Research On Fault Diagnosis Of Fan Gearbox Based On Support Vector Machine

With the rapid development of economy,traditional energy has been unable to meet the needs of countries,clean energy is more and more sought after by countries.China is one of the world’s most populous countries,and the demand for new energy is very urgent.The Chinese government is also increasing its investment in new energy every year.Among them,wind energy as an important clean energy is also an integral part.In China,the wind resource reserve is also very rich,especially in the southeast coast and northwest grassland desert zone.Because the wind resources are rich in the area,the environment is also very bad,which has a great test for wind power equipment,wind power equipment failure is also frequent.In order to minimize the occurrence of faults,it is very important to establish a complete set of prevention and protection fan equipment system.The perfect prevention and protection mechanism can not be separated from scientific artificial intelligence modeling.Because the modeling needs massive data support,but the actual wind power equipment fault data is difficult to obtain.Therefore,this paper establishes the design method of SVM fault diagnosis classifier based on wavelet packet decomposition according to the characteristics of wind power equipment fault.This paper improves the traditional method of gearbox fault diagnosis,which combines wavelet packet with support vector machine of particle swarm optimization.Firstly,data acquisition is needed.Due to the unsteady characteristics of the fault vibration signal of gearbox,the vibration signal needs to be transformed into an intuitive expression of energy distribution.In this paper,wavelet packet is selected as the signal decomposition method.The vibration signal is extracted and analyzed;the twin support vector machine and fuzzy support vector machine are structured by the method of binary tree,so as to achieve the effect of multi classification of gearbox fault samples.In order to verify the classification accuracy of the multi fault classifier established in this paper,the multi fault classifier established in this paper is tested by using the actual data of gearbox provided by Shanghai wind power group.According to the experimental results of this paper,the data shows that the multi fault classifier established in this paper has super high classification recognition degree and remarkable classification effect.This paper focuses on the problem of fault diagnosis of the gearbox of wind turbine.Combining wavelet packet decomposition with support vector machine,it mainly does the following work and analysis:(1)According to the different faults of the gearbox of the wind turbine,four typical state representations are selected and analyzed to explore the state representation of the gearbox fault from the perspective of time and frequency.In this paper,the fault signal extraction based on wavelet packet decomposition is adopted.Here,the vibration signal is transformed into an intuitive energy spectrum vector form,which is decomposed at a high frequency threshold,and then more intuitive and convenient State eigenvector of processing;(2)The obtained eigenvectors are fed into the fault diagnosis model.This paper improves the method of fault diagnosis based on wavelet packet decomposition and support vector machine.On the basis of classical support vector machine,two new types of support vector machine are optimized by the strategy of binary tree structure,which are twin support vector machine and fuzzy support vector machine.The fuzzy membership degree of fuzzy support vector machine is calculated by the fuzzy c-means algorithm,and their penalty factors are also calculated Particle swarm optimization improves the efficiency and accuracy of fault diagnosis.(3)At last,two optimized SVM are compared with their fault diagnosis results,and their advantages and disadvantages are analyzed by data,and a conclusion is drawn.