Identification Of Structural Excitation Source And Its Vibration Contribution Analysis

In the system of ship structure vibration especially excessive vibration seriouslyaffects the normal work of the system, threatens the safety of stealth, and damages to thestaff’s physical and mental health, therefore, it is particularly important to reduce structurevibration and the sound radiation through the certain method effectively. Structuralvibration excitation source identification and vibration contribution analysis find thereason resulting in structural vibration fundamentally, making the antivibration noisereduction work more pertinence and effectiveness. However, due to the complexity of shippower equipment structure and the installation of the equipment itself needs, and morethan one excitation source acting and influencing each other, characteristics of theexcitation source in most cases can not be obtained by direct measurement method, whichrequires the adoption of indirect measurement method for source identification analysis.The research situation of excitation source identification is reviewed, the basicprinciple of the principal component analysis (blind source identification) and the leastsquare method (not blind source identification) is stated and the related equationderived.In the study, a beam with elastic supports on both ends is chosen as the researchobject. The simulation and the experimental test are carried out and the validity of theproposed methods has been verified according to the estimation results. This study focueson the various factors’ effects on identification accuracy such as the differentsignal-to-noise ratio (SNR), the distribution of measured points, the number of measuredpoints,etc. the quantitative analysis of each principal component and the vibrationacceleration contribution of the exciting force in the measuring point position areanalyzed.In the light of some problems of the least the least square method a approachabout exciting frequencies judgement based on the segment similarity of the estimatedexciting source curves has been proposed. The analysis results indicated that with theapplication of the presented method, the error recognition rate was reduced effectively andthe recognition accuracy and effect were well improved.Two main tasks are completed inthis thesis as follows:(1) In the knowledge of the situation of the system output response, the principalcomponent analysis method is used for the blind exciting source identification in the simulation and experimental test. The study focuses on the rules about the identificationnumber of excitation sources,frequency spectrum characteristics,contribution rate andcumulative contribution rate of the principal component with different SNR,thedistribution of measured points and the number of measured points,which could provide areference for the blind excitation source identification with the principal componentanalysis method in application and its stable accurate source estimation results are veryimportant to improve the accuracy of the source estimation results with the least squaremethod.(2) In the knowledge of the situation of the system output response and systemfeatures, the least square method is used for the non-blind excitation source identificationin the simulation and experimental test.The vibration contribution rate in the measuredpoint is caculated for determing the main exciting force. This study focues on the variousfactors’ effects on identification accuracy with absolute error variance such as the SNR,the distribution of measured points, the number of measured points, the installationdeviation of accelerometers and the position offset of exciting forces, which could providea reference for the non-blind excitation source identification with the least square methodin application.(3) This paper proposed similarity decision method for traditional least squaremethod used amplitude for evaluation which had high difficulty, long time, high falsepositive rate and low accuracy rate for using the frequency of the largest amplitude asexcitation frequency. The similarity decision method could realize adaptive excitationfrequency decision with minimal supervision through writing related processing programfor reducing decision difficulty, saving time and improving accuracy based upon certainmathematical methods, comprehensive decision method which contained a majority ofexcitation frequency decision that based on level of similarity of estimation force curvesegmentation, and assist of amplitude decision.