The Dynamic Analysis And Intelligent Diagnosis Of Rubbing Fault Based On Rotor-Ball Bearing Coupling System

In order to improve the performance of aero-engine, reducing the clearance between rotor and stator is an important approach, but it increases the probability of rubbing between rotor and stator. Presently, in the process of the design, manufacture and application of aero-engine, rubbing fault has been become"a lion in the way", and it is badly threatening the flight safety. Therefore, a model of rotor-ball bearing coupling system is established aimming at the rubbing fault of aero-engine, and the dynamics mechanism analysis of the rubbing fault has been done. Further, the Support Vector Machine (SVM) is also used for the intelligent diagnosis of rubbing fault.Firstly, a dynamic model of rotor-ball bearing coupling system including imbalance-rubbing coupling fault is established aimming at the actual aero-engine model, and the Hertzian nonlinear contact and the radial clearance are considered in this model. And using the time waveforms, the orbits, the frequency spectra, the Poincarémaps and the bifurcation plots etc, the effects of rotating speed, radial clearance, impact stiffness coefficient, imbalance and rotor-stator clearance on system responses are studied.Secondly, the intelligent diagnosis of rubbing fault based on Support Vector Machine is studyed. First, the relational theories of Support Vector Machine are introduced. Next, the ZT-3 rotor test-bed and the aero-engine rotor experimental rig are used to obtain the rubbing fault samples on different running conditions. Based on the study of rubbing mechanism, multiple frequency analysis method is used for extracting the feature symptoms of rubbing fault, so the experiment samples of rubbing faults are obtained. Meanwhile, making use of the dynamic model of rotor-ball bearing coupling system with imbalance-rubbing coupling fault, the simulation rubbing fault samples at different rotating speeds and system dynamics parameters are obtained, and the simulation samples of rubbing fault is also gained. Last, the Support Vector Machine is trained by the simulation rubbing fault samples, and the trained classifier is used to diagnose the experiment rubbing fault samples. Meanwhile, the intelligent diagnosis results of the Support Vector Machine are contrasted with that of the Neural Network, and the results fully show the Support Vector Machine has large application potential in rubbing fault identification of aero-engine. Thirdly, in order to study the nonlinear characteristics of rubbing coupling fault further, in this paper, correlation dimension is used to analyze the nonlinear time series. First, the phase spaces of the nonlinear time series of the coupling faults are reconstructed; next, the correlation dimensions are computed; last, the correlation dimension characteristics of the nonlinear time series with different rubbing degree and on different system running conditions are analyzed, and the change rules of the coupling fault correlation dimensions are found.