Modal parameter accuracy evaluation method, system and storage medium
By constructing the posterior probability density function and objective function under Bayesian framework, the Hansen matrix was derived, and the uncertainty problem of modal parameter recognition under environmental excitation and forced vibration was solved, and the accuracy evaluation of modal parameters was achieved. , supports structural safety assessment and damage identification.
Patent Information
- Application Number
- CN202010568581.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2020-06-19
- Publication Date
- 2024-12-06
- Estimated Expiration
- 2040-06-19
AI Technical Summary
The existing technology cannot accurately evaluate the accuracy of structural modal parameters under the dual action of environmental excitation and forced vibration, resulting in great uncertainty in modal parameter identification, affecting structural safety performance evaluation and damage identification.
Under Bayesian framework, construct the posterior probability density function and objective function, deduce the Hansen matrix, and establish the relationship between the parameter covariance matrix and the Hansen matrix, obtain the optimal value of the modal parameters through optimization steps, and build the Hansen matrix , calculate the covariance matrix and coefficient of variation to achieve the accuracy evaluation of modal parameters.
The accuracy evaluation of modal parameters under the dual action of environmental excitation and forced vibration is realized, reducing the uncertainty of modal parameter identification, and providing reliable data and theoretical support for structural safety assessment and damage identification.
Smart Images

Figure CN111859250B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of data processing technology, and in particular to a method, system and storage medium for rapid modal parameter accuracy evaluation under environmental and forced excitation. Background Art
[0002] For structures such as bridges and buildings, they have some inherent dynamic characteristics, including natural frequency, damping ratio, vibration mode, etc. These properties are basically unchanged when the structure is not damaged. When designing the structure, these properties are also important indicators for controlling the structural characteristics. At the same time, after the structure is completed, data is collected from these structures through sensors, and the dynamic characteristics of the structure are identified based on these data, which are then applied to the safety assessment and damage identification of existing structures. In the process of structural data collection, due to factors such as environmental noise and data acquisition system errors, the collected data are inevitably mixed with a lot of noise, so that modal parameter identification based on these data will produce greater uncertainty. These uncertainties make it difficult to evaluate the structural safety performance and identify damage based on modal parameter changes. It is impossible to determine whether the parameter changes are caused by inaccurate identification or structural damage changes. Therefore, how to evaluate the accuracy of modal parameter identification is a technical problem that needs to be solved urgently. Forced vibration is common in gymnasiums and bridge structures, and the purpose is to excite sufficient vibration amplitude of the structure.
[0003] Forced vibration testing is one of the three common structural vibration testing methods. Existing data analysis methods based on forced vibration collection all treat the response generated by environmental excitation as noise. However, when the forced vibration load is not enough to excite the structure, the structural response generated by environmental excitation cannot be ignored, and the uncertainty of the resulting modal parameters will be large. At present, there is no relevant accuracy assessment technology for modal parameter accuracy assessment under the dual effects of environmental excitation and forced vibration.
[0004] Defects and deficiencies of background technology: The existing technology is mainly the finite difference method, which is an approximate algorithm. The result will have a large error and cannot accurately evaluate the accuracy of the forced vibration modal identification parameters considering environmental excitation. In addition, other methods simulate environmental excitation as white noise. This simplified method will have a large error when the forced vibration is not large enough. Since structures such as bridges and buildings are large structures, the forced excitation is often not large enough. There is no solution to how to evaluate the uncertainty of modal parameters under the dual effects of environmental excitation and forced vibration. Summary of the invention
[0005] The present invention solves the problem of parameter accuracy assessment in the process of structural modal parameter assessment based on test data under the dual effects of environmental excitation and forced vibration. Based on the posterior probability density function and the objective function established under the Bayesian framework, the accuracy assessment calculation theory is derived, the Hansen matrix is obtained, and the relationship between the parameter covariance matrix and the Hansen matrix is established. The calculation method of the parameter variation coefficient is obtained, and the parameter uncertainty assessment in the modal parameter identification of the test data under the dual effects of environmental excitation and forced vibration is realized.
[0006] The present invention provides a method for quickly evaluating the accuracy of modal parameters under environmental and forced excitation, comprising the following steps:
[0007] Step 1, optimization step: optimizing the acquired modal parameters to obtain the optimal values of the modal parameters;
[0008] Step 2, Hansen matrix construction step: construct the Hansen matrix according to the optimal values of the modal parameters;
[0009] Step 3, calculation step: calculate and obtain the covariance matrix of the modal parameters;
[0010] Step 4, evaluation step: calculate the coefficient of variation of the modal parameters to achieve the accuracy evaluation of the modal parameters.
[0011] As a further improvement of the present invention, the step 1, the optimization step, comprises sequentially performing the following steps:
[0012] Optimal value calculation steps: Optimize based on formula (12) to obtain the optimal value of the modal parameter;
[0013] Judgment step: judge whether the optimal value of the modal parameter converges. If so, execute the Hansen matrix construction step, otherwise return to execute the optimal value calculation step; formula (12) is as follows:
[0014]
[0015] In formula (12), S e represents the power spectral density of the prediction error, and Φ represents the vibration mode.
[0016] The present invention also provides a system for rapidly evaluating the accuracy of modal parameters under environmental and forced excitation, comprising:
[0017] Optimization module: used to optimize the acquired modal parameters to obtain the optimal values of the modal parameters;
[0018] Hansen matrix construction module: used to construct the Hansen matrix according to the optimal values of modal parameters;
[0019] Calculation module: used to calculate the covariance matrix of modal parameters;
[0020] Evaluation module: used to calculate the coefficient of variation of modal parameters and realize the accuracy evaluation of modal parameters.
[0021] The present invention also provides a computer-readable storage medium, which stores a computer program. The computer program is configured to implement the steps of the method for rapid modal parameter accuracy evaluation under environment and forced excitation described in the present invention when called by a processor.
[0022] The beneficial effects of the present invention are as follows: in view of the problem that existing methods cannot realize the accuracy evaluation of forced vibration modal identification parameters under environmental excitation, the present invention develops a new modal parameter uncertainty evaluation technology, which can realize the accuracy evaluation of modal parameters under the dual effects of environmental excitation and forced vibration, and provide data and theoretical support for the later development of damage identification and safety assessment technology based on vibration data and modal parameter changes. BRIEF DESCRIPTION OF THE DRAWINGS
[0023] Figure 1 It is a flow chart of the present invention. DETAILED DESCRIPTION
[0024] The present invention discloses a method for rapid modal parameter accuracy evaluation under environmental and forced excitation. The present invention is described in two major parts. The first part is an overview of the construction of the objective function, and the second part is an explanation of the parameter accuracy evaluation technology developed based on the objective function. The focus of the present invention is the second part.
[0025] 1. Bayesian framework to construct dual-action objective function of environmental excitation and forced vibration:
[0026] Structural acceleration response obtained based on environmental excitation and forced vibration tests (n represents the number of degrees of freedom of measurement) can be obtained by the following formula:
[0027]
[0028] Here i = 1, ..., N, N represents the number of measurement sample points; represents the theoretical response of the structure under forced vibration; represents the theoretical response of the structure under environmental excitation; e i Represents the prediction error between the theoretical response and the measured response.
[0029] The Fast Fourier Transform (FFT) of the measured structural response is defined as follows:
[0030]
[0031] In the formula, i 2=-1, indicating a complex number; Δt indicates the sampling interval.
[0032] Define a vector In a selected frequency band, FFT data can be used to k According to Bayes' theorem, under the condition of given data, the posterior probability density function of the modal parameter θ to be identified can be expressed as follows:
[0033] p(θ|{Z k})∝p(θ)p({Z k}|θ)(3)Here p(θ) is the prior probability density function; p({Z k}|θ) represents the likelihood function. Assuming that the prior probability density function follows a uniform distribution, the posterior probability density function can be obtained by the following formula: p(θ|{Z k})∝p({Z k}|θ)(4)
[0034] Under the condition of small Δt and long enough data, it can be proved that the likelihood function follows Gaussian distribution, that is,
[0035]
[0036] Here det represents the determinant; μ k (θ)=E[Z k |θ], E[·|θ] represents the conditional expectation given θ; C k (θ)∈R 2n ×2n Represents Z k The covariance matrix of
[0037]
[0038] In the formula, S e represents the power spectral density of the prediction error; Φ represents the vibration mode, H k The unit inside can be obtained by the following formula:
[0039]
[0040] in
[0041]
[0042] S ij is the unit in the power spectrum density matrix of the modal force under environmental excitation; f k Represents the frequency horizontal axis; β ik =f i / f k Represents a frequency ratio; f iis the i-th order natural frequency; ζ i is the i-th order damping ratio.
[0043] In formula (5), μ k (θ) can be expressed as
[0044]
[0045] Where Φ(i)∈R n represents the i-th order vibration mode; Φ(I,i) represents the component of the i-th order vibration mode corresponding to the forced excitation position I; S k Represents the frequency coordinate f k The corresponding fast Fourier transform (FFT) of the exciter acceleration response; i represents the ratio of the shaker mass to the modal mass in the i-th mode, which can be obtained by the following formula:
[0046]
[0047] In summary, the modal parameters that need to be identified include natural frequency, damping ratio, power spectral density of modal force, power spectral density of prediction error, vibration mode and the ratio of the exciter mass to the modal mass. Theoretically, the optimal value of the modal parameter can be obtained by optimizing the likelihood function in formula (5) to reach the maximum value. In order to facilitate optimization, we convert the maximum value problem into a minimum value problem by using the negative log-likelihood function method. The negative log-likelihood function is as follows:
[0048] L(θ)=-lnp({Z k}|θ) (11)
[0049] By optimizing the negative log-likelihood function to its minimum value, the optimal value of the modal parameter can be obtained. However, if the optimization is performed directly, ill-conditioned problems will occur in the calculation process. At the same time, the number of parameters that need to be optimized increases with the number of measurement degrees of freedom. Therefore, when the number of degrees of freedom is large, the calculation efficiency will be greatly reduced. In order to solve these problems, a fast and practical calculation method is developed. For the case of separated modes, the negative log-likelihood function is reconstructed as shown in the following formula:
[0050]
[0051] In formula (12), S e represents the power spectral density of the prediction error, Φ represents the vibration mode, S is the power spectral density of the modal force, N f is the number of frequency domain sample points in the selected frequency band; n is the number of test degrees of freedom, β k =f / f k; f is the natural frequency; f k is the frequency coordinate point; ζ is the damping ratio. The other parts are defined as follows:
[0052]
[0053] a k =(1+S e / SD k ) -1 (14)
[0054]
[0055] Where r represents the ratio of the vibrator mass to the modal force mass; S k Represents the frequency coordinate f k The corresponding fast Fourier transform (FFT) of the shaker acceleration response; i is the symbol for complex numbers, i 2 =-1; Φ(I) is the Ith variable of the vibration mode vector, corresponding to the position of the exciter; Re(.) means taking the real part of the internal imaginary variable; Im(.) means taking the imaginary part of the internal imaginary variable; represents the fast Fourier transform value of the test data, express Theoretical value.
[0056]
[0057] Based on the reconstructed negative log-likelihood function - formula (12), some parameters that need to be identified can be represented by other parameters, so only the remaining parameters need to be optimized, thereby establishing an iterative algorithm and greatly improving the computational efficiency. The modal parameter uncertainty assessment in the present invention is also constructed based on formula (12), as follows.
[0058] 2. Technical description of modal parameter uncertainty assessment:
[0059] When there are enough data and modal parameters that can be globally identified, the posterior probability density function can be approximated as a Gaussian distribution, which is equivalent to the approximate form of the second-order Taylor expansion of L(θ) in formula (12). The optimal value of the modal parameter θ can be used To express, considering the fact that the first-order term disappears at the optimal value, the second-order Taylor expansion can be expressed as follows:
[0060]
[0061] In the formula Indicates that L(θ) is The Hansen matrix at . At the same time,
[0062] p(θ|D)∝exp[-L(θ)](18)Substituting formula (13) into (14), the posterior probability density function can be changed to:
[0063]
[0064] This can be viewed as a Gaussian distribution, so:
[0065]
[0066] represents the posterior covariance matrix of the modal parameters. Based on this covariance matrix, the uncertainty assessment of the modal parameters can be achieved. Next, the Hansen matrix The method for obtaining the theoretical solution will be introduced in detail.
[0067] a. Difficulties and solutions in the calculation process
[0068] In the process of evaluating the posterior covariance matrix of modal parameters, the regularization constraints of vibration modes will lead to some computational difficulties, which are briefly introduced as follows. In the process of determining the optimal value of modal parameters, Φ∈R n Subject to the unit norm constraint, that is,
[0069] ||Φ||=1 (21) When determining the Hansen matrix of the negative log-likelihood function, this unit norm constraint can be achieved by explicitly writing the vibration mode in the normalized counterpart, i.e.
[0070]
[0071] Regardless of the value of Φ, we can prove that In order to properly consider the norm constraint in determining the Hansen matrix, the vibration mode Φ in equation (12) should be expressed as Instead, we can use λ to replace λ. In this way, we do not need to consider the influence of constraints when we derive the negative log-likelihood function with respect to the free parameter Φ in solving the Hansen matrix. Through the above steps, the Hansen matrix will be singular in the Φ direction and in each eigendirection (each mode), so there will be m zero eigenvalues. This will lead to computational problems when evaluating the covariance matrix according to formula (20). Let {λ i :i=1,...,n p}and To represent the eigenvalues (in ascending order) and eigenvectors of the Hansen matrix at the optimal value of the modal parameters, the Hansen matrix of the negative log-likelihood function can be written as:
[0072]
[0073] Because λ1=...=λ m= 0, the singular direction can be ignored, and the posterior covariance matrix is equal to the inverse of H, which can be obtained by the following formula:
[0074]
[0075] In the next section, the method of determining each unit in the Hansen matrix will be introduced in detail.
[0076] b. Solving the Hansen matrix for the separated mode case
[0077] Based on the above discussion, for the case of separated modes, the negative log-likelihood function in formula (12) should be written as:
[0078]
[0079] The corresponding modal parameter set when calculating the Hansen matrix is This matrix can be obtained by taking the second derivative of the negative log-likelihood function with respect to these parameters, and the dimension of the matrix is (n+5)×(n+5). The detailed calculation method of the Hansen matrix is as follows. For simplicity, the variables under differentiation are represented by superscripts in brackets. To save space, only the derivative of the negative log-likelihood function L with respect to the modal parameters is listed below.
[0080]
[0081] In the formula L (ζζ) The expression of L (ff) similar.
[0082]
[0083] L (ζS) The expression of L (fS) similar.
[0084]
[0085] The expression is similar.
[0086]
[0087] L (ζr) The expression of L (fr) similar.
[0088]
[0089] The expression is similar.
[0090]
[0091] L (rr) and The expression is similar.
[0092] Based on the above formula, the Hansen matrix of the negative log-likelihood function can be obtained in an analytical solution, which greatly improves the accuracy and efficiency of the calculation.
[0093] In summary, a method for rapid modal parameter accuracy assessment under environmental and forced excitation of the present invention comprises the following steps:
[0094] Step 1, optimization step: optimizing the acquired modal parameters to obtain the optimal values of the modal parameters;
[0095] Step 2, Hansen matrix construction step: construct the Hansen matrix according to the optimal values of the modal parameters;
[0096] Step 3, calculation step: calculate and obtain the covariance matrix of the modal parameters;
[0097] Step 4, evaluation step: calculate the coefficient of variation of the modal parameters to achieve the accuracy evaluation of the modal parameters.
[0098] The step 1, the optimization step, comprises performing the following steps in sequence:
[0099] Optimal value calculation steps: Optimize based on formula (12) to obtain the optimal value of the modal parameter;
[0100] Judgment step: judge whether the optimal value of the modal parameter converges, if so, execute the Hansen matrix construction step, otherwise return to execute the optimal value calculation step.
[0101] In step 2, the Hansen matrix construction step, the optimal values of the modal parameters are substituted into formulas (26)-(39), and the Hansen matrix is constructed based on formulas (26)-(39).
[0102] In the step 3, the calculation step, the covariance matrix of the modal parameters is obtained based on formula (23) and formula (24).
[0103] Appendix: Calculation of other variables used
[0104] Derivation of Ad
[0105]
[0106] Where T I ∈R n It means that the I-th item is the only non-zero item in this vector, and this non-zero item is equal to 1.
[0107]
[0108]
[0109] Where h k and D k The derivation is briefly mentioned.
[0110] Derivative of Bp
[0111]
[0112]
[0113] Ca k Derivation of
[0114] a k The derivative of can be obtained by the following general calculation method. Its derivative for any variables z1 and z2 is shown below:
[0115]
[0116] Based on the above formula, only and Need to calculate, it can be obtained by the following formula:
[0117]
[0118]
[0119] and D k The derivation is briefly mentioned.
[0120] Derivative of DW
[0121] W=Φ T D k Φ / Φ T Φ (77)
[0122]
[0123] Where e i ∈R m represents a vector whose i-th term is equal to 1 and other terms are equal to 0; δ ij is the Kronecker delta function, which is equal to 1 when i=j and 0 otherwise.
[0124] ED k Derivation of
[0125]
[0126]
[0127] Where e j ∈R m represents a vector whose jth item is equal to 1 and other items are equal to 0;
[0128]
[0129]
[0130] V=Φ(I) 2 ΦΦ T (111)
[0131]
[0132] The present invention also provides a system for rapidly evaluating the accuracy of modal parameters under environmental and forced excitation, comprising:
[0133] Optimization module: used to optimize the acquired modal parameters to obtain the optimal values of the modal parameters;
[0134] Hansen matrix construction module: used to construct the Hansen matrix according to the optimal values of modal parameters;
[0135] Calculation module: used to calculate the covariance matrix of modal parameters;
[0136] Evaluation module: used to calculate the coefficient of variation of modal parameters and realize the accuracy evaluation of modal parameters.
[0137] The optimization module includes running the following modules in sequence:
[0138] Optimal value calculation module: optimizes based on formula (12) to obtain the optimal value of the modal parameter;
[0139] Judgment module: judge whether the optimal value of the modal parameter converges. If so, run the Hansen matrix construction module, otherwise return to run the optimal value calculation module.
[0140] In the Hansen matrix construction module, the optimal values of the modal parameters are substituted into formulas (26)-(39), and the Hansen matrix is constructed based on formulas (26)-(39).
[0141] In the calculation module, the covariance matrix of the modal parameters is obtained based on formula (23) and formula (24).
[0142] The present invention also discloses a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and the computer program is configured to implement the steps of the method for rapid modal parameter accuracy evaluation under environment and forced excitation described in the present invention when called by a processor.
[0143] In order to solve the problem that existing methods are unable to achieve accuracy evaluation of forced vibration modal identification parameters under environmental excitation, the present invention develops a new modal parameter uncertainty evaluation technology to achieve accuracy evaluation of modal parameters under the dual effects of environmental excitation and forced vibration, providing data and theoretical support for the subsequent development of damage identification and safety assessment technology based on vibration data and modal parameter changes.
[0144] The above contents are further detailed descriptions of the present invention in combination with specific preferred embodiments, and it cannot be determined that the specific implementation of the present invention is limited to these descriptions. For ordinary technicians in the technical field to which the present invention belongs, several simple deductions or substitutions can be made without departing from the concept of the present invention, which should be regarded as falling within the protection scope of the present invention.
Claims
1. A method for evaluating the accuracy of modal parameters, characterized in that: The steps include: Step 1, optimization step: optimizing the acquired modal parameters to obtain the optimal values of the modal parameters; Step 2, Hansen matrix construction step: construct the Hansen matrix according to the optimal values of the modal parameters; Step 3, calculation step: calculate and obtain the covariance matrix of the modal parameters; Step 4, evaluation step: calculating the coefficient of variation of the modal parameters to achieve the accuracy evaluation of the modal parameters; the step 1, optimization step includes sequentially performing the following steps: Optimal value calculation steps: Optimize based on formula (12) to obtain the optimal value of the modal parameter; Judgment step: judge whether the optimal value of the modal parameter converges. If so, execute the Hansen matrix construction step, otherwise return to execute the optimal value calculation step; formula (12) is as follows: In formula (12), S e represents the power spectral density of the prediction error, Φ represents the vibration mode, S is the power spectral density of the modal force, N f is the number of frequency domain sample points in the selected frequency band; n is the number of test degrees of freedom; here β k =f / f k ; f is the natural frequency; f k is the frequency coordinate point; ζ is the damping ratio; the other parts are defined as follows: to k (1+S) e / SD k ) -1 ; (14) Where r represents the ratio of the vibrator mass to the modal force mass; S k Represents the frequency coordinate f k The corresponding fast Fourier transform (FFT) of the shaker acceleration response; i is the symbol for complex numbers, i 2 =-1; Φ(I) is the Ith variable of the vibration mode vector, corresponding to the position of the exciter; Re(.) means taking the real part of the internal imaginary variable; Im(.) means taking the imaginary part of the internal imaginary variable; represents the fast Fourier transform value of the test data, express Theoretical value of In the step 2, the Hansen matrix construction step, the optimal values of the modal parameters are substituted into formulas (26)-(39), and the Hansen matrix is constructed based on formulas (26)-(39); In the formula Φ is the vibration mode; ‖Φ‖ is the regularization of the vibration mode; L (ff) It represents the second derivative of the function L with respect to the parameter f; Representation function D k Take the derivative of parameter f once; Representation function D k Take the second derivative of parameter f; d (ff) represents the second derivative of function d with respect to parameter f; p (ff) It represents the quadratic derivative of function p with respect to parameter f; Where L (fζ) It means that the function L is differentiated once with respect to parameter f and then once with respect to ζ; Representation function D k Take the derivative of parameter f once; Representation function D k Take the derivative of parameter ζ once; Representation function D k After taking the derivative of parameter f once, take the derivative of ζ once again; d (fζ) It means that the function d is differentiated once with respect to the parameter f and then differentiated once with respect to ζ; p (fζ) It means that the function p is differentiated once with respect to the parameter f and then differentiated once with respect to ζ; Where L (fS) It means that the function L is differentiated once with respect to parameter f and then differentiated once with respect to S; Representation function D k Take the derivative of parameter f once; p (fS) It means that the function p is differentiated once with respect to the parameter f and then differentiated once with respect to S; In the formula It means that the function L is derived once with respect to the parameter f and then S e First derivative; d (f) represents the first derivative of function d with respect to parameter f; p (f) It means that the function p is derived once with respect to the parameter f; It means that the function p is derived once with respect to the parameter f and then S e One derivative; Where L (fr) It means that the function L is differentiated once with respect to parameter f and then once with respect to r; d (fr) It means that the function d is differentiated once with respect to the parameter f and then once with respect to r; p (fr) It means that the function p is differentiated once with respect to the parameter f and then once with respect to r; In the formula It means that the function L is differentiated with respect to the parameter f once and then One derivative; It means that the function d is differentiated with respect to the parameter f once and then One derivative; It means that the function p is differentiated with respect to the parameter f once and then One derivative; Where L (SS) represents the second derivative of function L with respect to parameter S; p (SS) It represents the quadratic derivative of function p with respect to parameter S; In the formula Represents the function L on parameter S e Second derivative; Represents the function p on parameter S e One derivative; Represents the function p on parameter S e Second derivative; In the formula It means that the function L is derived from the parameter S once and then e First derivative; p (S) It means that the function p is derived once with respect to the parameter S; It means that the function p is derived from the parameter S once and then e One derivative; Where L (Sr) It means that the function L is differentiated once with respect to parameter S and then once with respect to r; p (Sr) It means that the function p is differentiated once with respect to the parameter S and then once with respect to r; In the formula Represents the function L on parameter S e After taking the first derivative, take the first derivative with respect to r; d (r) represents the derivative of function d with respect to parameter r; p (r) It means that the function p is derived once with respect to the parameter r; Represents the function p on parameter S e After taking the first derivative, take the first derivative with respect to r; In the formula It means that the function L is derived once with respect to the parameter S and then One derivative; It means that the function p is derived once with respect to the parameter S and then One derivative; In the formula Represents the function L on parameter S e After taking the derivation once, One derivative; Represents the function d on the parameter One derivative; Represents a function ∈ on a parameter One derivative; Represents a function ∈ with respect to parameter S e After taking the derivation once, One derivative; In the formula Represents the function L on the parameter Second derivative; Represents the function d on the parameter Second derivative; Represents the function p on the parameter Second derivative; In the step 3, in the calculation step, the covariance matrix of the modal parameters is obtained based on formula (23) and formula (24); In formula (23), H represents the Hansen matrix, {λ i :i=1,...,n p }and They represent the ascending order of eigenvalues and eigenvectors of the Hansen matrix at the optimal value of the modal parameters; because λ1=...=λ m = 0, singular directions are ignored, and the posterior covariance matrix C is equal to the inverse of H, which is obtained by the following formula:
2. The modal parameter accuracy evaluation method according to claim 1, characterized in that: The modal parameters include natural frequency, damping ratio, power spectrum density of modal force, power spectrum density of prediction error, vibration mode and ratio of exciter mass to modal mass.
3. A system based on the modal parameter accuracy evaluation method according to claim 1, characterized in that: include: Optimization module: used to optimize the acquired modal parameters to obtain the optimal values of the modal parameters; Hansen matrix construction module: used to construct the Hansen matrix according to the optimal values of modal parameters; Calculation module: used to calculate the covariance matrix of modal parameters; Evaluation module: used to calculate the coefficient of variation of modal parameters and realize the accuracy evaluation of modal parameters; The optimization module includes running the following modules in sequence: Optimal value calculation module: optimizes based on formula (12) to obtain the optimal value of the modal parameter; Judgment module: judge whether the optimal value of the modal parameter converges. If so, run the Hansen matrix construction module, otherwise return to run the optimal value calculation module; In the Hansen matrix construction module, the optimal values of the modal parameters are substituted into formulas (26)-(39), and the Hansen matrix is constructed based on formulas (26)-(39); In the calculation module, the covariance matrix of the modal parameters is obtained based on formula (23) and formula (24).
4. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and the computer program is configured to implement the steps of the modal parameter accuracy evaluation method described in any one of claims 1 to 2 when called by a processor.
