Application Study On Wavelet Analysis And Fractal Method On Hydraulic Powerhouse Vibration Signal Processing

With the development trend of high water head and great capacity, executing antitype vibration test to the generator unit and workshop of great hydraulic powerhouse and processing the measured vibration signal is a significant work for the sake of steady operation.Former means included traditional amplitude-frequency statistical method, time-frequency analysis and correlation function analysis, etc. Despite the comprehensive application of the methods mentioned above together with the nearly perfect theory and experience, they can't describe the information taken by vibration signal entirely, the practices are not simple and visualized enough. How to expatiate on whether the generator unit and workshop of the hydraulic powerhouse operate steadily from a new point of view and diagnose the failure form shown by vibration signal become hot research problems at present.Based on the engineering fact of in-situ test on Lijiaxia hydraulic powerhouse, this paper introduces the fractal theory and wavelet analysis which are comprehensively applied in other domains. This can describe the essential attribute of the vibration more scientifically and effectively, and furthermore, it can indicate what kind of vibration is acceptable, under which condition the vibration is unusual which may bring failure and destruction.Wavelet coefficients obtained by CWT describe the self-resemblance feature of the vibration signal and validate the fractal characteristic. And then a concept named complexity measure is introduced for the purpose of comparing the original signal with the de-noised signal. It is discovered that the complexity measure reduces obviously after de-noising. It can be concluded from above that the noisy portion with chaos and fluctuation has been effectually filtered. The purpose of de-noising has been successfully achieved. Combining the theory of non-linear fractal geometry, this paper calculates the box dimension and correlation dimension of the vibration signal respectively. The author analyses the essential attribute of the vibration signal from the magnitude of the fractal dimension and its variation tendency with work condition. Consequently, the author points out some work conditions under which the generator unit and workshop structure can operate safely. Combining the fractal methods with wavelet analysis, this paper analyses the feature peak of the signal to gain some information which can't be gained by spectrum analysis. Finally, from the point of view of vibration source analysis and the frequency-spectrum analysis, failure...