Study On Gearbox Fault Diagnosis Methods Based On Signal Processing

Gearbox is one of the core parts of machinery equipment.It often works in poor and bad environment,so it's easy to fail in the process of operation.Once the gearbox fall failure,it will cause the production interrupted and the economic loss.Diagnosis method based on vibration signal analysis is one of the most effective methods in gearbox fault diagnosis.During Gearbox's normal operation,the vibration signal shows stability form;while fault occurs,the vibration signal is usually in the form of nonlinear and non-stationary performance.So,we must use the time-frequency analysis method.Traditional time-frequency analysis methods have the common shortcoming which is inadaptability.Adaptive time-frequency analysis methods have great advantage in processing non-linear and non-stationary signal.It can reflect the law of the instantaneous frequency changing with time,also can accurately response signal energy along with the change of time and frequency.Empirical mode decomposition method,Local mean decomposition and Extreme-point symmetric mode decomposition method are the most popular adaptive time-frequency analysis methods.In this paper,we introduce the main principle of adaptive time-frequency analysis method and algorithm characteristics and also their problems that exist.Some corresponding improved methods are put forward to overcome their shortcomings.And then these presented methods are applied to fault diagnosis of gearbox.Main research work and innovations of this paper are as follows:(1)The main failure part in the gearbox such as gear is studied,as well as its failure form,failure causes and failure mechanism of the vibration.(2)Introduce the principle and algorithm of empirical mode decomposition method.Study the the main defects of this method.To target the problem of mode mixture of empirical mode decomposition,analytic empirical mode decomposition method is proposed.Through the simulation signal decomposition,the effectiveness of the proposed method is verified preliminarily.(3)The principle of local mean decomposition method is introduced.Aiming at the existence of mode mixture in local mean decomposition,the method of wavelet local mean decomposition is put forward in order to decrease the mode mixing phenomenon of the method.use wavelet local decomposition method and the ensemble local mean decompositon to decompose simulation signals respectively,then use them to split the signal of the rotor-rub fault vibration repectively,the final result shows that the wavelet local mean decomposition method is practical and can be used in rotor fault diagnosis of gearbox.Moreover,compared with the ensemble local mean decomposition method,the method of wavelet local mean decompostion is more effective and the decomposition time is shorter as well as decomposition accuracy is better.(4)The extreme-point symmetric mode decomposition method is a new adaptive time-frequency analysis method and it has not been used in the mechanical fault diagnosis recently.In this paper,it is used in the fault diagnosis of gear.The extreme-point symmetric mode decomposition method combined with energy operator demodulation method is applied to the analysis of gear tooth-broken fault vibration signal in order to implement the fault diagnosis of gear's tooth-broken.The analysis results show that the method is effective and practical and it's useful for the actual fault diagnosis of gearbox.