Cavitation detection in a Kaplan turbine based on multifractal detrended fluctuation analysis of vibration signals

Abstract

Cavitation generally causes a decrease in efficiency and vibrations to a Kaplan turbine, but effectively identifying cavitation state is quite difficult. In this paper, a comprehensive test system consisting of a high-speed camera and a vibration testing system has been constructed based on a high-precision universal hydraulic machinery test stand. The vibration signals of the runner and the draft tube and cavitation images on the runner blade have been obtained for a Kaplan turbine under different cavitation states based on the system. In addition, the multifractal detrended fluctuation analysis (MF-DFA) method has been introduced to analyse the obtained vibration signals, in order to investigate the influence of runner blade cavitation on turbine vibration. Results show that the vibration signals of both the runner and the draft tube have multifractal characteristics. The Hurst exponent, the scaling exponent and the multifractal spectrum by MF-DFA are strongly dependent on turbine cavitation state. It is found that cavitation states including the incipient cavitation of the turbine could be effectively identified by using characteristic parameters extracted from the multifractal spectrum obtained by the MF-DFA of vibration signals. This work provides an effective method for cavitation identification of Kaplan turbines.