Deep Learning-based Bearing Diagnosis System for Wind Turbines
The bearings of wind turbines,as their key components,often fail due to wear and cracks under normal operation,resulting in the collected vibration data containing other interfering signals,and the traditional diagnostic methods have large error in fault detection,leading to poor accuracy of the diag-nostic results.To address this problem,a wind turbine bearing diagnosis system based on the DRSN-CW model is proposed.The system combines software and hardware design,algorithm theory and simulation results analysis,aiming to improve the accuracy and efficiency of bearing fault detection.The hardware de-sign is carried out by suitable sensors and data acquisition devices,and the software design including mod-ules of data preprocessing,feature extraction,model training and inference is constructed to realize the au-tomated bearing fault diagnosis process.DRSN-CW is selected as the base model,which combines soft valorization,residual network and attention mechanism to effectively learn important features in bearing signals.Simulation experiments are conducted using Case Western Reserve University bearing vibration da-ta,and several different fault diagnosis algorithms are analyzed.Experimental results show that the accura-cy of the DRSN-CW model outperforms other methods.
国家科技期刊平台
NSTL
万方数据
