Research On Vibration Fault Diagnosis Of Hydrelectric Generating Set With Nonlinear Factors Considered

Hydroelectric generating set is one of the important devices in power system, the vibration problems of which not only affects the normal operation of the set, but influences the stability of the whole power system. Therefore, carrying out broad and deep researches on vibration fault diagnosis of hydroelectric generating set is of great significance and practical value. This paper intended to studying problems concerning hydroelectric units vibrant fault diagnosis. The following contents are discussed:With nonlinear factors taken into consideration, the characteristics of nonlinear vibration of hydroelectric generating set are studied, the impacts these nonlinear vibration characteristics has on the vibration signal are analyzed, the mechanism of the vibration fault is explored, the diagnostic scheme is put forward and a thought for the determination of the vibration fault diagnosis methods is provided.The extraction of hydroelectric generating set vibration fault features is based on the wavelet analysis method and rough set theory. Firstly, the energy feature information from the vibration signal is obtained and the feature set of vibration fault is constructed. Secondly, facing the excessive dimension of vibration fault feature set, rough set theory is applied to extract the fault characteristics so as to eliminate the redundant information and strengthen the validity of the fault feature set.LS-SVM is used based on the particle swarm optimization to establish a model in regard to vibration fault diagnosis of hydroelectric generating set to acquire accurate diagnosis and classification of the fault patterns, with nonlinear factors considered.After employing mix programming by VC++6.0and Matlab toolbox, effective diagnosis of the vibration fault can be achieved.Taking Francis turbine generator set as the research object, its process and results of vibration fault diagnosis are analyzed. The simulations reveal that the method of vibration fault diagnosis of hydroelectric generating set on the base of rough set theory and LS-SVM works well and is capable of making instant, accurate diagnoses on the vibration fault patterns.