Research On Processing Method Of Vibration Signal Of Oil-Film Pivot Tilting-pad Bearing

The mechanical vibration of turbine rotor system is the main factor that affects the performance of underwater vehicle.The mechanical vibration of turbine rotor system when working may propagate to the outside through the basis and the shell to generate underwater noise.Oil-film pivot tilting-pad bearing has excellent vibration reduction characteristic for its double layers oil-film structure,which receives much attention in such fields as underwater vehicle which are sensitive to vibration noise.The study object of this thesis is oil-film pivot tilting-pad bearing,the starting point is the application of oil-film pivot tilting-pad bearing on underwater vehicle,the emphasis is the vibration signal processing method of oil-film pivot tilting-pad bearing,the purpose is to eliminate power-frequency interference and random noise.The main research contents include:? Carry out oil-film pivot tilting-pad bearing experiment,summarise the powerfrequency interference characteristics of experiment signals.? Analyse the characteristics of classical singular value decomposition method applied to power-frequency interference analysis of experiment signal,study singular value decomposition theory as the influence factors of singular values,the method to determine the number of effective singular values,the characteristics of singular vectors under difference Hankel matrix structures.These theory studies build the foundations for the application of singular value decomposition method to experiment signals power-frequency interference elimination.? Research on the power-frequency interference elimination algorithm based on singular vector spectrum estimation.The algorithm tightly combines singular values and singular vectors.First,the method optimizing difference spectral peek of singular values is adopted to choose the number of effective singular values to smooth away random noise.Second,the method of singular vectors spectrum estimation is used to extract power-frequency components from the denoised signal.The proposed algorithm can avoid of the demand for accurately estimating power frequency parameters,including amplitude,frequency and initial phase.In this way,it is reduced for the difficulty of power-frequency interference elimination of experiment signal and computing complexity.Put forward the judgement method for powerfrequency elimination of experiment signal.To verify the effectiveness of this proposed algorithm: Perform Hilbert transformation on the extracted power-frequency components.Compute internal products between the power-frequency components before HT and after HT seperately and experiment signals.Compare the computating results by the properties of correlation function,the former result is bigger for their strong relevance relationship and the later is less because they are uncorrelated.Apply Prony algorithm to solve the precision values of power-frequency paremeters,including amplitude,frequency and initial phase;Adopt the judgement method for power-frequency elimination to further verify the accuracy of powerfrequency paremeters.This proposed algorithm has obvious advantages over regular methods,such as filter,Flourier transformation,wavelet transformation and sparse decomposition.? Base on the proposed singular vector spectrum estimation algorithm,the SVD spectrum concept is proposed which is from the combination of singular values and singular vectors,which is convenient for the application of singular value decomposition method in feature extraction or noise deduction.? To apply the classical singular value decomposition method to power-frequency analysis,iteration algorithm of singular value decomposition is proposed.The singular value decomposition processing is iterated to extract the main components from experiment signals,gradually filtering out the power-frequency components by the spectrum features of the main components.To improve the computing efficiency of the algorithm,a new algorithm combining the classical singular value decomposition method with stationary wavelet transformation is proposed,in this way the number of iterations of singular value decomposition process reduces from 5 to 2,computational efficiency being improved.? Extract and contrast the power-frequency paremeters under all the experiment conditions with the universal regularities during the whole experiment process are drawn as that: The frequency and amplitude of power-frequency fundamental are within 48.75 ~ 48.95 Hz and 0.002 ~ 0.007 mm respectively;The frequency and amplitude of third harmonic are within 146.25 ~ 146.65 Hz and 0.001 ~ 0.005 mm respectively;The frequency of third harmonic is three times as that of the fundamental,the multiples varying within 2.99 ~ 3.01;For the same signal,the initial phase of fundamental component and that of third harmonic are different,the initial phase of fundamental component(or third harmonic)are different too.These universal regularities about power-frequency interference are conducive for deeply understanding the characteristics of experiment signals,for fully eliminating power-frequency interference,for the application of oil-film pivot tilting-pad bearing on underwater vehicle.