Differential-based Empirical Mode Decomposition And Its Application In Multi-fault Diagnosis Of Rotor

Rotary machines are key equipments in electronic power, petrochemical, metallurgy, mechanics, aeronautics and some military industry. With the development of modern industry and technology, and the further improvement of the automation, rotary machines are changing to the direction of large-scale, high-speed, continuity, concentration, and automation. In manufacturing, some faults of the rotor components occur unavoidably. These faults are harmful to the equipments and even the whole manufacturing process. They would cause economic lose and some of they even hurt human beings. Therefore, to ensure the large-scale machines working safely and efficiently, many experts do a lot of researches in prediction and diagnosis of faults in the rotor.In recent decades, many experts from different countries are doing researches on faults diagnosis technology about rotary machines. Among them, Empirical Mode Decomposition method is viewed as a great progress in nonlinear and non-stationary signal processing. On the basis of previous accomplishment, in this paper, Differential-based Empirical Mode Decomposition (DEMD) method, which is developed from EMD method, was put into practice in the identification of the faults signals of rotor. This effort will make full use of the ability of self-adaptive and time domain analysis of Hilbert-Huang Transform.This paper is concerned with several aspects as follows:1. This paper summarized the algorithm of Empirical Mode Decomposition (EMD), declaimed the process of signal processing based on Ensemble Empirical Mode Decomposition (EEMD), and briefly introduced the characteristics of Ensemble Empirical Mode Decomposition method. This paper verified the high frequency resolution and efficiency in faults identification of Differentia1l-based Empirical Mode Decomposition (DEMD) method by simulation. In the same way, this paper also shows the applicability of EEMD method in dealing with pulse signal.2. This paper summarized the vibration types and principles of crack, rub-impact and pedestal looseness,improved the mathematical models of the faults mentioned above, and verified the applicability of models by numerical simulation.3. This paper put DEMD method into identification of the vibration signals of single faults and coupling-faults mentioned above. By analyzing the signals from numerical modeling and actual faults signals and comparing with the results of EMD methods, this paper validated the applicability and efficiency of DMED method in the field of rotor fault diagnosis.In the field of rotor fault diagnosis, Hilbert-Huang Transform, which serves as a new signal processing tool, has gain great favor because of its advantage in signal detect and faults identification. However, there are still more researches need to do on the improvement of the method and combining it with other methods.