Research On Fault Diagnosis Of Rotating Machinery With Variable Speed Based On Order Analysis

Wind turbines, aircraft engines, high-speed rails and other large equipment often work under variable conditions, the key parts of which, such as bearings, gears, rotor system and other rotating parts, are fault-prone. in these kinds of system. Once the failure appears, it may affect operation of the equipment, and even lead to catastrophic accidents. Therefore, the condition monitoring and fault diagnosis of bearing, gear, rotor and other key parts are necessary. Vibration diagnosis is the most common and the most effective fault diagnosis method. But the traditional diagnosis method is mainly aimed at the stationary signal, which cannot be applied to the rotating machinery fault diagnosis under variable speed. The order analysis can well reflect the vibration related to speed. But there are still some problems concerned the fault diagnosis based on order analysis, such as: the accurate calculation of the speed curve, low fault identification accuracy caused by that the characteristics cannot effectively characterize the rotating machinery under variable speed, extensive error of the holographic waterfall caused by variable speed of the rotating machinery. Aiming at solving these problems, this paper concentrates on fault diagnosis of rotating machinery based on order analysis. The main research contents of the paper are as follows:Study on the accurate calculation of the speed curve. The speed curve calculation method based on first order digital differential and least square five point three smoothing is studied in this paper with tachometer. Firstly, Lagrange interpolation is used to obtain the precise moment of the pulse threshold value, and then the initial speed curve is obtained by first order digital differential. At last, the accurate rotational speed curve is achieved by the least square method five point three times. Aiming at solving the problem of low precision and poor noise immunity without tachometer, the speed curve calculation method based on time frequency fusion is proposed. At the beginning, vibration signal will be transformed into time-frequency domain through short-time Fourier transform and Wigner Ville distribution, then the two time-frequency matrix would be normalized before fusion, at last the speed curve would be acquired by the cost function.Aiming at solving the problem of low fault identification accuracy due to that the characteristics cannot effectively characterize the rotating machinery under variable speed, the method of fault diagnosis based on order resampling and manifold learning is put forward. Study on the calculation of resampling time, angle domain resampling and other the key technology of the order resampling process. By taking advantage of the order sampling technique, the non-stationary time interval signal is transformed into stationary signals. Study on the extraction of effective characteristics of variable speed rotating machinery. High-dimensional features which are constructed from the dimensionless time domain characteristics of the angle domain signal and IMF components after the EMD decomposition. Apply adaptive neighborhood local tangent space alignment manifold learning to reduce high dimensional statistical features, and the fault is going to beginning identified by the k-nearest neighbor classifier.In order to settle the problem of gross error of the holographic waterfall caused by variable speed of the rotating machinery, the fault diagnosis method based on the fusion of the order specific component is proposed. The Vold-Kalman order analysis based on angular displacement is studied, and the extraction of multiple-order component is realized by using the state equation and the observation equation as the basis model of the vibration signal. By taking Vold-Kalman order analysis, the time domain waveform of the required order specific component is extracted, and the amplitude and phase information of the X?Y channel key phase is obtained. Then the holographic parameters by which the holographic waterfall can make effective diagnosis, are calculated by the amplitude and phase information of each order.A set of rotating machinery fault diagnosis system based on order analysis is developed, the system is composed of data acquisition module?order analysis module? fault diagnosis module based on order resampling and holographic spectrum module. Each module is verified through the application.