CN103852269B - Bullet train runs kinetic parameter detection method - Google Patents

Abstract

The present invention relates to a kind of bullet train and run kinetic parameter detection method, comprise the steps: to set up the random vibration numerical simulation model of train dynamics state, complete power spectrum response simulation calculation；To line test sensors optimum placement, measuring train dynamic response time-domain information of each test point when circuit is runed, test signal extracts vehicle part dynamic response feature after frequency domain power analysis of spectrum processes；Coincideing by numerical simulation power spectrum and line test power spectrum, sets up the fundamental equation of kinetic parameter detection；Solve this fundamental equation, it is thus achieved that and output spectrum density coincide under the conditions of dynamics of vehicle detection parameter.The present invention is using the kinetic parameter of train as monitoring object, mixing on-the-spot test technology and dynamics simulation technology, the fundamental equation setting up kinetic parameter detection completes the detection of high speed train dynamics parameter, realize the long-term action to running state of high speed strain kinetic parameter to assess, for the safe and reliable offer technical guarantee of train operation.

Description

Bullet train runs kinetic parameter detection method

Technical field

The present invention relates to a kind of high speed train dynamics parameter detection method based on real measured data, particularly to a kind of pin The bullet train of long-term operation is run the detection method that kinetic parameter feature changes.

Background technology

In order to the dynamic response behavior that Accurate Prediction bullet train is under long service operating environment, bullet train The kinetic parameter detection of system obtains increasing concern.For the essence of problem, this technology belongs to military service row at a high speed Car system dynamics parameter identification technique, falls within the problem domains of Modeling Method for Train Dynamics correction.For conventional building work For journey structural model correction technique, at present from the object of Model Identification, choosing of identification parameter, the selection of test data, with And method for solving has obtained greater advance.But for bullet train system dynamics parameter detecting, due to row of being on active service The randomness of the experienced environmental load of car, and the complexity of train system itself, its parameter identification correlation technique is also in rising Step section.

Summary of the invention

Present invention is primarily aimed at solution the problems referred to above and deficiency, it is provided that a kind of bullet train runs kinetic parameter inspection Survey method, it is achieved assessing the long-term action of running state of high speed strain kinetic parameter, safe and reliable for train operation carries For technical guarantee.

For achieving the above object, the technical scheme is that

A kind of bullet train runs kinetic parameter detection method, comprises the steps:

Step A, set up the random vibration numerical simulation model of train dynamics state, complete power spectrum response emulation meter Calculate, the overall or partial dynamic response by the part classification output block such as car body, framework；

Step B, train line test sensor is optimized layout, measures bullet train and respectively survey when circuit is runed The dynamic response time-domain information of pilot, it is special that test signal extracts vehicle part dynamic response after frequency domain power analysis of spectrum processes Levy；

Step C, coincideing by Vehicular system random vibration numerical simulation power spectrum and line test power spectrum, set up dynamic The fundamental equation of mechanics parameter detection, S (x)=S^m,

Wherein, S^mFor the power spectral density value by the actual test of described step B, S (x) is to be calculated by correction model accordingly Power spectral density value；

Fundamental equation in step D, solution procedure C, it is thus achieved that and output spectrum density coincide under the conditions of vehicle power Learn detection parameter, complete vehicle system dynamics parameter detecting.

Further, bullet train according to claim 1 runs kinetic parameter detection method, it is characterised in that: In described step A, use flexible car body FEM (finite element) model, and by one being, two be connection system and bogie, coupled axles Composition rail vehicle kinetic model, applies infinite periodic structure to carry out orbital simulation, sets up typical orbit sub-structure model, should Carry out random track irregularity process by pseudo-excitation method, carry out vehicle-virtual Harmonic Analysis of track structure Coupled Dynamics, Complete power spectrum response simulation calculation eventually.

Further, in described step B, the preferred arrangement of described test sensor, it is specifically included in train axle box, turns to Frame and car body install the sensor for test, and described sensor at least includes acceleration transducer, and wherein, trailing or leading bogie adds Velocity test is divided into two passages vertical, horizontal, horizontal and vertical passage, bogie, car body before and after car body acceleration test point Test is divided into different acquisition units, each unit all to work alone.

Further, in described step B, message processing flow is axle box, bogie or car body test signal, pre-through signal Carry out A/D conversion after processing module, and carry out signal characteristic abstraction through signal processing analysis.

Further, described signal processing analysis and feature extraction use modern spectral estimation method, specifically include: first By the test data estimation of the described sensor acquisition laid is gone out the AR parameter model of signal, MA model or arma modeling, Output according still further to different parameters model completes the power Spectral Estimation of time-domain signal.

Further, in described step C, set up kinetic parameter detection fundamental equation step particularly as follows:

Setting up the parameter list system affecting dynamics of vehicle behavior, this parameter list is finally reflected by mapping relations In dynamics of vehicle FEM (finite element) model in step A；

Adjusted the design parameter of vehicle dynamics system by correction factor, and will dynamics of vehicle FEM (finite element) model be repaiied The quality of positive model, damping and stiffness matrix are expressed as the function of correction factor；

If M power spectral density value of the actual test of step B is expressed as, and corresponding by repairing The power spectral density value that positive model calculates is expressed as S (x)=(S₁(x),S₂(x),…,S_M(x))^T, in the ideal case, after correction The calculating power spectral value of model should be equal with measured power spectrum, obtains S (x)=S^m。

Further, in described step D, use the fundamental equation in step C described in the L-M Algorithm for Solving of trusted zones type, Obtain the least square solution of this fundamental equation.

To sum up content, bullet train of the present invention runs kinetic parameter detection method, with the power of bullet train Parameter, as monitoring object, mixing on-the-spot test technology and dynamics simulation technology, establishes the base of kinetic parameter detection This equation, application trusted zones type L-M derivation algorithm obtains the least square solution of parameter detecting fundamental equation, it is achieved bullet train moves The detection of mechanics parameter, it is achieved the long-term action of running state of high speed strain kinetic parameter is assessed, for the peace of train operation Entirely, technical guarantee is reliably provided.

Accompanying drawing explanation

Fig. 1 is flow chart of the present invention.

Detailed description of the invention

With detailed description of the invention, the present invention is described in further detail below in conjunction with the accompanying drawings:

As it is shown in figure 1, a kind of bullet train runs kinetic parameter detection method, make with the dynamics state of bullet train For monitoring object, mixing on-the-spot test technology and dynamics simulation technology, respectively survey when circuit is runed by measuring bullet train The Dynamic Response Information of pilot, and test signal extraction vehicle component dynamic response after frequency domain power analysis of spectrum processes is special Levy, coincideing by Vehicular system random vibration numerical simulation power spectrum and line test power spectrum, set up kinetic parameter detection Fundamental equation, reapply trusted zones type L-M derivation algorithm, it is thus achieved that the least square solution of parameter detecting fundamental equation, finally real The detection of existing high speed train dynamics parameter.

This detection method specifically includes following steps:

Step A, set up the random vibration numerical simulation model of train dynamics state, complete power spectrum response emulation meter Calculate, the overall or partial dynamic response by the part classification output block such as car body, framework.

In order to accurately reflect, car body is overall, local vibration state, uses flexible car body FEM (finite element) model, and by one being, two It is connection system and bogie, coupled axles composition rail vehicle kinetic model.Infinite periodic structure is applied to carry out track mould Intend, only need to set up the typical orbit sub-structure model that degree of freedom greatly reduces, and apply its periodic boundary condition, it is possible to Frequency Response to whole track structure.Application pseudo-excitation method carries out random track irregularity process, it is considered to the para-position of different wheel The place's of putting arbitrary excitation is the complete Coherence Mode of homology, has the virtual harmonic excitation of lagging phase by spectral factorization viewpoint structure.Complete Under the most virtual harmonic excitation effect, vehicle-virtual Harmonic Analysis of track structure Coupled Dynamics, and basic by pseudo-excitation method Principle completes power spectrum response simulation calculation, finally presses the part classification output block such as car body, framework entirety or local dynamic effect rings Should.

Vehicle is taken turns constituting by car body (FEM (finite element) model), 2 bogies and 4, and passing through one between them is to be with two Suspension arrangement connects.Track is considered as three-dimensional three layer scattering point-supported endless chain structure, including rail, sleeper and railway roadbed, Choose the sleeper under the rail between adjacent rail sleepers, and rail and railway roadbed as Substructure System.Rail uses spatial beam Discrete, each node has 5 degree of freedom；Sleeper is considered as rigid body, it is considered to it is vertical, horizontal and rotates 3 degree of freedom；Railway roadbed from Dissipate for rigid block, only consider its vertical vibration.

Three class track irregularities are modeled as zero-mean Stationary Gauss Random process, and the most uncorrelated, and its power spectrum divides Wei S_v(ω)、S_a(ω) and S_c(ω).Considering 4 wheels arbitrary excitation by homology track, they are the most relevant many Point inspiration problem, the vector of the track irregularity composition at four Wheel/Rail Contact Points is:

r_j(t)={r_j(t-t₁) r_j(t-t₂) r_j(t-t₃) r_j(t-t₄)}^T, (j=v, a, c) (1)

Wherein r_jT the spectral power matrix of (), can be to be expressed as form:

$S_{\mathrm{rr}}^j\left(\mathrm{\ω}\right)=\left[\begin{array}{cccc}1& e^{\mathrm{i\ω}(t_1-t_2)}& ...& e^{\mathrm{i\ω}(t_1-t_m)}\\ e^{\mathrm{i\ω}(t_2-t_1)}& 1& ...& e^{\mathrm{i\ω}(t_2-t_m)}\\ .& .& & .\\ .& .& & .\\ .& .& & .\\ e^{\mathrm{i\ω}(t_m-t_1)}& e^{\mathrm{i\ω}(t_m-u_2)}& ...& 1\end{array}\right]---\left(2\right)$

The response of coupled system is u (t), then according to traditional random vibration theory, and a class track irregularity independent role Time, coupled system response power spectrum is:

$S_{\mathrm{uu}}^i\left(\mathrm{\ω}\right){=H}^{*}\left(\mathrm{\ω}\right)S_{\mathrm{rr}}^j\left(\mathrm{\ω}\right)H^T\left(\mathrm{\ω}\right)---\left(3\right)$

Wherein, subscript * represents that complex conjugate, subscript T represent transposition.Then three class track irregularities act on down simultaneously, coupled systemes System response power spectrum should be superposition during three's independent role, it may be assumed that

$S_{\mathrm{uu}}\left(\mathrm{\ω}\right)=\underset{j=v,a,c}{\mathrm{\Σ}}{S}_{\mathrm{uu}}^j\left(\mathrm{\ω}\right)---\left(4\right)$

Obviously, directly calculate according to equation (3) and (4), need to solve the frequency response function matrix of coupled system, go forward side by side Company's multiplication of the big matrix of row.

According to dummy excitation law theory, it is constructed as follows dummy excitation:

${\tilde{r}}_j(\mathrm{\ω},t)={\left[\begin{array}{cccc}{e}^{-\mathrm{i\ω}{t}_1}& e^{-\mathrm{i\ω}{t}_2}& e^{-\mathrm{i\ω}{t}_2}& e^{-\mathrm{i\ω}{t}_4}\end{array}\right]}^T\sqrt{S_j\left(\mathrm{\ω}\right)}{e}^{\mathrm{i\ωt}},(j=v,a,c)---\left(5\right)$

This is the single-point harmonic excitation of a broad sense, and equation (3)-(4) can be re-written as:

$S_{\mathrm{uu}}\left(\mathrm{\ω}\right)=\underset{j=v,a,c}{\mathrm{\Σ}}{{\tilde{u}}_j}^{*}(\mathrm{\ω},t){{\tilde{u}}_j}^T(\mathrm{\ω},t)---\left(6\right)$

${\tilde{u}}_j(\mathrm{\ω},t)=H\left(\mathrm{\ω}\right){\tilde{r}}_j(\mathrm{\ω},t)---\left(7\right)$

Wherein,It is that structure is in dummy excitationStable state virtual responsive under Zuo Yong.

The power spectrum of response acceleration can be tried to achieve by following formula:

$S_{\stackrel{..}{u}\stackrel{..}{u}}\left(\mathrm{\ω}\right)={\mathrm{\ω}}^4{S}_{\mathrm{uu}}\left(\mathrm{\ω}\right)---\left(8\right)$

Three class track irregularities are converted into corresponding simple harmonic quantity dummy excitation by pseudo-excitation method, by complicated coupled system with Machine vibration problem is converted into the Solve problems of the stable state virtual responsive of broad sense single-point harmonic excitation, thus the reducing of high degree Operand.

Step B, train line test sensor is optimized layout, measures bullet train and respectively survey when circuit is runed The dynamic response time-domain information of pilot, it is special that test signal extracts vehicle part dynamic response after frequency domain power analysis of spectrum processes Levy.

Kinetics frequency domain character according to wheel shaft, bogie and car body, selects to be suitable for sensor type, i.e. at train axle Case, bogie and car body install the sensor at least including acceleration transducer type, and according to vehicle integral power scholarship and moral conduct are Information or local dynamic effect information have tested the optimization design of layout scheme.Trailing or leading bogie acceleration test be divided into vertical, Horizontal two passages, horizontal and vertical passage before and after car body acceleration test point.The test of bogie, car body is divided into different acquisition list Unit, each unit all works alone, and constitutes whole information test system.

Message processing flow is bogie or car body test signal, carries out A/D conversion, and warp after signal pre-processing module Signal processing analysis carries out signal characteristic abstraction, including using modern spectral estimation method, by the test number to sensor acquisition It is believed that breath estimates the argument sequence model of signal, and complete to gather time domain letter according to the output of different parameters series model Number power Spectral Estimation.

Using modern spectral estimation method, by bogie and flexible car body being laid the test data of sensor acquisition, estimating Count out test and gather AR, MA or ARMA parameter model etc. of signal, and further according to the signal output work of different parameters model Rate completes power Spectral Estimation.

The basic thought of this signal processing method, it is believed that test signal time sequence x (n) is that white noise passes through certain model Produce.By selecting certain model, the sample of signal data having been observed that or auto-correlation function are determined the ginseng of this forecast model Number, and then extract the power spectrum characteristic of signal.As follows:

$x\left(n\right)=\stackrel{p}{\mathrm{\Σ}}{b}_{l}w(n-l)-\stackrel{p}{\mathrm{\Σ}}{a}_{k}x(n-k)---\left(9\right)$

Wherein, w (n) is white noise sequence, carries out z-transform and obtains:

$\underset{k=0}{\overset{p}{\mathrm{\Σ}}}{a}_{k}X\left(z\right)z^{-k}=\underset{l=0}{\overset{p}{\mathrm{\Σ}}}{b}_{l}W\left(z\right)z^{-l}---\left(10\right)$

The transmission function of model is:

$H\left(z\right)=\frac{X\left(z\right)}{W\left(z\right)}=\frac{\underset{l=0}{\overset{p}{\mathrm{\Σ}}}{b}_l{z}^{-1}}{\underset{k=0}{\overset{p}{\mathrm{\Σ}}}{a}_k{z}^{-k}}=\frac{B\left(z\right)}{A\left(z\right)}---\left(11\right)$

Wherein, $A\left(z\right)=\underset{k=0}{\overset{p}{\mathrm{\Σ}}}{a}_k{z}^{-k}$ , $B\left(z\right)=\underset{l=0}{\overset{p}{\mathrm{\Σ}}}{b}_l{z}^{-l}$ 。

The power spectral density assuming input signal white noise is, then the power spectral density of model output is

$P_x\left(z\right)={\mathrm{\σ}}_{\mathrm{\ω}}^{2}H\left(z\right)H\left(z^{-1}\right)={\mathrm{\σ}}_{\mathrm{\ω}}^2\frac{B\left(z\right)B\left(z^{-1}\right)}{A\left(z\right)A\left(z^{-1}\right)}---\left(12\right)$

By z=e^iωSubstitute into above formula,

$P_x\left(z\right)={\mathrm{\σ}}_{\mathrm{\ω}}^2{\left|\frac{B\left(e^{\mathrm{i\ω}}\right)}{A\left(e^{\mathrm{i\ω}}\right)}\right|}^2---\left(13\right)$

When determiningWith coefficient a_k、b_lAfter, it is possible to estimate the power spectral density of stochastic signal time domain sequences.

AR, MA or arma modeling are had respectively:

AR model: $x\left(n\right)=-\underset{k=1}{\overset{p}{\mathrm{\Σ}}}{a}_{k}x(n-k)+w\left(n\right)---\left(14\right)$

MA model: $x\left(n\right)=\underset{l=1}{\overset{p}{\mathrm{\Σ}}}{b}_{l}w(n-l)---\left(15\right)$

Arma modeling: $x\left(n\right)=\underset{l=0}{\overset{p}{\mathrm{\Σ}}}{b}_{l}w(n-l)-\underset{k=0}{\overset{p}{\mathrm{\Σ}}}{a}_{k}x(n-k)---\left(16\right)$

Step C, coincideing by Vehicular system random vibration numerical simulation power spectrum and line test power spectrum, set up dynamic The fundamental equation of mechanics parameter detection.

Set up the parameter list system of dynamics of vehicle behavior of affecting, as include one be, two be connection system rigidity and The mechanics parameters etc. such as damping.This parameter list is finally reflected dynamics of vehicle finite element mould in such as step A by mapping relations In type.Adjusted the design parameter of vehicle dynamics system by correction factor, and will dynamics of vehicle FEM (finite element) model be revised The quality of model, damping and stiffness matrix are expressed as the function of correction factor:

$M=\underset{i=1}{\overset{N_e}{\mathrm{\Σ}}}{x}_{M_i}{M}_i^a,C=\underset{i=1}{\overset{N_e}{\mathrm{\Σ}}}{x}_{C_i}{C}_i^a,K=\underset{i=1}{\overset{N_e}{\mathrm{\Σ}}}{x}_{K_i}{K}_i^a---\left(17\right)$

Wherein, M, C, K are the quality of correction model, damping and stiffness matrix, N_eUnit for structural model is total, ,,It is respectively the quality of i-th unit, damping and stiffness modification.,,It is respectively initial model Quality, damping and stiffness matrix.

If M power spectral density value of the actual test of step B is expressed as, and corresponding by repairing The power spectral density value that positive model calculates is expressed as S (x)=(S₁(x),S₂(x),…,S_M(x))^T。

In the ideal case, after correction, the calculating power spectral value of model should be equal with measured power spectrum, i.e.

S(x)=S^m (18)

Wherein, x=(x₁,x₂,…,x_N)^TFor the correction factor vector being made up of quality, damping and stiffness modification.Side Journey (18) is fundamental equation based on power spectral density Modifying model.Solving equation (18), can obtain power spectral density and coincide Under the conditions of dynamics of vehicle detection parameter.

Fundamental equation in step D, solution procedure C, it is thus achieved that and output spectrum density coincide under the conditions of vehicle power Learn detection parameter, complete vehicle system dynamics parameter detecting.

The common method of the fundamental equation (18) in solution procedure C is method of least square, due in line test data one As containing noise error, cause this detection equation to show as ill-posed problem under Hadmadard meaning, equation (18) will be caused Without solving.Use the Levenberg-Marquardt(L-M of trusted zones type) Algorithm for Solving train dynamics parameter detecting the most square Journey, wherein Jacobi battle array J (x) is the sensitivity composition of substantial amounts of train dynamics response power spectrum, needs during iterative Repeatedly to calculate Jacobi matrix J (x).Giving the Sensitivity Analysis Method of random response power spectrum, the method is based on height Imitate what accurate pseudo-excitation method derived, and do not introduce any in derivation it is assumed that application the method is rung at random The sensitive analysis answered, it is possible to obtain higher computational efficiency and computational accuracy.

Use trusted zones type Levenberg-Marquardt(L-M) this nonlinear ill-posed problem of Algorithm for Solving, specifically Including:

Make r (x)=(r₁(x),r₂(x),…,r_i(x),…,r_M(x))^T, wherein(i=1, 2 ..., M).Open problems (18) translates into following least square problem

$\begin{array}{cc}\min & f\left(x\right)={\left|\right|r\left(x\right)\left|\right|}_2^2=r{\left(x\right)}^{T}r\left(x\right)=\underset{i=1}{\overset{M}{\mathrm{\Σ}}}{\left(\frac{S_i-S_i^m}{S_i^m}\right)}^2\\ s.t.& x_j^b\≤x_j\≤x_j^t,j=\mathrm{1,2},...,N\end{array}---\left(19\right)$

Here,、It is respectively jth correction factor x_jLower limit and the upper limit.

The Jacobi battle array making J (x) be r (x), its expression-form is as follows

$J\left(x\right)=\left[\begin{array}{cccc}\frac{\∂r_1\left(x\right)}{\∂x_1}& \frac{\∂r_1\left(x\right)}{\∂x_2}& ...& \frac{\∂r_1\left(x\right)}{\∂x_N}\\ \frac{\∂r_2\left(x\right)}{\∂x_1}& \frac{\∂r_2\left(x\right)}{\∂x_2}& ...& \frac{\∂r_2\left(x\right)}{\∂x_N}\\ ...& ...& ...& ...\\ \frac{\∂r_M\left(x\right)}{\∂x_1}& \frac{\∂r_M\left(x\right)}{\∂x_2}& ...& \frac{\∂r_M\left(x\right)}{\∂x_N}\end{array}\right]---\left(20\right)$

Nonlinear function r (x) is at x^kNeighbouring linear model is, useReplace R (x) in equation (19), and apply trust region method, then can obtain following trusted zones model

$\begin{array}{cc}\min & {\left|\right|r\left(x^k\right)+J\left(x^k\right)(x-x^k)\left|\right|}_2^2\\ s.t.& x_j^b\≤x_j\≤x_j^t,j=\mathrm{1,2},...,N\\ & {\left|\right|x-x^k\left|\right|}_2\≤{\mathrm{\Δ}}_k\end{array}---\left(21\right)$

Wherein, Δ_kFor Trust Region Radius.Above formula (21) is constrained least square problem, can expand into

$\begin{array}{cc}\min & q_k\left(x\right)={\left|\right|r\left(x^k\right)\left|\right|}^2+2r{\left(x^k\right)}^{T}J\left(x^k\right)(x-x^k)+{(x-x^k)}^{T}J{\left(x^k\right)}^{T}J\left(x^k\right)(x-x^k)\\ s.t.& x_j^b\≤x_j\≤x_j^t,j=\mathrm{1,2},...,N\\ & {\left|\right|x-x^k\left|\right|}_2\≤{\mathrm{\Δ}}_k\end{array}---\left(22\right)$

Make s=x-x^k, above-mentioned subproblem can be solved by equation below

(J(x^k)^TJ(x^k)+μ_kI)s=-J(x^k)^Tr(x^k) (23)

Wherein, μ_k＞ 0 is referred to as L-M parameter.Work as x=x^k+1Time, can be obtained fom the above equation

x^k+1=x^k-(J(x^k)^TJ(x^k)+μ_kI)^-1J(x^k)^Tr(x^k) (24)

By iterative problem (24), the least square solution of dynamics of vehicle detection fundamental equation (18) can be obtained, Complete vehicle system dynamics parameter detecting.

As it has been described above, combine the plan content given by accompanying drawing, similar technical scheme can be derived.In every case it is not take off From the content of technical solution of the present invention, any simple modification of above example being made according to the technical spirit of the present invention, etc. With change and modification, all still fall within the range of technical solution of the present invention.

Claims (3)

1. a bullet train runs kinetic parameter detection method, it is characterised in that comprise the steps:

Step A, set up the random vibration numerical simulation model of train dynamics state, complete power spectrum response simulation calculation, press The part classification output block such as car body, framework entirety or partial dynamic response；

The concrete flexible car body FEM (finite element) model that uses, and by one being, two be connection system and bogie, coupled axles composition row Car vehicle dynamic model, applies infinite periodic structure to carry out orbital simulation, sets up typical orbit sub-structure model, apply virtual Advocate approach carries out random track irregularity process, carries out vehicle-virtual Harmonic Analysis of track structure Coupled Dynamics, is finally completed Power spectrum response simulation calculation；

Step B, train line test sensor is optimized layout, measures bullet train each test point when circuit is runed Dynamic response time-domain information, test signal extracts vehicle part dynamic response feature after frequency domain power analysis of spectrum processes；

The preferred arrangement of described test sensor, is specifically included in train axle box, bogie and car body and installs the biography for test Sensor, described sensor at least includes acceleration transducer, and wherein, trailing or leading bogie acceleration test is divided into vertical, horizontal two Individual passage, horizontal and vertical passage before and after car body acceleration test point, the test of bogie, car body is divided into different acquisition units, often Individual unit all works alone；

Step C, coincideing by Vehicular system random vibration numerical simulation power spectrum and line test power spectrum, set up kinetics ginseng The fundamental equation of number detection, S (x)=S^m, concretely comprise the following steps；

Setting up the parameter list system affecting dynamics of vehicle behavior, this parameter list is finally reflected in such as step by mapping relations In rapid A in dynamics of vehicle FEM (finite element) model；

Adjusted the design parameter of vehicle dynamics system by correction factor, and dynamics of vehicle FEM (finite element) model will be revised mould The quality of type, damping and stiffness matrix are expressed as the function of correction factor；

If M power spectral density value of the actual test of step B is expressed asAnd accordingly by revising mould The power spectral density value that type calculates is expressed as S (x)=(S₁(x),S₂(x),…,S_M(x))^T, in the ideal case, revise rear mold The calculating power spectral value of type should be equal with measured power spectrum, obtains S (x)=S^m；

S^mFor the power spectral density value by the actual test of described step B, S (x) is the power spectral density calculated by correction model accordingly Value；

Step D, use the fundamental equation in step C described in the L-M Algorithm for Solving of trusted zones type, it is thus achieved that the minimum of this fundamental equation Two take advantage of solution, so obtain and output spectrum density coincide under the conditions of dynamics of vehicle detection parameter, complete Vehicular system and move Mechanics parameter detects.

Bullet train the most according to claim 1 runs kinetic parameter detection method, it is characterised in that: in described step In B, message processing flow is axle box, bogie or car body test signal, carries out A/D conversion after signal pre-processing module, and Signal characteristic abstraction is carried out through signal processing analysis.

Bullet train the most according to claim 2 runs kinetic parameter detection method, it is characterised in that: at described signal Reason is analyzed and feature extraction uses modern spectral estimation method, specifically includes: first passes through the described sensor to laying and adopts The test data estimation of collection goes out the AR parameter model of signal, MA model or arma modeling, according still further to the output of different parameters model Power completes the power Spectral Estimation of time-domain signal.