Research On Fault Diagnosis Of Wind Turbine Gearbox Based On Fiber Bragg Grating Detection Technology

The working environment of wind turbine is bad,and the disordered change of wind speed causes frequent start and stop of wind turbine,which is easy to cause greater damage to dynamic load parts.Among them,the gear box has the highest failure rate.Therefore,How to accurately reveal the state of health in the process of wind power gear box in the work is very important.In this paper,FBG sensing technology is used to collect data of fault signals generated by wind turbine gearbox,and fault diagnosis is carried out by using extreme learning machine,probabilistic neural network and support vector machine,and the effectiveness of the method is verified through experiments.The main contents and conclusions of this study are as follows:(1)According to the sensing characteristics of fiber Bragg grating,the waveguide structure,reflectivity and strain sensitivity are firstly introduced,Then the coupled mode theory and strain sensing characteristics of FBG are analyzed.Finally based on the edge of the FBG filter method,interferometer demodulation method and tunable wavelength demodulation principle of Febry-Perot filter method and module are introduced,for the construction of the experimental platform to provide a solid theoretical foundation and proved the feasibility of optical fiber dynamic diagnosis system design.(2)Aiming at the design of the optical fiber dynamic diagnosis system,the experimental platform of optical fiber dynamic monitoring was designed by using the optical fiber sensing technology which has strong sensitivity to strain.Firstly,the layout of calibration FBG and sensing FBG is designed,and the application program is designed according to the requirements of signal analysis.Through the combination of software and hardware,the optical fiber dynamic diagnosis system is built.(3)In order to extract the gearbox fault feature information effectively,two algorithms are designed for feature extraction.Firstly,the modal parameter K of VMD is analyzed,MPE parameters such as the embedding dimension and data length.The entropy value was calculated by combining VMD and MPE.Secondly,multi-component calculation is carried out by EEMD and evaluated by energy entropy value.Finally,the sequence containing the most fault feature information is selected by MPE and EEMD algorithm respectively to construct the feature vector set.(4)Through to the wind turbine gearbox fault diagnosis,first of all,using the method of spectral kurtosis analysis of signal,and the gear box of different fault types identification.Secondly,the VMD-MPE-ELM algorithm was used to diagnose the faults.The average correct probability of ELM was 99.225%,and the running time was 0.345 s.The principle of PNN and SVM algorithm is analyzed respectively,and the fault diagnosis of wind turbine is realized.The average correct probability is 97.46% and98.33% respectively.Secondly,a comparative analysis is carried out from the correct probability and running time.Finally,the effectiveness of the proposed method is verified by experiments.