Rapid multiphysics inversion method and apparatus for power device, device and storage medium

Abstract

A rapid multiphysics inversion method and apparatus for a power device, a device and a storage medium, relating to the technical field of power device inversion. The method comprises: using simulation software to construct a multiphysics simulation model of a power device; constructing a data set on the basis of a predefined input parameter set and output parameter set; by means of using a preset data order reduction method, performing order reduction on the data set, so as to obtain an order-reduced data set and an order reduction matrix; by means of using a preset inversion method, fitting the relationship between the order-reduced data set and the input parameter set, so as to obtain an inversion coefficient matrix; and on the basis of the order reduction matrix, the inversion coefficient matrix and an input parameter set to be inverted, performing rapid inversion on the multiphysics simulation model, so as to obtain a multiphysics inversion result of the power device. The present invention can obviously reduce the amount of calculation while satisfying calculation precision, achieves high calculation efficiency, and can adapt to the requirements for online monitoring and real-time simulation of power devices.

Description

Method, device, equipment and storage medium for rapid inversion of multi-physics fields of power equipment

[0001] This application claims priority to the Chinese patent application filed with the China Patent Office on February 23, 2024, with application number 202410202409.5 and invention name “Method, device, equipment and storage medium for rapid inversion of multi-physics fields of power equipment”, the entire contents of which are incorporated by reference into this application. Technical Field

[0002] The present invention relates to the field of power equipment inversion technology, and in particular to a method, device, equipment and storage medium for rapid multi-physical field inversion of power equipment. Background Art

[0003] With the development of new power systems, the digitization and intelligentization of power equipment is a necessary process and a key link in supporting the construction of new power systems. To ensure the safe and reliable operation of power equipment under complex operating conditions and multiple environmental factors, based on the development of modern sensor technology, the combination of big data, artificial intelligence, and deep learning with digital twin technology for the multi-physics field of power equipment is a key technology for realizing dedicated real-time monitoring of power equipment.

[0004] Multi-physics field real-time simulation is the basis of digital twin technology for power equipment. Traditional finite element methods consume large amounts of computing resources and take a long time to calculate, and cannot meet the technical requirements of real-time simulation for online monitoring of power equipment.

[0005] Summary of the Invention

[0006] The purpose of the present invention is to provide a method, device, equipment and storage medium for rapid inversion of multi-physical fields of power equipment to solve the technical problems of large computing resource requirements and low computing efficiency when calculating multi-physical fields of power equipment in the existing technology.

[0007] The purpose of the present invention can be achieved through the following technical solutions:

[0008] Solution 1: A fast multi-physics inversion method for power equipment, including:

[0009] Constructing a multi-physics field simulation model of the power equipment using simulation software, wherein the multi-physics field simulation model simulates at least the electric field, magnetic field, thermal field, fluid field, and stress field of the power equipment;

[0010] Constructing a dataset based on a predefined set of input and output parameters;

[0011] Using a preset data reduction method to reduce the order of the data set to obtain a reduced-order data set and a reduced-order matrix;

[0012] Using a preset inversion method to fit the relationship between the reduced-order data set and the input parameter set, to obtain an inversion coefficient matrix;

[0013] According to the reduced-order matrix, the inversion coefficient matrix and the input parameter set to be inverted, the multi-physics field simulation model is quickly inverted to obtain a multi-physics field inversion result of the power equipment.

[0014] Optionally, reducing the order of the data set by using a preset data reduction method to obtain a reduced order data set and a reduced order matrix includes:

[0015] The data set is reduced in order using an eigenorthogonal decomposition method to obtain a reduced-order data set and an eigenvector matrix; the eigenvector matrix is ​​a reduced-order matrix corresponding to the reduced-order data set.

[0016] Optionally, reducing the order of the data set by using an eigenorthogonal decomposition method to obtain a reduced-order data set and an eigenvector matrix includes:

[0017] S301: constructing a training matrix according to the data set and performing normalization processing to obtain a first matrix, constructing a covariance matrix according to the first matrix and solving to obtain eigenvalues ​​and eigenvectors, and arranging the eigenvalues ​​in descending order to obtain an eigenvalue sequence and a corresponding eigenvector sequence;

[0018] S302: Selecting a preset number of eigenvectors from the eigenvector sequence to form an eigenvector matrix, and obtaining a reduced-order data set based on the first matrix and the eigenvector matrix;

[0019] S303: Restoring the reduced-order data to the data dimension of the data set to obtain a restored data set, and calculating a relative error between the restored data set and the data set;

[0020] S304: When the relative error is greater than a preset error threshold, steps S302 to S304 are repeated until the relative error is less than the error threshold, thereby obtaining a reduced-order data set and an eigenvector matrix.

[0021] Optionally, the using a preset inversion method to fit the relationship between the reduced-order data set and the input parameter set to obtain an inversion coefficient matrix includes:

[0022] The response surface method is used to fit the functional relationship between the reduced-order data set and the input parameter set to obtain a response surface coefficient matrix; the response surface coefficient matrix is ​​an inversion coefficient matrix obtained using the response surface method.

[0023] Optionally, the performing rapid inversion on the multi-physics field simulation model according to the reduced-order matrix, the inversion coefficient matrix and the input parameter set to be inverted includes:

[0024] The multi-physics field simulation model is quickly inverted according to the eigenvector matrix, the response surface coefficient matrix and the input parameter set to be inverted.

[0025] Optionally, the using a response surface method to fit the functional relationship between the reduced-order data set and the input parameter set to obtain a response surface coefficient matrix includes:

[0026] Using the response surface method to fit the functional relationship between the reduced-order data set and the input parameter set, respectively solving the response surface coefficient for each column of the reduced-order data set to obtain a response surface coefficient column vector corresponding to each column;

[0027] The response surface coefficient column vectors are used to form a response surface coefficient matrix.

[0028] Optionally, constructing a data set according to a predefined input parameter set and output parameter set includes:

[0029] Within the value range of the input parameter set, performing a parametric scan on each of the input parameters according to a preset number of scans to obtain an input parameter set and a corresponding output parameter set after the parametric scan;

[0030] The input parameter sets and corresponding output parameter sets after all parametric sweeps are used as data sets.

[0031] Solution 2: A multi-physics field fast inversion device for power equipment, comprising:

[0032] A simulation model construction module is used to construct a multi-physics field simulation model of the power equipment using simulation software, wherein the multi-physics field simulation model simulates at least the electric field, magnetic field, thermal field, fluid field and stress field of the power equipment;

[0033] A data set construction module, used to construct a data set according to a predefined input parameter set and output parameter set;

[0034] A data order reduction module is used to reduce the order of the data set using a preset data order reduction method to obtain a reduced order data set and a reduced order matrix;

[0035] A relationship fitting module is used to fit the relationship between the reduced-order data set and the input parameter set using a preset inversion method to obtain an inversion coefficient matrix;

[0036] A fast inversion module is used to perform fast inversion on the multi-physics field simulation model according to the reduced-order matrix, the inversion coefficient matrix and the input parameter set to be inverted, so as to obtain a multi-physics field inversion result of the power equipment.

[0037] Solution 3: An electronic device comprising: a processor and a memory;

[0038] The memory stores a computer program, and the processor implements the steps of the method described in Solution 1 when executing the computer program.

[0039] Solution 4: A computer-readable storage medium having a computer program stored thereon, which implements the steps of the method described in Solution 1 when the computer program is executed by a processor.

[0040] The present invention provides a method, apparatus, device and storage medium for rapid inversion of the multi-physics fields of an electric power device, wherein the method comprises: constructing a multi-physics field simulation model of the electric power device using simulation software, wherein the multi-physics field simulation model simulates at least the electric field, magnetic field, thermal field, fluid field and stress field of the electric power device; constructing a data set according to an input parameter set and an output parameter set; reducing the order of the data set using a preset data reduction method to obtain a reduced-order data set and a reduced-order matrix; fitting the relationship between the reduced-order data set and the input parameter set using a preset inversion method to obtain an inversion coefficient matrix; and rapidly inverting the multi-physics field simulation model according to the reduced-order matrix, the inversion coefficient matrix and the input parameter set to be inverted to obtain a multi-physics field inversion result of the electric power device.

[0041] Based on the above technical solution, the beneficial effects brought about by the present invention are:

[0042] The present invention considers the coupling conditions of the electric-magnetic-thermal-mechanical-fluid multi-physics fields within the power equipment and constructs a multi-physics simulation model of the power equipment. From the perspective of data reduction and high-order data fitting of the model, after defining the input and output parameter sets of the data set, a training data set for a fast inversion method is constructed. The data set is reduced using the order reduction method, and the accuracy of the reduced and pre-reduced data sets is ensured through multiple iterations to obtain the reduced-order data set and the reduced-order matrix. The correspondence between the reduced-order data set and the input parameter set is established using the inversion method to obtain the inversion coefficient matrix. Finally, the reduced-order matrix and the inversion coefficient matrix are used to perform online fast inversion on any input parameter set to obtain the multi-physics inversion data of the power equipment. The present invention can significantly reduce the amount of calculation while meeting the calculation accuracy, has high calculation efficiency, realizes the fast inversion of the reduced-order fitting of the multi-physics field of the power equipment, and the inversion of the online simulation, can meet the needs of online monitoring and real-time simulation of the power equipment, and is conducive to the online fast calculation and real-time simulation of the multi-physics field of the power equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] FIG1 is a flow chart of a first embodiment of a method for rapid inversion of multi-physics fields of power equipment according to the present invention;

[0044] FIG2 is a flow chart of a second embodiment of a method for rapid inversion of multi-physics fields of power equipment according to the present invention;

[0045] FIG3 is a schematic diagram showing the results of a specific example of the method for rapid inversion of multi-physics fields of power equipment according to the present invention;

[0046] FIG4 is a schematic structural diagram of an embodiment of a multi-physics field rapid inversion device for power equipment according to the present invention. DETAILED DESCRIPTION

[0047] Explanation of terms:

[0048] Proper orthogonal decomposition (POD) is a mathematical method used to extract characteristic information from discrete data. Proper orthogonal decomposition decomposes the original data (space-time) into multiple spatial modes (called eigenmodes) and the time evolution coefficients (sequences) corresponding to each mode. The modes are mutually orthogonal. The order of the modes is sorted from high to low by the amount of energy they capture, which is given by the eigenvalues ​​corresponding to the eigenmodes. The expansion of this set of eigenmodes optimally captures the energy of the data. If truncated at the rth order, then any other orthogonal expansion of the same order will capture less energy than the first r modes of the POD. Proper orthogonal decomposition is also known as principal component analysis (PCA).

[0049] Inversion methods are a method for solving problems by transforming the solution to a problem into the solution of a known function. They have widespread applications in mathematics, physics, and other fields. By leveraging mathematical tools, inversion methods can represent the solution to a problem as a combination or transformation of several known functions, and then derive the properties of new functions from the properties of the known functions. In physics, inversion methods are widely used to solve problems such as partial differential equations, electromagnetic fields, and fluid dynamics. Their application can significantly simplify the complexity of a problem.

[0050] The response surface method uses a polynomial function to approximate an implicit limit-state function through a series of deterministic experiments. By selecting appropriate test points and using an iterative strategy, the failure probability of the polynomial function is guaranteed to converge to the true implicit limit-state function. When the true limit-state function is not nonlinear, a linear response surface method offers high approximation accuracy.

[0051] The embodiments of the present invention provide a method, apparatus, device and storage medium for rapid inversion of multi-physical fields of power equipment to solve the technical problems of large computing resource requirements and low computing efficiency when calculating multi-physical fields of power equipment in the prior art.

[0052] To facilitate understanding of the present invention, the present invention will be described more fully below with reference to the accompanying drawings. Preferred embodiments of the present invention are shown in the drawings. However, the present invention may be embodied in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided to provide a more thorough and comprehensive disclosure of the present invention.

[0053] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this invention pertains. The terms used in this specification of the present invention are for the purpose of describing specific embodiments only and are not intended to limit the present invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0054] The coupled relationships between the electric, magnetic, thermal, mechanical, and fluidic multi-physics fields within power equipment are strong, and the distribution of typical defect physics fields under different operating conditions is complex. By performing finite element calculations on various typical multi-physics field distributions and using a fast inversion algorithm for physics field reduction fitting, we can achieve online monitoring of the internal state of power equipment and fault diagnosis. Research on fast calculation and online simulation technologies for multi-physics fields in power equipment will provide technical methods for the construction of new power systems.

[0055] The application scenario of the embodiment of the present invention is to perform rapid inversion of the multi-physics field of the power equipment, and realize online rapid inversion calculation of the multi-physics field of the power equipment. Taking into account the electromagnetic-magnetic-thermal-mechanical-fluid multi-physics field coupling conditions inside the power equipment, from the perspective of data reduction and high-order data fitting, after defining the input and output parameter sets of the data set, a parametric scan is performed to obtain a training data set before reduction; the field quantity or independent variable of the multi-physics field of the simulation model is reduced in data order using a data reduction method, and the accuracy of the reduced-order data set and the pre-reduction data set is guaranteed through multiple iterations to obtain a reduced-order matrix and a reduced-order data set; the corresponding relationship between the reduced-order data set and the input parameter set is established through an inversion method to obtain an inversion coefficient matrix; finally, the reduced-order matrix and the inversion coefficient matrix are used to perform online inversion on any input parameter set to obtain inversion data of the multi-physics field of the power equipment. The embodiment of the present invention can significantly reduce the amount of calculation while meeting the calculation accuracy, and can realize rapid calculation and real-time simulation of the multi-physics field of the power equipment.

[0056] Referring to FIG1 , the present invention provides a first embodiment of a method for rapid inversion of multi-physics fields of power equipment, comprising:

[0057] S100: constructing a multi-physics field simulation model of the power equipment using simulation software, wherein the multi-physics field simulation model simulates at least the electric field, magnetic field, thermal field, fluid field, and stress field of the power equipment;

[0058] S200: constructing a data set according to the input parameter set and the output parameter set;

[0059] S300: Using a preset data reduction method to reduce the order of the data set to obtain a reduced order data set and a reduced order matrix;

[0060] S400: fitting the relationship between the reduced-order data set and the input parameter set using a preset inversion method to obtain an inversion coefficient matrix;

[0061] S500: Performing rapid inversion on the multi-physics simulation model according to the reduced-order matrix, the inversion coefficient matrix and the input parameter set to be inverted, to obtain a multi-physics inversion result of the power equipment.

[0062] In step S100, simulation software (such as finite element simulation software) can be used to construct a multi-physics field simulation model of the power equipment, which simulates at least the electric field, magnetic field, thermal field, fluid field and stress field of the power equipment.

[0063] In one embodiment, the multi-physical fields of the power equipment include but are not limited to: the electric field, magnetic field, thermal field, fluid field, stress field, etc. of the power equipment.

[0064] It should be noted that the multi-physics simulation model can be constructed through commercial software and self-programming, with the aim of obtaining multi-physics data of power equipment.

[0065] In step S200 , a data set is constructed according to a predefined input parameter set and an output parameter set.

[0066] Specifically, the input parameter set of the training set (data set) is defined as: X = [X1, X2, ..., X k ...,X q ], where X k =[x k (1) ,x k (2) ,…x k (n) ] T ;X k For the kth group of input parameters, x k (1) ~x k (n) The n input parameters (key parameter variables) in the kth group of input parameters.

[0067] At the same time, define the value range of the above key input parameters X m =[x min ,x max ], where x min 、x max The input parameters X are k The value range of is the minimum vector and the maximum vector.

[0068] It is understood that the input parameter set of the dataset is the set of key parameters of the power equipment that need to be input into the fast inversion algorithm. The input parameters in the input parameter set are key parameter variables of the power equipment, including but not limited to: thermal conductivity of the material, operating input current, elastic modulus of the material, and other parameters. The training set is used for fast inversion calculations.

[0069] Specifically, the output parameter set of the fast inversion calculation method training set is defined as: F(X k )=[f1(X k ), f2(X k ),…,f p (X k )] T , where f p (X k ) is when the input parameter is X k The independent variable or physical field quantity of the pth point in the finite element calculation, including but not limited to: temperature value, stress value, electric field intensity value, etc.

[0070] It can be understood that the output parameter set of the data set is a set of indicators to be inverted for the power equipment in the fast inversion algorithm.

[0071] In one embodiment, constructing a data set based on an input parameter set and an output parameter set includes:

[0072] Within the value range of the input parameter set, each input parameter is parametrically scanned according to a preset number of scans to obtain an input parameter set after the parametric scan and a corresponding output parameter set;

[0073] The input parameter sets and corresponding output parameter sets after all parametric sweeps are taken as data sets.

[0074] Specifically, according to the input parameters and output parameters of the data set defined above, parametric scanning is performed within the value range of the input parameters. The number of scans for each input parameter is m, and the total number of scans is defined as q = m × n. The input parameters and output parameters obtained from all parametric scans form a data set for the fast inversion algorithm.

[0075] It should be noted that parametric scanning refers to changing the different input parameters of a multi-physics field simulation model (finite element model), such as voltage, current, thermal conductivity of the material, etc., to calculate the physical field distribution when different parameters are obtained, and all the above physical field distributions are combined to obtain a data set.

[0076] In step S300, a preset data reduction method is used to reduce the order of the data set to obtain a reduced-order data set and a reduced-order matrix.

[0077] In one embodiment, a data set is reduced in order using a preset data reduction method to obtain a reduced-order data set and a reduced-order matrix, including:

[0078] The data set is reduced in order using the eigenorthogonal decomposition method to obtain a reduced-order data set and an eigenvector matrix; the eigenvector matrix is ​​a reduced-order matrix corresponding to the reduced-order data set.

[0079] The specific process of reducing the order of a data set using the proper orthogonal decomposition method includes:

[0080] (1) Construct a training matrix based on the data set and perform normalization processing to obtain a first matrix, construct a covariance matrix based on the first matrix and solve it to obtain eigenvalues ​​and eigenvectors, and arrange them in descending order according to the eigenvalues ​​to obtain an eigenvalue sequence and a corresponding eigenvector sequence;

[0081] (2) selecting a preset number of eigenvectors from the eigenvector sequence to form an eigenvector matrix, and obtaining a reduced-order data set based on the first matrix and the eigenvector matrix;

[0082] (3) Restore the reduced-order data to the data dimension of the dataset to obtain a restored dataset, and calculate the relative error between the restored dataset and the dataset;

[0083] (4) When the relative error is greater than the preset error threshold, steps (2) to (4) are repeated until the relative error is less than the error threshold, thereby obtaining the reduced-order data set and eigenvector matrix.

[0084] Specifically, the eigenorthogonal decomposition method is used to reduce the dimension of the data set and establish a training matrix D = [F(X1), F(X2), ..., F(X q )], D is a matrix of size p×q, and the data in row i and column j of the training matrix D is the input variable X j The independent variable or physical field quantity of the finite element calculation point i. The expression of D is shown in formula (1):

[0085] The training matrix D is decentralized (i.e., normalized) to obtain the matrix U. The specific processing method is shown in formula (2):

[0086] Among them, u ij is the data of row i and column j in matrix U, d ij is the data of row i and column j in the training matrix D, d ik is the data in the i-th row and k-th column of the training matrix D, 1≤k≤q.

[0087] Then, the eigenvalues ​​and eigenvectors of the covariance matrix C are obtained based on the obtained matrix U. The expression of the covariance matrix C is shown in formula (3):

[0088] Solve the equation shown in formula (4):

[0089] Cv=vdiag(λ); (4)

[0090] Get the eigenvalues ​​and eigenvectors of the covariance matrix C, and arrange them in descending order according to the size of the eigenvalues. The sorted eigenvector sequence and eigenvalue sequence are: v = [v1, v2, ..., v p ] and λ=[λ1,λ2,...λ p ] T .

[0091] Next, select the first preset number of eigenvectors from the eigenvector sequence. For example, take the first α-order eigenvectors (the first α-order principal components) to reduce the order of the original training data, and form the first α-order eigenvectors into an eigenvector matrix Reduced dataset The expression is shown in formula (5):

[0092] The reduced dataset Expressed as: in, Restore the reduced data set to the original dimension data to obtain the restored data set Its expression is shown in formula (6):

[0093] Calculate the relative error between the restored dataset and the initially constructed (original) dataset. The relative error expression is shown in formula (7):

[0094] Among them, e% is the relative error between the restored dataset and the dataset; p is the number of rows in the restored dataset and the dataset; q is the number of columns in the restored dataset and the dataset; To restore the data of row i and column j in the dataset; U ij is the data in row i and column j in the dataset.

[0095] In one embodiment, the relative error between the restored dataset and the original dataset is used to measure the dimensionality reduction ability of the selected first α-order principal components on the original dataset, and the relative error between the original dataset and the restored dataset is used as an evaluation indicator of the dimensionality reduction ability.

[0096] It should be noted that the first α-order principal components refer to the eigenvector matrix composed of the first α-order eigenvectors. The principal component is the part that can reflect the characteristics of the original data and is part of the eigenvector. α is determined by controlling the error. Steps (2) and (3) are the process of determining the order of the principal components.

[0097] Set a certain error interval, i.e., a preset error threshold e%. For example, the error threshold can be set to 0.5% or 0.1%. The setting of the error threshold can be customized according to the actual accuracy requirements. When e% ≤ 0.5%, add 1 to α and jump to step (2). Repeat steps (2) to (4) until the relative error is less than the error threshold. When e% ≤ 0.5%, complete the order reduction of the data set and retain the eigenvector matrix composed of the first α-order eigenvectors. and the reduced dataset The eigenvector matrix As the reduced-order matrix of the fast inversion model, the next step is then carried out.

[0098] In step S400, a preset inversion method is used to fit the relationship between the reduced-order data set and the input parameter set to obtain an inversion coefficient matrix.

[0099] In one embodiment, a preset inversion method is used to fit the relationship between the reduced-order data set and the input parameter set to obtain an inversion coefficient matrix, including:

[0100] The response surface method is used to fit the functional relationship between the reduced-order data set and the input parameter set to obtain a response surface coefficient matrix; the response surface coefficient matrix is ​​the inversion coefficient matrix obtained using the response surface method.

[0101] Specifically, the response surface method is used to fit the reduced-order data set With the input parameter set X=[X1,X2,...,X q ], and solve the response surface coefficient for each column in the reduced-order data set to obtain the response surface coefficient column vector corresponding to each column. In a preferred embodiment, the response surface coefficient column vector β is calculated using formula (8) i =[β1,β2,…,β n ] T :

[0102] Then, the response surface coefficient column vectors are used to form a response surface coefficient matrix. The expression of the response surface coefficient matrix is ​​shown in formula (9):

[0103] β=[β1,β2,...,β α ];(9)

[0104] In step S500, a multi-physics simulation model is rapidly inverted according to the reduced-order matrix, the inversion coefficient matrix and the input parameter set to be inverted to obtain a multi-physics inversion result of the power equipment.

[0105] In one embodiment, a multi-physics simulation model is rapidly inverted based on a reduced-order matrix, an inversion coefficient matrix, and a set of input parameters to be inverted, including:

[0106] The multi-physics simulation model is quickly inverted based on the eigenvector matrix, the response surface coefficient matrix and the input parameter set to be inverted.

[0107] Specifically, the input parameter set to be inverted (given) is set as: X δ =[x δ (1) ,x δ (2) ,…x δ (n) ] T And use the calculation method shown in formula (10) to perform rapid inversion to obtain the inversion data Y = [y1, y2, ..., y p ] T :

[0108] In this embodiment, the eigenvector matrix (reduced-order matrix) for implementing the reduced-order inversion algorithm can be obtained by using the eigenorthogonal decomposition, and the response surface coefficient matrix obtained by the response surface method is input into the eigenvector matrix and the response surface coefficient matrix according to the value range of the input parameters set in step S200. The original multi-physics field simulation model is inverted to obtain the corresponding inversion data.

[0109] The embodiment of the present invention provides a fast inversion method for multi-physics fields of power equipment. The method considers the coupling conditions of the electric-magnetic-thermal-mechanical-fluid multi-physics fields within the power equipment to construct a multi-physics simulation model of the power equipment. From the perspective of data reduction and high-order data fitting of the model, after defining the input and output parameter sets of the data set, a training data set for the fast inversion method is constructed. The data set is reduced in order using a reduction method. The accuracy of the reduced and unreduced data sets is ensured through multiple iterations to obtain the reduced-order data set and the reduced-order matrix. The corresponding relationship between the reduced-order data set and the input parameter set is established through an inversion method to obtain an inversion coefficient matrix. Finally, the reduced-order matrix and the inversion coefficient matrix are used to perform online fast inversion on any input parameter set to obtain the multi-physics field inversion data of the power equipment. The embodiment of the present invention can significantly reduce the amount of calculation while meeting the calculation accuracy, has high calculation efficiency, realizes fast inversion of reduced-order fitting of multi-physics fields of power equipment, and inversion of online simulation, can meet the needs of online monitoring and real-time simulation of power equipment, and is conducive to the online fast calculation and real-time simulation of multi-physics fields of power equipment.

[0110] Referring to FIG. 2 , the present invention provides a second embodiment of a method for rapid multi-physics field inversion of power equipment, including:

[0111] Step 1: Establish a three-dimensional multi-physics simulation model of the power equipment to calculate the multi-physics coupling field of the power equipment. The multi-physics field includes but is not limited to the electric field, magnetic field, thermal field, fluid field, and stress field;

[0112] Step 2: Define the input parameter X of the fast inversion calculation method training set k =[x k (1) ,x k (2) ,…x k (n) ] T , and define the value ranges of the above key input parameters;

[0113] Step 3: Define the output parameter F(X k )=[f1(X),f2(X),…,f p (X)] T , including but not limited to temperature values, stress values, and electric field strength values;

[0114] Step 4: Perform a parametric scan within the range of input parameter values ​​based on the input and output parameters defined in Steps 2 and 3, and form a data set for the fast inversion algorithm with all the output parameters and input parameters obtained from the parametric scan;

[0115] Step 5: Based on the data set generated in step 4, the intrinsic orthogonal decomposition-response surface method is used to train the inversion calculation model;

[0116] In step 5, the training matrix is ​​established and normalized to obtain the matrix U. The eigenvectors and eigenvalues ​​of the covariance matrix C are solved according to the matrix U, and the eigenvalues ​​are sorted in descending order to obtain the sorted eigenvectors v and eigenvalues ​​λ. The eigenvectors corresponding to the first α order eigenvalues ​​are taken to form the eigenvector matrix And obtain the reduced-order data set Restore the reduced data set to the original dimension data to obtain the restored data set Calculate the relative error e% between the original data set and the restored data set. When the relative error is less than the preset error threshold (such as 0.5%), retain the eigenvector matrix composed of the first α-order eigenvectors. and the reduced dataset Proceed to step 6.

[0117] Step 6 After the intrinsic orthogonal decomposition-response surface method training in step 5, the eigenvector matrix for implementing the reduced-order inversion algorithm can be obtained. As well as the response surface coefficient matrix β, according to the input parameter range set in step 2, input the input parameter set (physical parameter set) to be inverted, and use the eigenvector matrix The inversion is performed in the response surface coefficient matrix β to obtain the inversion data Y.

[0118] The embodiment of the present invention provides a method for rapid inversion of multi-physics fields of power equipment, which is a method for rapid calculation and real-time simulation of multi-physics fields of power equipment. First, a three-dimensional multi-physics field simulation model of the power equipment is established to calculate the multi-physics field coupling field of the power equipment, wherein the multi-physics field includes but is not limited to the electric field, magnetic field, thermal field, fluid field, stress field, etc.; after defining the input key parameter set and its value range and the output key parameter set of the data set, a three-dimensional finite element simulation model is used to perform parametric scanning on the above parameters to obtain a training data set before order reduction. The original data set is reduced in order using the proper orthogonal decomposition method. By changing the order α of the reduced-order matrix, the accuracy of the reduced-order model is controlled, and while achieving rapid calculation, the calculation error of the inversion algorithm is guaranteed to be less than 0.5%. Based on the proper orthogonal decomposition method, the reduced-order data set and the eigenvector matrix are obtained. The response surface method is used to establish the corresponding relationship between the reduced-order data set and the input parameter set to obtain the response surface coefficient matrix. The response surface coefficient matrix is ​​the inversion coefficient matrix obtained using the response surface method. Finally, the eigenvector matrix and the response surface coefficient matrix are used to perform physical field inversion on any input parameter.

[0119] The embodiment of the present invention considers the coupling conditions of the electromagnetic, magnetic, thermal, force and fluid multi-physics fields inside the power equipment while meeting the calculation accuracy, and is a method for fast calculation and real-time simulation of the multi-physics fields of the power equipment. The method starts from the perspective of data reduction and high-order data fitting of the data set, performs parametric scanning after defining the input and output parameters of the data set, and obtains a training data set before reduction; uses the intrinsic orthogonal decomposition method to reduce the data order of the multi-physics field quantities or independent variables, and through multiple iterations, ensures the accuracy of the reduced data set and the data set before reduction, obtains the reduced-order eigenvector matrix and the reduced-order data set; establishes the corresponding relationship between the reduced-order data set and the input parameters through the response surface method, and obtains the response surface method coefficient matrix; finally, uses the reduced-order eigenvector matrix and the response surface method coefficient matrix to perform online inversion on any input parameters. The invention can significantly reduce the amount of calculation and realize the fast inversion of the online simulation of the power equipment.

[0120] Please refer to Figure 3. In the specific example of the present invention, an application software for fast calculation and real-time simulation can be used. The software panel of the application software can have the following functions: working folder setting function, calculation fitting operation function, status bar display function, and physical quantity field diagram display function.

[0121] The working folder setting function can include: inputting the computer's working folder, determining the number of header rows of the data, reading the geometry, grid, field quantity, unit and other information in the data file, and importing data through the software interface.

[0122] The calculation and fitting operation functions may include: setting the fitting order, initializing all parameters, fitting result operation, drawing result operation, saving result file and other operations.

[0123] The status bar display functions may include: description file import information, fitting error information, calculation time record, export file information, image drawing information and other functions.

[0124] The physical quantity field diagram display function can include: physical quantity field diagram and its legend, optional display of units and titles, support for mouse interactive clicking and extraction of physical quantity values, etc.

[0125] It should be noted that what is obtained in Figure 3 is a contour map of the physical field distribution of the power equipment. Different areas represent the values ​​of the corresponding data scale in the data on the right. The figure shows a physical field distribution situation.

[0126] Referring to FIG4 , the present invention provides an embodiment of a multi-physics field fast inversion device for power equipment, comprising:

[0127] A simulation model construction module 11 is used to construct a multi-physics field simulation model of the power equipment using simulation software, wherein the multi-physics field simulation model simulates at least the electric field, magnetic field, thermal field, fluid field and stress field of the power equipment;

[0128] A data set construction module 22 is used to construct a data set according to a predefined input parameter set and output parameter set;

[0129] A data order reduction module 33 is configured to reduce the order of the data set using a preset data order reduction method to obtain a reduced order data set and a reduced order matrix;

[0130] A relationship fitting module 44 is used to fit the relationship between the reduced-order data set and the input parameter set using a preset inversion method to obtain an inversion coefficient matrix;

[0131] The fast inversion module 55 is used to perform fast inversion on the multi-physics simulation model according to the reduced-order matrix, the inversion coefficient matrix and the input parameter set to be inverted, so as to obtain the multi-physics inversion result of the power equipment.

[0132] Optionally, the data order reduction module reduces the order of the data set using a preset data order reduction method to obtain a reduced order data set and a reduced order matrix, including:

[0133] The data order reduction module reduces the order of the data set using an eigenorthogonal decomposition method to obtain a reduced-order data set and an eigenvector matrix; the eigenvector matrix is ​​a reduced-order matrix corresponding to the reduced-order data set.

[0134] Optionally, the data order reduction module reduces the order of the data set using an eigenorthogonal decomposition method to obtain a reduced-order data set and an eigenvector matrix, including:

[0135] S301: constructing a training matrix according to the data set and performing normalization processing to obtain a first matrix, constructing a covariance matrix according to the first matrix and solving to obtain eigenvalues ​​and eigenvectors, and arranging the eigenvalues ​​in descending order to obtain an eigenvalue sequence and a corresponding eigenvector sequence;

[0136] S302: Selecting a preset number of eigenvectors from the eigenvector sequence to form an eigenvector matrix, and obtaining a reduced-order data set based on the first matrix and the eigenvector matrix;

[0137] S303: Restoring the reduced-order data to the data dimension of the data set to obtain a restored data set, and calculating a relative error between the restored data set and the data set;

[0138] S304: When the relative error is greater than a preset error threshold, steps S302 to S304 are repeated until the relative error is less than the error threshold, thereby obtaining a reduced-order data set and an eigenvector matrix.

[0139] Optionally, the relationship fitting module uses a preset inversion method to fit the relationship between the reduced-order data set and the input parameter set to obtain an inversion coefficient matrix, including:

[0140] The relationship fitting module uses the response surface method to fit the functional relationship between the reduced-order data set and the input parameter set to obtain a response surface coefficient matrix; the response surface coefficient matrix is ​​an inversion coefficient matrix obtained using the response surface method.

[0141] Optionally, the fast inversion module performs fast inversion on the multi-physics field simulation model according to the reduced-order matrix, the inversion coefficient matrix and the input parameter set to be inverted, comprising:

[0142] The fast inversion module performs fast inversion on the multi-physics field simulation model according to the eigenvector matrix, the response surface coefficient matrix and the input parameter set to be inverted.

[0143] Optionally, the relationship fitting module uses a response surface method to fit the functional relationship between the reduced-order data set and the input parameter set to obtain a response surface coefficient matrix, including:

[0144] The relationship fitting module uses the response surface method to fit the functional relationship between the reduced-order data set and the input parameter set, and solves the response surface coefficient for each column of the reduced-order data set to obtain a response surface coefficient column vector corresponding to each column;

[0145] The response surface coefficient column vectors are used to form a response surface coefficient matrix.

[0146] Optionally, the data set construction module constructs a data set according to a predefined input parameter set and output parameter set, including:

[0147] The data set construction module performs parametric scanning on each of the input parameters within the value range of the input parameter set according to a preset number of scans to obtain an input parameter set and a corresponding output parameter set after the parametric scanning;

[0148] The input parameter sets and corresponding output parameter sets after all parametric sweeps are taken as data sets.

[0149] The embodiment of the present invention provides a multi-physics field rapid inversion device for power equipment. Considering the coupling conditions of the electric-magnetic-thermal-mechanical-fluid multi-physics fields within the power equipment, a multi-physics field simulation model of the power equipment is constructed. Starting from the perspective of data reduction and high-order data fitting of the model, after defining the input and output parameter sets of the data set, a training data set for the rapid inversion method is constructed. The data set is reduced in order using the order reduction method. The accuracy of the reduced order data set and the pre-reduction data set is ensured through multiple iterations to obtain the reduced order data set and the reduced order matrix. The correspondence between the reduced order data set and the input parameter set is established through the inversion method to obtain the inversion coefficient matrix. Finally, the reduced order matrix and the inversion coefficient matrix are used to perform online rapid inversion on any input parameter set to obtain the multi-physics field inversion data of the power equipment. The embodiment of the present invention can significantly reduce the amount of calculation while meeting the calculation accuracy, has high calculation efficiency, realizes the rapid inversion of the reduced order fitting of the multi-physics field of the power equipment, and the inversion of the online simulation, can meet the needs of online monitoring and real-time simulation of the power equipment, and is conducive to the online rapid calculation and real-time simulation of the multi-physics field of the power equipment.

[0150] In addition, the present invention also provides an electronic device, comprising: a processor and a memory;

[0151] The memory stores a computer program, and the processor implements the steps of the method when executing the computer program.

[0152] At the same time, the present invention also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of the method described above are implemented.

[0153] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0154] In the embodiments provided in this application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of units is only a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interface, device or unit, which can be electrical, mechanical or other forms.

[0155] Units described as separate components may or may not be physically separate, and components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0156] In addition, the functional units in the various embodiments of the present invention may be integrated into a single processing unit, each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0157] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the various embodiments of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, and other media that can store program code.

[0158] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A fast inversion method for multi-physics fields of power equipment, characterized in that: include: Constructing a multi-physics field simulation model of the power equipment using simulation software, wherein the multi-physics field simulation model simulates at least the electric field, magnetic field, thermal field, fluid field, and stress field of the power equipment; Constructing a dataset based on a predefined set of input and output parameters; Using a preset data reduction method to reduce the order of the data set to obtain a reduced-order data set and a reduced-order matrix; Using a preset inversion method to fit the relationship between the reduced-order data set and the input parameter set, to obtain an inversion coefficient matrix; According to the reduced-order matrix, the inversion coefficient matrix and the input parameter set to be inverted, the multi-physics field simulation model is quickly inverted to obtain a multi-physics field inversion result of the power equipment.

2. The multi-physics field rapid inversion method for power equipment according to claim 1, characterized in that: The method of reducing the order of the data set by using a preset data reduction method to obtain a reduced order data set and a reduced order matrix includes: The data set is reduced in order using an eigenorthogonal decomposition method to obtain a reduced-order data set and an eigenvector matrix; the eigenvector matrix is ​​a reduced-order matrix corresponding to the reduced-order data set.

3. The multi-physics field rapid inversion method for power equipment according to claim 2, characterized in that: The method of reducing the order of the data set by using the eigenorthogonal decomposition method to obtain the reduced order data set and eigenvector matrix includes: S301: constructing a training matrix according to the data set and performing normalization processing to obtain a first matrix, constructing a covariance matrix according to the first matrix and solving to obtain eigenvalues ​​and eigenvectors, and arranging the eigenvalues ​​in descending order to obtain an eigenvalue sequence and a corresponding eigenvector sequence; S302: Selecting a preset number of eigenvectors from the eigenvector sequence to form an eigenvector matrix, and obtaining a reduced-order data set based on the first matrix and the eigenvector matrix; S303: Restoring the reduced-order data to the data dimension of the data set to obtain a restored data set, and calculating a relative error between the restored data set and the data set; S304: When the relative error is greater than the preset error threshold, repeat step S302 The process proceeds to S304 until the relative error is less than the error threshold, thereby obtaining a reduced-order data set and an eigenvector matrix.

4. The method for rapid inversion of multi-physics fields of power equipment according to claim 2, characterized in that: The method of fitting the relationship between the reduced-order data set and the input parameter set using a preset inversion method to obtain an inversion coefficient matrix includes: The response surface method is used to fit the functional relationship between the reduced-order data set and the input parameter set to obtain a response surface coefficient matrix; the response surface coefficient matrix is ​​an inversion coefficient matrix obtained using the response surface method.

5. The method for rapid inversion of multi-physics fields of power equipment according to claim 4, characterized in that: The method further comprises: performing a rapid inversion on the multi-physics field simulation model according to the reduced-order matrix, the inversion coefficient matrix, and the input parameter set to be inverted, comprising: The multi-physics field simulation model is quickly inverted according to the eigenvector matrix, the response surface coefficient matrix and the input parameter set to be inverted.

6. The method for rapid inversion of multi-physics fields of power equipment according to claim 4, characterized in that: The method of using the response surface method to fit the functional relationship between the reduced-order data set and the input parameter set to obtain a response surface coefficient matrix includes: Using the response surface method to fit the functional relationship between the reduced-order data set and the input parameter set, respectively solving the response surface coefficient for each column of the reduced-order data set to obtain a response surface coefficient column vector corresponding to each column; The response surface coefficient column vectors are used to form a response surface coefficient matrix.

7. The method for rapid inversion of multi-physics fields of power equipment according to claim 1, characterized in that: The step of constructing a data set according to a predefined input parameter set and an output parameter set includes: Within the value range of the input parameter set, performing a parametric scan on each of the input parameters according to a preset number of scans to obtain an input parameter set and a corresponding output parameter set after the parametric scan; The input parameter sets and corresponding output parameter sets after all parametric sweeps are taken as data sets.

8. A multi-physics field fast inversion device for power equipment, characterized in that: include: The simulation model building module is used to build a multi-physics field simulation model of the power equipment using simulation software. The multi-physics field simulation model at least analyzes the electric field, magnetic field, thermal field, Simulate fluid field and stress field; A data set construction module, used to construct a data set according to a predefined input parameter set and output parameter set; A data order reduction module is used to reduce the order of the data set using a preset data order reduction method to obtain a reduced order data set and a reduced order matrix; A relationship fitting module is used to fit the relationship between the reduced-order data set and the input parameter set using a preset inversion method to obtain an inversion coefficient matrix; A fast inversion module is used to perform fast inversion on the multi-physics field simulation model according to the reduced-order matrix, the inversion coefficient matrix and the input parameter set to be inverted, so as to obtain a multi-physics field inversion result of the power equipment.

9. An electronic device, characterized in that: include: processor and memory; The memory stores a computer program, and the processor implements the steps of the method according to any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images