LAUSR

Francis turbine is a core energy conversion part and low-frequency vibrations caused by unsteady vortex rope in the draft tube may damage the unit through resonance. Therefore, a more comprehensive assessment and monitoring of the status of Francis turbine is of great significance for unit stability. In this hydraulic machinery, mechanical impulses and hydraulic excitation are coexisting. The former manifests as a narrow low-frequency band with high kurtosis, while the latter represents a wide high-frequency band with energy associated with the vortex rope. Based on that, a novel method coined Convkurgram is proposed to suppress the extreme high kurtosis mechanical components and extract the hydraulic carrier wave efficiently. Besides, it takes multiple information frequency bands into consideration in case of contamination waves exist. Finally, the method is evaluated by means of actual vibration signals collected from the hydroelectric generating unit under normal and cavitation conditions. The results show that the proposed method recognizes the characteristic frequency bands of mechanical components and hydraulic carriers as well as extracts the shaft frequency and blade passing frequency successfully.