Research On Fault Diagnosis Of Fan Rolling Rearing Based On Optimized VMD Algorithm

In recent years,the wind power industry has developed rapidly with its features of clean,safe and stable.The proportion of wind power installed capacity and power generation is increasing year by year.However,with the rapid development of wind power industry,the maintenance and repair of wind turbine endpoint effect has the characteristics of high cost,long time and low accuracy.Events that cause major accidents caused by rolling bearing often frequently occurring,which seriously affects the unit’s power generation efficiency and power quality.Therefore,researching on fault diagnosis of wind turbine rolling bearing has important theoretical significance and engineering value.Firstly,the principle and vibration frequency of the rolling bearing of wind turbine generators are introduced.And the principle of the variational mode decomposition is introduced,too.Compared with the empirical mode decomposition algorithm in the simulation signal,the components decomposed by the VMD algorithm contain more fault feature information than that decomposed by the EMD algorithm.The VMD can overcome the shortcomings of the EMD algorithm in modal overlap and end endpoint effect.Secondly,according to the high noise characteristics of the rolling bearing vibration signal,the Morlet wavelet threshold denoising method suppresses the rolling bearing vibration noise signal.The extended particle swarm optimization(EPSO)is used to optimize the parameter combination[K,α]in VMD,and compared with the genetic algorithm(GA)to verify the effectiveness of EPSO.The intrinsic mode function is extracted using the optimized VMD algorithm,the kurtosis criterion is used to select the best IMF,which has the most fault information.AND it is used for envelope demodulation,and energy entropy(He IMF_i)is constructed as a fault feature vector.Finally,in order to solve the problem of low fault diagnosis rate of wind turbine rolling bearing,a diagnostic method based on EPSO and support vector machine classification is proposed.Using the EPSO algorithm to optimize the parameters of the offline support vector regression machine,train and optimize the model.The measured data is used to verify the effectiveness of the proposed algorithm comparing the support vector regression machine model of genetic algorithm.The improved optimized variational mode decomposition algorithm proposed in this paper realizes the diagnosis of damage degree and damage position of wind turbine rolling bearing.The algorithm improves the fault diagnosis rate of rolling bearing,and the accuracy rate is up to 95.7%.