Research On Diagnosis Method Of Wind Turbine Gearbox Based On VMD Multi-feature Extraction And Multi-kernel SVM

According to statistics,the failure of the gearbox in the wind turbine accounts for about 4% of the total,but it causes the longest downtime.The resulting operation and maintenance costs and the power generation losses caused by the temporary shutdown are the largest,which seriously affects the efficiency of power generation and wind power.Economic benefits.Therefore,it is very important to study the fault diagnosis method of wind turbine gearbox.When a gear fails,its vibration signal tends to be non-stationary.In this paper,in order to effectively identify the faults of the gears,the optimized variational modal decomposition and multi-feature fusion method are adopted in the fault feature extraction.The whale optimization algorithm and multi-core support vector machine are used in the pattern recognition.Combined method.The main research contents of this article are as follows:First,the effectiveness of the VMD method is proved by comparing the EMD method with the VDM method,and it is proved that the VMD method can avoid the problem of modal aliasing in the EMD to a certain extent.In order to better use the VMD method,the singular value difference spectrum theory is used to reconstruct and reduce the noise of each component of the VMD.By comparing the reconstructed components of the VMD with the original components,it can be seen that the frequency domain characteristics of each component in the frequency domain after noise reduction are more prominent,and there are fewer interference frequencies.Secondly,the theory of whale optimization algorithm and support vector machine algorithm are introduced in detail.In the process of using the support vector machine,in order to solve the problem that the relevant parameters are difficult to select,a whale optimization algorithm is introduced to optimize the parameters of the support vector machine.And use part of UCI data set to do classification experiment.The final result shows that the support vector machine after the whale optimization algorithm has better classification accuracy.Finally,analyze the gear vibration signals collected in the laboratory,calculate and select the best combination of eigenvalues ??to construct a feature vector,andinput the optimized support vector machine for pattern training and classification recognition.And the actual fan gear data is analyzed,the final result shows the effectiveness of the method.