Research On Neutrosophic Theory Based Fault Diagnosis Methods For Wind Turbine Gearbox

As an inexhaustible source of clean energy,wind energy has received extensive attention in the field of new energy.With the rapid development of the wind power industry,the risk of failure and maintenance of wind turbines has also increased.As the core component of the doubly-fed wind turbine,the gearbox is prone to mechanical failure due to the severe working conditions.Therefore,it is significant to carry out the study on fault diagnosis of wind turbine gearboxes.Therefore,this paper focuses on the fault diagnosis of wind turbine gearboxes,and combines the intelligent diagnosis algorithm with the neutrosophic theory.The main research and analysis are as follows:(1)In-depth study of the structure of the wind turbine gearbox and common faults.According to the research status of fault diagnosis,the common fault diagnosis methods are summarized.(2)Based on the analysis of the operating characteristics,failure mechanism and vibration fault characteristics of the gearbox,the nonlinearity,non-stationary,difficult feature extraction and quantification of the gearbox vibration signal under complex conditions are studied.The quantitative description of signal feature extraction based on wavelet packet is studied.The method of reconstructing the components of the frequency band of interest can effectively extract and accurately describe the characteristics of the vibration signal of the gearbox.(3)With the advantage of the neutrosophic theory in dealing with the inconsistency and uncertainty of the information,the traditional neighbor-supervised learning algorithm is improved,and the neutrosophic k-nearest neighbor based fault diagnosis method for gearbox is proposed.The classification and identification of the gearbox faults have achieved good diagnostic results.At the same time,the application of the neutrosophic partition in the hybrid fault is illustrated,which provides a new research idea for the intelligent diagnosis technology of the gearbox fault of the wind turbine.(4)Aiming at the shortcomings of the neighbor learning algorithm which is susceptible to the outliers,an improved neutrosophic KNN fault diagnosis algorithm based on the principle of fusion neighboring neighbors and kernel method is proposed,which effectively identifies the outliers and eliminates the outlier information for the diagnosis algorithm.The accurate identification of the fault type is realized in the high-dimensional feature space,and the fault diagnosis algorithm is further optimized,which improves the reliability and applicability of the algorithm in fault diagnosis.(5)Finally,the fault diagnosis and condition monitoring system of the gearbox was built for the experimental scheme.The hardware composition and software architecture system of the system were briefly explained,and the design scheme was verified by simple experiment.