Optimizing Fault Diagnosis in Wind Turbine Generators with CALO-Optimized Wavelet Kernel Extreme Learning Machine

Abstract

Focusing on the fault characteristics of generator equipment in wind turbines, this study introduces a fault diagnosis approach that leverages an enhanced ant lion optimization algorithm (CALO) to optimize a wavelet kernel extreme learning machine (WKELM). Initially, the feature vector of SCADA data is extracted using the guaranteed bureau projection algorithm. Subsequently, the CALO algorithm incorporates a Cauchy mutation operator to enhance the traditional ant lion algorithm, thereby boosting its global optimization capability. The CALO algorithm is then employed to fine-tune the parameters of the wavelet kernel extreme learning machine, aiming to enhance the diagnostic precision and convergence rate of the method. To validate the efficacy of the proposed diagnostic model, experimental tests were conducted using real-world operational data from wind turbine equipment at a wind farm in northwest China. The simulation results indicate that the CALO-WKELM diagnostic model effectively identifies various generator faults, fulfilling the requirements for generator fault diagnosis.