Diagnosis Technology Investigation Of Typical Faults Of Drive Chain In Servo Motor System

Drive chain in servo motor system is widely used in wind power generation,CNC machine tools,heavy machinery and other fields.But the phenomenon that the equipment break down due to its fault is very common in actual situations.In recent years,greater demands have been placed on the reliability and safety of equipment by the advancement of "intelligent manufacturing engineering".The ability of health monitoring and fault identification of drive chain in motor system has become one of the significant signs of the next generation high-grade serve motor driver.Therefore,fault diagnosis and health maintenance technologies have also ushered in new development opportunities.The diagnosis method based on motor drive system,which employs the motor driver as an intelligent sensor to acquire current,motor speed,identified mechanical parameters and other signals as the fault carrier for the monitor and diagnosis of drive chain in motor system,is the latest research hotspot in this field.It doesn't need to install additional sensors,which can effectively reduce the cost.At the same time,the diagnosis method based on the motor drive system is a non-intrusive fault diagnosis method,which will not damage the equipment,nor affect its reliability and service life.In this paper the three typical faults of drive chain in motor system,transmission gear fault,motor bearing fault,and installation misalignment fault,are systematically investigated,aiming to fully tap the potential of diagnosis methods based on motor drive system.Firstly,key technologies,such as the selection of optimal fault carrier,the optimization of fault feature extraction algorithm,and fault judgment independent of health data,are explored in the steady state to improve the diagnosis performance.Then,the algorithms are developed in the spatial domain to realize the fault diagnosis of drive chain in motor system under transient conditions,which can further enhance its industrial application value.Specific research contents include:Aiming at the difficulty of fault feature extraction in gear local fault diagnosis based on motor current signal analysis(MCSA),a dual parameter optimization resonance sparse decomposition(RSSD)scheme based on artificial bee colony algorithm is proposed in this paper.Based on the established motor-gear integrated electromechanical model,the influence of gear fault on motor phase current is analyzed.To solve the problem that the gear local fault components,which are modulated by the current fundamental wave and the meshing frequency related components,exist in different frequency bands,resonance sparse signal decomposition(RSSD)of motor current is proposed for feature extraction.Furthermore,to avoid the influence of artificial selection of RSSD parameters,dual parameters optimized RSSD based on artificial bee colony(ABC)algorithm is employed.Thus,the performance of MCSA for gear local fault diagnosis is further improved and its effective operation range is further expanded.The effectiveness of the algorithm is verified on the gear broken-tooth fault diagnosis experimental platform.In order to improve the diagnosis effect of motor bearing local fault which is relatively weak,fault feature extraction method and drive algorithm preprocessing scheme are studied.Firstly,from the perspective of fault carrier,the mechanisms of bearing local fault diagnosis based on MCSA and motor speed signature analysis(MSSA)are analyzed.On this basis,the spectral kurtosis scheme is employed to extract the highfrequency impact component caused by bearing local fault,and the detection performance of MCSA and MSSA are comprehensively compared on the established bearing crack diagnosis platform.In accordance with the theoretical analysis,the performance of MSSA is better than that of MCSA.An adaptive signal reconstruction method based on ensemble empirical mode decomposition(EEMD)algorithm is proposed to improve the traditional spectral kurtosis algorithm,which can realize the noise reduction and further improve the performance of MSSA.Finally,aiming at the problem that the signal-to-noise ratio of MSSA is reduced due to the periodic speed ripples caused by current measurement errors or other faults,a noise pre-reduction scheme combined with drive algorithm is proposed.With the improvement of drive algorithm,the noise component in speed signal can be suppressed actively.Aiming at the problem that the installation misalignment fault often occurs in initial installation and there is no health data for reference,this paper proposes an innovative detection scheme for initial installation misalignment based on MSSA.Firstly,the influence of parallel misalignment and angular misalignment on motor speed signal is analyzed by modeling.On this basis,model coefficient identification scheme is applied to extract speed periodic ripples caused by misalignment faults,which can avoid the inaccurate FFT results caused by asynchronous sampling.Then,the change regularity of installation misalignment fault related components with the motor speed is further analyzed,and the trend is combined with the ratio of fault components to realize the initial installation misalignment detection and misalignment type identification based on MSSA without healthy data.Finally,the effectiveness of the fault diagnosis technology for initial installation misalignment is verified through simulation and experiment.Aiming at the problem that the loss of repeatability of fault components leads to the failure of conventional fault diagnosis algorithms in transient state,fault diagnosis of the above three faults type is studied in the spatial domain according to the complexity of the algorithm.Firstly,the failure mechanism of conventional schemes under transient conditions is analyzed.Considering the fact that the fault feature is always the periodic function of motor angle,the spatial domain resampling algorithm is studied to recover the periodicity of fault features.On this basis,the synchronous average scheme of spatial domain kinematic error signal is studied;speed order spectrum analysis is applied to diagnose the installation misalignment;and the spectral kurtosis algorithm is improved in spatial domain as the order spectral kurtosis algorithm for bearing fault detection.The effectiveness of proposed schemes is verified by ramp acceleration experiments on corresponding fault diagnosis platforms,and fault feature extraction under transient operation conditions is successfully realized.