Research On Planetary Gears Fault Diagnosis In Heavy Noise

As a transmission mode in mining machinery equipment, planetary gear transmission takes on an important mission in the whole production process. Planetary gear transmission has many advantages compared with ordinary gear transmission, such as small size, large carrying capacity and stable work. However, high power and high speed planetary gear transmission structure is more complex and requires high manufacturing precision.Due to the harsh field environment, the changing conditions and the nonlinear characteristics of the mechanical equipment, the vibration signals often have the characteristics of strong noise background, nonlinear and nonstationary. These features severely affect the accuracy of the device obtained status information. Therefore, this paper proposes an intelligent diagnosis method for planetary gear using local mean decomposition(LMD) and Discrete Hidden Markov Models(DHMM), including Local Mean Decomposition, the energy difference spectrum of singular value, multi-scale sample entropy(MSE), and the Discrete Hidden Markov Model.The local mean decomposition algorithm, a new time-frequency analysis method, is described. When the gear failure, vibration signals are usually accompanied by a surge signal and with varying degrees of modulation phenomenon. The demodulation information of fault vibration signal modulation can be gained by LMD. Compare with empirical mode decomposition(EMD) method, the similarities and differences are analyzed between the two. The advantages and deficiencies of LMD are analyzed, and then the endpoint-based extension technology of including support vector regression(SVR) extension and plus cosine window functions are used for end effect improvements.The basic principles of the method of singular value decomposition(SVD) is introduced. The energy difference spectrum of singular value is used to confirm the order of the reconstructed signal which is used to distinguish between the useful signal component and noise component. The integrity of the information is protected, and SVD inverse transform is used to reconstruct signal spectrum. The denoised and reconstructed signals are achieved. The signal noise reduction method is studied based on LMD and energy difference spectrum of singular value, and MSE is used to extract feature information.The basic theory of hidden markov models(HMM) is introduced. The multiple observation samples training problems and algorithm underflow phenomena are studied and improved. The application of DHMM theory in fault state identification is studied. The experimental data is collected under the strong noise background. The fault diagnosis algorithm based on LMD and DHMM is studied, including improved LMD, energy difference spectrum of singular value, MSE and DHMM, and the effectiveness of the proposed method is verified by different types of experimental signals.The intelligent fault diagnosis system for planetary gear based on LabVIEW and Matlab is developed. On line monitoring and data analysis of planetary gear vibration signal are realized. Data monitoring and diagnostic results of the storage, query and recording functions are completed. The actual signal of the planet gear is analyzed and the results show that the platform has friendly interface, convenient operation and good performance.At the end of this thesis, the summarizations of the research and expectation of the related technology development are presented.