Study On Faults Diagnosis Method Of A Rub-impact Rotor With Unbalanced Excitation

As the core component of many rotating machinery such as turbogenerator, aircraft engine, turbine, wind power generator, centrifugal compressor and so on, the rotor system plays a very important role. With the increasingly higher demand of high speed and high efficiency of the rotor system, the gap between the rotor and the stator is becoming smaller and smaller, and the frequency of rub-impact fault is increasing consistently. The rub-impact fault of the rotor system may affect the life of the equipment, and sometimes may even bring about catastrophic consequences to the related personnel. Therefore, it is urgent to monitor and analyze the rub-impact fault of the rotor timely and accurately.Mass unbalance is the most common factor to lead to rub-impact fault. Therefore, this paper takes the rotor rub-impact fault with unbalanced excitation as the object of study, mainly studies the denoising preprocessing method and fault feature extraction method of the fault vibration signal, and then puts it into practice and applies it in the fault diagnosis of the rub-impact fault experiment platform.For the uniformity, randomness and low SNR of the rotor rub-impact fault signal, an improved denoising algorithm of Singular Value Decomposion complexed with Wavelet Transform is proposed to reduce the impact of the selection of the the algorithm denoising order. At the same time, the advantages of the wavelet threshold denoising algorithm and the modulus maximum denoising algorithm are combined and become an improved algorithm. Then, the obove two methods are organically combined to form a complex method so that the noise interference and the signal to noise ratio of the signal are improved.According to the fact that rotor rub impact fault signal is nonstationary and nonlinear, Intrinsic Time-scale Decomposition, as a new local wave decomposition algorithm, is introduced to the feature extraction of fault signal. In view of the existing pseudo component and the end effect, the correlation coefficient is introduced to remove the pseudo component and an extension method based on feature points matching is proposed to restrain the end effect in this paper, which form an improved algorithm. Applying this method to the rotor rub impact fault diagnosis, the fault characteristics of rotor rub impact fault can be extracted accurately. By comparing with the algorithm, the improved method is proved to be simple and has no need to calculate the high order matrix, and the operation efficiency is high.A dynamic model of single-span single-disk rotor bearing system with local rub impact fault is established, and the simulation results are obtained, which give timedomain waveform chart, spectrum chart, orbit and Poincare map of fault signal under different speed conditions with certain unbalancd excitation. In addition, the experiment system of rotor rub-impact fault is built and the fault characteristics are extracted with the methods previously studied. With the comparison of the simulation results, the correctness of the model is verified, and at the same time, the effectiveness of the improved denoising algorithm of Singular Value Decomposion complexed with Wavelet Transform and the improved method of fault feature extraction proposed in this paper are verified.