A non-probabilistic sensitivity analysis method for high-power wind turbine gearboxes

Abstract

The application provides a high-power wind power gear box non-probability sensitivity analysis method, belongs to the technical field of gear box sensitivity analysis, obtains displacement response sequences corresponding to each group of structure parameters, obtains parameter non-probability covariances of each structure parameter from non-probability variances of each parameter and non-probability correlation coefficients of each parameter, and then obtains response non-probability variances and non-probability correlation coefficient matrices of each displacement response data; based on contributions of non-probability variances of each parameter to each normalized response non-probability variance, contributions of non-probability covariances of each parameter to each normalized response non-probability variance, and the non-probability correlation coefficient matrix, the independent sensitivity index and the correlation sensitivity index of each structure parameter under the displacement response sequence are calculated, and then the total sensitivity index corresponding to the structure parameter is obtained, each total sensitivity index is arranged in ascending order or descending order, and the sensitivity degree sequence of each structure parameter to the load sharing characteristic is obtained, so that the optimization design direction of the load sharing characteristic of the wind power gear box is provided, and the optimization efficiency is improved.

Description

Technical Field

[0001] This invention relates to the field of sensitivity analysis technology for wind turbine gearboxes, and specifically to a nonprobabilistic sensitivity analysis method for high-power wind turbine gearboxes. Background Technology

[0002] With the development of the energy industry, the installed capacity of wind power is constantly increasing. At the same time, the industry has put forward greater challenges to the stability, reliability and lifespan of wind power systems. Among the many wind power system failures, most are caused by the failure of the gearbox transmission system. The load sharing characteristics of the gearbox can well measure its transmission performance, but there are many structural parameters that affect the load sharing characteristics of the gearbox, and the degree of influence is also different.

[0003] In existing technologies, global sensitivity analysis based on probability variance requires a large number of samples to obtain accurate results. Sensitivity analysis considering parameter correlation requires converting the parameter distribution into a standard normal distribution and performing a decorrelation process through orthogonalization, which is a cumbersome process. Therefore, for wind turbine gearboxes, considering the uncertainty and correlation of gearbox structural parameters under small sample conditions and clarifying the degree of sensitivity of structural parameters to their load-sharing characteristics has become an urgent problem to be solved. Summary of the Invention

[0004] In view of the above-mentioned defects or deficiencies in the prior art, the present invention aims to provide a non-probabilistic sensitivity analysis method for high-power wind turbine gearboxes, comprising the following steps: Multiple sets of structural parameters are input into a gearbox proxy model, which has multiple model nodes. The gearbox proxy model outputs displacement response sequences corresponding to each set of structural parameters, and each set of displacement response sequences includes displacement response data of each model node. The non-probability variance of each structural parameter is calculated based on the uncertainty of the structural parameters, and the non-probability correlation coefficient between each pair of structural parameters is calculated based on the correlation of the structural parameters. The non-probability covariance of each structural parameter is obtained from the non-probability variance and the non-probability correlation coefficient of each parameter, and then the response non-probability variance and response non-probability correlation coefficient of each displacement response data are obtained. The non-probability correlation coefficients of each response are assembled to obtain a non-probability correlation coefficient matrix; the displacement response data are normalized by the non-probability variance of the response to obtain normalized displacement response data, and then the normalized response non-probability variance of each normalized displacement response data is obtained. Based on the contribution of the non-probability variance of each parameter to the non-probability variance of each normalized response, and the non-probability correlation coefficient matrix, the independent sensitivity index of each structural parameter under the displacement response sequence is calculated. Based on the contribution of the non-probability covariance of each parameter to the non-probability variance of each normalized response, and the non-probability correlation coefficient matrix, the relevant sensitivity index of each structural parameter under the displacement response sequence is calculated. Calculate the total sensitivity index corresponding to each of the structural parameters, wherein the total sensitivity index is the sum of the independent sensitivity index and the related sensitivity index; Arrange the total sensitivity indices in ascending or descending order to obtain a sequence of the sensitivity of each structural parameter to the load-sharing characteristics.

[0005] According to the technical solution provided by the present invention, before calculating the non-probability variance of each structural parameter based on the uncertainty of the structural parameters, and before calculating the non-probability correlation coefficient between every two structural parameters based on the correlation of the structural parameters, the method includes at least the following steps: Based on the prior information of the structural parameters, a nonprobabilistic ellipsoidal model is used to measure multiple sets of samples of the structural parameters. The uncertainty of the structural parameters is obtained through formula (1), and the correlation of the structural parameters is obtained through formula (2). Formula (1) in, Represents structural parameters The interval, Represents structural parameters The lower bound Represents structural parameters The upper realm, and Representing structural parameters respectively The midpoint and radius of the interval, n Indicates the number of structural parameters; Formula (2) in, Represents structural parameters and structural parameters The parameter is the non-probability correlation coefficient. d , c These are the envelope structure parameters. Sample and structural parameters The length of the semi-axis of the smallest ellipse in the sample.

[0006] According to the technical solution provided by the present invention, after obtaining the non-probability covariance of each of the structural parameters, the response non-probability variance of each of the displacement response data is obtained through the following steps: The non-probability covariance matrix of the parameters is obtained by assembling the non-probability covariances of each parameter; The response nonprobabilistic variance of each displacement response data is obtained using formula (3): Formula (3) in, Displacement response data The response variance is not a probability. m Indicates the number of displacement response data. It involves performing the trace operation on a matrix. Displacement response data First-order partial derivatives of structural parameters, Displacement response data The Hessian matrix of the structural response function relating to structural parameters, where T denotes the transpose of the matrix. The parametric nonprobabilistic covariance matrix represents the structural parameters.

[0007] According to the technical solution provided by the present invention, the normalized displacement response data is calculated by formula (6): Formula (6) in, Displacement response data The normalized displacement response data.

[0008] According to the technical solution provided by the present invention, the contribution of the non-probability variance of each parameter to the non-probability variance of each normalized response is calculated by formula (4): Formula (4) in, Represents structural parameters The non-probability variance of the parameters affects the displacement response data. The contribution of the normalized response nonprobabilistic variance. Displacement response data Normalized displacement response data with respect to the structural response function of structural parameters, Displacement response data Normalized displacement response data for structural parameters The first-order partial derivative, Displacement response data Normalized displacement response data for structural parameters Structural parameters The second-order partial derivative, Represents structural parameters The parameter is not the probability variance. Represents structural parameters The parameter is the non-probability variance.

[0009] According to the technical solution provided by the present invention, a method for calculating the relevant sensitivity index of each structural parameter under the displacement response sequence includes at least the following steps: The contribution of the parametric nonprobability covariance of each of the structural parameters to the nonprobability variance of each of the normalized responses is calculated using formula (5): Formula (5) in, Represents structural parameters The nonprobabilistic covariance of the parameters affects the displacement response data. The contribution of the normalized response nonprobabilistic variance. Displacement response data Normalized displacement response data with respect to the structural response function of structural parameters, Displacement response data Normalized displacement response data for structural parameters The first-order partial derivative, Displacement response data Normalized displacement response data for structural parameters Structural parameters The second-order partial derivative, Displacement response data Normalized displacement response data for structural parameters Structural parameters The second-order partial derivative, Displacement response data Normalized displacement response data for structural parameters Structural parameters The second-order partial derivative, Represents structural parameters The parameter is not the probability variance. Represents structural parameters Structural parameters The parameter is the non-probability covariance. Represents structural parameters Structural parameters The parameter is the non-probabilistic covariance; The independent sensitivity indices of each structural parameter under the normalized displacement response data are calculated using formula (7): Formula (7) in, Represents structural parameters Independent sensitivity indicators, Displacement response data The normalized response non-probability variance This represents the non-probability correlation coefficient matrix. Represents structural parameters The projection of the non-probability variance of the parameters onto the non-probability variance of the normalized response. The definition is as follows:

[0010] in, Structural parameters The non-probability variance of the parameters affects the displacement response data. The contribution of the normalized response nonprobabilistic variance.

[0011] According to the technical solution provided by the present invention, the relevant sensitivity index of each structural parameter under the displacement response sequence is calculated by formula (8), which is as follows: Formula (8) in, Represents structural parameters Relevant sensitivity indicators, Represents structural parameters The projection of the nonprobability covariance of the parameters onto the nonprobability variance of the normalized response. The definition is as follows:

[0012] in, Structural parameters The nonprobabilistic covariance of the parameters affects the displacement response data. The contribution of the normalized response nonprobabilistic variance.

[0013] According to the technical solution provided by the present invention, the gearbox proxy model is constructed through the following steps: Obtain the finite element model of the gearbox and multiple sets of structural parameter samples, and set multiple finite element model nodes at the root of a meshing tooth in the internal gear ring of the gearbox finite element model. Dynamic loads are applied to the gearbox finite element model to obtain displacement response data samples at each node of the finite element model under the action of each set of structural parameter samples. The displacement response data samples at each node of the finite element model form a displacement response sequence sample. Using the structural parameter samples and displacement response data samples, a gearbox proxy model is constructed to characterize the data relationship between the structural parameters and displacement response data.

[0014] According to the technical solution provided by the present invention, after constructing the gearbox proxy model for characterizing the data relationship between the structural parameters and displacement response data, the following steps are included: Calculate the load-sharing coefficient of the gearbox proxy model; Obtain the gearbox entity corresponding to the gearbox proxy model, and attach a uniaxial strain gauge at the root of a meshing tooth in the internal gear ring of the gearbox entity; A dynamic load is applied to the gearbox entity, causing deformation displacement at the root of one meshing tooth in the internal gear ring of the gearbox entity. The uniaxial strain gauge converts the displacement response data of each entity node corresponding to each model node into a strain signal. The actual load-averaging coefficient is calculated through the strain signal. The actual load-averaging coefficient is used to verify the accuracy of the load-averaging coefficient of the gearbox proxy model.

[0015] According to the technical solution provided by the present invention, the calculation of the actual load-equalizing factor through the strain signal includes the following steps: The strain signal is filtered to remove interference peaks, resulting in multiple effective peak values; The strain signal was obtained at the first p Calculate the mean of the peak values ​​for each of the effective peak values. The actual load factor is calculated using formula (9). ; Formula (9) in, p =1+ q , q =0,1,2,3…, This indicates the number of planetary gears meshing with the gear under test in the gearbox entity. r This indicates the number of teeth on the gear to be tested in the gearbox entity. h This represents the number of solid nodes for each tooth of the gear under test, obtained by analyzing the number of nodes for the first tooth. e tooth f Peak mean of each node Summation yields , max means taking the maximum value.

[0016] The beneficial effects of this scheme are as follows: This method can clarify the sensitivity of each structural parameter to the load-sharing characteristics of wind turbine gearboxes under small sample conditions, based on the uncertainty and correlation of structural parameters. This makes it easier for engineers to select suitable structural parameters for analysis when performing sensitivity analysis on the load-sharing characteristics of wind turbine gearboxes, simplifying the analysis process. It can also determine sensitive parameters for the optimized design of the load-sharing characteristics of wind turbine gearboxes, thereby improving optimization efficiency. Attached Figure Description

[0017] Figure 1 The flowchart illustrates the steps of the nonprobabilistic sensitivity analysis method for high-power wind turbine gearboxes provided by this invention. Detailed Implementation

[0018] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and not intended to limit it. Furthermore, it should be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings.

[0019] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0020] As mentioned in the background section, to address the problems in the existing technology, this invention proposes a non-probabilistic sensitivity analysis method for high-power wind turbine gearboxes. Please refer to [link / reference]. Figure 1 As shown, it includes the following steps: S100. Input multiple sets of structural parameters into the gearbox proxy model. The gearbox proxy model has multiple model nodes. The gearbox proxy model outputs the displacement response sequence corresponding to each set of structural parameters. Each set of displacement response sequences includes the displacement response data of each model node. S101. Calculate the non-probability variance of each of the structural parameters based on the uncertainty of the structural parameters, and calculate the non-probability correlation coefficient between each pair of structural parameters based on the correlation of the structural parameters; S102. From the non-probability variance of each parameter and the non-probability correlation coefficient of each parameter, the non-probability covariance of each structural parameter is obtained, and then the non-probability variance of the response and the non-probability correlation coefficient of each displacement response data are obtained. S103. Assemble the non-probability correlation coefficients of each response to obtain a non-probability correlation coefficient matrix; normalize the displacement response data through the non-probability variance of the response to obtain normalized displacement response data, and then obtain the normalized response non-probability variance of each normalized displacement response data. S104. Based on the contribution of the non-probability variance of each parameter to the non-probability variance of each normalized response, and the non-probability correlation coefficient matrix, calculate the independent sensitivity index of each structural parameter under the displacement response sequence. S105. Based on the contribution of the non-probability covariance of each parameter to the non-probability variance of each normalized response, and the non-probability correlation coefficient matrix, calculate the correlation sensitivity index of each structural parameter under the displacement response sequence. S106. Calculate the total sensitivity index corresponding to each of the structural parameters, wherein the total sensitivity index is the sum of the independent sensitivity index and the related sensitivity index; S107. Arrange the total sensitivity indices in ascending or descending order to obtain the sensitivity sequence of the structural parameters to the load-sharing characteristics for each group.

[0021] Specifically, multiple sets of structural parameters are input, and multiple sets of corresponding displacement response sequences are output to obtain the gearbox proxy model.

[0022] Specifically, the non-probability correlation coefficient matrix It is expressed as follows:

[0023] Here, an element of the matrix For example, this element represents displacement response data. With displacement response data The non-probabilistic correlation coefficient between the responses is calculated in the same way as other elements, and this element is obtained through formula (10): Formula (10) in, Displacement response data With displacement response data The non-probabilistic covariance of the responses between them Displacement response data The response variance is not a probability. Displacement response data First-order partial derivatives of structural parameters, Displacement response data The Hessian matrix of the structural response function relating structural parameters. Displacement response data The response variance is not a probability. Displacement response data First-order partial derivatives of structural parameters, Displacement response data The Hessian matrix of the structural response function with respect to structural parameters. This performs the trace operation on the matrix, and T represents the transpose operation on the matrix. The parametric nonprobabilistic covariance matrix represents the structural parameters.

[0024] Specifically, the total sensitivity index can represent the sensitivity of the structural parameter to the load sharing characteristics of the gearbox. The obtained sensitivity sequence of each structural parameter to the load sharing characteristics can be used by engineers in the later sensitivity analysis of the load sharing characteristics of wind turbine gearboxes to select suitable structural parameters for analysis, simplify the analysis process, and also determine sensitive parameters for the optimized design of the load sharing characteristics of wind turbine gearboxes, thereby improving optimization efficiency.

[0025] In a preferred embodiment, the gearbox proxy model is constructed through the following steps: Obtain the finite element model of the gearbox and multiple sets of structural parameter samples, and set multiple finite element model nodes at the root of a meshing tooth in the internal gear ring of the gearbox finite element model. Dynamic loads are applied to the gearbox finite element model to obtain displacement response data samples at each node of the finite element model under the action of each set of structural parameter samples. The displacement response data samples at each node of the finite element model form a displacement response sequence sample. By using multiple sets of structural parameter samples and displacement response data samples, a gearbox proxy model is constructed to characterize the data relationship between the structural parameters and displacement response data.

[0026] Specifically, based on the actual design parameters and rated operating conditions of the gearbox, a finite element model of the wind turbine gearbox is established, and the load sharing coefficient of the gearbox is calculated through the finite element model. The load sharing coefficient characterizes the load sharing characteristics of the gearbox.

[0027] Specifically, the structural parameters affecting the load sharing characteristics of the gearbox are determined based on the calculated load sharing coefficient. The nominal values ​​of each structural parameter are determined based on prior data. The uncertainty and correlation of the structural parameters are measured using a nonprobabilistic ellipsoidal model.

[0028] Specifically, the structural parameters affecting the load-sharing characteristics of a gearbox include the material parameters of the gear shafts in the primary transmission system and Young's modulus. ,density Compared to Poisson Their nominal values ​​are respectively MPa, 7850 kg / m 3 The material parameters of the internal gear ring are elastic modulus, density, and Poisson's ratio, with nominal values ​​of 0.3 and 0.3, respectively. MPa, 7800 kg / m 3 0.3, ideal transmission backlash b The value is 1.05 mm, and the corresponding uncertainty intervals are as follows: MPa kg / m 3 , , MPa kg / m 3 , , mm.

[0029] Specifically, the number of model nodes is 33, and correspondingly, the number of response non-probability variances is also 33.

[0030] Specifically, the finite element model of the wind turbine gearbox can be modeled using Ansys, Romax, and Abaqus software. The structural parameters can be measured using interval models, or, if there is a correlation, using multidimensional ellipsoidal models, etc.

[0031] In a preferred embodiment, after constructing the gearbox proxy model for characterizing the data relationship between the structural parameters and the displacement response data, the following steps are included: Calculate the load-sharing coefficient of the gearbox proxy model; Obtain the gearbox entity corresponding to the gearbox proxy model, and attach a uniaxial strain gauge at the root of a meshing tooth in the internal gear ring of the gearbox entity; A dynamic load is applied to the gearbox entity, causing deformation displacement at the root of one meshing tooth in the internal gear ring of the gearbox entity. The uniaxial strain gauge converts the displacement response data of each entity node corresponding to each model node into a strain signal. The actual load-averaging coefficient is calculated through the strain signal. The actual load-averaging coefficient is used to verify the accuracy of the load-averaging coefficient of the gearbox proxy model.

[0032] In a preferred embodiment, calculating the actual load-equalizing factor using the strain signal includes the following steps: The strain signal is filtered to remove interference peaks, resulting in multiple effective peak values; The strain signal was obtained at the first p Calculate the mean of the peak values ​​for each of the effective peak values. The actual load factor is calculated using formula (9); Formula (9) in, p =1+ q , q =0,1,2,3…, This indicates the number of planetary gears meshing with the gear under test in the gearbox entity. r This indicates the number of teeth on the gear to be tested in the gearbox entity. h This represents the number of solid nodes for each tooth of the gear under test, obtained by analyzing the number of nodes for the first tooth. e tooth f Peak mean of each node Summation yields , max means taking the maximum value.

[0033] Specifically, the strain signals are measured at 33 nodes at the root of a meshing tooth in the gear ring inside the gearbox, and the strain signals at each node are processed.

[0034] Specifically, based on the structural parameter samples and displacement response data samples, the gearbox proxy model is constructed using the optimal chaotic polynomial.

[0035] Specifically, the highest order is The optimal chaotic polynomial of order is shown in equation (11): Formula (11) in, M This indicates the total number of items, specifically: , This indicates that the polynomial's nth... z The coefficient of each sub-item , Represents structural parameters The corresponding number Gegenbauer sub-items, .

[0036] In a preferred embodiment, before calculating the non-probability variance of each structural parameter based on the uncertainty of the structural parameters, and before calculating the non-probability correlation coefficient between every two structural parameters based on the correlation of the structural parameters, at least the following steps are included: Based on the prior information of the structural parameters, a nonprobabilistic ellipsoidal model is used to measure multiple sets of samples of the structural parameters. The uncertainty of the structural parameters is obtained through formula (1), and the correlation of the structural parameters is obtained through formula (2). Formula (1) in, Represents structural parameters The interval, Represents structural parameters The lower bound Represents structural parameters The upper realm, and Representing structural parameters respectively The midpoint and radius of the interval, where n represents the number of structural parameters; Formula (2) in, Represents structural parameters and structural parameters The parameter is the non-probability correlation coefficient. d , c These are the envelope structure parameters. Sample and structural parameters The length of the semi-axis of the smallest ellipse in the sample.

[0037] In a preferred embodiment, after obtaining the parametric nonprobability covariance of each of the structural parameters, the response nonprobability variance of each of the displacement response data is obtained through the following steps: The non-probability covariance matrix of the parameters is obtained by assembling the non-probability covariances of each parameter; The response nonprobabilistic variance of each displacement response data is obtained using formula (3): Formula (3) in, Displacement response data The response variance is not a probability. m Indicates the number of displacement response data. It involves performing the trace operation on a matrix. Displacement response data First-order partial derivatives of structural parameters, Displacement response data The Hessian matrix of the structural response function relating to structural parameters, where T denotes the transpose of the matrix. The parametric nonprobabilistic covariance matrix represents the structural parameters.

[0038] The nonprobabilistic covariance matrix of structural parameters It is expressed as follows:

[0039] Example of assembling a parametric nonprobabilistic covariance matrix: the off-diagonal elements in the 1st row and 2nd column. Represents structural parameters With structural parameters The parameter is the non-probability covariance. Represents structural parameters and The parameter is the non-probability correlation coefficient, and the diagonal element in the second row and second column is... Represents structural parameters The parameter is not the probability variance. Represents structural parameters The radius of the interval, Represents structural parameters The radius of the interval.

[0040] In a preferred embodiment, the contribution of the non-probability variance of each parameter to the non-probability variance of each normalized response is calculated using formula (4): Formula (4) in, Represents structural parameters The non-probability variance of the parameters affects the displacement response data. The contribution of the normalized response nonprobabilistic variance. Displacement response data Normalized displacement response data with respect to the structural response function of structural parameters, Displacement response data Normalized displacement response data for structural parameters The first-order partial derivative, Displacement response data Normalized displacement response data for structural parameters Structural parameters The second-order partial derivative, Represents structural parameters The parameter is not the probability variance. Represents structural parameters The parameter is the non-probability variance.

[0041] In a preferred embodiment, the method for calculating the relevant sensitivity index of each structural parameter under the displacement response sequence includes at least the following steps: The contribution of the parametric nonprobability covariance of each of the structural parameters to the nonprobability variance of each of the normalized responses is calculated using formula (5): Formula (5) in, Represents structural parameters The nonprobabilistic covariance of the parameters affects the displacement response data. The contribution of the normalized response nonprobabilistic variance. Displacement response data Normalized displacement response data with respect to the structural response function of structural parameters, Displacement response data Normalized displacement response data for structural parameters The first-order partial derivative, Displacement response data Normalized displacement response data for structural parameters Structural parameters The second-order partial derivative, Displacement response data Normalized displacement response data for structural parameters Structural parameters The second-order partial derivative, Displacement response data Normalized displacement response data for structural parameters Structural parameters The second-order partial derivative, Represents structural parameters The parameter is not the probability variance. Represents structural parameters Structural parameters The parameter is the non-probability covariance. Represents structural parameters Structural parameters The parameter is the non-probabilistic covariance; The independent sensitivity indices of each structural parameter under the normalized displacement response data are calculated using formula (7): Formula (7) in, Represents structural parameters Independent sensitivity indicators, Displacement response data The normalized response non-probability variance This represents the non-probability correlation coefficient matrix. Represents structural parameters The projection of the non-probability variance of the parameters onto the non-probability variance of the normalized response. The definition is as follows:

[0042] in, Structural parameters The non-probability variance of the parameters affects the displacement response data. The contribution of the normalized response nonprobabilistic variance.

[0043] In a preferred embodiment, the normalized displacement response data is calculated using formula (6): Formula (6) in, Displacement response data The normalized displacement response data.

[0044] In a preferred embodiment, the relevant sensitivity index of each structural parameter under the displacement response sequence is calculated by formula (8), as follows: Formula (8) in, Represents structural parameters Relevant sensitivity indicators, Represents structural parameters The projection of the nonprobability covariance of the parameters onto the nonprobability variance of the normalized response. The definition is as follows:

[0045] in, Structural parameters The nonprobabilistic covariance of the parameters affects the displacement response data. The contribution of the normalized response nonprobabilistic variance.

[0046] This invention targets the nonlinear complex system of wind turbine gearboxes. Under small sample conditions, it accurately evaluates the impact of the uncertainty and correlation of gearbox structural parameters on the load-sharing characteristics, obtains the sensitivity sequence of each structural parameter to the load-sharing characteristics, provides direction for the subsequent optimization design of the load-sharing characteristics of wind turbine gearboxes, and improves optimization efficiency.

[0047] This article uses specific examples to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. The above descriptions are only preferred embodiments of the present invention. It should be noted that due to the limitations of textual expression, and the objective existence of infinite specific structures, those skilled in the art can make several improvements, modifications, or changes without departing from the principles of the present invention, and can also combine the above technical features in an appropriate manner; these improvements, modifications, changes, or combinations, or the direct application of the inventive concept and technical solution to other situations without modification, should all be considered within the scope of protection of the present invention.

Claims

1. A nonprobabilistic sensitivity analysis method for high-power wind turbine gearboxes, characterized in that, Includes the following steps: Multiple sets of structural parameters are input into a gearbox proxy model, which has multiple model nodes. The gearbox proxy model outputs displacement response sequences corresponding to each set of structural parameters, and each set of displacement response sequences includes displacement response data of each model node. The non-probability variance of each structural parameter is calculated based on the uncertainty of the structural parameters, and the non-probability correlation coefficient between any two structural parameters is calculated based on the correlation of the structural parameters. The non-probability covariance of each structural parameter is obtained from the non-probability variance and the non-probability correlation coefficient of each parameter, and then the response non-probability variance and response non-probability correlation coefficient of each displacement response data are obtained. The non-probability correlation coefficients of each response are assembled to obtain a non-probability correlation coefficient matrix; the displacement response data are normalized by the non-probability variance of the response to obtain normalized displacement response data, and then the normalized response non-probability variance of each normalized displacement response data is obtained. Based on the contribution of the non-probability variance of each parameter to the non-probability variance of each normalized response, and the non-probability correlation coefficient matrix, the independent sensitivity index of each structural parameter under the displacement response sequence is calculated. Based on the contribution of the non-probability covariance of each parameter to the non-probability variance of each normalized response, and the non-probability correlation coefficient matrix, the correlation sensitivity index of each structural parameter under the displacement response sequence is calculated. Calculate the total sensitivity index corresponding to each of the structural parameters, wherein the total sensitivity index is the sum of the independent sensitivity index and the related sensitivity index; Arrange the total sensitivity indices in ascending or descending order to obtain a sequence of the sensitivity of each structural parameter to the load-sharing characteristics.

2. The non-probabilistic sensitivity analysis method for high-power wind turbine gearboxes according to claim 1, characterized in that, Before calculating the non-probability variance of each structural parameter based on the uncertainty of the structural parameters, and before calculating the non-probability correlation coefficient between any two structural parameters based on the correlation of the structural parameters, the process includes at least the following steps: Based on the prior information of the structural parameters, a nonprobabilistic ellipsoidal model is used to measure multiple sets of samples of the structural parameters. The uncertainty of the structural parameters is obtained by formula (1), and the correlation of the structural parameters is obtained by formula (2). Official (1) in, Represents structural parameters The interval, Represents structural parameters The lower bound Represents structural parameters The upper realm, and Representing structural parameters respectively The midpoint and radius of the interval, n Indicates the number of structural parameters; Official (2) in, Represents structural parameters and structural parameters The parameter is the non-probability correlation coefficient. d , c These are the envelope structure parameters. Sample and structural parameters The length of the semi-axis of the smallest ellipse in the sample.

3. The non-probabilistic sensitivity analysis method for high-power wind turbine gearboxes according to claim 2, characterized in that, After obtaining the parametric nonprobabilistic covariance of each of the structural parameters, the response nonprobabilistic variance of each of the displacement response data is obtained through the following steps: The non-probability covariance matrix of the parameters is obtained by assembling the non-probability covariances of each parameter; The response nonprobability variance of each displacement response data is obtained using formula (3): Official (3) in, Displacement response data The response non-probability variance m Indicates the number of displacement response data. It involves performing the trace operation on a matrix. Displacement response data First-order partial derivatives of structural parameters, Displacement response data The Hessian matrix of the structural response function relating to structural parameters, where T denotes the transpose of the matrix. The parametric nonprobabilistic covariance matrix represents the structural parameters.

4. The non-probabilistic sensitivity analysis method for high-power wind turbine gearboxes according to claim 3, characterized in that, The normalized displacement response data are calculated using formula (6): Official (6) in, Displacement response data The normalized displacement response data.

5. The nonprobabilistic sensitivity analysis method for high-power wind turbine gearboxes according to claim 4, characterized in that, The contribution of the non-probability variance of each parameter to the non-probability variance of each normalized response is calculated using formula (4): Official (4) in, Represents structural parameters The non-probability variance of the parameters affects the displacement response data. The contribution of the normalized response nonprobabilistic variance. Displacement response data Normalized displacement response data with respect to the structural response function of structural parameters, Displacement response data Normalized displacement response data for structural parameters The first-order partial derivative, Displacement response data Normalized displacement response data for structural parameters Structural parameters The second-order partial derivative, Represents structural parameters The parameter is not the probability variance. Represents structural parameters The parameter is the non-probability variance.

6. The nonprobabilistic sensitivity analysis method for high-power wind turbine gearboxes according to claim 4, characterized in that, A method for calculating the relevant sensitivity indices of each structural parameter under the displacement response sequence includes at least the following steps: The contribution of the parametric nonprobability covariance of each structural parameter to the nonprobability variance of each normalized response is calculated using formula (5): Official (5) in, Represents structural parameters The nonprobabilistic covariance of the parameters affects the displacement response data. The contribution of the normalized response nonprobabilistic variance. Displacement response data Normalized displacement response data with respect to the structural response function of structural parameters, Displacement response data Normalized displacement response data for structural parameters The first-order partial derivative, Displacement response data Normalized displacement response data for structural parameters Structural parameters The second-order partial derivative, Displacement response data Normalized displacement response data for structural parameters Structural parameters The second-order partial derivative, Displacement response data Normalized displacement response data for structural parameters Structural parameters The second-order partial derivative, Represents structural parameters The parameter is not the probability variance. Represents structural parameters Structural parameters The parameter is the non-probability covariance. Represents structural parameters Structural parameters The parameter is the non-probabilistic covariance; The independent sensitivity indices of each structural parameter under the normalized displacement response data are calculated using formula (7): Official (7) in, Represents structural parameters Independent sensitivity indicators, Displacement response data The normalized response non-probability variance This represents the non-probability correlation coefficient matrix. Represents structural parameters The projection of the non-probability variance of the parameters onto the non-probability variance of the normalized response. The definition is as follows: in, Structural parameters The non-probability variance of the parameters affects the displacement response data. The contribution of the normalized response nonprobabilistic variance.

7. The nonprobabilistic sensitivity analysis method for high-power wind turbine gearboxes according to claim 6, characterized in that, The relevant sensitivity index of each structural parameter under the displacement response sequence is calculated by formula (8), which is as follows: Official (8) in, Represents structural parameters Relevant sensitivity indicators, Represents structural parameters The projection of the nonprobability covariance of the parameters onto the nonprobability variance of the normalized response. The definition is as follows: in, Structural parameters The nonprobabilistic covariance of the parameters affects the displacement response data. The contribution of the normalized response nonprobabilistic variance.

8. The nonprobabilistic sensitivity analysis method for high-power wind turbine gearboxes according to claim 1, characterized in that, The gearbox proxy model is constructed through the following steps: Obtain the finite element model of the gearbox and multiple sets of structural parameter samples, and set multiple finite element model nodes at the root of a meshing tooth in the internal gear ring of the gearbox finite element model. Dynamic loads are applied to the gearbox finite element model to obtain displacement response data samples at each node of the finite element model under the action of each set of structural parameter samples. The displacement response data samples at each node of the finite element model form a displacement response sequence sample. Using the structural parameter samples and displacement response data samples, a gearbox proxy model is constructed to characterize the data relationship between the structural parameters and displacement response data.

9. The nonprobabilistic sensitivity analysis method for high-power wind turbine gearboxes according to claim 8, characterized in that, After constructing the gearbox proxy model to characterize the data relationship between the structural parameters and displacement response data, the following steps are included: Calculate the load-sharing coefficient of the gearbox proxy model; Obtain the gearbox entity corresponding to the gearbox proxy model, and attach a uniaxial strain gauge at the root of a meshing tooth in the internal gear ring of the gearbox entity; A dynamic load is applied to the gearbox entity, causing deformation displacement at the root of one meshing tooth in the internal gear ring of the gearbox entity. The uniaxial strain gauge converts the displacement response data of each entity node corresponding to each model node into a strain signal. The actual load-averaging coefficient is calculated through the strain signal. The actual load-averaging coefficient is used to verify the accuracy of the load-averaging coefficient of the gearbox proxy model.

10. The nonprobabilistic sensitivity analysis method for high-power wind turbine gearboxes according to claim 9, characterized in that, The calculation of the actual load-sharing factor using the strain signal includes the following steps: The strain signal is filtered to remove interference peaks, resulting in multiple effective peak values; The strain signal was obtained at the first p Calculate the mean of the peak values ​​for each of the effective peak values. The actual load factor is calculated using formula (9). ; Official (9) in, p =1+ q , q =0,1,2,3…, This indicates the number of planetary gears meshing with the gear under test in the gearbox entity. r This indicates the number of teeth on the gear to be tested in the gearbox entity. h This represents the number of solid nodes for each tooth of the gear under test, obtained by analyzing the number of nodes for the first tooth. e tooth f Peak mean of each node Summation yields , max means to take the maximum value.

Images