Research On Bearing Fault Diagnosis Of Wind Turbines Based On Multi-resolution Singular Value Decomposition

Among the fault types of wind turbines,bearing faults,gearbox faults,high-speed shaft faults and low-speed shaft faults are the most prone to fault types,rolling bearings are more prone to failure.The non-smooth running condition of rolling bearings and the interference noise brought by the mutual impact of other parts of the equipment drown out the weak information of the early fault characteristics,which brings a great challenge to the fault diagnosis of rolling bearings.For the fault diagnosis of rolling bearings in wind turbines,this paper introduces the Multi-resolution singular value decomposition(MRSVD)method into the fault diagnosis of wind turbines and studies the fault diagnosis of rolling bearings in wind turbines based on MRSVD algorithm.The main research content of this paper is as follows.(1)Research on signal denoising based on multi-resolution singular value decomposition(MRSVD)algorithm,and on this basis,propose a multi-substructure MRSVD denoising method,through simulation signal analysis The noise reduction effect of this noise reduction model is selected,the optimal noise reduction model is selected,and then it is applied to the measured data of the wind turbine bearing fault to eliminate the noise in the rolling bearing vibration signal and highlight the fault characteristics.(2)Aiming at the defect that the traditional Random Forest(RF)parameter selection cannot make the model performance optimal,this paper proposes an improved random forest algorithm(P-RF),that is,the use of particle swarm optimization(Partical Swarm Optimization).,PSO)optimize the two key parameters of the random forest algorithm to make it have better learning ability and better diagnostic performance.Then,the MRSVD decomposition method of multi-substructure is combined with the optimized random forest algorithm,and a fault diagnosis model of rolling bearing based on MRSVD and P-RF is proposed.(3)In order to verify the effectiveness of the proposed fault diagnosis model,the fault diagnosis model is applied to the measured data of the wind farm,and the diagnosis results of the rolling bearing in different fault states and different fault types are tested respectively,and the results are compared with Various fault feature extraction methods are compared with the combined diagnosis model of machine learning.The results show that the diagnosis model proposed in this paper has high accuracy and can effectively diagnose the faults of rolling bearings of wind turbines.