Multilayer cyclic periodicity-induced symplectic matrix decomposition for incipient fault diagnosis of wind turbine gearboxes
MA Xiaojing
JIA Haikun
ZHOU Chuandi
YU Xiang
ZHAO Na
WANG Dong
ZHANG Bo
TENG Wei
Abstract：[Objective]Aiming at the problem that wind turbine gearboxes operate in harsh environments,their incipient fault features are easily masked by strong background noise,and traditional signal processing algorithms are difficult to adaptively extract fault features,an improved matrix decomposition method was proposed to achieve accurate identification of incipient faults in wind turbine gearboxes under complex working conditions.[Methods]Firstly,a multilayer periodic cyclic matrix was constructed by utilizing the time-domain periodic impact characteristics of rotating machinery faults to enhance fault periodic features and disperse aperiodic noise;secondly,multilayer symplectic matrix decomposition was performed on the matrix to separate fault components and interference components in the signal;then,a periodically constrained kernel mapping induced operator was designed to directionally screen fault-related principal components in each layer;finally,the screened principal components were weighted and averaged to reconstruct a fault feature signal with high signal-to-noise ratio,and verification was carried out through simulated signals,test bench signals and wind turbine field-measured signals.[Results]Verification shows that the proposed method requires no manual parameter setting and has excellent adaptive performance;the periodic spectral energy level indices of simulated,test bench and field-measured signals are as low as 1.84,2.12 and 1.56 respectively,which are significantly better than comparative algorithms such as the empirical mode decomposition,the variational mode decomposition and the symplectic geometry mode decomposition.It can effectively suppress background noise and discrete frequency interference,and the accuracy of fault feature extraction is significantly improved,providing a reference for incipient fault diagnosis of wind turbine gearboxes.
Keywords：Wind turbine gearboxIncipient fault diagnosisPeriodicity-inducedSymplectic matrix decomposition
Publication Date：2026-05-31
Online Publishing Date：2026-09-12(First online date of this platform, not the publication date of the document)
Pages：10( 76-85 )