Research On Fault Diagnosis Of Motor Rolling Bearing Based On CEEMDAN And CSSA-SVM

As an important component that affects the normal operation of rotating electric machines,rolling bearings are also very prone to failure.Therefore,early fault detection of the working state of the motor bearings appears importance awfully to ensure the normal operation of rotary motor,as a result,the fault detection of rolling bearing has great practical significance.This paper takes the feature extraction and identification of various conditions of motor bearing normal and abnormal state as the discuss target.First of all,this paper analyzes the physical structure of motor rolling bearing,the reason and the feature when motor rolling bearing goes out of order,at the same time,it discusses frequency performance of different situations.This paper draws Complete Ensemble Empirical Mode Decomposition with Adaptive Noise(CEEMDAN)into fault feature extraction of motor rolling bearing,then the eigenvectors of different working states of motor rolling bearing are obtained.The intrinsic mode function(IMF)of rolling bearing signals in different working states is obtained by using ceemdan signal mode decomposition method,and then the energy moment of IMF is calculated respectively,and the eigenvectors of rolling bearing in different working states of different motors are obtained.Afterwards,this paper draws Chaos Sparrow Search Algorithm &Support Vector Machine(CSSA-SVM)into identification of different working states of motor rolling bearing.From a mathematical point of view,it is proved that CSSA algorithm is superior to SSA algorithm in solving accuracy,then,the CSSA algorithm is used to optimize the penalty factor and kernel function in SVM,which makes the accuracy of the algorithm recognition better.CEEMDAN and CSSA-SVM are used for analyzing the actual measurement data of different working states of motor rolling bearings.Compared with other methods,the recognition results of the method show colossal gigantic effectiveness and superiority.The thesis has 34 pictures,9 tables,and 59 references.