Research On Analysis Method Of Characteristics Of Typical Defects For Aircraft Rolling Bearing

As an important component of the aviation industry, the air bearing is one of itscomponents easily damaged, whose reliability has been a research focus and emphasis.In order to assess the real-time performance air bearing effectively. Bearing defectdetection system is very important, In view of the traditional bearing defect analysisthe status quo, unable to deal with complex vibration signal, This paper introduced thewavelet analysis method, which overcomes the drawback of traditional analysismethod, and can complete the non-stationary signal analysis effectively. In addition,this article also puts forward the energy eigenvector extraction method based onwavelet analysis, and finally realize the effective air rolling bearing defects analysis.Aimming at the bad characteristics of the aviation equipments workingenvironment, the operational requirement of air rolling bearing is analyzed in thispaper, the typical defects caused before use and its basic characteristics of air rollingbearing are studied. The frequency characteristics of vibration signals of rollingbearings with defects are studied, the theoretical model of typical defects and thecharacteristics of vibration signals under the condition of different defects are obtainedwhich provides the theoretical basis for the rolling bearing defects recognition.For the nonlinear of air rolling bearing system and nonstationary characteristic ofvibration signals, the avelet theory and its de-noising algorithm are studied, the de-noising preprocessing of rolling bearing simulation signals is completed, the feasibilityof wavelet theory is verified. The extraction method of energy feature vector from thevibration signals is studied based on the principle of wavelet packet. And theconstruction of the dynamic monitoring testbed and simulation test of rolling bearingwith defects are completed, the feature vector of rolling bearing with defects isobtained, which provided sample datas for the recognition of system defects types.In this paper, the main technical parameters of support vector machine (SVM) is stuilded, SVM classifier is built by combining the energy eigenvector extracted, theoptimal parameters of classifier is obtained. By comparing the combinationmechanism of usual algorithms like “one to many”,“one to one” and the SVMdecision tree, the advantages and disadvantages of them are analyzed. A new algorithmbased on the SVM decision tree support vector machine and voting theory is proposed,the energy eigenvector extracted are used for training and testing the support vectormachine, finally, reached the ideal classification accuracy by recycled a few trainingsamples, completed the aviation defect analysis of rolling bearing.