Fault Diagnosis Of Key Components In Planetary Gearboxes

Planetary gearboxes(PG)are key components of many types of mechanical equipment.They are used to transfer torque and rotation and are extensively used in large-scale machines such as helicopters,wind turbines and so on.Due to the operating characteristics of low speed,heavy load and harsh working environment,once their key components such as gears and bearings are suffering faults,the equipments are prone to failure,even lead to major accidents,and finally result in huge economic losses and casualties.Therefore,the hot topic of fault diagnosis is to develop ways of revealing fault characteristics of key components in PG at early stage so that reliable and safe operation of equipments can be ensured.In this regard,this paper focusing on the investigation of fault diagnosis of key components in PG based on vibration analysis,the main topics can be summarized as follows:(1)A revised phenomenological model of the planetary gearbox systemInvestigation and modeling of the vibration responses of gears when they are healthy is the basis of understanding the vibration attributes of the whole PG.Although traditional phenomenological models,to some extent,can describe the vibration responses of the gearbox system,some hidden problems do exist and remain unrevealed.For example,the structural analysis of the vibration spectrum,the fundamental frequency selection,and the neutralization of the vibrations of multiple planetary gears.In this section,the reasons for the above problems are thoroughly investigated and corrections to reported interpretations are made.Finally,the transfer path function is reconstructed to overcome the deficiency of traditional models.Simulation and experimental studies both demonstrate the advantages of the revised model over traditional ones on the vibration responses of the planetary gear systems.This section is also a fundamental work for revealing the fault symptoms.(2)Interpretation of fault characteristic frequencies and its application in fault diagnosisA limitation of traditional spectrum analysis is that it must compare a large number of sidebands for fault diagnosis.A more natural diagnosis method is proposed that uses the amplitudes of a fewer characteristic frequencies than conventional methods.In this paper,the amplitudes of the characteristic frequencies of the vibrations from a planetary gearbox are further interpreted based on a phenomenological fault model.A mapping relationship between these amplitudes and the fault types is established.This mapping relation shows that the health status of different components in the planetary gearbox can be recognized by choosing the sidebands induced by corresponding components.In addition,considering the inevitable speed fluctuation even when the rotational speed is set as a constant,a bandwidth setting strategy is proposed to calculate the energy among a certain bandwidth rather than picking the single peak point of the characteristic frequencies.Experimental studies show that the proposed method can successfully diagnose faults of different components in a planetary gearbox.(3)Effect on dynamic fault meshing behavior and its pattern revealingThe amplitude of a gear fault characteristic frequency contains rich fault information.In planetary gear systems,the meshing behavior of a faulty tooth is the primary source of fault information.However,the meshing positions of a faulty tooth are timevarying and this phenomenon leads to difficulties in capturing the fault information.The motion pattern of the fault meshing behavior has not been revealed in reported studies.A method of determining the motion periods of the fault meshing position is proposed under this topic.The motion period expression is validated through geometrical demonstrations of a specific planetary gearbox with specified parameters.The result shows that the motion periods depend only on the geometric parameters of the gears so that it is an inherent characteristic of the gearbox system.(4)Application of fault meshing pattern in fault diagnosisThe motion pattern of fault meshing positions has not been taken into account in traditional fault modeling.We have introduced the meshing period of the faulty tooth into these models under this topic.The modulation effects of this meshing period are discussed.Simulation and experimental studies both demonstrate that the revised model is able to reflect the real fault-induced vibration than reported ones.The sidebands caused by the fault-meshing period can be regarded as a preliminary judgment in early fault diagnosis.Furthermore,after exploring the internal link between the motion period and the fault-induced vibration a suggestion of the minimum data length is put forward through experimental studies under different working conditions for effective fault diagnosis.