbine rolling bearings.At the same time,they can provide a reference for solving similar problems in the diagnosis of rotating machinery faults.