The Improved EEMD Analysis Method And Its Application To Fault Diagnosis

Due to the increasement of complexity,the equipment condition monitoring and fault diagnosis method research are of great significance.This paper mainly focuses on the improvement research and application of ensemble empirical mode decomposition.This paper proposes the diagnosis method which based on the improved EEMD,kurtosis theory and energy operator demodulation.Firstly,the paper studies the demodulation principles of EMD and EEMD.EEMD is a basement function self-adapted method and has a better filtering characteristics and adaptability in the resolution compared with the traditional time domain and frequency domain analysis methods.But in practical application,due to the particularities of the fault signals and strong background noise,EEMD method appears some imperfections,such as mode mixing and end effect.Mode decomposition is a newly proposed method which needs to be further improved.Secondly,the paper researches the modified measures for the decomposition imperfections such as mode mixing,end effect and stopping criteria.The paper proposes a high frequency harmonic auxiliary method,and gives the method of choosing coefficient of amplitude and high frequency parameters.And a data prediction and error compensation method is proposed based on BP neural network to guarantee the integrity of valid data and restrain the end effect.A new iterative stopping criteria is studied based on the cross correlation coefficient,and effectively reduces the calculation time and redundant results.Then the effectiveness of these improvements is verified by experimental simulation signals.Thirdly,the paper proposes the fault diagnosis method based on the modified EEMD,kurtosis theory and energy operator demodulation.Then applied this method to the extraction and recognition of weak fault features in rolling bearing.The energy of fault signal is very weak,so the kurtosis value which is sensitive to the wave peak points is used to select the significant signals,and the cross correlation function is used to avoid the influence of noise.In order to get accurate fault signal information,the fast-kurtograme helps to get the best band filter parameters.Energy operator demodulation has better accuracy and smaller envelope edge flying wing,makes a great effect to obtain accurate fault frequency.The application in weak faults diagnosis of rolling bearing has verified the effectiveness of the proposed method.Lastly,on the basis of theoretical research,an EEMD diagnosis module is developed by Visual C++and Matlab,and has been integrated in the fault diagnosis system.Its practicability and effectiveness has been tested through the actual examples.