Vector Spectrum Of The Transient Signal Characterization And Fault Identification

The large-size rotary machines for modern industrial production, such as compressors,steam turbines,generators, are key devices in the major fields of national economy,which are industry of chemistry, petroleum, power and so on. If they were out of order, huge economic lost will be not only caused, but also the life safety of workers on the spot will be threatened badly. For the sake of controlling devices' running condition and reducing or even avoiding the happening of these devices' malfunction, the key units mentioned above are monitored online and diagnosed. It has brought much obvious economic and social benefits. The system for treating starting-stopping progress is one indispensable part of rotary machine's monitoring and diagnosis system,and it is very important too. However, the common analytical method used widely now on analyzing machines' starting-stopping progress is based on the data collected just from one channel on certain section of rotors, which is not complete enough and ignores the connection among the sensors and even much bad for accurate diagnosis. For changing this condition, the theory of full-information has been introduced. Based on the Vector Spectrum theory, The method on characterizing and identifying the instantaneous signal is quested and studied.Some issues on the starting-stopping progress and the methods for describing them and the full-information instantanueity technique are introduced briefly. The typical mechanics for fault and characteristic for vibration of the rotors is studied systematiclly. the feature of common instantaneity analysis and methods based on Full-spectrum theory and Holospectrum theory is introduced. All the work done above lay solid foundation for putting forward the instantaneity analytical method for fault identification based on the Vector-spectrum theory.The information-fusion theoretical basis of two channels on one section of a rator is studied. Combining with the fast algorithm based on vector-spectrum theory, all the parameter for instantaneity analysis is deduced,and two kinds of instantaneity methods based on the vector-spectrum theory are established. The One is Full Vector Bode(FVB),and the other is Full Vector Cascade(FVC).Then the feasibility is demonstrated not only by the specific machines' starting-stopping instances , but also by the mock experiment. As it shows: FVB and FVC is feasible for instantaneity analysis and can display accurately the critical speed of a rotor and the characteristic when passing the critical area. FVB and FVC can be used for starting-stopping analysis in place of common Bode figure and cascade figure.According to the fault statistic and combined with the experimental condition we have, the typical fault which happened on machines' starting-stopping progress is mocked, such as the unbalancing condtion ,the collision- friction condition between rotary part and still part, the oil whipping(or oil whirling), the rapping of rotors' bearing system. Then FVB and FVC is used to analyze them. Compared with common fault analitical methods,the result is that FVB and FVC is sensitive for some of fault. the characteristic of vibration is much more clear and the result of analysis is much more accurate.