Application And Research On Rotating Machinery Fault Diagnosis Basing On Two Dimensional Hidden Markov Model

The research theme of this dissertation is the 2D-HMM rotating machinery fault diagnosis, combined with national nature science fund project: "Application on faults diagnosis of rotating machinery in hidden Markov models" (No. 50075079). The basic algorithm and implementing technology for vibration signal filtering, feature extraction and fault classification basing on 2D-HMM are proposed by uniting theoretical analysis, computer simulation and experiment test and proved to be effective through simulation experiment and practical data.In the first chapter, the studying signification and the current situation of the rotating machine fault diagnosis are summarized. In the analysis of HMM for fault diagnosis, the new fault diagnosis strategy basing on 2D-HMM is given out. The research meaning, primary contents and innovations of this paper are illustrated too.The HMM's theory and algorithm are introduced shortly in the second chapter. Subsequently the model types topology structure, mainly parameters and primary arithmetic of the 2D-HMM are discussed in detail. The problem in the practical application of the model and the upswing measure for these problems are also set forwarded. Finally the comparison research of HMM and 2D-HMM from the model structure parameters description and arithmetic complexity is presented too.Starting with Kalman filter, the 2D-HMM filtering method is proposed in the third chapter and the operation principle and carryout technology are described detailedly too. It is proved to be validity through comparing 2D-HMM HMM and Kalman filtering method by using simulation data. By means of analyzing the advantaged and disadvantaged of 2D-HMM filter, the new filtering strategy combining 2D-HMM and wavelet is proposed and testified to be effective through the simulating and practical signals.In the chapter 4, Basing on analyzing 2D-HMM describing speech signal, the 2D-HMM feature extraction method is given out. Whereafter the operation principle specific algorithm and realizing technology are discussed in detail from three aspect of observation serial from parameters determinateness and spectrogram segmentation. Moreover the new method is tested by the simulating signal. Finally, basing on the analysis of the feature's strongpoint and shortcoming, the algorithm that fuses new featureand the traditional one in the frame of SOFM is proposed.In the short review of HMM fault diagnosis, the 2D-HMM fault diagnosis method and specific performing technology are put forward in the chap 5. The new method is proved to be effective by vibration signal from Bently rotor kits. Furthermore it is checked to be validly by the multi mode bearing vibration data in the practical application occasion.In the chapter 6, the developing environments and mixture programming implementation methods of MATLAB and C++ Builder are introduced briefly. Meanwhile the fault diagnosis software prototype and their essential buildup and main function block are given out.All work in this dissertation is summed up in the final chapter and the future research about 2D-HMM application is prospected.