Load identification method and system based on hybrid iterative regularization

Abstract

The present invention provides a load identification method and system based on hybrid iterative regularization, belonging to the field of load identification technology. A Green kernel function transfer matrix is ​​subjected to TSVD decomposition, and the decomposed singular values ​​are divided into larger singular values ​​and smaller singular values, wherein the larger singular values ​​are singular values ​​greater than a set value, and the smaller singular values ​​are singular values ​​less than or equal to the set value; a Gaussian high-pass filter function is used as a regularization matrix for Tikhonov regularization, and the smaller singular values ​​are corrected to obtain a TSVD-Tikhonov-Gaussian hybrid regularization result; and an LSQR algorithm is used to solve the TSVD-Tikhonov-Gaussian hybrid regularization result to achieve more accurate identification of different types of loads.

Description

Technical Field

[0001] The present invention relates to the technical field of load identification, and in particular to a load identification method and system based on hybrid iterative regularization. Background Art

[0002] The statements in this section merely provide background art related to the present invention and do not necessarily constitute prior art.

[0003] In practical engineering, obtaining accurate knowledge of structural input loads is crucial for analyzing structural reliability and safety, as well as improving structural design. Dynamic load identification is a typical inverse problem in engineering applications. However, in many cases, measurement noise and ill-posed system matrices are the largest sources of error in load identification techniques. Most studies have employed regularization methods to overcome the ill-posed nature of the system transfer matrix, but this approach still has limitations for complex systems and those with significant measurement noise. Therefore, improving the noise immunity and well-posedness of load identification techniques is crucial for accurate load identification.

[0004] The inventors have found that existing regularization methods are mainly divided into direct methods and iterative methods. Direct methods include Tikhonov regularization method and truncated singular value (TSVD). Wherein the essence of TSVD algorithm is to abandon a part of the smaller singular values ​​that are obviously affected by the error and then obtain the approximate solution of the original problem, but the TSVD algorithm also loses the resolution of the smaller singular values ​​to the solution estimate while abandoning the smaller singular values. Tikhonov regularization method mainly adds a filter factor to each item to correct the singular value to suppress the influence of noise on the solution, but it also causes larger singular values ​​to be corrected, so that the approximate solution deviates from the true solution and this method only reduces the condition number of the coefficient matrix, and still retains the morbid characteristics of the original transfer matrix. Whether it is TSVD or Tikhonov regularization method, there is the problem of over-determination, which will cause greater errors on certain peaks and is not suitable for load identification with sparsity such as impact loads. Iterative methods include Newton iteration method, Landweber iteration method and least squares QR (LSQR) iteration method. When the Newton iteration method is used to deal with inverse problems, multiple matrix inversion operations are involved in the iteration process. Therefore, it is easy to cause large errors due to the serious ill-posedness of the problem, which ultimately affects the accuracy of the dynamic load identification results. Summary of the Invention

[0005] In order to address the shortcomings of the existing technology, the present invention provides a load identification method and system based on hybrid iterative regularization. Based on the Green kernel function method, a load identification equation is established, and TSVD and Tikhonov-Gaussian regularization are combined to avoid the discarding of small singular values ​​by TSVD and the correction of large singular values ​​by Tikhonov regularization. The LSQR iterative method is used to solve large sparse matrices to achieve more accurate identification of different types of loads.

[0006] In order to achieve the above object, the present invention adopts the following technical solutions:

[0007] A first aspect of the present invention provides a load identification method based on hybrid iterative regularization.

[0008] A load identification method based on hybrid iterative regularization includes the following steps:

[0009] Obtain parameter data of the object to be identified by the payload;

[0010] According to the acquired parameter data, the Green kernel function transfer matrix is ​​established;

[0011] Perform TSVD decomposition on the Green kernel function transfer matrix and divide the decomposed singular values ​​into larger singular values ​​and smaller singular values. The larger singular values ​​are singular values ​​greater than the set value, and the smaller singular values ​​are singular values ​​less than or equal to the set value.

[0012] The Gaussian high-pass filter function is used as the regularization matrix of Tikhonov regularization to correct the small singular values ​​and obtain the TSVD-Tikhonov-Gaussian hybrid regularization result;

[0013] The LSQR algorithm is used to solve the TSVD-Tikhonov-Gaussian mixed regularization result and obtain the load identification result.

[0014] As an optional implementation, the relationship between the Green kernel function transfer matrix, the measured response, and the input load is:

[0015]

[0016] Among them, Y1···Y M are the dynamic responses of each measurement point, F1···F M is the load at different action positions, It represents the Green kernel function transfer matrix between the load position i and the measurement point, where M is less than or equal to N.

[0017] As an optional implementation method, the Gaussian high-pass filter function is used as the regularization matrix of Tikhonov regularization to obtain the improved Tikhonov-Gaussian filter function:

[0018]

[0019] Among them, λ is the regularization parameter, σ i Pass the singular values ​​of the matrix decomposition to the Green kernel function.

[0020] Furthermore, the smaller singular values ​​are corrected, including:

[0021] The hybrid filter factor is recorded as:

[0022]

[0023] Where diag represents a diagonal matrix,

[0024] Combined with the cutoff parameter r, if σ r >λ≥σ r+1 ,but:

[0025]

[0026] As an optional implementation, TSVD-Tikhonov-Gaussian hybrid regularization results include:

[0027]

[0028] Among them, λ is the regularization parameter, σ i is the singular value obtained after the Green kernel function transfer matrix decomposition, μ i is the singular value after the singular value decomposition of the regularized matrix, ν i is the right singular vector obtained after decomposition of the Green kernel function transfer matrix, Y is the measurement response, F λ is the input load under the regularization parameter λ.

[0029] As an optional implementation, the LSQR algorithm is used to solve the TSVD-Tikhonov-Gaussian hybrid regularization results to obtain the load identification results, including:

[0030] According to the condition number of the Green kernel function transfer matrix, select the appropriate transfer matrix and measurement response, perform initialization, Lanczos bidiagonalization and QR decomposition, and finally update and iterate;

[0031] The number of iterations starts from 1 and the optimal number of iterations is determined. The minimum relative error between the measured response value and the fitted value is used as the convergence criterion to obtain the optimal regularized solution.

[0032] A second aspect of the present invention provides a load identification system based on hybrid iterative regularization.

[0033] A load identification system based on hybrid iterative regularization, comprising:

[0034] The data acquisition module is configured to: acquire parameter data of the object to be recognized by the payload;

[0035] The transfer matrix building module is configured to: build a Green kernel function transfer matrix based on the acquired parameter data;

[0036] The transfer matrix decomposition module is configured to: perform TSVD decomposition on the Green kernel function transfer matrix, and divide the decomposed singular values ​​into larger singular values ​​and smaller singular values, wherein the larger singular values ​​are singular values ​​greater than a set value, and the smaller singular values ​​are singular values ​​less than or equal to the set value;

[0037] The hybrid regularization module is configured to use a Gaussian high-pass filter function as the regularization matrix of Tikhonov regularization to correct small singular values ​​and obtain a TSVD-Tikhonov-Gaussian hybrid regularization result;

[0038] The load identification module is configured to use the LSQR algorithm to solve the TSVD-Tikhonov-Gaussian hybrid regularization result to obtain the load identification result.

[0039] A third aspect of the present invention provides a computer-readable storage medium having a program stored thereon, which, when executed by a processor, implements the steps of the load identification method based on hybrid iterative regularization as described in the first aspect of the present invention.

[0040] A fourth aspect of the present invention provides an electronic device, comprising a memory, a processor, and a program stored in the memory and executable on the processor. When the processor executes the program, the steps of the load identification method based on hybrid iterative regularization as described in the first aspect of the present invention are implemented.

[0041] Compared with the prior art, the present invention has the following beneficial effects:

[0042] 1. The load identification method and system based on hybrid iterative regularization described in the present invention establish a load identification equation based on the Green kernel function method, combining TSVD and Tikhonov-Gaussian regularization to avoid the discarding of small singular values ​​by TSVD and the correction of large singular values ​​by Tikhonov regularization. The LSQR iterative method is used to solve large sparse matrices to achieve more accurate identification of different types of loads.

[0043] 2. The load identification method and system based on hybrid iterative regularization described in the present invention solve the problem of loss of solution estimation caused by TSVD discarding small singular values ​​and Tikhonov regularization correcting large singular values.

[0044] 3. The load identification method and system based on hybrid iterative regularization described in the present invention establish a new TSVD-Tikhonov-LSQR hybrid iterative regularization algorithm, which is not only suitable for the identification of sinusoidal loads and random loads, but also for the identification of loads with sparse properties such as impact loads; at the same time, the new hybrid iterative regularization algorithm is not limited to the identification of inverse load problems, but also includes inverse analysis such as image processing and temperature field reconstruction, and has broad prospects and high engineering application value.

[0045] Advantages of additional aspects of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] The accompanying drawings, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0047] Figure 1 A schematic flow chart of a load identification method based on a TSVD-Tikhonov-LSQR hybrid iterative regularization method provided in Example 1 of the present invention.

[0048] Figure 2 A schematic diagram of the Green kernel function response provided in Example 1 of the present invention.

[0049] Figure 3 Schematic diagram of the LSQR hybrid iterative regularization process provided in Example 1 of the present invention. DETAILED DESCRIPTION

[0050] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0051] It should be noted that the following detailed descriptions are exemplary and intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which the present invention belongs.

[0052] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, they indicate the presence of features, steps, operations, devices, components and / or combinations thereof.

[0053] In the absence of conflict, the embodiments of the present invention and the features thereof may be combined with each other.

[0054] Example 1:

[0055] Embodiment 1 of the present invention provides a load identification method based on hybrid iterative regularization, including the following process:

[0056] S1: Establish a simply supported beam finite element model, apply a unit pulse load, obtain the kernel function response time history, establish the Green kernel function response matrix, and obtain the load identification model;

[0057] S2: Perform truncated singular value decomposition (TSVD) on the Green kernel function response matrix and use the L-curve method to determine the truncation parameter;

[0058] S3: Select the Gaussian high-pass filter function as the regularization matrix and perform Tikhonov-Gaussian regularization on the Green kernel function response matrix;

[0059] S4: According to steps 2 and 3, establish the TSVD-Tikhonov-Gaussian regularization algorithm;

[0060] S5: According to step 4, a TSVD-Tikhonov-LSQR hybrid regularization algorithm is established based on the LSQR iterative algorithm;

[0061] S6: Solve the load identification model according to the hybrid iterative regularization algorithm in step 5.

[0062] Specifically, S1 includes:

[0063] In this embodiment, a simply supported beam is taken as an example, a finite element model of a simply supported beam is established, a unit pulse load is applied to the load node of the simply supported beam structure, and the Green kernel function response time history between the load application point and the response measurement point is calculated, as shown in FIG. Figure 2 As shown, the Green kernel function response matrix is ​​established. The cantilever beam vibration differential equation is:

[0064]

[0065] Where: M, C, K represent the mass, damping and stiffness matrices of the system respectively, x(t) represents the acceleration response, velocity response, and displacement response of the system, respectively. F(t) is the input load of the system, and n is the number of degrees of freedom.

[0066] According to the characteristics of the pulse function, the integral of the product of the pulse function and any other function is equal to the function value at the pulse moment, which can be expressed as:

[0067]

[0068] Where f(τ) represents the external load.

[0069] Assume that the Green function from the load application point to the response measurement point is the response of the system under the action of unit pulse load δ(t). According to the superposition principle, the relationship between the excitation and the system response can be written as a convolution form:

[0070]

[0071] Among them, H(t-τ) represents the Green kernel function of the structural impulse response, y(t) represents the dynamic response of the system, and the response data can be displacement, velocity, acceleration, etc.

[0072] For a system with zero initial conditions, the sampling time interval is Δt, the number of sampling points is m, and the convolution integral form of Equation (3) in the time domain is discretized into a linear equation system:

[0073]

[0074] The discrete form of Equation (4) can be given by the algebraic equation of the system

[0075] Y=HF (5)

[0076] Equation (5) represents the linear discrete equation for a single-input single-output system. i (i=1,2,...,M) The total response of the system acting at different positions i in the same time can be expressed as, at each known measurement point Y jSuperposition of the dynamic responses of (j=1,2,...,N).

[0077]

[0078] in, It represents the Green kernel function matrix between the load position i and the measurement point. Therefore, the equivalent form of formula (6) can be written as:

[0079]

[0080] Where M and N represent the number of load points and response measurement points, respectively, and it is necessary to ensure that M≤N. Equation (7) represents the linear discrete equation model of the multi-input multi-output system:

[0081] S2 specifically includes:

[0082] Considering that the small singular values ​​of the transfer matrix H will amplify the impact of the measurement response noise on the solution, which will lead to unstable solutions, the small singular value vectors obtained by decomposing the matrix H are removed by the truncated singular value method, and only the first r large singular values ​​are retained. In this way, the singular value decomposition form of the matrix H is changed to:

[0083]

[0084] Where: r represents the cutoff parameter of TSVD, which is determined by the L-curve method; u i σ i v i are the left singular vectors, singular values ​​and right singular vectors obtained after the singular value decomposition of the transfer matrix H.

[0085] Specifically, S3 includes:

[0086] (1) Tikhonov regularization

[0087] Solve Equation (5) by least squares to obtain the residual norm objective function:

[0088] J(H)=||Y-HF|| 2 (9)

[0089] Where: Represents the square of the 2-norm of a vector

[0090] Add a regularization term Ω(F) to constrain the solution F in equation (9):

[0091] J λ (H)=||Y-HF|| 2 +λ 2 Ω(F) (10)

[0092] Where λ is the regularization parameter. Regularization term Ω(F) = ||RF|| 2 (R is the regularization matrix), Equation (10) is Tikhonov regularization, and its solution is

[0093] F λ =(H T H+λ 2 R T R) -1 H T Y (11)

[0094] Using the matrix to decompose the generalized singular values ​​of (H, R), we can get

[0095]

[0096] Where μ i is the singular value after the singular value decomposition of the regularized matrix R, is the Tikhonov regularized filter function.

[0097] (2) Regularization matrix R

[0098] The Gaussian high-pass filter function is used as the regularization matrix of Tikhonov regularization, and its specific form is:

[0099]

[0100] Substituting equation (13) into equation (11), we get the improved Tikhonov-Gaussian filter function

[0101]

[0102] S4 specifically includes:

[0103] The TSVD in S2 retains only the first r large singular values ​​and discards the smaller ones; the Tikhonov-Gaussian regularization in S3 corrects all singular values. We now improve TSVD by splitting the transfer matrix H into two parts: the larger singular values ​​and the smaller ones. Based on S3, the smaller singular values ​​are corrected using Gaussian filtering.

[0104] The hybrid filter factor is recorded as:

[0105]

[0106] Where diag represents a diagonal matrix,

[0107] Combined with the truncation parameter r in S2, if σr >λ≥σ r+1 ,but

[0108]

[0109] Formula (16) not only retains the smaller singular values, but also only modifies the smaller singular values.

[0110] The solution of TSVD-Tikhonov-Gaussian mixed regularization is:

[0111]

[0112] TSVD-Tikhonov-Gaussian mixed regularization is:

[0113]

[0114] S5 specifically includes:

[0115] (1) LSQR algorithm

[0116] The LSQR method is a method particularly suitable for solving large, sparse matrix linear equations. It converts any sparse matrix equation into an equation with a square coefficient matrix, then uses the Lanczos diagonalization algorithm to build a lower diagonal matrix and solve the equation with the least squares solution:

[0117] The least squares solution of formula (9) is:

[0118] H T HF=H T Y (19)

[0119] Formula (19) can also be written as:

[0120]

[0121] Where r = Y-HF, the initial value is

[00] T Then, the Lanczos method is used to construct the Krylov subspace K(H,Y0)=span{Y0,HY0,...,H m-1 Y0}, where Y0 = [Y,0] T .

[0122] Through direct calculation, we can know that:

[0123]

[0124] In addition, two sets of unit vectors {u k},{w k} represents this set of orthogonal bases:

[0125]

[0126] Let β = ||Y||2, then formula (20), the Lanczos iteration process can be expressed as:

[0127]

[0128] Among them, β k >0 and α k+1 The choice of >0 is to ensure that ||u k+1 ||2=||w k+1 ||2=1, that is:

[0129] β k =||Hw k -α k u k ||2,α k+1 =||H T u k+1 -β k w k ||2,Assume U m =[u1,u2,...,u m ],W m =[w1,w2,...,w m ],

[0130] The iterative process is a bidiagonalization process:

[0131]

[0132] Among them, e m+1 =[0,0,...,1] T ∈R m+1 , and matrix The column vector group of K 2m+1 A basis of (H,Y0).

[0133] Next, in the subspace K 2m+1 Find the approximate solution of formula (20) in (H, Y0), which is denoted as So that the residual norm of the original system of equations Reach minimum.

[0134] First, Written as From formula (22), we can see

[0135]

[0136] Assume T mThe Lanczos method is applied to the normal equation H T HF=H T The symmetric tridiagonal matrix obtained on Y is because Therefore, the norm of formula (23) is:

[0137]

[0138] Therefore, minimizing the residual norm is equivalent to solving the following least squares problem

[0139]

[0140] (2)TSVD-Tikhonov-LSQR hybrid regularization algorithm

[0141] The TSVD-Tikhonov-Gaussian hybrid regularization algorithm (18) in S4 can be abbreviated as

[0142] H1F λ =Y1 (26)

[0143] in, Y1=H T Y. The above LSQR algorithm is used to solve Equation (26). Combining the TSVD-Tikhonov-Gaussian mixed regularization with the LSQR algorithm constitutes the TSVD-Tikhonov-LSQR mixed iterative regularization algorithm for solving ill-posed problems.

[0144] S6 specifically includes:

[0145] First, input the transfer matrix H and measurement response Y in step 1; select the appropriate transfer matrix and measurement response based on the condition number of H; then perform initialization, Lanczos bidiagonalization, and QR decomposition according to step 5; finally, update and iterate.

[0146] The number of iterations k starts from 1 and determines the optimal number of iterations k op , based on the difference between the measured response value Y and the fitted value Y k The minimum relative error between is used as the convergence criterion to obtain the optimal regular solution. Figure 3 .

[0147]

[0148] Example 2:

[0149] Embodiment 2 of the present invention provides a load identification system based on hybrid iterative regularization, comprising:

[0150] The data acquisition module is configured to: acquire parameter data of the object to be recognized by the payload;

[0151] The transfer matrix building module is configured to: build a Green kernel function transfer matrix based on the acquired parameter data;

[0152] The transfer matrix decomposition module is configured to: perform TSVD decomposition on the Green kernel function transfer matrix, and divide the decomposed singular values ​​into larger singular values ​​and smaller singular values, wherein the larger singular values ​​are singular values ​​greater than a set value, and the smaller singular values ​​are singular values ​​less than or equal to the set value;

[0153] The hybrid regularization module is configured to use a Gaussian high-pass filter function as the regularization matrix of Tikhonov regularization to correct small singular values ​​and obtain a TSVD-Tikhonov-Gaussian hybrid regularization result;

[0154] The load identification module is configured to use the LSQR algorithm to solve the TSVD-Tikhonov-Gaussian mixed regularization result to obtain the load identification result.

[0155] The working method of the system is the same as the load identification method based on hybrid iterative regularization provided in Example 1, and will not be repeated here.

[0156] Example 3:

[0157] Embodiment 3 of the present invention provides a computer-readable storage medium having a program stored thereon. When the program is executed by a processor, the program implements the steps of the load identification method based on hybrid iterative regularization as described in Embodiment 1 of the present invention.

[0158] Example 4:

[0159] Embodiment 4 of the present invention provides an electronic device, including a memory, a processor, and a program stored in the memory and executable on the processor. When the processor executes the program, the steps of the load identification method based on hybrid iterative regularization as described in Embodiment 1 of the present invention are implemented.

[0160] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of hardware embodiments, software embodiments, or embodiments combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage and optical storage, etc.) containing computer-usable program code.

[0161] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0162] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0163] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0164] Those skilled in the art will appreciate that all or part of the processes in the above-described method embodiments can be implemented by instructing the relevant hardware through a computer program. The program can be stored in a computer-readable storage medium, and when executed, the program can include the processes in the above-described method embodiments. The storage medium can be a magnetic disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).

[0165] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art will readily appreciate that various modifications and variations of the present invention are possible. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention are intended to be within the scope of protection of the present invention.

Claims

1. A load identification method based on hybrid iterative regularization, characterized by: The following processes are included: Obtain parameter data of the object to be identified by the payload; According to the acquired parameter data, the Green kernel function transfer matrix is ​​established; Perform TSVD decomposition on the Green kernel function transfer matrix and divide the decomposed singular values ​​into larger singular values ​​and smaller singular values. The larger singular values ​​are singular values ​​greater than the set value, and the smaller singular values ​​are singular values ​​less than or equal to the set value. The Gaussian high-pass filter function is used as the regularization matrix of Tikhonov regularization to correct the small singular values ​​and obtain the TSVD-Tikhonov-Gaussian hybrid regularization result; The LSQR algorithm is used to solve the TSVD-Tikhonov-Gaussian mixed regularization results and obtain the load identification results; The Gaussian high-pass filter function is used as the regularization matrix of Tikhonov regularization to obtain the improved Tikhonov-Gaussian filter function: Among them, λ is the regularization parameter, σ i Transfer the singular values ​​of matrix decomposition to Green kernel function; Corrections for smaller singular values ​​include: The hybrid filter factor is recorded as: in, represents a diagonal matrix, Combined truncation parameters ,like ,but: 。 2. The load identification method based on hybrid iterative regularization according to claim 1, characterized in that: The relationship between the Green kernel function transfer matrix, measurement response and input load is: Among them, Y1···Y M are the dynamic responses of each measurement point, F1···F M is the load at different action positions, Indicates the load acting position and the Green kernel function transfer matrix between the measurement points, M is less than or equal to N.

3. The load identification method based on hybrid iterative regularization according to claim 1, characterized in that: TSVD-Tikhonov-Gaussian hybrid regularization results, including: in, λ is the regularization parameter, σ i The singular values ​​obtained after the Green kernel function transfer matrix decomposition, μ i is the singular value after the singular value decomposition of the regularized matrix, ν i is the right singular vector obtained after the Green kernel function transfer matrix decomposition, Y To measure the response, F λ is the regularization parameter λ The input load is .

4. The load identification method based on hybrid iterative regularization according to claim 1, characterized in that: The LSQR algorithm is used to solve the TSVD-Tikhonov-Gaussian mixed regularization results and obtain the load identification results, including: According to the condition number of the Green kernel function transfer matrix, the appropriate transfer matrix and measurement response are selected, and initialization, Lanczos bidiagonalization and QR decomposition are performed, and finally update iterations are performed; The number of iterations starts from 1 and the optimal number of iterations is determined. The minimum relative error between the measured response value and the fitted value is used as the convergence criterion to obtain the optimal regularized solution.

5. A load identification system based on hybrid iterative regularization, using the load identification method based on hybrid iterative regularization according to any one of claims 1 to 4, characterized in that: include: The data acquisition module is configured to: acquire parameter data of the object to be recognized by the payload; The transfer matrix building module is configured to: build a Green kernel function transfer matrix based on the acquired parameter data; The transfer matrix decomposition module is configured to: perform TSVD decomposition on the Green kernel function transfer matrix, and divide the decomposed singular values ​​into larger singular values ​​and smaller singular values, wherein the larger singular value is a singular value greater than a set value, and the smaller singular value is a singular value less than or equal to the set value; The hybrid regularization module is configured to use a Gaussian high-pass filter function as the regularization matrix of Tikhonov regularization to correct small singular values ​​and obtain a TSVD-Tikhonov-Gaussian hybrid regularization result; The load identification module is configured to use the LSQR algorithm to solve the TSVD-Tikhonov-Gaussian mixed regularization result to obtain the load identification result.

6. The load identification system based on hybrid iterative regularization according to claim 5, characterized in that: TSVD-Tikhonov-Gaussian hybrid regularization results, including: in, λ is the regularization parameter, σ i The singular values ​​obtained after the Green kernel function transfer matrix decomposition, μ i is the singular value after the singular value decomposition of the regularized matrix, ν i is the right singular vector obtained after the Green kernel function transfer matrix decomposition, Y To measure the response, F λ is the regularization parameter λ The input load is .

7. A computer-readable storage medium having a program stored thereon, characterized in that: When the program is executed by a processor, the steps of the load identification method based on hybrid iterative regularization according to any one of claims 1 to 4 are implemented.

8. An electronic device comprising a memory, a processor, and a program stored in the memory and executable on the processor, wherein: When the processor executes the program, the steps of the load identification method based on hybrid iterative regularization are implemented as described in any one of claims 1 to 4.

Images