Research On Fault Diagnosis Of Transmission Gearbox Based On Acoustic Signal

Drive gearboxes are the most commonly used mechanical equipment,directly related to various aspects of industrial production,and carry out important theoretical value and practical significance for the fault diagnosis and state monitoring research of this type of equipment.In response to the shortcomings of the traditional vibration signal fault diagnosis,the non-contact detection of the acoustic signal is easy to collect,and the signal is easy to collect,and the early fault can be found,and the target device can be used for online real-time monitoring.The author launches the transmission gearbox based on the sound signal.First,the typing type and mechanism of the drive gearbox is introduced.From the theoretical basis of acoustics,the advantages of sound signals in fault diagnosis are set forth through the analysis and research of literature.This section describes how many technological mature sound signal fault diagnoses is introduced to their principles and features.At the same time,the analysis of the noise signal of the drive gearbox is analyzed.Secondly,based on research purposes and research background,the experimental scheme is designed based on existing experimental equipment,and the acquisition of the acoustic signal is carried out,and the vibration signal is used synchronously to compare the fault diagnosis effect based on the acoustic signal transmission gearbox.In the experiment,the main acquisition analysis object includes the collection and analysis of the rolling bearing fault signal and the gearbox gear engagement fault signal.Thirdly,the pretreatment method of the noise reduction and feature extraction of the acoustic signal of the drive gearbox is studied.A signal noise reduction performance index system is created: through signal-to-noise ratio,signal-to-noise ratio gain evaluation,according to the result of the measured signal processing noise.And compare the empirical modal decomposition algorithm and the characteristic modal extraction effect of its optimization algorithm,and exist in an intrinsic modal component obtained by empirical modal decomposition algorithm(EMD)and set empirical modal decomposition algorithm(EEMD).Mixed "phenomenon,selecting a particular modal component in which the fault classification effect is poor,using an adaptive noise complete set Experience Modal Decomposing Algorithm(CEEMDAN)can effectively decompose time domain waveform data including a homogeneous scale feature band,There is no "modal aliasing" phenomenon,and the vibration signal is verified in the modal component diagram of the vibration signal,and the noise of the sound signal is realized,and the precise extraction of the target feature is realized.Finally,the study of the method of classifying the sound acoustic signal of the drive gearbox will use a one-dimensional residual neural network such a deep learning method to extract the fault acoustic signal to realize the fault classification of the drive gearbox.By comparative study,the transmission gearbox fault classifier for combined with wavelet packet decomposition and noise reduction and CEEMDAN feature modal extraction pretreatment and one-dimensional disability neural network fault identification are applied to the fault feature classification,and the transmission can be effectively transmitted through the sound signal.Gearbox faults are classified,and the sample classification accuracy is 95.1%.