Measurement and Signal Processing of Incipient Cavitation in Vortex Zone of Francis Turbine

Abstract

In order to reduce the influence of cavitation on the efficiency of the turbine, it is of great significance to extract the cavitation pressure signal. Generally, the cavitation inception coefficient is determined by manually using the vibration characteristic curve. However, the vibration characteristic curve can only be obtained when the cavitation is serious to a certain extent. Therefore, measures should be taken to extract the incipient cavitation signal. The bubbles generated by the cavitation will have a big impact at the moment of burst. The impact is impulsive, but it is usually covered by other water flow noise. In this paper, a method for extracting the cavitation pressure pulsation signal based on short-time singular value decomposition (ST-SVD) is proposed. In the proposed method, the short-time slip vector sequence is firstly acquired by circulating interception. Then, by applying SVD to the sequence, the singular value diagonal matrix can be obtained. Based on it, a feature extraction method based on main singular ratio is proposed, which can be used as an effective index to reflect the impulsive characteristics of the reconstructed sequences. To verify the performance of the proposed method for extracting the cavitation pressure signal, the cavitation test experiment on a turbine was performed and the results showed the proposed method can successfully reduce the noise and extract the cavitation signal.