Method Research On Fault Diagnosis Of Interturn Short Circuit In Rotor Windings And Rotor Supporting Axles Of Generator

Along with our country's electric power industry enters into the new level of big power network and big unit, the synchronous generator acts as the heart of the power network. Its operational reliability directly affects the power network's safety running and the development of national economy. Therefore, study of generator's fault diagnosis method has important practical significance.Two kinds of typical faults of turbo generator are selected to research, which are the interturn short circuit in rotor windings and the mechanical vibration faults of rotor supporting axles. The main contents are included as follows:According to the thought of complex system's task part-fusion, the synthesis diagnosis system structure of the turbo generator is designed. And three typical characteristic extraction method of fault symptom information are studied. Simultaneously the synthesis diagnosis method is researched, which is based on the homologous multi-characteristic information and the different kinds of multiple source information.To deal with the fault of interturn short circuit in rotor windings, wavelet analysis theory is applied to extract the singularity characteristics of the induced electromotive force on the detecting coil, which is fixed in the air gap between stator and rotor. Then according to these singularity characteristics, the detection of interturn short circuit fault and the accurate orientation of the fault trough are realized. The simulation indicates that the method of wavelet analysis not only can detect the fault but also can locate accurately the trough position of the fault.To deal with the mechanical vibration faults of rotor supporting axles, when the relationship between faults and symptoms is understood well enough and the standard information group is known, a diagnosis method based on the fuzzy multi-layer theroy is adopted after analyzing the limitation of the diagnostic method of the fuzzy general judgement. The instance's contrastive analysis indicates that this method can carry on the diagnosis effectively to either the single fault or multi-fault of the turbo generator, and obviously surpasses the diagnostic method of the fuzzy general judgement.To deal with the mechanical vibration faults of rotor supporting axles, when the standard information group is lacked, a method of clustering analyze based on the fuzzy equivalent matrix and a method of SOM neural network are studied. But the cluster process of this cluster analytic method is extremely tedious when the known fault patterns are quite many. At the same time, the SOM neural network method can't be easy to make the correct judgement when the multi-fault is happened. Therefore, a diagnosis method combined the fuzzy clustering analysis and the SOM neural network is put forward in this article. The example analysis indicates that the method can carry on the diagnosis not only to the single fault but also to the multi-faults.Finally the research work is summarized, and a perspective on the future research in this domain is pointed out.