Research On Mechanical Fault Diagnosis Method Based On Probability Amplitude Demodulation

This paper is under the assistance of the National Natural Science Foundation of China(51261024,51075372,51265039,50775208), Key state Laboratory of mechanical transmission Open Fund(No. SKLMT-KFKT-201514), Jiangxi Provincial Department of Education Science and technology project(No. GJJ12405) and, Key Laboratory of the digital signal and image processing Technology of Guangdong Province(No. 2014GDDSIPL-01). Probability amplitude demodulation algorithm has been introduced into the mechanical fault diagnosis based on its unique advantages, and a mechanical fault diagnosis method which is based on probability amplitude demodulation is proposed. Simulation and experimental studies have been carried out, and very good results have been achieved. The main contents of this paper include the following aspects:1. Discuss the necessity and significance of the research that probability amplitude demodulation is introduced into mechanical fault diagnosis, and discusses the current research state of demodulation method in mechanical fault diagnosis at home and abroad, the basic idea of the probability amplitude demodulation is discussed in detail, and on this basis, the research contents and main innovative points of this paper are presented.2. The thesis discusses the principle and algorithm of simple probability amplitude demodulation, the method using probability theory, treat demodulation as an inference problem, using the prior knowledge reasoning the problem of uncertainty, so as to achieve the purpose of reasoning and demodulation. On this basis, a mechanical fault diagnosis method based on probability amplitude demodulation is proposed, and the simulation and experimental research are carried out. The simulation results show that the proposed method has a good processing ability to the amplitude modulated signal and can effectively decompose the modulation frequency. Compared with the traditional Hilbert envelope demodulation method, proposed method has obvious advantages. The envelope demodulation by the proposed method and the original signal are well fitted, which is obviously superior to the traditional Hilbert envelope demodulation method. Finally, the proposed method is applied to the fault diagnosis of rolling bearing. The experimental results show that the proposed method is very effective and can effectively extract the fault characteristics of rolling bearing.3. This paper discusses the principle and algorithm of the amplitude demodulation of Gauss process, which is an extension of the simple probability amplitude demodulation method. First, the method is able to recover the missing parts of the modulation signal, using the maximum a posteriori probability inference, and using the fast Fu Liye transform to improve the computation speed. In addition, Gaussian probability amplitude demodulation combined with the fast Fourier transform, the maximum inspection probability and Lanczos algorithm, infered The Laplasse approximation error line which is on the maximum a posteriori probability of the original envelope, and efficiently compute the maximum posterior probability. Finally, the method uses approximate maximum likelihood to obtain all the free parameters in the model. The simulation results show that the proposed method has good performance on the modulation and amplitude signal, can effectively decompose the frequency modulation and amplitude modulation signal. Finally, the proposed method is applied to the fault diagnosis of rolling bearing. The experimental results show that the proposed method is very effective, can effectively extract the fault features of rolling bearing.4. The principle and algorithm of cascade demodulation is discusse. the method takes simple probability amplitude demodulation as frame, and get the original sub envelope using Gauss probability amplitude demodulation in each stage. Mechanical fault signal is often in multiple time scale demodulation. Simple probability amplitude demodulation and Gaussian probability amplitude can only demodulate demodulation signal containing a time scale, and cascade demodulation can demodulate machimnery fault signal contains multiple time scales. Therefore, this method is more practical, adapt to more widely. This method can make estimation of all variables, obtain multiple time scale of the demodulation and cut the top envelope unnecessary. Different from other probability amplitude demodulation method, cascade demodulation decomposition results are a number of different time scale of the envelope and a rapidly changing carrier, wherein the change of carrier is the fastest in all decomposition. The simulation results show that the proposed method has good processing ability to the sound signal which contain multiple time scale, can decompose a number of different time scales of envelope and a carrier, and cut the top of no significance sub envelope. Finally, the proposed method is applied to the fault diagnosis of rolling bearing. The experimental results show that the proposed method is very effective and can effectively extract the fault characteristics of rolling bearing.This paper discusses the principle and algorithm of mutual frequency probability amplitude demodulation, which is an extension of the simple probability amplitude demodulation. In the existing probability amplitude demodulation model, each filter demodulation is independent of each other, and modulation of machinery fault signal usually associated with different frequency channels, the method using dependencies between different frequency channels, between the different sub bands of demodulation simulate the interdependent relationship, to establish a model to the envelope of the different frequency channels. This model can reduce the dimensionality, avoid over fitting and calculation is more efficient. Simulation results show that the proposed method has good processing ability to the sound signal. Different channels of time-domain and frequency-domain signal are related, each channel of the envelope and carrier can well reflect the characteristics of the sub channel signals. Finally, the proposed method is applied to the fault diagnosis of rolling bearing. The experimental results show that the proposed method is very effective and can effectively extract the fault characteristics of rolling bearing.