ASTFA Method And Its Application To Rotating Machinery Trend Analysis And Life Prediction

Nowadays,trend analysis and remaining life prediction method for rotating machinery mainly depends on extraction of fault feature parameters.Time-frequency analysis methods have been widely used to extract fault feature parameters of rotating machinery.And several commonly adopted time-frequency methods are wavelet decomposition,empirical mode decomposition(EMD)and local mode decomposition(LMD).However,those aforementioned methods all have some drawbacks which are difficult to resolve,thus making research on new time-frequency methods become a hotspot in the research area of trend analysis and remaining life prediction of rotating machinery.This dissertation has investigated the method of adaptive and sparsest time-frequency analysis(ASTFA)and its application to the research field of trend analysis and remaining life prediction.Firstly,ASTFA method has been modified in order to overcome the drawbacks;Secondly,the modified ASTFA method has been applied to extract the fault feature parameters of rolling bearings,thereby predicting remaining life of rolling bearings;Lastly,quantitative diagnosis of gear crack fault has been realized by combining the modified ASTFA method and the dynamic model of a gear transmission system.The main research contents of the dissertation are listed as follows:1.Aimed at the drawback that ASTFA method selects the initial value of the phase function blindly,the dissertation has proposed three initial value search methods,namely the whole phase initial value search based ASTFA method,the genetic algorithm initial value search based ASTFA method and the gold section initial value search based ASTFA method.Then these three methods have been treated as representatives,and similarities and differences of these methods have also been analyzed,thusly providing a gist for selecting proper initial search method in the practical applications.2.In consideration of the phenomenon that the intrinsic mode function(IMF)obtained by ASTFA arrays irregularly,the dissertation has studied this phenomenon and utilized principle mode analysis(PMA)to modify the ASTFA method.Then the PMA based ASTFA method(PMA-ASTFA)is appplied to rotor fault diagnosis.Results show that PMA-ASTFA is suitable for frequency recognization of rotor fault,thus proving the effectiveness of the presented method.3.The modified ASTFA method has been adopted to the trend analysis and life prediction of rolling bearings.Firstly,fault feature parameters of life-cycle data sample of rolling bearings have been extracted;Secondly,fault feature parameters which can manifest trend state well in the trend analysis have been selected to predict residual life of rolling bearing by combining the modified ASTFA method and life prediction model.Analysis results have indicated that the life prediction method has reference value for life management of actual faulty rolling bearings.4.Quantitative diagnosis of gear root crack has been conducted by combining dynamic modeling and time-frequency method.Based on the experimental rig of rotating machinery,the dynamic model of a gear transmission system has been established.The dynamic response of the gear transmission system has been obtained via inner stimuli caused by time-varying mesh stiffness and time-varying friction factor.And the influence of gear root crack and surface frictional coefficient on the gear dynamic response has also been investigated.Fault feature parameters which are sensitive to gear root crack in the dynamic simulation analysis have been applied to quantitative diagnosis of gear root crack.Diagnosis errors verify the effectiveness of the modified ASTFA method based quantitative diagnosis method for gear crack fault.