Wind turbine bearing fault diagnosis method based on energy valley optimization algorithm optimization multi kernel extreme learning machine

A wind turbine bearing fault diagnosis method based on Variational Mode Decomposition (VMD), and Energy Valley Optimization Algorithm Optimization Multi Kernel Extreme Learning Machine (EVO-MKELM) is proposed to address the difficulty in accurately identifying the types of bearing faults in wind turbines under complex operating conditions. Firstly, VMD is used to decompose the vibration signal of wind turbine bearings and obtain the modal components containing the bearing state characteristics. Secondly, in order to improve the accuracy of the proposed feature quantity while avoiding feature redundancy, the correlation coefficient method is used to select modal components with close correlation, solve the Multi Scale Fuzzy Entropy (MFE) values of multiple modal components with high correlation, and use the obtained MFE values to construct a feature set. Then, with the goal of achieving the highest diagnostic accuracy, EVO is used to optimize the key kernel parameters of MKELM and establish an EVO-MKELM diagnostic model; Finally, input the feature set into EVO-MKELM for recognition and diagnosis. The experimental results show that the overall diagnostic accuracy of the proposed diagnostic method has reached 98.67%, achieving the goal of high-precision diagnosis.