Abstract:

This paper is a summary of the research projects on vibration specification, signal analysis and vibration diagnosis of the engine with the object of discussing the prospects of vi bration diagnosing.This paper analyses the mechanisms of three vibration analysis methods.the normal spectrum nalysis, time series analysis and modal parameter recognition mathematically and physically, and reviews the prospects of these methods used for vibration diagnosing of mechanical faults of the rotating machinery.This paper reviews also the various global identification methods:diagnosing by discrimination Parameters, fuzzy logic and others.Finally, the following tentative conclusions can be drawn:The time series analysis, as a tool to recognize the "forming filter" ,is similar to the norms spectrum analysis method but it is in the time domain, therefore there are another kinds of methods to increase the signal noise ratio.People can use the fuzzy logic to normalize the discrete characteristic frequency component of a spectrum or the parameters of a time series model and create suitable normalized crit eria wit both physical and mathematical meaning to diagnose the mechnical condition of a rotating machine.

Key words: vibration, diagnostics, rotating machinery