FEATURE EXTRACTION FOR VIBRATION ANALYSIS OF CAVITATION IN KAPLAN WATER TURBINES

摘要

Several intelligent cavitation indicators obtained from vibration measurements have been compared in a Kaplan turbine. The indicators are based on the nonlinear scaling of features: one of the features is rms value and the other is either kurtosis or peak value. Indicators obtained from acceleration x~((2)) and higher derivatives x~((3)) and x~((4)) were tested by comparing the calculated indices with the sound of the recorded acceleration signals and analysing the signals with an oscilloscope in a wide power range. The results were compared in four frequency ranges with the knowledge-based cavitation index and previous studies. The indicators detect the normal operating conditions, which are free of cavitation, and also provide a clear indication of cavitation already at an early stage. The indices obtained from x~((4)) are the best alternative though also the index obtained from x~((3)) provides good results throughout the power range. Acceleration provided a good fit with the data but was less sensitive than higher derivatives. Automatic monitoring can be based on steps: detecting normal conditions, cavitation and the type of cavitation. The indicator also provides warnings of possible risk on short periods of cavitation. Uncertainties can be taken into account by extending the feature calculations and classification rules to fuzzy set systems.