Singular Analysis Of Reconstruction Of The Attractor, Fracal In Time-Frequency Domain And Its Application On Vibration Analysis Of Reciprocating Machines

In modern society, it is an important problem to state control and fault diagnosis of mechanical equipment and has being paid more and more attention to, for it has great economy and project value. In recent years, many scholars have made a lot of job on it, and put forword some new methods of fault diagnosis. Time-field analysis, frequency-field analysis and correlation function analysis in traditional, have been improved in application of fault diagnosis. And gray system, obscure subjection degree, expert system, wavelet analysis and neural network, etc. have been applied in running state of qualitative analysis system. But condition monitoring of reciprocating machines through the analysis of their vibrations has been recognized to be a difficult issue, essentially because of the strong nonlinearity of the vibration signals. Chaos and nonlinear dynamic theory is a new direction of fault diagnosis. Base on the analysis above, some methods in nonlinear dynamics and its engineering-oriented application is researched. The main content of the thesis is listed as follows:1. Time-Frequency analysis characteristics of the Local wave mothod which involved instantaneous frequency characteristics, Time-Frequency distributing characteristics, the mathematic expressions of IMFs and the adaptive frequency multiresolution analysis characteristics is analyzed and sumed.2. Some bassic theory of the nonlinear dynamics which involved Reconstruction theory, the choice of the time lag and the embedding dimension, the chaos in complex system and the the classic chaos system is introduced. The method based on Star. characteristic analysis of reconstruction of the attractor is put forward. 3. A improved decomposed method of local wave basedon reconstruction of the attractor SVD analysis is put forword. The method analyzes the process of denoising and phase choice. The emulational example and engineering example show that this method improves the decomposed ability of the local method and makes the feature extraction more meaningful and exact.4. A method of the IMFs reconstruct and singular entropy analysis of reconstruction of the attractor is put forword. The compare with the method in existence shows that the method is exact and practical.5. A new idea of fractal- Multifracals in Time-Frequency Domain is put forword. And this method is used in practical application. General dimension is calculated from vibration signals in the local wave time-frenquency domain by least square method. The multifracal method is applied to packing leakage fault diagnosis of reciprocating compressor. And sensitive demension of this fault is gained. The results shows that vibration signal multifracals in time domain is only sentive in early stage of packing leakage of reciprocating compressor., and the general demension of vibration signal multifracals in time-frenquency domain is increased with leakage degree. It can differentiate the fault degree clearly. The example proves that this integrated method is feasible.6. The key technology of condition monitoring of reciprocating machines is introduced, A the project of online monitoring of reciprocating Compressors is designed.