Research On Fault Diagnosis Of Rolling Bearing Of Wind Turbine Based On Improved Self-adaptive Morphological Method

With the continuous development of the social economy,environmental protection is getting more and more attention from people all over the world.People are gradually replacing the use of non renewable energy such as coal,oil,natural gas to other clean energy.This makes the wind power industry rise rapidly and enter a new normal state of steady growth in this context.In China,the development of new energy is more urgent.With the rapid improvement of wind power generation technology,the single machine capacity of wind turbine is increasing,and the proportion of wind power in the power market is rising.The consequent augment of the complexity and cost of the wind turbines,coupled with the harsh environmental conditions of the wind farm,has greatly increased the cost of wind power generation.Therefore,it is of great significance to carry out research on condition monitoring and fault diagnosis methods for wind power generation units to reduce the operation and maintenance costs of wind power generation units and improve the operational economic benefits of wind power generation units.Vibration signal as a carrier of fault information of transmission chain,how to effectively extract signals from the vibration data can effectively reflect the operating state of the wind turbine transmission chain,which is particularly important for wind turbine generator fault diagnosis research.Taking the rolling bearing of wind turbine as the research object,a fault diagnosis algorithm combining a mathematical morphology method and spectral correlation analysis are proposed in this paper.The main contents are as follows:(1)With the non-linear and non-steady characteristics of vibration signals of rotating machinery in wind turbines,this paper proposes a fault diagnosis method for rolling bearing based on the height of self-adaptive and open-close operator self-adaptive of triangular structuring elements of mathematical morphology.Taking the data of Case Western Reserve University as the research object,this method uses signal-based triangular structuring element and morphological closed operator to train known fault signals.After that,the obtained of the height average of triangular structuring elements of each fault type and the average of the optimal weighting factor are applied to the morphological processing of unknown signals.Correlation analysis is made between the unknown test signal and the training signal’s morphological spectrum which is obtained by the fast Fourier transform.Finally,according to the size of correlationcoefficient,the paper identifies and determines the classification of signal faults.This method verifies the effectiveness of adaptive open-close combinatorial operators in fault feature extraction.But for some individual fault types,the result is still interfered.(2)In view of the existence of interference terms in the simulation results of the above algorithms,this paper introduces the concept of composite structuring elements according to the characteristics of the signal itself.This paper also proposes a new W-type structuring element,which improves the self-adaptive morphological algorithm of structuring element height and open-closed operator.After simulation experiments,the results show that the interference situation can be significantly improved.(3)The improved adaptive mathematics morphology algorithm is applied to the actual fan case for experiment.The results show that this method can effectively extract the fault features in the signal and has higher recognition rate and reliability than the traditional mathematical morphology algorithm.