Research On Denoising Of Hydropower Unit Vibration Signal Based On Quadratic Decomposition And Approximate Entropy

As the core equipment for hydraulic and electric energy conversion,the operational stability of hydropower unit is guaranteed to be critical.Once the stable operating environment of hydropower units is destroyed,the quality of power transmission and the frequency of the grid will be affected,causing unnecessary economic losses and even threatening personal safety.The unit fault or accident information is mostly reflected by the hydropower unit vibration signal.Based on the processing and analysis of the vibration signal,the operating condition of the unit can be mastered in real time,which is of great significance to the operational stability of the unit is improved.As the structure of hydropower units becomes more complex,the mechanical and electromagnetic excitation coupling between internal structures becomes more intense.At the same time,with the influence of hydraulic excitation,the non-stationary and nonlinear characteristics of vibration signals are continuously enhanced,the vibration characteristic band is annihilated in a large amount of background noise.In order to identify the true operating state of the unit and accurately extract the feature quantity in the signal,the vibration signal must be denoised.How to improve the noise reduction performance of the hydropower unit vibration signal and optimize the noise reduction effect are studied in this paper.Firstly,the existing signal denoise techniques are summarized,and then the hydroelectric vibration signal characteristics and their commonly used analytical processing methods are discussed in this paper.Secondly,for the non-stationary and nonlinear vibration signals of hydropower units,the basic principles and specific implementation processes of two signal processing methods based on modal decomposition,EMD and VMD,are introduced.A vibration signal denoising method based on modal decomposition and autocorrelation analysis is applied.The target signal is decomposed into a series of IMFs with different frequency components by EMD and VMD.The effective IMF components are selected according to autocorrelation analysis and aggregated into the de-noised vibration signal.Both EMD and VMD are suitable for signal denoising processing are evidenced by simulated noise reduction tests.Endpoint effects and modal aliasing problems in EMD are overcame by VMD,the waveform distortion phenomenon after the vibration signal denoising is suppressed to some extent.Thirdly,in order to ensure the reliability and stability of the denoising effect,the approximate entropy theory is introduced.According to the approximate entropy characteristics,it is used as an important index to filter the effective IMF component,and the basic criteria for filtering effective IMF components are established.The approximate entropy of a series of IMFs components obtained by modal decomposition of the target signal is calculated.By comparing with the given approximate entropy threshold,the effective IMF component is filtered for signal reconstruction to achieve the purpose of noise reduction.The effective modal component screening criterion based on the approximate entropy theory is proved to be effective and feasible by the simulated denoising test,it can accurately and efficiently discriminate the noise signal component and the vibration signal component,which is suitable for signal denoising analysis;the better noise reduction effect is obtained by the vibration signal denoising method based on modal decomposition and approximate entropy.Finally,in order to extract the characteristic frequency bands in the signal more accurately and optimize the denoising effect,combined with the advantages of EMD and VMD in signal processing,a method of denoising the hydropower unit vibration signal based on quadratic decomposition and approximate entropy is proposed.EMD is the primary decomposition of vibration signals,the obvious strong high-frequency components in the IMF component are eliminated,most of the background noise is filtered out,and the retained IMF component is accumulated to obtain a denoised signal,and the low-band signal is enhanced.VMD is the second decomposition of the signal after primary denoising,and the effective IMF component is filtered by the approximate entropy to reconstruct the signal to obtain the final denoised vibration signal.The simulation comparison experiment and engineering example analysis show that the proposed method has better denoising effect and superior denoising performance.The number of tests for decomposing the scale K value in the VMD algorithm is reduced,and the K value determination process is simplified.The feature extraction ability of VMD for periodic signal is improved.The proposed method can provide a basis for hydropower unit condition maintenance and health management.