CN115270633A - Prediction method, system, device and medium for 3D physics of fuel cell - Google Patents

Abstract

The invention discloses a prediction method, a system, equipment and a medium of a three-dimensional physical field of a fuel cell, wherein the prediction method comprises the following steps: acquiring variable input parameters of random working conditions of the fuel cell to be predicted; the variable input parameters of the random working conditions of the fuel cell to be predicted are used as the input of a preset digital twin model of the three-dimensional multi-physical field of the fuel cell, and the digital twin result of the three-dimensional multi-physical field is output to obtain the prediction result of the three-dimensional multi-physical field of the fuel cell to be predicted; the method can realize the rapid and accurate prediction of three-dimensional multi-physical fields in the fuel cell on the basis of a given fuel cell numerical model, and meet the demand of multi-physical field prediction in the on-line control scene of the proton exchange membrane fuel cell; the method realizes rapid and accurate prediction of three-dimensional multi-physical fields in the fuel cell based on the intrinsic orthogonal decomposition method, and performs prediction on weight coefficients under any random working condition, so that the prediction efficiency and the accuracy of prediction results are high.

Description

Fuel cell three-dimensional physical field prediction method, system, equipment and medium

technical field

The invention belongs to the technical field of fuel cells, and in particular relates to a prediction method, system, equipment and medium of a fuel cell three-dimensional physical field.

Background technique

With global climate change and the acceleration of the carbon neutral agenda, the energy industry is shifting from fossil fuel dominance to renewable energy dominance at an increasing speed, however, the intermittency of renewable energy power generation limits its further promotion and application ; Hydrogen energy is expected to be used as a secondary energy source to solve the intermittent problem of renewable energy power generation; as one of the hydrogen energy utilization devices, proton exchange membrane fuel cells may become the most widely used engine replacement in the future. Commercial applications of proton exchange membrane fuel cells are currently developing rapidly around the world.

It is of great significance to fully understand the real-time state inside the proton exchange membrane fuel cell, and to optimize its online operation control and life evaluation; because there are a large number of micro-scale structures in the proton exchange membrane fuel cell, it is difficult to use conventional experimental methods to analyze them. The internal physical field conducts in-situ tests; for this reason, scholars have established a series of proton exchange membrane fuel cell performance prediction models; among them, the rapid prediction models suitable for the online control process of proton exchange membrane fuel cells are usually more simplified; calculation The fluid dynamics model can obtain the detailed three-dimensional physical field in the proton exchange membrane fuel cell, but its simulation speed is slow, and it is difficult to be used for online prediction of the internal physical field of the proton exchange membrane fuel cell.

Contents of the invention

Aiming at the technical problems existing in the prior art, the present invention provides a prediction method, system, equipment and medium for the three-dimensional physical field of a fuel cell, so as to solve the problem that the simulation speed of the existing three-dimensional multi-physical field for the proton exchange membrane fuel cell is slow , the technical problem that the online prediction of the three-dimensional multi-physics field inside the fuel cell cannot be realized.

In order to achieve the above object, the technical scheme adopted in the present invention is:

The invention provides a method for predicting the three-dimensional physical field of a fuel cell, comprising:

Obtain the variable input parameters of the random operating conditions of the fuel cell to be predicted;

The variable input parameters of the stochastic operating conditions of the fuel cell to be predicted are used as the input of the preset digital twin model of the three-dimensional multi-physics field of the fuel cell, and the digital twin result of the three-dimensional multi-physics field is output, that is, the fuel cell to be predicted is obtained 3D multiphysics prediction results;

Wherein, the construction process of the digital twin model of the preset fuel cell three-dimensional multi-physics field includes:

Solve the pre-built 3D multi-physics coupling model of the fuel cell to obtain several twin snapshots;

For each type of three-dimensional physical field in the fuel cell, several basis functions are constructed by using the intrinsic orthogonal decomposition method according to several twin snapshots, and the weights of each basis function direction of each three-dimensional physical field under each twin snapshot working condition are obtained. coefficient;

Using the weight coefficients of each basis function direction of each three-dimensional physical field under each twin snapshot working condition as a sample set, based on a pre-selected machine learning algorithm, interpolation algorithm or regression algorithm, the preset fuel cell three-dimensional multi-dimensional Digital twins of physical fields.

Further, the variable input parameters of the stochastic operating conditions of the fuel cell to be predicted include cathode pressure, anode pressure, cathode humidity, anode humidity, cathode stoichiometric ratio, anode stoichiometric ratio and operating temperature of the fuel cell to be predicted.

Further, the construction process of the digital twin model of the preset three-dimensional multi-physics field of the fuel cell is as follows:

Construct a 3D multi-physics coupled model of the fuel cell;

Acquiring model parameters of the fuel cell according to the three-dimensional multi-physics coupling model of the fuel cell;

Select variable parameters from the model parameters of the fuel cell to obtain variable input parameters of several snapshot working conditions;

The variable input parameters of several snapshot working conditions are used as the input of the three-dimensional multi-physics coupling model of the fuel cell, and several twin snapshots are obtained through simulation;

For each type of three-dimensional physical field in the fuel cell, according to several twin snapshots, generate a snapshot matrix under several snapshot conditions;

For each type of three-dimensional physical field in the fuel cell, use the singular value decomposition method to perform matrix decomposition and matrix transformation processing on the snapshot matrix under several snapshot conditions, and obtain several basis functions;

Project several twin snapshots to each basis function direction, and reconstruct to obtain the weight coefficients of each basis function direction of each three-dimensional physical field under each twin snapshot working condition;

Using the weight coefficients of each basis function direction of each three-dimensional physical field under each twin snapshot working condition as a sample set, based on a pre-selected machine learning algorithm, interpolation algorithm or regression algorithm, the preset fuel cell three-dimensional multi-dimensional Digital twins of physical fields.

Further, the model parameters of the fuel cell include geometric parameters, physical parameters, electrochemical parameters and operating parameters of the fuel cell.

Further, the fuel cell three-dimensional multi-physics coupling model is a three-dimensional two-phase non-isothermal numerical simulation model based on continuous liquid pressure.

Further, the variable input parameters of several snapshot working conditions are used as the input of the three-dimensional multi-physics coupling model of the fuel cell, and the process of obtaining several twin snapshots through simulation is as follows:

Using C language, write the coupling model of the three-dimensional multi-physics field of the fuel cell to be predicted into ANSYS Fluent software, take the variable input parameters of several snapshot working conditions as input, and simulate the output to obtain several twin snapshots; among them, several twin snapshots The output is in the format of a Tecplot file.

Further, the digital twin model of the preset fuel cell three-dimensional multi-physics field is:

Among them, r is the number of the physical quantity; f′ _r is the three-dimensional physical field of the rth physical quantity under the variable input parameters of the fuel cell to be predicted; b′ _r,j is the three-dimensional physical field of the rth physical quantity in the jth basis function ψ _r,j is the jth basis function of the rth physical quantity; j is the number of the above basis functions; l _r is the truncation order of the rth physical quantity.

The present invention also provides a prediction system for the three-dimensional physical field of the fuel cell, including:

An acquisition module, configured to acquire variable input parameters of fuel cell stochastic operating conditions to be predicted;

The prediction module is used to use the variable input parameters of the random operating conditions of the fuel cell to be predicted as the input of the preset digital twin model of the three-dimensional multi-physics field of the fuel cell, and output the digital twin result of the three-dimensional multi-physics field, namely Obtain the prediction results of the three-dimensional multi-physics field of the fuel cell to be predicted;

Wherein, the construction process of the digital twin model of the preset fuel cell three-dimensional multi-physics field includes:

Solve the pre-built 3D multi-physics coupling model of the fuel cell to obtain several twin snapshots;

For each type of three-dimensional physical field in the fuel cell, several basis functions are constructed by using the intrinsic orthogonal decomposition method according to several twin snapshots, and the weights of each basis function direction of each three-dimensional physical field under each twin snapshot working condition are obtained. coefficient;

Using the weight coefficients of each basis function direction of each three-dimensional physical field under each twin snapshot working condition as a sample set, based on a pre-selected machine learning algorithm, interpolation algorithm or regression algorithm, the preset fuel cell three-dimensional multi-dimensional Digital twins of physical fields.

The present invention also provides a prediction device for the three-dimensional physical field of the fuel cell, comprising

memory for storing computer programs;

The processor is configured to realize the steps of the method for predicting the three-dimensional physical field of the fuel cell when executing the computer program.

The present invention also provides a computer-readable storage medium, the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method for predicting the three-dimensional physical field of the fuel cell are realized.

Compared with prior art, the beneficial effect of the present invention is:

The present invention provides a prediction method and system for the three-dimensional physical field of a fuel cell, which realizes the online prediction of the internal three-dimensional physical field of the fuel cell through several twin snapshots obtained based on off-line simulation; The coupling model of the fuel cell is solved to obtain several twin snapshots, and offline snapshots are provided; using the intrinsic orthogonal decomposition method, several basis functions are constructed for each type of three-dimensional physical field in the fuel cell, and each twin snapshot working condition is obtained The weight coefficients of each basis function direction of each three-dimensional physical field are used to realize the establishment of the basis function coordinate system describing the three-dimensional physical field of the fuel cell, and several twin snapshots under the three-dimensional Cartesian coordinate system are projected into the basis function coordinate system to obtain The weight coefficients of each basis function direction of each three-dimensional physical field in each snapshot working condition are obtained, and then the re-characterization of the physical field in the fuel cell is realized. The characterization method of the fuel cell transforms the prediction problem of the three-dimensional physical field of the fuel cell into the prediction problem of the weight coefficient, so that the prediction problem of the three-dimensional physical field of the fuel cell can be reduced on a large scale, and the prediction time can be reduced to the order of seconds; at the same time, Using the pre-selected machine learning algorithm model, interpolation algorithm model and or regression algorithm model as the basic framework of the prediction model, it meets the characteristics of building a prediction model based on fewer snapshots and realizes multiple input parameters, and solves the three-dimensional physical field of fuel cells There are many input parameters in the prediction problem but the number of snapshots should not be too many; the prediction process is simple, and the simulation speed of the three-dimensional multi-physics field for the proton exchange membrane fuel cell is fast, and the online prediction of the three-dimensional multi-physics field inside the fuel cell can be realized. The prediction efficiency and prediction result accuracy are high.

Description of drawings

Fig. 1 is the construction flowchart of the digital twin model of the preset fuel cell three-dimensional multi-physics field in the present invention;

Fig. 2 is the global relative error result diagram of electronic potential field, temperature field, film water content field and liquid water saturation field in the digital twin results of each random working condition in the embodiment;

Fig. 3 is a comparison diagram of the simulation results of a random working condition liquid water saturation field and the digital twin prediction results in the embodiment;

Fig. 4 is the comparison chart of the simulation result and digital twin prediction result of electric potential field of a certain random working condition in the embodiment;

Fig. 5 is a comparison diagram of the simulation results and the digital twin prediction results of the temperature field of a certain random working condition in the embodiment.

Detailed ways

In order to make the technical problems, technical solutions and beneficial effects solved by the present invention clearer, the present invention will be further described in detail below in conjunction with specific embodiments. It should be understood that the specific embodiments described here are only used to explain the present invention, not to limit the present invention.

The invention provides a method for predicting a three-dimensional physical field of a fuel cell, comprising the following steps:

Step 1. Obtain the variable input parameters of the random operating conditions of the fuel cell to be predicted; wherein, the variable input parameters of the random operating conditions of the fuel cell to be predicted include the cathode pressure, anode pressure, cathode humidity, and anode humidity of the fuel cell to be predicted. Humidity, cathode stoichiometry, anode stoichiometry, and operating temperature.

Step 2. The variable input parameters of the fuel cell random operating conditions to be predicted are used as the input of the preset digital twin model of the three-dimensional multi-physics field of the fuel cell, and the digital twin result of the three-dimensional multi-physics field is output, that is, the to-be 3D Multiphysics Prediction Results for Predicting Fuel Cells.

Wherein, the construction process of the digital twin model of the preset fuel cell three-dimensional multi-physics field includes:

Solve the pre-built 3D multi-physics coupling model of the fuel cell to obtain several twin snapshots;

For each type of three-dimensional physical field in the fuel cell, several basis functions are constructed by using the intrinsic orthogonal decomposition method according to several twin snapshots, and the weights of each basis function direction of each three-dimensional physical field under each twin snapshot working condition are obtained. coefficient;

Taking the weight coefficients of each basis function direction of each three-dimensional physical field under each twin snapshot working condition as a sample set, and based on a pre-selected machine learning algorithm, interpolation algorithm or regression algorithm, the preset fuel cell three-dimensional Multiphysics digital twins.

As shown in Figure 1, the construction process of the digital twin model of the preset fuel cell three-dimensional multi-physics field described in the present invention specifically includes the following steps:

Step 201, constructing a three-dimensional multi-physics coupling model of the fuel cell; wherein, the three-dimensional multi-physics coupling model of the fuel cell is a three-dimensional two-phase non-isothermal numerical simulation model based on continuous liquid pressure.

Step 202, according to the three-dimensional multi-physics coupling model of the fuel cell, obtain the model parameters of the fuel cell.

Wherein, the model parameters of the fuel cell include geometric parameters, physical parameters, electrochemical parameters and operating parameters of the fuel cell; specifically, the geometric parameters include the flow channel length, ridge width ratio and layer thickness of the fuel cell; The physical parameters include layer conductivity, diffusion coefficient, and density electrode thermal conductivity of the fuel cell; the electrochemical parameters include the reference exchange current, reference concentration, and platinum/carbon ratio of the fuel cell; the operating parameters include the cathode pressure of the fuel cell , anode pressure, cathode humidity, anode humidity, cathode stoichiometric ratio, anode stoichiometric ratio and operating temperature.

Step 203, select variable parameters from the model parameters of the fuel cell, and use the experimental design method to determine variable input parameters for several snapshot working conditions when the values of the variable parameters change; wherein, several snapshots Variable input parameters for operating conditions include: cathode pressure, anode pressure, cathode humidity, anode humidity, cathode stoichiometric ratio, anode stoichiometric ratio, and operating temperature of the fuel cell.

Step 204, using the variable input parameters of several snapshot working conditions as the input of the three-dimensional multi-physics coupling model of the fuel cell, through simulation, to obtain the three-dimensional multi-physics simulation results under several snapshot working conditions, that is, to obtain several twin snapshot.

Among them, the specific process is as follows:

Using C language, write the coupling model of the three-dimensional multi-physics field of the fuel cell to be predicted into ANSYS Fluent software, take the variable input parameters of several snapshot working conditions as input, and simulate the output to obtain several twin snapshots; among them, several twin snapshots The output is in the format of a Tecplot file.

Step 205 , for each type of three-dimensional physical field in the fuel cell, according to several twin snapshots, generate snapshot matrices under several snapshot working conditions.

Step 206 , for each type of three-dimensional physical field in the fuel cell, use the singular value decomposition method to perform matrix decomposition and matrix transformation processing on the snapshot matrices under several snapshot working conditions to obtain several basis functions.

Step 207 : Project several twin snapshots to each basis function direction, and reconstruct to obtain the weight coefficients of each basis function direction of each three-dimensional physical field under each twin snapshot working condition.

Step 208: Taking the weight coefficients of the basis function directions of each three-dimensional physical field under each twin snapshot working condition as a sample set, and based on a pre-selected machine learning algorithm, interpolation algorithm or regression algorithm, obtain the preset A 3D multiphysics digital twin of a fuel cell.

Wherein, the digital twin model of the preset fuel cell three-dimensional multi-physics field is:

Among them: r is the serial number of the physical quantity; f′ _r is the three-dimensional physical field of the r-th physical quantity under the variable input parameters of the fuel cell to be predicted; b′ _r,j is the three-dimensional physical field of the r-th physical quantity in the j-th basis function ψ _r,j is the jth basis function of the rth physical quantity; j is the number of the above basis functions; l _r is the truncation order of the rth physical quantity. The truncation order refers to the number of basis functions that need to be considered in this type of three-dimensional physical field.

The present invention also provides a prediction system for the three-dimensional physical field of the fuel cell, which includes an acquisition module and a prediction module; the acquisition module is used to obtain variable input parameters of the fuel cell random working conditions to be predicted; The variable input parameters of the random operating conditions of the fuel cell to be predicted are used as the input of the preset digital twin model of the fuel cell 3D multiphysics, and the output is the digital twin result of the 3D multiphysics field, that is, the 3D multiphysics of the fuel cell to be predicted is obtained. The prediction result of the field; wherein, the construction process of the digital twin model of the preset three-dimensional multi-physics field of the fuel cell includes: solving the coupling model of the pre-built three-dimensional multi-physics field of the fuel cell to obtain several twin snapshots; For each type of three-dimensional physical field in the fuel cell, use the intrinsic orthogonal decomposition method according to several twin snapshots to construct several basis functions, and obtain the weight coefficients of each basis function direction of each three-dimensional physical field under each twin snapshot working condition ; Using the weight coefficients of each basis function direction of each three-dimensional physical field under each twin snapshot working condition as a sample set, based on a pre-selected machine learning algorithm, interpolation algorithm or regression algorithm, the preset fuel cell three-dimensional Multiphysics digital twins.

The present invention also provides a prediction device for the three-dimensional physical field of the fuel cell, comprising: a memory for storing a computer program; and a processor for implementing the steps of the method for predicting the three-dimensional physical field of the fuel cell when executing the computer program.

When the processor executes the computer program, the steps of the method for predicting the above-mentioned three-dimensional physical field of the fuel cell are realized, for example: obtaining variable input parameters of the random operating conditions of the fuel cell to be predicted; The variable input parameters are used as the input of the preset digital twin model of the three-dimensional multi-physics field of the fuel cell, and the digital twin result of the three-dimensional multi-physics field is output, that is, the prediction result of the three-dimensional multi-physics field of the fuel cell to be predicted is obtained; Describe the construction process of the preset fuel cell 3D multiphysics digital twin model, including: solving the pre-built fuel cell 3D multiphysics coupling model to obtain several twin snapshots; for each type of fuel cell 3D In the physical field, according to several twin snapshots, the intrinsic orthogonal decomposition method is used to construct several basis functions, and the weight coefficients of each basis function direction of each three-dimensional physical field under each twin snapshot working condition are obtained; each twin snapshot The weight coefficients of each basis function direction of each three-dimensional physical field under working conditions are used as a sample set, and based on a pre-selected machine learning algorithm, interpolation algorithm or regression algorithm, the preset digital twin model of the three-dimensional multi-physics field of the fuel cell is obtained.

Or, when the processor executes the computer program, it realizes the functions of the modules in the above system, for example: the acquisition module is used to obtain the variable input parameters of the random operating conditions of the fuel cell to be predicted; the prediction module is used to convert the The variable input parameters of the random operating conditions of the fuel cell to be predicted are used as the input of the preset digital twin model of the fuel cell 3D multiphysics, and the output is the digital twin result of the 3D multiphysics field, that is, the 3D multiphysics of the fuel cell to be predicted is obtained. The prediction result of the field; wherein, the construction process of the digital twin model of the preset three-dimensional multi-physics field of the fuel cell includes: solving the coupling model of the pre-built three-dimensional multi-physics field of the fuel cell to obtain several twin snapshots; For each type of three-dimensional physical field in the fuel cell, use the intrinsic orthogonal decomposition method according to several twin snapshots to construct several basis functions, and obtain the weight coefficients of each basis function direction of each three-dimensional physical field under each twin snapshot working condition ; Using the weight coefficients of each basis function direction of each three-dimensional physical field under each twin snapshot working condition as a sample set, based on a pre-selected machine learning algorithm, interpolation algorithm or regression algorithm, the preset fuel cell three-dimensional Multiphysics digital twins.

Exemplarily, the computer program may be divided into one or more modules/units, and the one or more modules/units are stored in the memory and executed by the processor to complete the present invention. The one or more modules/units may be a series of computer program instruction segments capable of accomplishing preset functions, and the instruction segments are used to describe the execution process of the computer program in the fuel cell three-dimensional physical field prediction device . For example, the computer program can be divided into an acquisition module and a prediction module, and the specific functions of each module are as follows: the acquisition module is used to obtain the variable input parameters of the random operating conditions of the fuel cell to be predicted; the prediction module is used to The variable input parameters of the stochastic operating conditions of the fuel cell to be predicted are used as the input of the preset digital twin model of the three-dimensional multi-physics field of the fuel cell, and the digital twin result of the three-dimensional multi-physics field is output, that is, the fuel cell to be predicted is obtained 3D multiphysics prediction results.

The prediction equipment for the three-dimensional physical field of the fuel cell may be computing equipment such as desktop computers, notebooks, palmtop computers, and cloud servers. The prediction device for the three-dimensional physical field of the fuel cell may include, but not limited to, a processor and a memory. Those skilled in the art can understand that the above is an example of the prediction equipment of the three-dimensional physical field of the fuel cell, and does not constitute a limitation on the prediction equipment of the three-dimensional physical field of the fuel cell, and may include more components than the above, or combine some components, Or different components, for example, the prediction device for the three-dimensional physical field of the fuel cell may also include input and output devices, network access devices, buses, and the like.

The so-called processor can be a central processing unit (Central Processing Unit, CPU), and can also be other general-purpose processors, digital signal processors (Digital Signal Processor, DSP), application specific integrated circuits (Application Specific Integrated Circuit, ASIC), off-the-shelf programmable gate arrays (Field- ProgrammableGateArray, FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc., the processor is the control center of the prediction equipment of the three-dimensional physical field of the fuel cell, and connects the entire Various parts of the device for prediction of fuel cell 3D physics.

The memory can be used to store the computer programs and/or modules, and the processor implements the fuel by running or executing the computer programs and/or modules stored in the memory and calling the data stored in the memory. Prediction of various functions of the device by three-dimensional physics of the battery.

The memory may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system, at least one application program required by a function (such as a sound playback function, an image playback function, etc.) and the like; the storage data area may store Data created based on the use of the mobile phone (such as audio data, phonebook, etc.), etc. In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, internal memory, plug-in hard disk, smart memory card (SmartMediaCard, SMC), secure digital (SecureDigital, SD) card, flash memory card (FlashCard), at least one magnetic disk storage device, flash memory device, or other volatile solid-state storage device.

The present invention also provides a computer-readable storage medium, the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method for predicting the three-dimensional physical field of a fuel cell are realized .

If the integrated modules/units of the fuel cell three-dimensional physical field prediction system are realized in the form of software function units and sold or used as independent products, they can be stored in a computer-readable storage medium.

Based on such an understanding, the present invention realizes all or part of the process in the above-mentioned method for predicting the three-dimensional physical field of the fuel cell, and it can also be completed by instructing related hardware through a computer program, and the computer program can be stored in a computer-readable memory In the medium, when the computer program is executed by a processor, the steps of the above method for predicting the three-dimensional physical field of the fuel cell can be realized. Wherein, the computer program includes computer program code, and the computer program code may be in the form of source code, object code, executable file or preset intermediate form.

The computer-readable storage medium may include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a removable hard disk, a magnetic disk, an optical disk, a computer memory, and a read-only memory (Read-Only Memory, ROM) , random access memory (RandomAccessMemory, RAM), electric carrier signal, telecommunication signal and software distribution medium, etc.

It should be noted that the content contained in the computer-readable storage medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable Storage media excludes electrical carrier signals and telecommunication signals.

Example

Take the prediction process of the three-dimensional multi-physics field of the hydrogen fuel proton exchange membrane fuel cell as an example; wherein, the hydrogen fuel proton exchange membrane fuel cell is a parallel flow channel.

This embodiment provides a method for predicting the three-dimensional physical field of a fuel cell, including the following steps:

Step 1. Construct a three-dimensional multi-physics coupling model of the fuel cell; wherein, the fuel cell three-dimensional multi-physics coupling model is a three-dimensional two-phase non-isothermal numerical simulation model based on liquid pressure continuum; A three-dimensional two-phase non-isothermal numerical simulation model, including twelve conservation equations and several sub-models; wherein, the twelve conservation equations include: mass conservation equations, three momentum conservation equations, energy conservation equations, three component conservation equations Equations, electron potential conservation equations, proton potential conservation equations, membrane water conservation equations and liquid water pressure conservation equations; several sub-models include agglomeration sub-models, gas-liquid phase transition sub-models and membrane-water phase transition sub-models; wherein, the Three-dimensional multiphysics fields include, but are not limited to, electron potential fields, temperature fields, liquid water saturation fields, and film water content fields.

Step 2. According to the three-dimensional multi-physics coupling model of the fuel cell, the model parameters of the fuel cell are obtained; wherein, the model parameters of the fuel cell include geometric parameters, physical parameters, electrochemical parameters and operating parameters of the fuel cell .

Among them, the geometric parameters include the flow channel length, ridge width ratio and layer thickness of the fuel cell; the physical parameters include the layer conductivity, diffusion coefficient, and density electrode thermal conductivity of the fuel cell; the electrochemical parameters include the reference exchange current of the fuel cell, Reference concentration and platinum/carbon ratio; operating parameters include cathode pressure, anode pressure, cathode humidity, anode humidity, cathode stoichiometry, anode stoichiometry, and operating temperature of the fuel cell.

Step 3. Select variable parameters from the model parameters of the fuel cell, and adopt an experimental design method to determine variable input parameters for several snapshot working conditions when the values of the variable parameters change; wherein, several snapshots The variable input parameters of the working conditions include: cathode pressure, anode pressure, cathode humidity, anode humidity, cathode stoichiometric ratio, anode stoichiometric ratio and operating temperature of the fuel cell; in this embodiment, based on the method of paired experimental design, PICT (Pairwise Independent Combinatorial Testing) software was used to carry out experimental design, and a total of 139 sets of variable input parameters for snapshot conditions were designed.

Step 4. Using the variable input parameters of several snapshot working conditions as the input of the three-dimensional multi-physics coupling model of the fuel cell, several twin snapshots are obtained through simulation.

Among them, the simulation process is as follows:

Using C language, the fuel cell 3D multi-physics coupling model is written into ANSYS Fluent software, and the variable input parameters of several snapshot working conditions are used as input, and the simulation output obtains the 3D multi-physics simulation results under several snapshot working conditions; Wherein, the three-dimensional multi-physics simulation results under the several snapshot working conditions are output in the format of a Tecplot file.

In this embodiment, based on ANSYS Fluent software, a user-defined function (UDF) is written in C language, and the variable input parameters of several snapshot working conditions are used as input to realize the solution of the coupling model of the three-dimensional multi-physics field of the fuel cell; based on The ANSYS Fluent software interfaces with the file of the Tecplot software, analyzes the grid file at the same time, extracts the simulation results of the Tecplot software, and obtains the 3D multi-physics simulation results under several snapshot conditions output in the Tecplot file format.

Step 5. According to the simulation results of three-dimensional physical fields under several snapshot working conditions, for each three-dimensional physical field, generate snapshot matrices under several snapshot working conditions; specifically, realize the deconstruction of grid files and simulation results based on C++ self-programming, Realize the generation of snapshot matrix based on the simulation results; among them, under several snapshot conditions, the snapshot matrix of the rth 3D physical field is:

Among them: r is the number of the physical quantity or 3D physical field; F _r is the snapshot matrix of the rth 3D physical field; m is the number of grid nodes; _n is the number of snapshots; The value of the rth physical quantity on the i node; where, i is 1~m, and j is 1~n.

Step 6. For each three-dimensional physical field, based on the singular value decomposition method, perform matrix decomposition and matrix transformation on the snapshot matrix under several snapshot working conditions to obtain multiple three-dimensional basis functions; wherein, according to the singular value decomposition method, for For each 3D physical field, the snapshot matrix under several snapshot conditions can be decomposed into the following formula:

Among them: r is the serial number of the physical quantity; F _r is the snapshot matrix of the rth physical quantity; U _r is the left singular matrix formed by the left singular vector of the rth physical quantity; Σ _r is the singular value matrix of the rth physical quantity; V _r is the right singular matrix formed by the right singular vector of the rth physical quantity; ^* is the conjugate transpose of the matrix.

In this embodiment, the right singular matrix is obtained by using the unilateral Jacobi algorithm; wherein, the unilateral Jacobi algorithm is implemented based on C++; the specific process is as follows:

(1) Calculate the initial kernel matrix:

为第r个物理量第k次Jacobi变换后的核矩阵；F_r为第r个物理量的快照矩阵；·^*为矩阵的共轭转置。Among them: r is the serial number of the physical quantity; k is the Jacobi transformation number;

is the kernel matrix of the rth physical quantity after the kth Jacobi transformation; F _r is the snapshot matrix of the rth physical quantity; · ^* is the conjugate transposition of the matrix.

(2) Loop through all combinations of p and q that satisfy 1≤p<q≤n, and obtain the second-order principal subform of the kernel matrix as follows, and perform steps (3) and (4):

为第r个物理量第k次Jacobi变换后的核矩阵的第p行第p列；

为第r个物理量第k次Jacobi变换后的核矩阵的第p行第q列；

为第r个物理量第k次Jacobi变换后的核矩阵的第q行第p列；

为第r个物理量第k次Jacobi变换后的核矩阵的第q行第q列。Wherein: r is the numbering of physical quantity; p is the row number of described second-order main sub-form; q is the column number of described second-order main sub-form;

is the p-th row and p-th column of the kernel matrix after the k-th Jacobi transformation of the r-th physical quantity;

is the pth row and qth column of the kernel matrix after the kth Jacobi transformation of the rth physical quantity;

is the qth row and pth column of the kernel matrix after the kth Jacobi transformation of the rth physical quantity;

is the qth row and qth column of the kernel matrix after the kth Jacobi transformation of the rth physical quantity.

(3) Calculation of Jacobi matrix J _{r, k} :

Wherein: r is the numbering of physical quantity; p is the row numbering of described second-order principal subform; q is the column number of described second-order principal subform; θ _{r, k} is the rotation angle of the kth Jacobi transformation of the rth physical quantity; J _{r, k} is the matrix representation of the kth Jacobi transformation of the rth physical quantity; sin is the sine of the angle; cos is the cosine of the angle; cot is the cotangent of the angle.

(4) Update the kernel matrix:

为第r个物理量第k+1次Jacobi变换后的核矩阵；

为第r个物理量第k次Jacobi变换后的核矩阵；J_r,k为第r个物理量第k次Jacobi变换的矩阵表示；·^*为矩阵的共轭转置。Where: r is the serial number of the physical quantity;

is the kernel matrix of the rth physical quantity after the k+1th Jacobi transformation;

is the kernel matrix after the k-th Jacobi transformation of the r-th physical quantity; J _r,k is the matrix representation of the k-th Jacobi transformation of the r-th physical quantity; · ^* is the conjugate transposition of the matrix.

(5) Repeat step (2) continuously until the kernel matrix is transformed into a diagonal matrix.

(6) Calculation of the right singular vector:

Among them: r is the serial number of the physical quantity; N _r is the number of Jacobi transformations required to transform the kernel matrix into a diagonal matrix for the rth physical quantity; J _r,k (k takes 1 to N _r ) is the kth time Matrix representation of the Jacobi transform.

(7) Snapshot rotation:

ψ′ _r = F _r V _r

Among them: r is the serial number of the physical quantity; ψ′ _r is the matrix composed of unnormalized basis function vectors of the rth physical quantity; F _r is the snapshot matrix of the rth physical quantity; V _r is the right singular matrix of the rth physical quantity .

(8) Sort the column vectors of the matrix ψ′ _r according to their two norms (that is, the singular value or the amount of information contained) ||ψ′ _r,i || ₂ from large to small, and obtain the basis after normalization function:

Among them: r is the number of the physical quantity; i is the number of the basis function; ψ _r,i is the i-th basis function of the r-th physical quantity; ψ′ _r,i is the i-th unnormalized Basis function; ||·|| is the second norm of the vector.

Step 7. For each three-dimensional physical field, project the three-dimensional physical field under each snapshot condition to each basis function direction, and reconstruct to obtain the weights of each basis function direction of each three-dimensional physical field under each snapshot condition coefficient; the specific process is as follows:

Project the three-dimensional physical field under each snapshot condition to each basis function direction to obtain the corresponding weight coefficient of each snapshot condition in each basis function direction; where each snapshot condition corresponds to each basis function direction The weight coefficient is:

为第r个物理量第i个基函数的共轭转置；f_r,k为第k个快照中的第r个物理量的三维物理场。Among them: r is the serial number of the physical quantity; b _{r, i, k} represent the weight coefficients of the rth physical quantity projected on the i-th basis function in the k-th snapshot;

is the conjugate transpose of the i-th basis function of the r-th physical quantity; f _r,k is the three-dimensional physical field of the r-th physical quantity in the k-th snapshot.

Step 8. Taking the weight coefficients of each basis function direction of each three-dimensional physical field under each twin snapshot working condition as a sample set, based on a pre-selected machine learning algorithm, interpolation algorithm or regression algorithm, to obtain the preset fuel The digital twin model of the three-dimensional multi-physics field of the battery; preferably, the pre-selected machine learning algorithm is an artificial neural network algorithm, a multivariate adaptive regression spline algorithm or a deep learning algorithm.

In this embodiment, the preset truncation error TE% is selected, if l _r satisfies that the first l _r basis functions of the rth physical quantity ignore the amount of information less than TE% and the first _lr -1 of the rth physical quantity The basis function ignores the amount of information exceeding TE%, so l _r is called the truncation order under the rth physical quantity truncation error TE%. Among them, the preset digital twin model of the three-dimensional multi-physics field of the fuel cell is:

Among them: r is the serial number of the physical quantity; f′ _r is the three-dimensional physical field of the r-th physical quantity under the variable input parameters of the fuel cell to be predicted; b′ _r,j is the three-dimensional physical field of the r-th physical quantity in the j-th basis function ψ _r,j is the jth basis function of the rth physical quantity; j is the number of the above basis functions; l _r is the truncation order of the rth physical quantity.

Step 9. Obtain the variable input parameters of the random working conditions of the fuel cell to be predicted;

Step 10, using the variable input parameters of the random working conditions of the fuel cell to be predicted as the input of the preset digital twin model of the three-dimensional multi-physics field of the fuel cell, and outputting the result of the digital twin of the three-dimensional multi-physics field, that is, the to-be Predicting 3D multiphysics prediction results for fuel cells.

Test verification:

Taking 20 sets of randomly variable input parameters, TE%=0.01% as an example, the prediction effect of the three-dimensional multi-physics digital twin technology will be described below. The global relative error is defined by a two-norm, as shown in the following formula:

指三维数字孪生结果中第r个物理量的三维物理场；f_r,j指相同可变输入参数下的数值仿真结果中第r个物理量的三维物理场；δ_r为第r个物理量的全局相对误差。Where: r is the serial number of the physical quantity;

refers to the 3D physical field of the rth physical quantity in the 3D digital twin results; f _{r, j} refers to the 3D physical field of the rth physical quantity in the numerical simulation results under the same variable input parameters; δ _r is the global relative value of the rth physical quantity error.

As shown in Figure 2, Figure 2 shows the global relative error results of the electronic potential field, temperature field, membrane water content field and liquid water saturation field in the digital twin results of each random working condition in the embodiment ; It can be seen from Figure 2 that the global relative error of most working conditions is within 15%, which verifies the reliability of the three-dimensional multi-physics digital twin method provided by this embodiment.

As shown in accompanying drawings 3-5, accompanying drawing 3 provides the simulation results of a certain random working condition liquid water saturation field in the embodiment and the comparison chart of digital twin prediction results, and accompanying drawing 4 provides the embodiment A comparison diagram of the simulation results of the electronic potential field of a random working condition and the prediction results of the digital twin, and the comparison diagram of the simulation results of the temperature field of a random working condition in the embodiment and the prediction results of the digital twin are shown in Figure 5; It can be seen from the accompanying drawings 3-5 that the method for predicting the three-dimensional physical field of the fuel cell described in this embodiment can more accurately capture the global and local characteristics of each physical field, and the prediction accuracy is relatively high.

The prediction method of the three-dimensional physical field of the fuel cell described in this embodiment satisfies the requirement of fast and accurate prediction of the three-dimensional multi-physics field in the fuel cell based on a given numerical model of the fuel cell and based on a large number of off-line simulation results, and satisfies the requirements of the fuel cell on-line The demand for multi-physics prediction in the control scenario; among them, based on the intrinsic orthogonal decomposition technology, the fast and accurate prediction of the three-dimensional multi-physics field in the fuel cell is realized, and the pre-selected machine learning algorithm, interpolation algorithm or regression algorithm is used to analyze any stochastic The prediction of the weight coefficients under the conditions is carried out, and then the predicted weight coefficients and basis functions are used to quickly and accurately obtain the multi-physics prediction results under any random working conditions through the method of modal superposition.

It should be noted that the prediction method of the fuel cell three-dimensional physical field described in this embodiment is also applicable to other geometric structures, other types of fuel cells and other fuel cell models; for example: serpentine flow channel, solid oxide Fuel cells, three-dimensional single-phase thermal model, etc.

For the description of relevant parts of a fuel cell three-dimensional physical field prediction system, equipment, and computer-readable storage medium provided in this embodiment, please refer to the corresponding part of a fuel cell three-dimensional physical field prediction method described in this embodiment Detailed description will not be repeated here.

The prediction method, system, equipment and medium of the three-dimensional physical field of the fuel cell described in the present invention can predict the three-dimensional multi-physical field in the fuel cell within a few seconds, which verifies that the present invention can be applied to the online prediction of the internal physical field of the fuel cell predict.

The above-mentioned embodiment is only one of the implementation manners capable of realizing the technical solution of the present invention, and the scope of protection claimed by the present invention is not only limited by this embodiment, but also includes within the technical scope disclosed in the present invention, anyone familiar with this technology Changes, substitutions and other implementations that can easily occur to those skilled in the art.

Claims (10)

1. A method for predicting the three-dimensional physical field of a fuel cell, comprising: Obtain the variable input parameters of the random operating conditions of the fuel cell to be predicted; The variable input parameters of the stochastic operating conditions of the fuel cell to be predicted are used as the input of the preset digital twin model of the three-dimensional multi-physics field of the fuel cell, and the digital twin result of the three-dimensional multi-physics field is output, that is, the fuel cell to be predicted is obtained 3D multiphysics prediction results; Wherein, the construction process of the digital twin model of the preset fuel cell three-dimensional multi-physics field includes: Solve the pre-built 3D multi-physics coupling model of the fuel cell to obtain several twin snapshots; For each type of three-dimensional physical field in the fuel cell, several basis functions are constructed by using the intrinsic orthogonal decomposition method according to several twin snapshots, and the weights of each basis function direction of each three-dimensional physical field under each twin snapshot working condition are obtained. coefficient; Using the weight coefficients of each basis function direction of each three-dimensional physical field under each twin snapshot working condition as a sample set, based on a pre-selected machine learning algorithm, interpolation algorithm or regression algorithm, the preset fuel cell three-dimensional multi-dimensional Digital twins of physical fields. 2. The prediction method of a kind of fuel cell three-dimensional physical field according to claim 1, it is characterized in that, the variable input parameters of the stochastic working conditions of the fuel cell to be predicted include the cathode pressure and the anode pressure of the fuel cell to be predicted , cathode humidity, anode humidity, cathode stoichiometric ratio, anode stoichiometric ratio and operating temperature. 3. A method for predicting a fuel cell three-dimensional physical field according to claim 1, wherein the construction process of the digital twin model of the preset fuel cell three-dimensional multi-physics field is as follows: Construct a 3D multi-physics coupled model of the fuel cell; Acquiring model parameters of the fuel cell according to the three-dimensional multi-physics coupling model of the fuel cell; Select variable parameters from the model parameters of the fuel cell to obtain variable input parameters of several snapshot working conditions; The variable input parameters of several snapshot working conditions are used as the input of the three-dimensional multi-physics coupling model of the fuel cell, and several twin snapshots are obtained through simulation; For each type of three-dimensional physical field in the fuel cell, according to several twin snapshots, generate a snapshot matrix under several snapshot conditions; For each type of three-dimensional physical field in the fuel cell, use the singular value decomposition method to perform matrix decomposition and matrix transformation processing on the snapshot matrix under several snapshot conditions, and obtain several basis functions; Project several twin snapshots to each basis function direction, and reconstruct to obtain the weight coefficients of each basis function direction of each three-dimensional physical field under each twin snapshot working condition; Using the weight coefficients of each basis function direction of each three-dimensional physical field under each twin snapshot working condition as a sample set, based on a pre-selected machine learning algorithm, interpolation algorithm or regression algorithm, the preset fuel cell three-dimensional multi-dimensional Digital twins of physical fields. 4 . The method for predicting the three-dimensional physical field of a fuel cell according to claim 3 , wherein the model parameters of the fuel cell include geometric parameters, physical parameters, electrochemical parameters, and operating parameters of the fuel cell. 5 . The method for predicting a three-dimensional physical field of a fuel cell according to claim 3 , wherein the coupling model of the three-dimensional multi-physics field of the fuel cell is a three-dimensional two-phase non-isothermal numerical simulation model based on continuous liquid pressure. 6. The prediction method of a kind of fuel cell three-dimensional physical field according to claim 3, it is characterized in that, the variable input parameters of some snapshot operating conditions, as the input of the coupling model of the three-dimensional multi-physics field of fuel cell, through The process of simulating and obtaining several twin snapshots is as follows: Using C language, write the coupling model of the three-dimensional multi-physics field of the fuel cell to be predicted into ANSYS Fluent software, take the variable input parameters of several snapshot working conditions as input, and simulate the output to obtain several twin snapshots; among them, several twin snapshots The output is in the format of a Tecplot file. 7. The prediction method of a kind of fuel cell three-dimensional physical field according to claim 3, is characterized in that, the digital twin model of preset fuel cell three-dimensional multi-physics field is:

Among them, r is the serial number of the physical quantity; f _r ′ is the three-dimensional physical field of the r-th physical quantity under the variable input parameters of the fuel cell to be predicted; b _r ′ _,j is the three-dimensional physical field of the r-th physical quantity in the j-th basis function ψ _r,j is the jth basis function of the rth physical quantity; j is the number of the above basis functions; l _r is the truncation order of the rth physical quantity. 8. A prediction system for a fuel cell three-dimensional physical field, characterized in that it comprises: An acquisition module, configured to acquire variable input parameters of fuel cell stochastic operating conditions to be predicted; The prediction module is used to use the variable input parameters of the random operating conditions of the fuel cell to be predicted as the input of the preset digital twin model of the three-dimensional multi-physics field of the fuel cell, and output the digital twin result of the three-dimensional multi-physics field, namely Obtain the prediction results of the three-dimensional multi-physics field of the fuel cell to be predicted; Wherein, the construction process of the digital twin model of the preset fuel cell three-dimensional multi-physics field includes: Solve the pre-built 3D multi-physics coupling model of the fuel cell to obtain several twin snapshots; For each type of three-dimensional physical field in the fuel cell, several basis functions are constructed by using the intrinsic orthogonal decomposition method according to several twin snapshots, and the weights of each basis function direction of each three-dimensional physical field under each twin snapshot working condition are obtained. coefficient; Using the weight coefficients of each basis function direction of each three-dimensional physical field under each twin snapshot working condition as a sample set, based on a pre-selected machine learning algorithm, interpolation algorithm or regression algorithm, the preset fuel cell three-dimensional multi-dimensional Digital twins of physical fields. 9. A prediction device for a fuel cell three-dimensional physical field, characterized in that it includes memory for storing computer programs; A processor, configured to implement the steps of the method for predicting the three-dimensional physical field of a fuel cell according to any one of claims 1-7 when executing the computer program. 10. A computer-readable storage medium, the computer-readable storage medium stores a computer program, characterized in that, when the computer program is executed by a processor, the fuel cell according to any one of claims 1-7 is realized Steps in a prediction method for 3D physics.

Images