Research On SVM Fault Diagnosis Of Fan Bearing Based On Feature Dimension Reduction

Wind energy is an important part of the clean,clean energy industry,and governments around the world are increasing their investment in wind turbines every year.However,due to the extremely harsh working environment of the wind farm,the failure rate of the wind equipment will increase.Bearings are one of the key components of wind turbine transmission equipment,and their maintenance costs are very large,which has caused great economic losses for wind farm enterprises.Therefore,it is very important to make the fan camp the basis for fault diagnosis.In this article,the Western Reserve University Bearing data set will be used for corrective research and functional reduction will be made from the perspective of managing data globally and locally and the information that actually reflects incorrect information will be completely refined.Excavated.Work with the following areas:(1)Functional dimension reduction algorithm based on random projection and orthogonal locally sensitive discrimination analysis,focusing on complex types of bearing defects and large amounts of original data.The algorithm begins by storing global data,high-dimensional original data is randomly presented to a low-dimensional space through a subdivision,and then generates a mapping matrix that eventually captures the locally sensitive low-dimensional properties.Dimensional mapping indirectly shows the effect of dimension reduction and proves that each evaluation indicator method improves the validity of model validation by several,reduces the effect of data shift and reduces the model validation time.(2)Support vector machines have great advantages in solving classification problems,but are greatly affected by parameter selection.To improve the diagnostic accuracy of the support vector machine model,this article introduces the wheel algorithm to optimize the support vector machine parameters worldwide,designs the parameter correction process,regulates the optimal parameter synthesis,and creates a vector correction support vector.Machine diagnostic model.And compared to various other diagnostic models,the diagnostic model developed in this article confirms that it is characterized by several classes and multidimensional properties.