Fault Analysis And Research On Vibration Signal Of Main Bearing Of Large Wind Turbine

As environmental pollution problems become more and more serious,people are gradually recognizing the importance of clean energy such as wind energy.Large wind turbines are constantly evolving along with failures.Due to the large structure of the large wind turbines and the harsh working environment,the large wind turbines are also subjected to large loads.The main bearings are also important components of the wind turbines.Failures will result in failure to work properly,which will cause great economic losses.There will be casualties,so it is necessary to analyze the fault and quickly solve the problem of the main bearing fault,so that the wind farm can operate normally and obtain economic value.Because the vibration signal acquisition is relatively easy,this paper mainly analyzes the fault of the main bearing vibration signal.The specific research content is divided into the following parts:For the working environment of the main bearing of large wind turbines,the collected vibration signal will contain a lot of noise,which will cover the characteristic signal of the main bearing.It will be denoised by wavelet and wavelet packet method,and introduce the signal-to-noise ratio and mean square.Root error to compare the noise reduction effect.The effect of noise reduction is verified by experiments,and the fault analysis of the main bearing vibration signal is fully prepared.In view of the sudden change of the main bearing failure,the feature extraction is performed by the multiscale permutation entropy(MPE)method with strong detection ability.Firstly,rolling body fault signals were used to select multi-scale alignment entropy parameters;secondly,multi-scale alignment entropy was used for feature extraction of experimental signals;then,neural network method and Extreme learning machine(ELM)method were used for fault location identification and comparison.Finally,the MPE-ELM method is used to analyze the fault signal of the main bearing of a large wind turbine,and the fault location can be accurately identified.For the main bearing in the event of a fault,there will be a pulse excitation problem.First,the signal is subjected to wavelet packet denoising processing,then the empirical mode decomposition(EMD)method is used for signal decomposition,and the Hilbert transform is used for envelope analysis.Only the initial degree of bearing failure can be diagnosed.Based on this,the EMD-ELM method is used to analyze the degree of failure.After denoising,the signal is decomposed by method,the feature vector is established by the relationship of energy,and the fault is identified by ELM and neural network method,and compared.Finally,the EMD-ELM method is used to analyze the fault signal of the main bearing of the large wind turbine,which can accurately identify the fault degree.