Research On Virtual Characteristics Analyzer System Of Rotating Machinery Based On Time-Frequency Analysis

Rotating machinery takes up the major part in all machinery equipments and plays an important role in industry enterprises. It is very important to ensure the running safety of rotating machinery for both enterprises and our society. Characteristics analysis plays an important role in faults find and diagnosis of rotating machinery. Vibration measurement and test is a popular method in all characteristics analysis methods. The level for getting information depends on the analysis functions associated with the measurement instruments. Up to now, most measurement instruments often used are very good to analysis stationary vibration of rotating machinery, but is lacking in non-stationary vibration characteristics analysis to some extend, for example, the vibration caused by rotating machinery during run-up or coast down. This means the non-stationary vibration analysis functions current used in measurement instruments are not satisfied the requirement by the advanced development of rotating machinery which with high speed and high precision. So, it is urgent to continue perfect the techniques current used and add new analysis functions.In this dissertation, time-frequency analysis methods were introduced in the characteristics analysis of rotating machinery. While a good study to classic faults characteristics of rotating machinery, traditional methods for characteristics analysis and the time-frequency analysis for non-stationary signals, an important research were carried on order analysis, a very powerful tool for characteristics analysis of rotating machinery. Some new methods to perform order analysis were proposed and successful done based on the relationship between the instantaneous frequency of vibration signal and the rotating speed of machinery.Order tracking of rotating machinery based on instantaneous frequency estimation was proposed in our research. In this method, the speed corresponding instantaneous frequency estimation was attained from multi-components vibration signals by peak searching on time-frequency plane, curve fitting was employed to obtain the instantaneous frequency law, and order tracking sampling was done for order analysis. By this method, some special hardware, such as speed counter, keyphasor device, etc. are avoid, and the corresponding installation limited required by traditional order tracking methods is not exist too. It makes the order analysis simplified and can be performed at a low cost. By deep study and advanced research on traditional order tracking filtering method, a new order tracking filtering method based on instantaneous frequency estimation and zero-phase distortion digital filtering was proposed. It overcomes the problems of hardware requirement for speed counting and analog tracking filter to do tracking filtering in past time. It is an effective adding to traditional order tracking filtering methods.Order component extraction based on time-frequency filtering was also researched. Gabor transform, a popular time-frequency analysis method, is employed to do the order component extraction by signal synthesis. So, the order component selected can by treat alone by further analysis. The tacholess order analysis, such as order spectrum, order matrix and tracking order spectrum, order component extraction etc., was programmed based on these mentioned above. All of these functions provided a good foundation for the developing of virtual characteristics analyzer of rotating machinery.A virtual instrument, rotating machinery characteristics analyzer, was programmed with some powerful time-frequency analysis functions and all order analysis functions introduced above. So, it is enabled the abilities to analysis non-stationary vibration of rotating machinery and eliminate the weak points of current used measurement and test instruments. The advantage of instruments developing by the new virtual instruments developing concept is also well showed by it.Many actual tests for different test objects were done with the virtual characteristics analyzer...