Research On Order Analysis Of Rotating Machinery Based On Adaptive Frequency Tracking Filtering

Rotating machinery is widely used in manufacturing, aerospace, petrochemical,metallurgy, electric power and other industries, in order to guarantee its long-term andstable operation, avoid incidental or sudden failure led to huge losses and catastrophicaccidents, the vibration analysis for its operating status is common used and the mosteffective means. Conventional methods in the analysis of stationary signals increasinglyimproved, but is still a lack of comprehensive analysis on the process of typicalnon-stationary signals during acceleration and deceleration, also difficult to adapt to thedevelopment of high-speed, high-precision and automation rotating machinery. Orderanalysis is the better way to solve this problem, which convert the non-stationary signalto be stable through the time-angular domain transformation.This paper systematically studied the key issues of computed order tracking,revised the method of the speed pulse arrival time. A calculation of precise speed curveis achieved through the least square five points-three times method based on the initialspeed which is computed with the first-order numerical differentiation. Theimplemented method with a small amount of calculation can effectively improve thespeed accuracy to meet demand subsequent analysis. Angular-domain resampling,pre-processing and order spectral analysis have been studied.With a comparison between the commonly used order component extractionmethod, the Vold-Kalman adaptive filtering method based on the angular displacementis studied and revised. By constructing the signal model based on the structural and dataequations, the filtering arithmetic is modified with higher computational efficiency,which can extract and separate multi-order components of complex rotating machineryin the time-domain directly, and decoupling the close and crossing orders at the sametime. Proposed a method with the combination of EEMD and VKF, which suppress theinterference of each order component, offer the ability to separate the rolling bearingfault order components and the corresponding modulation components.Rotating machinery bode diagram is implemented based on zero-phase distortionfrequency tracking band-pass filtering. To restrain the boundary effct, data segmentationfiltering with overlap is proposed, the average rotating frequency as the centerfrequency of each segment. The·boder diagram will be realized with accurate amplitudeand phase, which are obtained from the filtered1st order component. On the basis of theoretical studies, designed a set of rotating machinery-orienteduniversal data acquisition and order analysis system based on VC6.0developmentplatform, and the hardware platform for test is builded, vibration and pulse signal duringoperation will be sampled with multi-sampling modes precisely. In the order analysismodule, the following functions are realized: the speed curve, the angular domainresampling, order spectrums, three-dimensional spectrum map, order extraction methodbased on Gabor transform and Vold-Kalman tracking filter, Bode diagram, etc. Actualtest verifies that this system is practicality and effectiveness.