Research Of De-correlation And Indetification Of Chief Excitataion Sources Among The Vibration Signals In Vehicleon

With the tremendous improvement of people’s life, consumers pay more attention to the NVH(Noise Vibration and Harshness) issues of vehicle. As a complex mechanical structure, there aremultiple excitation sources in the car. The signals collected in the location of the main excitationsources are the response signals of these points, i.e. every excitation cause the vibrations or noisetogether. While the correlation among these signals has not been taken into consideration during theidentification of the chief excitation sources, which will affect the result of the identification in acertain extent. Therefore, the analysis of the main excitation sources based on the de-correlation of theobtained excitation source signals has important theoretical significance and values of the vibrationand noise projects.To deal with the correlation of obtained vibration signals, this paper conducted a study ofde-correlation methods and the way to identify the chief vibration sources. The main content of thepaper are summarized as follows:(1) The de-correlation methods of signals have been reviewed as well as the vibration excitationsources identification ways. The existing problems and shortcomings have been pointed out.(2) The basic concepts of correlation and spectral analysis have been introduced systematically.Then the basic problems of the energy source identification have been expounded as well as theclassification of the correlation among the vibration sources. A research of the importance of thede-correlation has been conducted as well.(3) The idea of clustering analysis has been introduced to determine the number of the majorexcitation sources based on the whitening in the theory of blind separation. Then this paper proposeda de-correlation method named average delay whitening de-correlation methods. In this method, theaverage based on the theory of joint diagonalization has been made among several correlationmatrixes in different non-zero delays, and then the matrix to de-correlate the signals which isorthogonal has been extracted to achieve the separation of correlated signals. In this paper, a detailedanalysis process has been obtained after elaborating the basic theory of this method. The analysis ofthe simulated signals and the really vehicle vibration signals shows that this method has satisfiedseparation result in complex correlated vibration source signals.(4) The reason of the correlation among the signals has been analyzed systematically and thecomposition of the relationship has been given after taking the transfer function into consideration. Then another de-correlation method has been raised, which is named the transfer functionde-correlation method. The transfer function matrix has been obtained by conducting the modalanalysis at the location of every source in the vibration system, the separation complete by theapplication of this matrix. The theoretical analysis shows that this method can recover the amplitudeof the signal. After the processing of the simulated signals and tested signals, the usefulness of themethod has been validated.(5) the hierarchy analytical method has been used to indentify the main sources of the system onthe basis of the de-correlation of the obtained source signals. Traditional hierarchical judgment matrixhas been improved by combing the normally coherence function between the de-correlated sourcesignals and the output response signal and the1to9scale method. So the new method of judgmentmatrix has been given to do the quantitative analysis. Simulation analysis verifies the effectiveness ofthe method. Finally, this method has been applied to the project aimed to solve the abnormal vibrationof the cab in a truck and try to find out the main sources of this vibration. The improved results verifythat this method has a certain reference value to solve engineering problems.