Studies On Quantitative Diagnosis Of Vibration Faults Of Rotating Machinery On Distance Of The Information Entropy

This dissertation is based on the project provided by National Natural Science Fund(NNSF), the project name is: Research on the quantitative characteristics of fusion information entropy of signal induced by faulty vibration in large rotating machinery(50105504). Firstly, the status quo of vibration faults diagnosis of steam turbine-generator unit is demonstrated, and the domestic and overseas research and application situation are summarized. From this, the shortages of qualitative analysis and the necessary of quantitative analysis are induced. Then,the information fusion basic conception and singular spectrum entropy in time domain,power spectrum entropy in frequency domain,wavelet energy spectrum entropy and wavelet space feature entropy in time-frequency domain are introduced. Secondly,rotor model fault simulation experiments are described in detail. Structure of experiment shelf, monitoring system and experiment courses are described,too. After the analysis for a large number of experimental data, the range of four kinds of information entropy value of six kinds of rotor models faults are received. Regard each kind of entropy as one dimension, these four kinds of entropy make up a four-dimensional space, then four entropy value centers are received through counting average of the range of four kinds of information entropy value of each fault. One point is only confirmed in above-mentioned four-dimensional space, which is defined as a point for the fault entropy of the fault in four-dimensional space. Similarly, other faults entropy points in four-dimensional space can be gotten. On this basis, diagnose index in ration of integration is proposed, fusing quantity diagnose index, which is named distance of the information entropy. Finally, the deficiency of the diagnosis method on quantitative fusion at present are pointed out, that is just paying attention to fusing to the characteristic amount of each land of the signal under the single state, but neglecting the law of changing in the course of fault signal. This kind of diagnosis method will become the future research direction of quantitative fault diagnosing technology on fusion, which is through defining information exergy that can describe the law of changing in the course to study the diagnosis in the course of information fusion.