Application Of Enhanced Difference Product Morphological Filtering In The Fault Diagnosis Of Wind Generator Bearing

At present,all countries are in the era of green clean energy comprehensive development and utilization of the background.Wind turbine,the representative product of wind energy development technology,are now in large use,but because they operate in complex working conditions all year round,their own damage risks are inevitably increased.At the same time,in the face of restrictive factors such as low maturity of wind power technology and poor operation and maintenance conditions in the past,most of the active wind turbines are in a period of high failure.As an important component of wind generator,bearing’s healthy state during operation will directly affect the smooth performance of the whole machine.Therefore,in view of the noisy operating environment of large wind turbines,the vibration signal collected by sensors on the generator bearing is easily polluted by harmonic interference and strong noise,which imposes great difficulty on the subsequent fault feature extraction.So,the wind generator bearing is determined as the research object,and based on the construction of the mathematical morphology filter operator and the selection method of the optimal structural element scale,a new fault feature extraction method is proposed.The aims to enhance the accuracy of extracting fault feature information of wind generator bearings,so as to carry out effective fault diagnosis on it.In order to extract weak fault feature information of rolling bearing under strong noise,a morphological filtering method based on single-scale adaptive enhanced difference product is proposed.First of all,on the basis of the existing theory of mathematical morphology,the construction of morphological operators is innovated,and an enhance difference product morphological filter operator is proposed.Then,aiming at the blind empirical selection of the optimal structural element scale,a dimensionless adaptive selection strategy is proposed,which is called the kurtosis feature energy product.Finally,the single-scale adaptive enhanced difference product morphological filtering method is applied to the simulated signal and the experimental signal of the faulty bearing.The comparison with other methods shows that the proposed method has better feature information extraction performance.Aiming at the problem that the single-scale morphological filtering method cannot completely and accurately extract the feature information of rolling bearings,the thesis proposed an adaptive multi-scale enhanced difference product morphological filtering method.First,based on the superior performance of the enhanced difference product morphological filter operator,a corresponding multi-scale representation is constructed.At the same time,the multiscale weighting coefficient method is introduced to improve the problem of easy loss of feature information in the process of multi-scale signal reconstruction.Then,in order to solve the problem of poor selection of the optimal structural element scale of the multi-scale reconstruction signal,another dimensionless adaptive evaluation criterion is proposed,which is called the feature energy kurtosis.Finally,the multi-scale adaptive enhanced difference product morphological filtering method is applied to the simulation signal and the experimental signal of the faulty bearing.The comparison and analysis with other methods show the accuracy and superiority of this method for extracting bearing feature information.In order to verify the effectiveness of the single-scale and multi-scale enhanced difference product morphological filtering methods proposed in this thesis in the application of wind turbines.By analyzing specific experimental fault signals of wind generator bearings and comparing with other methods.The experimental results show the superiority and engineering applicability of the two morphological filtering methods proposed in this thesis.