Proceedings of the CSEE · 2006 · 10 citations · 0 references
ComputingComputer ScienceMachine LearningTime Series AnalysisVibrationClassificationDiagnosisDiagnosticsFault DetectionFault DiagnosisSupport Vector MachinesBinary ClassificationFrequency Domain ProcessingMachine Learning-based Fault DiagnosisOnline Fault DetectionSupport Vector ClassificationSupport Vector Machine ClassificationWavelet AnalysisWavelet Packet Transform
A new method of vibrant fault diagnosis was proposed for hydro-turbine generating unit based on wavelet packet analysis and support vector machine(SVM). A multi-class classifier was developed based on binary classification. Collecting the characteristics of this signal in frequency domain, and then using them as learning samples to train the constructed classifier so as to realize the mapping relationship between the fault and the spectrum characteristic, this method can be used for diagnosis of the unit faults efficiently. The simulation experiment shows that the proposed method has good classification ability and robust performances. This method is suitable for vibration multi-fault online diagnosis of hydro-turbine generating unit. A new way is provided in intelligence diagnosis of vibration fault of hydro-turbine generating unit.