Performance prediction model determination method, performance prediction method and device of fuel cell

Abstract

The invention discloses a performance prediction model determination method, a performance prediction method and a performance prediction device of a fuel cell, mainly aiming at modeling simulation and performance prediction of the fuel cell, the method comprises the following steps: constructing a fitting coefficient training database of the fuel cell based on a semi-empirical physical model and a computational fluid mechanics model of the fuel cell, and training a neural network model based on the data in the fitting coefficient training database to obtain a target neural network model of the fuel cell, coupling the neural network model and a semi-empirical physical model, and obtaining a performance prediction model of the hybrid physical neural network architecture. And the performance prediction model is used for outputting a fitting coefficient of the semi-empirical physical model based on the input operation parameters, taking the fitting coefficient as the input of the semi-empirical physical model, and finally outputting performance parameters. Therefore, by effectively combining a machine learning method with a physical model, a hybrid physical-data driven performance prediction model with an efficient calculation characteristic and a high degree of detail can be formed.

Description

Technical Field

[0001] The present application belongs to the technical field of electrochemical fuel cells, and in particular relates to a method for determining a performance prediction model of a fuel cell, a performance prediction method, and a device. Background Art

[0002] In today's increasingly popular world of clean energy, hydrogen energy, with its exceptional cleanliness, high enthalpy, and ease of long-term storage, has attracted significant attention from both industry and academia. Proton exchange membrane fuel cells (PEMFCs), which convert hydrogen into electricity and heat through electrochemical reactions, have been widely used in the automotive, marine, and aerospace industries.

[0003] However, in ordinary research, when using fuel cell numerical models to calculate the performance parameters of fuel cells, there is an unreconcilable balance between the level of detail of the fuel cell numerical models and the computational efficiency, that is, models with a high level of detail often have lower computational efficiency.

[0004] Therefore, how to effectively improve the simulation calculation efficiency of the model while ensuring a high level of detail in the model is a technical problem that needs to be solved urgently. Summary of the Invention

[0005] The embodiments of the present application provide a method for determining a performance prediction model of a fuel cell, a performance prediction method and an apparatus, which can ensure that the numerical model has high computational efficiency at a high level of detail.

[0006] In a first aspect, an embodiment of the present application provides a method for determining a performance prediction model of a fuel cell, the method comprising:

[0007] Using preset operating parameters as input to a pre-built computational fluid dynamics model of the fuel cell, obtaining a first performance parameter of the fuel cell output by the computational fluid dynamics model according to the input preset operating parameters;

[0008] constructing a fuel cell performance parameter training database based on the preset operating parameters and the first performance parameters corresponding to the preset operating parameters;

[0009] Inputting the data in the performance parameter training database into a pre-built semi-empirical physical model of the fuel cell, and performing fitting calibration on the fitting coefficients in the semi-empirical physical model to obtain first fitting coefficients of the semi-empirical physical model corresponding to the preset operating parameters, wherein the semi-empirical physical model is used to reflect the mapping relationship between the operating parameters, fitting coefficients, and performance parameters of the fuel cell;

[0010] constructing a fitting coefficient training database according to the preset operating parameters in the performance parameter training database and the first fitting coefficient corresponding to the preset operating parameters;

[0011] Training a preset neural network model based on the data in the fitting coefficient training database to obtain a target neural network model, wherein the target neural network model is used to output corresponding fitting coefficients based on the input operating parameters;

[0012] The target neural network model and the semi-empirical physical model are fused to obtain a performance prediction model of the fuel cell, and the output of the target neural network model in the performance prediction model is used as the fitting coefficient input of the semi-empirical physical model.

[0013] In some embodiments, before the preset operating parameters are used as input for the pre-built computational fluid dynamics model of the fuel cell, the method includes: using the pre-built initial computational fluid dynamics model of the fuel cell, performing simulation under the experimental operating parameters in the experimental database, and obtaining second performance parameters corresponding to the experimental operating parameters; the experimental database includes experimental operating parameters and experimental performance parameters corresponding to the experimental operating parameters; comparing the second performance parameters corresponding to the experimental operating parameters with the corresponding experimental performance parameters to determine the accuracy of the initial computational fluid dynamics model; if the accuracy of the initial computational fluid dynamics model does not reach the preset accuracy threshold, adjusting the initial computational fluid dynamics model, and returning to the step of using the pre-built initial computational fluid dynamics model of the fuel cell, performing simulation under the experimental operating parameters in the experimental database, and obtaining the second performance parameters corresponding to the experimental operating parameters, until the accuracy of the initial computational fluid dynamics model reaches the preset accuracy threshold, and the computational fluid dynamics model is obtained.

[0014] In some embodiments, the above-mentioned fitting calibration of the fitting coefficients in the semi-empirical physical model to obtain the first fitting coefficients of the semi-empirical physical model corresponding to the preset operating parameters includes: using the least squares method to fit and calibrate the fitting coefficients in the semi-empirical physical model to obtain the first fitting coefficients of the semi-empirical physical model corresponding to the preset operating parameters.

[0015] In some embodiments, the above method includes at least one of the following items: the semi-empirical physical model includes an electrical performance calculation sub-model, and the electrical performance calculation sub-model is used to reflect the mapping relationship between operating parameters, electrical performance parameters and a second fitting coefficient, and the second fitting coefficient includes at least one fitting coefficient; the semi-empirical physical model includes a thermal performance calculation sub-model, and the thermal performance calculation sub-model is used to reflect the mapping relationship between operating parameters, thermal performance parameters and a third fitting coefficient, and the third fitting coefficient includes at least one fitting coefficient, and the third fitting coefficient is different from the second fitting coefficient; the semi-empirical physical model includes a tail gas characteristic calculation sub-model, and the tail gas characteristic calculation sub-model is used to reflect the mapping relationship between operating parameters, tail gas characteristic parameters and a fourth fitting coefficient, and the fourth fitting coefficient includes at least one fitting coefficient, and the fourth fitting coefficient is different from the second fitting coefficient and the third fitting coefficient.

[0016] In a second aspect, an embodiment of the present application provides a fuel cell performance prediction method, the method comprising:

[0017] Obtaining actual operating parameters of the fuel cell;

[0018] Inputting the actual operating parameters into the performance prediction model of the fuel cell according to the first aspect;

[0019] Using the target neural network model in the performance prediction model, the fitting coefficients in the semi-empirical physical model are predicted based on the actual operating parameters to obtain target fitting coefficients;

[0020] The actual performance parameters of the fuel cell output by the semi-empirical physical model are obtained by taking the actual operating parameters and the corresponding target fitting coefficients as inputs of the semi-empirical physical model in the performance prediction model.

[0021] In a third aspect, an embodiment of the present application provides a device for determining a performance prediction model of a fuel cell, the device comprising:

[0022] a parameter acquisition module, configured to use preset operating parameters as input to a pre-built computational fluid dynamics model of the fuel cell, and acquire a first performance parameter of the fuel cell output by the computational fluid dynamics model corresponding to the input preset operating parameters;

[0023] a first training data building module, configured to build a fuel cell performance parameter training database based on the preset operating parameters and the first performance parameters corresponding to the preset operating parameters;

[0024] a coefficient acquisition module, configured to input the data in the performance parameter training database into a pre-built semi-empirical physical model of the fuel cell, and perform fitting calibration on the fitting coefficients in the semi-empirical physical model to obtain first fitting coefficients of the semi-empirical physical model corresponding to the preset operating parameters, wherein the semi-empirical physical model is configured to reflect a mapping relationship between the operating parameters, fitting coefficients, and performance parameters of the fuel cell;

[0025] a second training data construction module, configured to construct a fitting coefficient training database based on the preset operating parameters in the performance parameter training database and the first fitting coefficients corresponding to the preset operating parameters;

[0026] A model training module is used to train a preset neural network model based on the data in the fitting coefficient training database to obtain a target neural network model, wherein the target neural network model is used to output corresponding fitting coefficients based on the input operating parameters;

[0027] A model fusion module is used to fuse the target neural network model and the semi-empirical physical model to obtain a performance prediction model of the fuel cell, and the output of the target neural network model in the performance prediction model is used as the fitting coefficient input of the semi-empirical physical model.

[0028] In a fourth aspect, an embodiment of the present application provides a fuel cell performance prediction device, the device comprising:

[0029] A parameter acquisition module is used to obtain the actual operating parameters of the fuel cell;

[0030] a model input module, configured to input the actual operating parameters into the performance prediction model of the fuel cell according to the first aspect;

[0031] a coefficient determination module, configured to predict the fitting coefficients in the semi-empirical physical model based on the actual operating parameters using a target neural network model in the performance prediction model to obtain target fitting coefficients;

[0032] The performance calculation module is used to obtain the actual performance parameters of the fuel cell output by the semi-empirical physical model by taking the actual operating parameters and the corresponding target fitting coefficients as inputs of the semi-empirical physical model in the performance prediction model.

[0033] In a fifth aspect, an embodiment of the present application provides an electronic device, the electronic device comprising: a processor and a memory storing computer program instructions;

[0034] When the processor executes the computer program instructions, it implements the fuel cell performance prediction model determination method as described in the first aspect; or, it implements the fuel cell performance prediction method as described in the second aspect.

[0035] In a sixth aspect, an embodiment of the present application provides a computer storage medium, on which computer program instructions are stored. When the computer program instructions are executed by a processor, the method for determining a performance prediction model of a fuel cell as described in the first aspect is implemented; or, the method for determining a performance prediction model of a fuel cell as described in the second aspect is implemented.

[0036] In the seventh aspect, an embodiment of the present application provides a computer program product. When the instructions in the computer program product are executed by the processor of an electronic device, the electronic device executes the method for determining the performance prediction model of the fuel cell as described in the first aspect; or, implements the performance prediction method of the fuel cell as described in the second aspect.

[0037] The fuel cell performance prediction model determination method, performance prediction method, device, equipment, computer storage medium, and computer program product of the embodiments of the present application construct a fuel cell fitting coefficient training database based on a semi-empirical physical model and a computational fluid dynamics model of the fuel cell, train a neural network model based on the data in the fitting coefficient training database, obtain a target neural network model for the fuel cell, couple the neural network model and the semi-empirical physical model to obtain a performance prediction model with a hybrid physical neural network architecture, and output fitting coefficients of the semi-empirical physical model based on input operating parameters. The fitting coefficients are used as input to the semi-empirical physical model to ultimately output performance parameters. In this way, by effectively combining machine learning methods with the fuel cell physical model, a hybrid physical-data-driven performance prediction model with highly efficient computational characteristics and a high level of detail can be formed. In addition, the corresponding target fitting coefficients can be obtained for the current actual operating parameters, and then the semi-empirical physical model calculations can be performed based on the target fitting coefficients, thereby improving the effectiveness of obtaining the actual performance parameters of the fuel cell. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0039] Figure 1 1 is a flow chart of a method for determining a fuel cell performance prediction model provided in an embodiment of the present application;

[0040] Figure 2 Schematic diagram of the processing flow of the fuel cell performance prediction model provided in the embodiment of the present application;

[0041] Figure 3 1 is a flow chart of a fuel cell performance prediction method provided in an embodiment of the present application;

[0042] Figure 4 Schematic diagram of the structure of a fuel cell performance prediction model determination device provided in an embodiment of the present application;

[0043] Figure 5 Schematic diagram of the structure of a fuel cell performance prediction device provided in an embodiment of the present application;

[0044] Figure 6 It is a structural diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0045] The features and exemplary embodiments of various aspects of the present application will be described in detail below. In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain the present application, rather than to limit the present application. For those skilled in the art, the present application can be implemented without the need for some of these specific details. The following description of the embodiments is merely to provide a better understanding of the present application by illustrating the examples of the present application.

[0046] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article, or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or device. In the absence of further limitations, the elements defined by the phrase "comprising..." do not exclude the presence of other identical elements in the process, method, article, or device comprising the elements.

[0047] To solve the problems of the prior art, the present invention provides a method, apparatus, device, computer storage medium, and computer program product for determining a fuel cell performance prediction model. The method for determining a fuel cell performance prediction model provided by the present invention is first introduced below.

[0048] Figure 1 FIG1 shows a flow chart of a method for determining a fuel cell performance prediction model according to an embodiment of the present application. Figure 1As shown, the method specifically includes the following steps S101 to S104:

[0049] In step S101 , preset operating parameters are used as inputs of a pre-built computational fluid dynamics model of a fuel cell, and first performance parameters of the fuel cell outputted by the computational fluid dynamics model according to the input preset operating parameters are obtained.

[0050] Step S102 : constructing a fuel cell performance parameter training database based on preset operating parameters and first performance parameters corresponding to the preset operating parameters.

[0051] In step S103 , the data in the performance parameter training database is input into a pre-built semi-empirical physical model of the fuel cell, and the fitting coefficients in the semi-empirical physical model are fitted and calibrated to obtain first fitting coefficients of the semi-empirical physical model corresponding to the preset operating parameters.

[0052] Step S104: constructing a fitting coefficient training database according to the preset operating parameters in the performance parameter training database and the first fitting coefficients corresponding to the preset operating parameters.

[0053] Step S105 , training a preset neural network model based on the data in the fitting coefficient training database to obtain a target neural network model.

[0054] Step S106 , fusing the target neural network model and the semi-empirical physical model to obtain a fuel cell performance prediction model.

[0055] The semi-empirical physical model is used to reflect the mapping relationship between the operating parameters, fitting coefficients, and performance parameters of the fuel cell, and can be used to describe the electrochemical behavior and thermodynamic characteristics of the fuel cell. The fuel cell can be a proton exchange membrane fuel cell (PEMFC), a solid oxide fuel cell (SOFC), etc.

[0056] The operating parameters of the above-mentioned fuel cell can be understood as parameters that affect the operating state of the fuel cell. The operating parameters of the fuel cell may include at least one of the load current, anode stoichiometric ratio, cathode stoichiometric ratio, inlet pressure, anode relative humidity, cathode relative humidity, coolant inlet temperature, coolant temperature rise, etc.

[0057] The above fitting coefficients can be understood as coefficients that can be fixed values ​​in the semi-empirical physical model and can be adjusted accordingly according to different operating parameters.

[0058] The above-mentioned performance parameters can be key indicators for evaluating the working efficiency, output capacity and stability of fuel cells. The performance parameters may include at least one of electrochemical performance parameters, thermal performance parameters, tail gas characteristic parameters, etc. The electrochemical performance parameters may include the total voltage of the fuel cell stack, the thermal performance parameters may include the net output thermal power of the coolant, the tail gas characteristic parameters may include at least one of the tail gas flow rate, the tail gas specific heat capacity and the tail gas temperature. The tail gas flow rate may include at least one of the anode tail gas mass flow rate and the cathode tail gas mass flow rate, the tail gas specific heat capacity may include at least one of the anode tail gas specific heat capacity and the cathode tail gas specific heat capacity, and the tail gas temperature may include at least one of the anode tail gas flow temperature and the cathode tail gas flow temperature.

[0059] Since the output performance (ie, performance parameters) of the fuel cell varies under different operating parameters, there is a mapping relationship between the operating parameters and the performance parameters.

[0060] In step S101, different preset operating parameters can be input into a pre-built computational fluid dynamics model of the fuel cell, and the computational fluid dynamics model of the fuel cell is used to perform simulations under different preset operating parameters to obtain first performance parameters corresponding to the different preset operating parameters.

[0061] In step S102, different preset operating parameters and first performance parameters corresponding to the different preset operating parameters may be used to construct a fuel cell performance parameter training database, in which the preset operating parameters and the first performance parameters have a corresponding relationship.

[0062] In this embodiment, simulation using a computational fluid dynamics model of a fuel cell can effectively obtain a large number of performance parameters under preset operating parameters, and can quickly build a performance parameter training database.

[0063] In step S103, the above-mentioned performance parameter training database may include different parameter data, and the parameter data include preset operating parameters and first performance parameters corresponding to the preset operating parameters. For example, the performance parameter training database includes preset operating parameter A, preset operating parameter B, preset operating parameter C, first performance parameter A corresponding to preset operating parameter A, first performance parameter B corresponding to preset operating parameter B, and first performance parameter C corresponding to preset operating parameter C. The different parameter data are preset operating parameter A-first performance parameter A, preset operating parameter B-first performance parameter B, and preset operating parameter C-first performance parameter C.

[0064] The parameter data in the performance parameter training database may be input into the semi-empirical physical model through a streaming input method, for example, the parameter data in the performance parameter training database may be input into the semi-empirical physical model one by one or in batches.

[0065] Since the semi-empirical physical model is used to reflect the mapping relationship between the operating parameters, fitting coefficients and performance parameters of the fuel cell, the value of the fitting coefficient in the semi-empirical physical model can be determined based on the operating parameters and performance parameters. In this embodiment, a preset method is used to fit and calibrate the fitting coefficient in the semi-empirical physical model to obtain a first fitting coefficient corresponding to the preset operating parameters.

[0066] In step S104, training data is obtained by establishing a correspondence between the preset operating parameters in the performance parameter training database and the corresponding first fitting coefficients, and a fitting coefficient training database is formed based on the training data corresponding to each preset operating parameter in the performance parameter training database.

[0067] For example, the performance parameter training database includes preset operating parameters A, preset operating parameters B, and preset operating parameters C. The first fitting coefficient corresponding to the preset operating parameter A is the first fitting coefficient A, the first fitting coefficient corresponding to the preset operating parameter B is the first fitting coefficient B, and the first fitting coefficient corresponding to the preset operating parameter C is the first fitting coefficient C. Then the preset operating parameter A-first fitting coefficient A is the training data A, the preset operating parameter B-first fitting coefficient B is the training data B, and the preset operating parameter C-first fitting coefficient C is the training data C. The training data A, training data B, and training data C constitute the fitting coefficient training database.

[0068] The above-mentioned target neural network model can reflect the mapping relationship between operating parameters and fitting coefficients. The above-mentioned target neural network model is used to output corresponding fitting coefficients based on the input operating parameters. The corresponding fitting coefficients can be calculated based on the input operating parameters by the target neural network model, and the fitting coefficients corresponding to the input operating parameters can be output.

[0069] In step S105 , the neural network model may be trained using any of the following methods: optimizing the objective function, iterative algorithm, minimizing the loss function, optimizing parameters using the gradient descent method, and the like.

[0070] The above-mentioned neural network model can be any one of a convolutional neural network (CNN), a feedforward neural network (FNN), a recurrent neural network (RNN), a backpropagation neural network (BPNN), etc.

[0071] In this embodiment, the training data in the training database can be divided into a training set and a verification set in a ratio of 3:1. The training set is used to train the neural network model, and the verification set is used to verify the performance of the trained neural network model to obtain the optimal target neural network model.

[0072] In step S106, the target neural network model and the semi-empirical physical model can be fused through the model stacking method to obtain a performance prediction model of the fuel cell. The performance prediction model of the fuel cell integrates the target neural network model and the semi-empirical physical model, so that the corresponding target fitting coefficient can be obtained based on the input operating parameters through the target neural network, and the target performance parameters can be calculated based on the target fitting coefficient through the semi-empirical physical model.

[0073] In an embodiment of the present application, a fuel cell performance prediction model is obtained by integrating a target neural network model and a semi-empirical physical model, which can effectively combine machine learning methods with fuel cell physical models to form a hybrid physical-data-driven performance prediction model with extremely efficient computing characteristics and a high degree of detail.

[0074] In some embodiments, the semi-empirical physical model includes an electrical performance calculation sub-model, which is used to reflect the mapping relationship between the operating parameters, the electrical performance parameters, and the second fitting coefficients. The second fitting coefficients include at least one fitting coefficient.

[0075] The electrical performance parameter can be the voltage of a single cell in the stack or the total voltage of the stack. The voltage of a single cell in the stack can be calculated using the following formula (1):

[0076] E=E ner -E act -E ohm -E con (1);

[0077] Where E represents the voltage of the battery stack, E ner represents the Nernst voltage, E act represents activation polarization, E ohm Indicates ohmic polarization, E con represents concentration polarization.

[0078] The total stack voltage is the sum of the voltages of all the individual cells in the fuel cell stack.

[0079] The Nernst voltage can be calculated as follows:

[0080]

[0081] Where △G represents the Gibbs free energy, which can be taken as 2.37×10 5 J / mol, F represents the Faraday constant, which can be taken as 96485C / mol, and T represents the operating temperature of the fuel cell. They represent the partial pressure of hydrogen and oxygen respectively.

[0082] The partial pressure of the above-mentioned inlet hydrogen Inlet oxygen partial pressure They can be calculated using the following formulas (3) and (4):

[0083]

[0084] Among them, Rh a With Rh c Respectively represent the relative humidity of the anode inlet gas and the cathode inlet gas, p a With p c They represent the anode inlet pressure and the cathode inlet pressure respectively, i represents the current density, p sat represents the saturated water vapor pressure, and T represents the fuel cell operating temperature.

[0085] The above saturated water vapor pressure p sat It can be calculated as follows:

[0086]

[0087] Wherein, T represents the operating temperature of the fuel cell.

[0088] The above activation polarization E act It can be calculated as follows:

[0089]

[0090] Where i represents the current density, i o represents the reference current density, ξ1, ξ2, ξ3 represent the second fitting coefficients, ξ1, ξ2, ξ3 are different fitting coefficients, represents the local oxygen concentration, T represents the fuel cell operating temperature, T ref is the reference temperature, which can be taken as 298.15K.

[0091] The above local oxygen concentration It can be calculated as follows (7):

[0092]

[0093] in, represents the partial pressure of oxygen in the intake air, and T represents the operating temperature of the fuel cell.

[0094] The above ohmic polarization E ohm It can be calculated as follows (8):

[0095]

[0096] Among them, ξ4 represents the second fitting coefficient, ω m is the ionic conductivity of the proton exchange membrane, ξ4, ξ1, ξ2, and ξ3 are different fitting coefficients.

[0097] The ionic conductivity ω of the above proton exchange membrane m It can be calculated as follows (9):

[0098]

[0099] Where λ is the membrane water content and T is the fuel cell operating temperature.

[0100] The above concentration polarization Econ can be calculated as follows (10):

[0101] E con =ξ5·exp(ξ6·i) (10);

[0102] Among them, ξ5 and ξ6 represent the second fitting coefficients, ξ5, ξ6, ξ4, ξ1, ξ2, and ξ3 are different fitting coefficients, and i represents the current density.

[0103] In some embodiments, the electrical performance parameter may also include the fuel cell electrical power P e , fuel cell power P e It can be calculated by the following formula (11):

[0104] P e =E·I·N (11);

[0105] Among them, P e It represents the electric power output of the fuel cell, E represents the voltage of a single cell in the stack, I represents the current, and N represents the number of cells in the stack.

[0106] In some embodiments, the semi-empirical physical model includes a thermal performance calculation sub-model, which is used to reflect the mapping relationship between operating parameters, thermal performance parameters and a third fitting coefficient. The third fitting coefficient includes at least one fitting coefficient, and the third fitting coefficient is different from the second fitting coefficient.

[0107] Thermal performance parameters can include the heat power ΔP removed by cooling water t，co , can be calculated by the following formula (12):

[0108]

[0109] in, represents the third fitting coefficient, P t，ge Indicates the thermal power generated by the electrochemical reaction in the battery stack, P t，loss Indicates the heat power lost by the fuel cell stack in the air.

[0110] The heat power P generated by the electrochemical reaction in the above stack t，ge It can be calculated by the following formula (13):

[0111]

[0112] Where I represents the current, N represents the number of cells in the stack, and ΔH represents the enthalpy of hydrogen, which can be taken as 2.86×10 5 J / mol, E represents the voltage of the battery stack, and F represents the Faraday constant, which can be taken as 96485C / mol.

[0113] The heat power P lost by the above stack in the air t，loss It can be calculated by the following formula (14):

[0114] P t,loss =|h·f·(T am -T)+σ·γ·f·(T am 4 -T 4 )| (14);

[0115] Where h represents the convection heat transfer coefficient, which can be taken as 9W / (m 2 ·K), f represents the surface area of ​​the battery stack, T am represents the ambient temperature, and σ represents the Stefan-Boltzmann constant, which can be taken as 5.67×10 -8 W / (m 2 ·K 4 ), γ represents the emissivity, which can be taken as 0.9, and T represents the operating temperature of the fuel cell.

[0116] Thermal performance parameters can also include coolant flow rate, which can be calculated by the following formula (15):

[0117]

[0118] Where ΔP t，co Indicates the heat power taken away by cooling water, C coIndicates the specific heat capacity of the coolant, ΔT co Indicates the coolant temperature rise.

[0119] Thermal performance parameters can also include anode inlet heat power Cathode inlet heat power Anode outlet thermal power and cathode outlet heat power At least one of the following.

[0120] The above-mentioned anode inlet heat power It can be calculated as follows (16):

[0121]

[0122] in, represents the anode inlet specific heat capacity, represents the anode inlet gas flow rate, T a in Indicates the anode inlet air temperature.

[0123] The above cathode inlet heat power It can be calculated as follows (17):

[0124]

[0125] in, represents the cathode inlet gas specific heat capacity, represents the cathode inlet gas flow rate, T c in Indicates the cathode inlet gas temperature.

[0126] The above-mentioned anode outlet heat power It can be calculated as follows (18):

[0127]

[0128] in, represents the specific heat capacity of the anode outlet gas, represents the anode gas flow rate, T a out Indicates the anode outlet gas temperature.

[0129] The above cathode outlet heat power It can be calculated as follows (19):

[0130]

[0131] in, represents the cathode outlet gas specific heat capacity, Indicates cathode outlet gas flow rate, T c out Indicates cathode outlet gas temperature.

[0132] In some embodiments, the semi-empirical physical model includes a tail gas characteristic calculation sub-model, which is used to reflect the mapping relationship between operating parameters, tail gas characteristic parameters and a fourth fitting coefficient. The fourth fitting coefficient includes at least one fitting coefficient, and the fourth fitting coefficient is different from the second fitting coefficient and the third fitting coefficient.

[0133] The tail gas characteristic parameters may include the anode tail gas mass flow rate Cathode tail gas mass flow rate Specific heat capacity of anode tail gas Specific heat capacity of cathode tail gas Anode tail exhaust temperature T a out , cathode tail exhaust temperature T c out At least one of the following.

[0134] The above-mentioned anode tail gas mass flow rate It can be calculated as follows (20):

[0135]

[0136] in, represents the anode inlet gas flow rate, is the hydrogen consumption rate, It represents the transport rate of water molecules across the membrane.

[0137] The above-mentioned anode gas reaction consumption rate It can be calculated as follows (21):

[0138]

[0139] in, It represents the molar mass of hydrogen, which can be taken as 0.002 kg / mol, I represents the current, N represents the number of cells in the stack, and F represents the Faraday constant.

[0140] The above water molecule transmembrane transport rate It can be calculated as follows (22):

[0141]

[0142] Among them, m per Indicates the water molecule penetration rate caused by pressure difference, m osm Represents the electroosmotic migration rate of water molecules.

[0143] The water molecule penetration rate m caused by the above pressure difference per It can be calculated as follows (23):

[0144]

[0145] Among them, ρ m represents the density of the proton exchange membrane, EW represents the equivalent mass of the membrane, η represents the transmembrane permeability coefficient, κ l represents the dynamic viscosity of water, p l，CCL With p l，ACL are the capillary pressure of liquid water in the cathode catalyst layer and the capillary pressure of liquid water in the anode catalyst layer, respectively. m Represents the thickness of the proton exchange membrane, and N represents the number of cells in the stack.

[0146] The dynamic viscosity of water κ l It can be calculated as follows (24):

[0147]

[0148] Wherein, T represents the operating temperature of the fuel cell.

[0149] The capillary pressure of liquid water in the cathode catalyst layer is p l，CCL , the capillary pressure of liquid water in the anode catalyst layer p l，ACL It can be calculated as follows (25) and (26):

[0150] p l =

[0151] 101325p g -δ l ·cosθ·(ε / k) 0.5 ·[1.42(1-s)-2.12(1-s) 2 +1.26(1-s) 3 ] θ<90°(25);

[0152] p l =

[0153] 101325p g -δ l ·cosθ·(ε / k) 0.5 (1.42s-2.12s 2 +1.26s 3 ) θ>90° (26);

[0154] Among them, p l For p l，CCL or p l，ACL , p g represents the inlet pressure, θ represents the contact angle, ε represents the porosity, δ l represents the surface tension coefficient, k is the thermal conductivity, and s is the liquid water saturation.

[0155] The above surface tension coefficient δ l It can be calculated as follows (27):

[0156] δ l =-0.0001676T+0.1218 (27);

[0157] Wherein, T represents the operating temperature of the fuel cell.

[0158] The thermal conductivity k can be calculated as follows:

[0159]

[0160] Among them, d x d y Represents the thickness of adjacent layer x, the thickness of adjacent layer y, k x 、k y denote the thermal conductivity of the adjacent layer x and the thermal conductivity of the adjacent layer y, respectively.

[0161] The above liquid water saturation s can be calculated as follows (29):

[0162]

[0163] Among them, c sat represents the saturated water vapor concentration, is the water vapor concentration, ρ l is the density of liquid water,

[0164] Represents the molar mass of water.

[0165] The above saturated water vapor concentration c sat It can be calculated as follows (30):

[0166]

[0167] Among them, R represents the gas constant, which can be taken as 8.314 J / (mol·K), p sat Indicates the saturated water vapor pressure.

[0168] The above water vapor concentration It can be calculated as follows (31):

[0169]

[0170] Among them, V CL represents the volume of the catalyst layer, λ sat represents the saturated membrane water content, T represents the fuel cell operating temperature, R represents the gas constant, Rh a Indicates the relative humidity of the anode inlet air, psat represents the saturated water vapor pressure, ρ m represents the density of the proton exchange membrane, EW represents the equivalent mass of the membrane, and λ is the membrane water content.

[0171] The above saturated membrane water content λ sat It can be calculated as follows (32):

[0172] λ sat =0.043+17.18w-39.85w 2 +36w 3 (32);

[0173] Where w represents water activity, which can be approximated to relative humidity.

[0174] The electroosmotic migration rate m of the water molecules mentioned above osm It can be calculated as follows (33):

[0175]

[0176] in, represents the molar mass of water, which can be taken as 0.018 kg / mol, μ represents the fourth fitting coefficient, which can be the electroosmotic migration coefficient, A represents the active area of ​​the fuel cell, F represents the Faraday constant, which can be taken as 96485 C / mol, i represents the current density, and N represents the number of cells in the stack.

[0177] The above cathode tail gas mass flow rate It can be calculated as follows (34):

[0178]

[0179] in, represents the cathode reaction gas consumption rate, represents the cathode reaction gas consumption rate, It represents the transport rate of water molecules across the membrane.

[0180] The cathode reaction gas consumption rate It can be calculated as follows (35):

[0181]

[0182] in, represents the molar mass of oxygen, which can be taken as 0.032 kg / mol, I represents the current, A represents the active area of ​​the fuel cell, N represents the number of cells in the stack, and F represents the Faraday constant, which can be taken as 96485 C / mol. Represents the molar mass of water.

[0183] The above-mentioned anode tail gas specific heat capacity It can be calculated as follows (36):

[0184]

[0185] in, are the specific heat capacities of hydrogen and water vapor, respectively. They respectively represent the mass proportion of hydrogen in the tail exhaust mixed gas and the mass proportion of water vapor in the tail exhaust mixed gas.

[0186] The mass proportion of hydrogen in the above exhaust mixed gas Mass proportion of water vapor in the exhaust mixed gas They can be calculated as follows:

[0187]

[0188] Among them, α a represents the anode stoichiometric ratio, I represents the current, N represents the number of cells in the stack, and F represents the Faraday constant. represents the molar mass of hydrogen, Indicates the anode tail gas mass flow rate.

[0189] The above cathode tail gas specific heat capacity It can be calculated as follows (39):

[0190]

[0191] in, They represent the specific heat capacity of oxygen and nitrogen respectively. They respectively represent the mass proportion of oxygen in the tail exhaust mixed gas, the mass proportion of water vapor in the tail exhaust mixed gas, and the mass proportion of nitrogen in the tail exhaust mixed gas.

[0192] The mass ratio of oxygen in the above exhaust mixed gas It can be calculated as follows (40):

[0193]

[0194] Among them, α c represents the cathode stoichiometric ratio, I represents the current, N represents the number of cells in the stack, and F represents the Faraday constant. represents the molar mass of oxygen, Indicates the cathode tail gas mass flow rate.

[0195] The mass proportion of water vapor in the above exhaust mixed gas It can be calculated as follows (41):

[0196]

[0197] in, represents the mass ratio of water vapor entering the cathode, represents the cathode inlet gas flow rate, represents the molar mass of water, i represents the current density, N represents the number of cells in the stack, A represents the active area of ​​the fuel cell, and F represents the Faraday constant. represents the transport rate of water molecules across the membrane, Indicates the cathode tail gas mass flow rate.

[0198] The mass ratio of water vapor entering the cathode can be calculated as follows:

[0199]

[0200] Among them, Mair represents the molar mass of air, which can be taken as 0.029 kg / mol. Indicates the molar mass of water, Rh c Indicates the relative humidity of cathode inlet gas, p sat represents the saturated water vapor pressure, p c Indicates cathode inlet pressure.

[0201] The mass proportion of nitrogen in the above exhaust mixed gas It can be calculated as follows (43):

[0202]

[0203] in, Indicates the mass proportion of oxygen in the tail exhaust mixed gas, Indicates the mass proportion of water vapor in the exhaust mixed gas.

[0204] The above-mentioned anode tail gas flow temperature T a out It can be calculated as follows (44):

[0205]

[0206] in, Indicates the thermal power of the anode tail exhaust mixed gas, represents the specific heat capacity of the anode tail gas, Indicates the anode tail gas mass flow rate.

[0207] The thermal power of the anode tail exhaust mixed gas It can be calculated as follows (45):

[0208]

[0209] in, represents the fourth fitting coefficient, Unlike μ, P t，ge Indicates the thermal power generated by the electrochemical reaction in the battery stack, P t，loss Indicates the heat power lost by the stack in the air, Indicates the anode outlet thermal power.

[0210] The above cathode tail gas flow temperature T c out It can be calculated as follows (46):

[0211]

[0212] in, Indicates the thermal power of the cathode tail exhaust mixed gas, represents the specific heat capacity of cathode tail gas, Indicates the cathode tail gas mass flow rate.

[0213] The above cathode tail exhaust mixed gas thermal power It can be calculated as follows (47):

[0214]

[0215] in, represents the third fitting coefficient, represents the fourth fitting coefficient, P t，ge Indicates the thermal power generated by the electrochemical reaction in the battery stack, P t，loss Indicates the heat power lost by the stack in the air, Indicates the cathode outlet thermal power.

[0216] In some embodiments, before using the preset operating parameters as input to the pre-built computational fluid dynamics model of the fuel cell, the following steps may also be included but are not limited to:

[0217] A pre-built initial computational fluid dynamics model of the fuel cell is used to perform simulation under experimental operating parameters in the experimental database to obtain second performance parameters corresponding to the experimental operating parameters.

[0218] The above-mentioned experimental database includes experimental operation parameters and experimental performance parameters corresponding to the experimental operation parameters.

[0219] The steps for building an initial computational fluid dynamics model for a fuel cell can be as follows: first, establish a geometric model based on the fuel cell type (such as a proton exchange membrane fuel cell), then simplify the geometric model based on the research objectives. This can be created using CAD software (such as SolidWorks, AutoCAD) or built-in tools in CFD software. The geometric model is divided into discrete grid cells, with finer grids used in key areas (such as electrode surfaces and reaction areas). Then, physical models are set up, such as electrochemical reaction models, fluid flow models, heat transfer models, and mass transfer models. Finally, boundary conditions and initial conditions are defined.

[0220] The boundary conditions and initial conditions of the initial computational fluid dynamics model are defined as experimental operating parameters and then a simulation is performed, and performance parameters corresponding to the experimental operating parameters are output to obtain second performance parameters.

[0221] The second performance parameter corresponding to the experimental operating parameter is compared with the corresponding experimental performance parameter to determine the accuracy of the initial computational fluid dynamics model.

[0222] In one embodiment, the accuracy of the initial computational fluid dynamics model may be determined by establishing a corresponding relationship between the error between the second performance parameter and the experimental performance parameter and the model accuracy.

[0223] Specifically, the second performance parameter and experimental performance parameter corresponding to each experimental operating parameter in the experimental database are obtained. For each experimental operating parameter in the experimental database, the error between the second performance parameter and the experimental performance parameter corresponding to the experimental operating parameter is calculated. Then, the average value of the error corresponding to each experimental operating parameter in the experimental database is calculated. Based on the average value, the corresponding model accuracy is determined from the corresponding relationship between the error and the model accuracy to obtain the accuracy of the initial computational fluid dynamics model.

[0224] The accuracy of the initial computational fluid dynamics model is matched with a preset accuracy threshold. If the accuracy of the initial computational fluid dynamics model reaches the preset accuracy threshold, the initial computational fluid dynamics model is determined as the final computational fluid dynamics model of the fuel cell (i.e., the computational fluid dynamics model of the fuel cell used in step S101). If the accuracy of the initial computational fluid dynamics model does not reach the preset accuracy threshold, the initial computational fluid dynamics model is adjusted, and the process returns to the step of using the pre-built initial computational fluid dynamics model of the fuel cell to perform simulation under the experimental operating parameters in the experimental database to obtain the second performance parameter corresponding to the experimental operating parameters, until the accuracy of the initial computational fluid dynamics model reaches the preset accuracy threshold, thereby obtaining the computational fluid dynamics model.

[0225] In addition, if there are experimental operation parameters in the experimental database whose errors are greater than the preset error threshold, it is directly determined that the accuracy of the initial computational fluid dynamics model does not reach the preset accuracy threshold.

[0226] In one embodiment, the initial computational fluid dynamics model may be adjusted by adjusting hyperparameters of the initial computational fluid dynamics model.

[0227] In this embodiment, the initial computational fluid dynamics model is adjusted to obtain a target fuel cell CFD model (ie, the computational fluid dynamics model of the fuel cell) that can meet accuracy requirements, thereby improving the effectiveness of the target fuel cell CFD model.

[0228] In some embodiments, step S101 may include but is not limited to the following steps:

[0229] The least square method is used to fit and calibrate the fitting coefficients in the semi-empirical physical model to obtain the first fitting coefficients of the semi-empirical physical model corresponding to the preset operating parameters.

[0230] In this embodiment, the least square method is simple to calculate and easy to implement. By using the least square method to perform fitting calibration on the fitting coefficients in the semi-empirical physical model, the use of computing resources can be reduced.

[0231] In order to better understand the above method, an embodiment of the present invention provides a complete embodiment of a method for determining a performance prediction model of a fuel cell.

[0232] Taking a proton exchange membrane fuel cell as an example, the fuel cell physical properties and design parameters can be set according to Table 1 below.

[0233] Table 1

[0234]

[0235]

[0236] A semi-empirical physical model of a proton exchange membrane fuel cell is constructed. The semi-empirical physical model includes an electrical performance calculation sub-model, a thermal performance calculation sub-model, and a tail gas characteristic calculation sub-model. The electrical performance calculation sub-model is used to calculate the total voltage of the fuel cell stack, the thermal performance parameters are used to calculate the net output thermal power of the coolant, and the tail gas characteristic calculation sub-model is used to calculate the anode tail gas mass flow rate, cathode tail gas mass flow rate, anode tail gas specific heat capacity, cathode tail gas specific heat capacity, anode tail gas temperature, and cathode tail gas temperature. The construction of the semi-empirical physical model can refer to the above equations (1) to (46).

[0237] An initial CFD model of a proton exchange membrane fuel cell was constructed using ANSYS Fluent software, and the model's preset accuracy threshold was calibrated. The initial CFD model was used to perform simulations under the experimental operating parameters in the experimental database to obtain a second performance parameter corresponding to the experimental operating parameters. The second performance parameter and experimental performance parameter corresponding to each experimental operating parameter in the experimental database were obtained. For each experimental operating parameter in the experimental database, the error between the second performance parameter and the experimental performance parameter corresponding to the experimental operating parameter was calculated. The average of the errors corresponding to each experimental operating parameter was then calculated. Based on the average, the corresponding model accuracy was determined from the preset relationship between the error and model accuracy to obtain the accuracy of the initial CFD model. The accuracy of the initial CFD model was then matched with the preset accuracy threshold. If the accuracy of the initial CFD model met the preset accuracy threshold, the initial CFD model was determined as the target fuel cell CFD model. If the accuracy of the initial CFD model does not reach the preset accuracy threshold, the initial CFD model is adjusted and the accuracy of the adjusted initial CFD model is re-determined, that is, the adjusted initial CFD model is used to re-execute the simulation under the experimental operating parameters in the experimental database to obtain the second performance parameters corresponding to the experimental operating parameters, until the accuracy of the initial CFD model reaches the preset accuracy threshold, thereby obtaining the target fuel cell CFD model.

[0238] Using the quasi-horizontal orthogonal experimental design method, a quasi-horizontal orthogonal experimental table was constructed with preset operating parameters such as coolant inlet temperature, coolant temperature rise, anode stoichiometric ratio, cathode stoichiometric ratio, anode relative humidity, cathode relative humidity, and inlet pressure as input variables. This formed the input parameter end of the fuel cell operating parameter-thermoelectric performance database (i.e., the performance parameter training database). Refer to Table 2 below, which shows the quasi-horizontal orthogonal experimental table used to construct the fuel cell operating parameter-thermoelectric performance database, which shows 49 sets of preset operating parameters.

[0239] Table 2

[0240]

[0241]

[0242] According to the input parameter design of the quasi-horizontal orthogonal test table, the performance parameters of the fuel cell under different preset operating parameters are obtained, and the fuel cell operating parameter-thermoelectric performance database (i.e., performance parameter training database) is constructed, which can be expressed as:

[0243]

[0244] in, is the coolant inlet temperature, ΔT co is the coolant temperature rise, αa , α c are the anode stoichiometric ratio and the cathode stoichiometric ratio, respectively, and Rh a , Rh c are the relative humidity of the anode inlet gas and the cathode inlet gas, respectively, p in is the inlet pressure (atm), E s is the total voltage of the stack, ΔP t，co is the net heat output of the coolant (W), and are the tail gas mass flow rates of the anode and cathode respectively (kg / s), and are the tail gas specific heat capacity of the anode and cathode respectively (J / (kg·K)), T a out With T c out are the tail gas flow temperatures (K) of the anode and cathode, respectively.

[0245] Based on the fuel cell operating parameter-thermoelectric performance database (i.e., performance parameter training database), the fitting coefficients of the semi-empirical physical model are fitted and calibrated by the least squares method, and the fitting coefficient calibration values ​​(i.e., the first fitting coefficients) under different preset operating parameter settings are solved to construct the operating parameter-fitting coefficient database (i.e., fitting coefficient training database). The first fitting coefficients include ξ1, ξ2, ξ3, ξ4, ξ5, ξ6, η, μ, The operating parameter-fitting coefficient database can be expressed as:

[0246] Based on MATLAB numerical calculation software, the back propagation neural network model with 10×2 hidden layer structure is used to construct the operating parameters. With the fitting coefficients (ξ1, ξ2, ξ3, ξ4, ξ5, ξ6, eta, μ, ), that is, the neural network model is trained based on the training data in the fitting coefficient training database to obtain the target neural network model.

[0247] By integrating the semi-empirical physical model of the fuel cell with the operating parameter-fitting coefficient dynamic mapping neural network model (i.e., the target neural network model), a performance prediction model of the fuel cell is obtained.

[0248] Reference Figure 2 , Figure 2 This is a schematic diagram of the processing flow of the fuel cell performance prediction model provided in this embodiment.

[0249] Based on the above method, by constructing a neural network model mapping between operating parameters and the fitting coefficients in the semi-empirical physical model, the fitting coefficients in the semi-empirical physical model can be adaptively adjusted for multiple operating conditions, providing important technical support for improving the generalization prediction accuracy of the fuel cell model under multiple operating conditions.

[0250] Furthermore, machine learning methods offer advantages such as high-dimensional modeling and efficient computing, significantly improving the detail and computational efficiency of fuel cell models. Performance prediction models avoid the problem of purely data-driven machine learning methods lacking the inherent physical mechanisms of fuel cells and struggling to achieve basic physical conditions such as system energy conservation. Therefore, effectively combining machine learning methods with fuel cell physics has resulted in a highly computationally efficient and highly detailed data-driven fuel cell physical model.

[0251] Reference Figure 3 , Figure 3 This is a flow chart of a fuel cell performance prediction method provided in an embodiment of the present application. This method is an application of the above-mentioned fuel cell performance prediction model, such as Figure 3 As shown, the fuel cell performance prediction method specifically includes the following steps S201 to S204:

[0252] Step S201: obtaining actual operating parameters of the fuel cell.

[0253] Step S202 : inputting actual operating parameters into a fuel cell performance prediction model.

[0254] Step S203 , predicting the fitting coefficients in the semi-empirical physical model based on the actual operating parameters using the target neural network model in the performance prediction model of the fuel cell to obtain target fitting coefficients.

[0255] Step S204 , using the actual operating parameters and the corresponding target fitting coefficients as inputs of a semi-empirical physical model in a performance prediction model of the fuel cell, thereby obtaining the actual performance parameters of the fuel cell output by the semi-empirical physical model.

[0256] Based on the above method, it is possible to obtain the corresponding target fitting coefficient for the current operating parameters, and then perform semi-empirical physical model calculations based on the target fitting coefficients, thereby improving the effectiveness of obtaining the actual performance parameters of the fuel cell.

[0257] In order to better implement the above-mentioned fuel cell performance prediction model determination method, the embodiment of the present application provides a fuel cell performance prediction model determination device, referring to Figure 4 , Figure 4 A schematic diagram of the structure of a fuel cell performance prediction model determination device provided in an embodiment of the present application, such as Figure 4As shown, the fuel cell performance prediction model determination device 40 specifically includes the following:

[0258] The parameter acquisition module 401 is configured to use preset operating parameters as input to a pre-built computational fluid dynamics model of the fuel cell, and to acquire first performance parameters of the fuel cell output by the computational fluid dynamics model according to the input preset operating parameters.

[0259] The first training data building module 402 is configured to build a fuel cell performance parameter training database based on preset operating parameters and first performance parameters corresponding to the preset operating parameters.

[0260] The coefficient acquisition module 403 is used to input the data in the performance parameter training database into the pre-built semi-empirical physical model of the fuel cell, and to fit and calibrate the fitting coefficients in the semi-empirical physical model to obtain the first fitting coefficients of the semi-empirical physical model corresponding to the preset operating parameters. The semi-empirical physical model is used to reflect the mapping relationship between the operating parameters, fitting coefficients and performance parameters of the fuel cell.

[0261] The second training data construction module 404 is configured to construct a fitting coefficient training database according to the preset operating parameters in the performance parameter training database and the first fitting coefficients corresponding to the preset operating parameters.

[0262] The model training module 405 is used to train a preset neural network model based on the data in the fitting coefficient training database to obtain a target neural network model. The target neural network model is used to output corresponding fitting coefficients based on the input operating parameters.

[0263] The model fusion module 406 is used to fuse the target neural network model and the semi-empirical physical model to obtain a performance prediction model of the fuel cell. The output of the target neural network model in the performance prediction model is used as the fitting coefficient input of the semi-empirical physical model.

[0264] In some embodiments, the parameter acquisition module 401 is specifically used to: use a pre-built initial computational fluid dynamics model of the fuel cell to perform simulation under the experimental operating parameters in the experimental database to obtain a second performance parameter corresponding to the experimental operating parameters; the experimental database includes experimental operating parameters and experimental performance parameters corresponding to the experimental operating parameters; compare the second performance parameters corresponding to the experimental operating parameters with the corresponding experimental performance parameters to determine the accuracy of the initial computational fluid dynamics model; if the accuracy of the initial computational fluid dynamics model does not reach the preset accuracy threshold, adjust the initial computational fluid dynamics model, and return to the step of using the pre-built initial computational fluid dynamics model of the fuel cell to perform simulation under the experimental operating parameters in the experimental database to obtain the second performance parameter corresponding to the experimental operating parameters, until the accuracy of the initial computational fluid dynamics model reaches the preset accuracy threshold and the computational fluid dynamics model is obtained.

[0265] In some embodiments, the coefficient acquisition module 403 is specifically used to: use the least squares method to perform fitting calibration on the fitting coefficients in the semi-empirical physical model to obtain first fitting coefficients of the semi-empirical physical model corresponding to the preset operating parameters.

[0266] In some embodiments, the semi-empirical physical model in the above-mentioned coefficient acquisition module 403 includes at least one of the following items: the semi-empirical physical model includes an electrical performance calculation sub-model, which is used to reflect the mapping relationship between operating parameters, electrical performance parameters and the second fitting coefficient, and the second fitting coefficient includes at least one fitting coefficient; the semi-empirical physical model includes a thermal performance calculation sub-model, which is used to reflect the mapping relationship between operating parameters, thermal performance parameters and the third fitting coefficient, and the third fitting coefficient includes at least one fitting coefficient, and the third fitting coefficient is different from the second fitting coefficient; the semi-empirical physical model includes a tail gas characteristic calculation sub-model, which is used to reflect the mapping relationship between operating parameters, tail gas characteristic parameters and the fourth fitting coefficient, and the fourth fitting coefficient includes at least one fitting coefficient, and the fourth fitting coefficient is different from the second fitting coefficient and the third fitting coefficient.

[0267] Based on the above-mentioned fuel cell performance prediction model determination device, by effectively combining machine learning methods with fuel cell physical models, a hybrid physical-data driven performance prediction model with extremely efficient computing characteristics and a high degree of detail can be formed.

[0268] In order to better implement the above-mentioned fuel cell performance prediction method, the present application embodiment provides a fuel cell performance prediction device, referring to Figure 5 , Figure 5 A schematic diagram of the structure of a fuel cell performance prediction device provided in an embodiment of the present application is shown in FIG. Figure 5As shown, the fuel cell performance prediction device 50 specifically includes the following:

[0269] The parameter acquisition module 501 is used to acquire the actual operating parameters of the fuel cell.

[0270] A model input module 502 is used to input actual operating parameters into the fuel cell performance prediction model;

[0271] The coefficient determination module 503 is used to predict the fitting coefficients in the semi-empirical physical model based on the actual operating parameters through the target neural network model in the performance prediction model to obtain the target fitting coefficients.

[0272] The performance calculation module 504 is configured to obtain the actual performance parameters of the fuel cell output by the semi-empirical physical model by using the actual operating parameters and the corresponding target fitting coefficients as inputs to the semi-empirical physical model in the performance prediction model.

[0273] Based on the above-mentioned fuel cell performance prediction device, it is possible to obtain the corresponding target fitting coefficient for the current operating parameters, and then perform semi-empirical physical model calculations based on the target fitting coefficients, thereby improving the effectiveness of obtaining the actual performance parameters of the fuel cell.

[0274] Figure 6 A schematic diagram of the hardware structure of an electronic device provided in an embodiment of the present application is shown.

[0275] The electronic device may include a processor 601 and a memory 602 storing computer program instructions.

[0276] Specifically, the processor 601 may include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or may be configured to implement one or more integrated circuits of the embodiments of the present application.

[0277] The memory 602 may include a large capacity memory for data or instructions. By way of example and not limitation, the memory 602 may include a hard disk drive (HDD), a floppy disk drive, a flash memory, an optical disk, a magneto-optical disk, a magnetic tape, or a universal serial bus (USB) drive, or a combination of two or more of these. Where appropriate, the memory 602 may include removable or non-removable (or fixed) media. Where appropriate, the memory 602 may be inside or outside the integrated gateway disaster recovery device. In a specific embodiment, the memory 602 is a non-volatile solid-state memory.

[0278] In some embodiments, the memory 602 may include read-only memory (ROM), random access memory (RAM), magnetic disk storage media devices, optical storage media devices, flash memory devices, electrical, optical, or other physical / tangible memory storage devices. Thus, in general, the memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., memory devices) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors), it is operable to perform the operations described with reference to the methods according to the present disclosure.

[0279] The processor 601 reads and executes computer program instructions stored in the memory 602 to implement any one of the fuel cell performance prediction model determination methods or fuel cell performance prediction methods in the above embodiments.

[0280] In one example, the electronic device may further include a communication interface 603 and a bus 610. Figure 6 As shown, the processor 601, the memory 602, and the communication interface 603 are connected via a bus 610 and communicate with each other.

[0281] The communication interface 603 is mainly used to implement communication between various modules, devices, units and / or equipment in the embodiments of the present application.

[0282] Bus 610 includes hardware, software or both, and the performance prediction device of fuel cell or the performance prediction model of fuel cell determine the parts of equipment are coupled to each other.For example, and not limitation, bus may include accelerated graphics port (AGP) or other graphics bus, enhanced industry standard architecture (EISA) bus, front side bus (FSB), hypertransport (HT) interconnection, industry standard architecture (ISA) bus, infinite bandwidth interconnection, low pin count (LPC) bus, memory bus, micro channel architecture (MCA) bus, peripheral component interconnection (PCI) bus, PCI-Express (PCI-X) bus, serial advanced technology attachment (SATA) bus, video electronics standard association local (VLB) bus or other suitable bus or two or more of these combinations. Where appropriate, bus 610 may include one or more buses. Although the present application embodiment describes and shows a specific bus, the application considers any suitable bus or interconnection.

[0283] The electronic device can execute the fuel cell performance prediction model determination method or fuel cell performance prediction method in the embodiment of the present application, thereby realizing the combination Figure 1 and Figure 4 Described fuel cell performance prediction model determination device, or realizes the combination Figure 3 and Figure 5A fuel cell performance prediction apparatus is described.

[0284] In addition, in conjunction with the fuel cell performance prediction model determination method or fuel cell performance prediction method in the above-mentioned embodiments, embodiments of the present application may provide a computer storage medium for implementation. The computer storage medium stores computer program instructions; when the computer program instructions are executed by a processor, any of the fuel cell performance prediction model determination method or fuel cell performance prediction method in the above-mentioned embodiments is implemented.

[0285] In combination with the fuel cell performance prediction model determination method or fuel cell performance prediction method in the above-mentioned embodiments, an embodiment of the present application also provides a computer program product. When the instructions in the computer program product are executed by the processor of an electronic device, the electronic device implements the fuel cell performance prediction model determination method or fuel cell performance prediction method in the above-mentioned embodiments.

[0286] It should be understood that the present application is not limited to the specific configurations and processes described above and illustrated in the figures. For the sake of brevity, a detailed description of known methods is omitted here. In the above embodiments, several specific steps are described and illustrated as examples. However, the method process of the present application is not limited to the specific steps described and illustrated. Those skilled in the art can make various changes, modifications, and additions, or change the order of the steps after understanding the spirit of the present application.

[0287] The functional blocks shown in the above-described block diagram can be implemented as hardware, software, firmware or a combination thereof. When implemented in hardware, they can be, for example, electronic circuits, application specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, etc. When implemented in software, the elements of the present application are programs or code segments that are used to perform the required tasks. The program or code segment can be stored in a machine-readable medium, or transmitted on a transmission medium or a communication link by a data signal carried in a carrier wave. "Machine-readable medium" can include any medium that can store or transmit information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROMs, flash memories, erasable ROMs (EROMs), floppy disks, CD-ROMs, optical disks, hard disks, optical fiber media, radio frequency (RF) links, etc. The code segment can be downloaded via a computer network such as the Internet, an intranet, etc.

[0288] It should also be noted that the exemplary embodiments mentioned in this application describe some methods or systems based on a series of steps or devices. However, this application is not limited to the order of the above steps. In other words, the steps can be performed in the order mentioned in the embodiments, or in a different order, or several steps can be performed simultaneously.

[0289] Aspects of the present disclosure have been described above with reference to the flowcharts and / or block diagrams of the methods, devices (systems) and computer program products according to the embodiments of the present disclosure. It should be understood that each box in the flowchart and / or block diagram and the combination of each box in the flowchart and / or block diagram can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer or other programmable data processing device to produce a machine so that these instructions executed by the processor of the computer or other programmable data processing device enable the implementation of the function / action specified in one or more boxes of the flowchart and / or block diagram. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor or a field programmable logic circuit. It is also understood that each box in the block diagram and / or flowchart and the combination of the boxes in the block diagram and / or flowchart can also be implemented by dedicated hardware that performs the specified function or action, or can be implemented by a combination of dedicated hardware and computer instructions.

[0290] The above description is only a specific embodiment of the present application. Those skilled in the art will clearly understand that for the convenience and brevity of description, the specific working processes of the systems, modules and units described above can refer to the corresponding processes in the aforementioned method embodiments, and will not be repeated here. It should be understood that the scope of protection of the present application is not limited thereto. Any person skilled in the art can easily think of various equivalent modifications or replacements within the technical scope disclosed in the present application, and these modifications or replacements should be included in the scope of protection of the present application.

Claims

1. A method for determining a fuel cell performance prediction model, characterized in that: The method comprises: Using preset operating parameters as input to a pre-built computational fluid dynamics model of the fuel cell, obtaining a first performance parameter of the fuel cell output by the computational fluid dynamics model according to the input preset operating parameters; constructing a fuel cell performance parameter training database based on the preset operating parameters and the first performance parameters corresponding to the preset operating parameters; Inputting the data in the performance parameter training database into a pre-built semi-empirical physical model of the fuel cell, and performing fitting calibration on the fitting coefficients in the semi-empirical physical model to obtain first fitting coefficients of the semi-empirical physical model corresponding to the preset operating parameters, wherein the semi-empirical physical model is used to reflect the mapping relationship between the operating parameters, fitting coefficients, and performance parameters of the fuel cell; constructing a fitting coefficient training database according to the preset operating parameters in the performance parameter training database and the first fitting coefficient corresponding to the preset operating parameters; Training a preset neural network model based on the data in the fitting coefficient training database to obtain a target neural network model, wherein the target neural network model is used to output corresponding fitting coefficients based on the input operating parameters; The target neural network model and the semi-empirical physical model are fused to obtain a performance prediction model of the fuel cell, and the output of the target neural network model in the performance prediction model is used as the fitting coefficient input of the semi-empirical physical model.

2. The method according to claim 1, characterized in that Before using the preset operating parameters as input to the pre-built computational fluid dynamics model of the fuel cell, the method includes: Using a pre-built initial computational fluid dynamics model of the fuel cell, a simulation is performed under experimental operating parameters in an experimental database to obtain a second performance parameter corresponding to the experimental operating parameters; the experimental database includes the experimental operating parameters and the experimental performance parameters corresponding to the experimental operating parameters; comparing the second performance parameter corresponding to the experimental operating parameter with the corresponding experimental performance parameter to determine the accuracy of the initial computational fluid dynamics model; If the accuracy of the initial computational fluid dynamics model does not reach the preset accuracy threshold, the initial computational fluid dynamics model is adjusted, and the process returns to executing the step of using the pre-built initial computational fluid dynamics model of the fuel cell to perform simulation under the experimental operating parameters in the experimental database to obtain the second performance parameters corresponding to the experimental operating parameters, until the accuracy of the initial computational fluid dynamics model reaches the preset accuracy threshold and the computational fluid dynamics model is obtained.

3. The method according to claim 1, characterized in that The fitting and calibration of the fitting coefficients in the semi-empirical physical model to obtain first fitting coefficients of the semi-empirical physical model corresponding to the preset operating parameters includes: The least square method is used to perform fitting calibration on the fitting coefficients in the semi-empirical physical model to obtain first fitting coefficients of the semi-empirical physical model corresponding to the preset operating parameters.

4. The method according to claim 1, wherein The method comprises at least one of the following: The semi-empirical physical model includes an electrical performance calculation sub-model, which is used to reflect the mapping relationship between the operating parameters, the electrical performance parameters and the second fitting coefficients, wherein the second fitting coefficients include at least one fitting coefficient; The semi-empirical physical model includes a thermal performance calculation sub-model, the thermal performance calculation sub-model is used to reflect the mapping relationship between the operating parameters, the thermal performance parameters and the third fitting coefficient, the third fitting coefficient includes at least one fitting coefficient, and the third fitting coefficient is different from the second fitting coefficient; The semi-empirical physical model includes a tail gas characteristic calculation sub-model, which is used to reflect the mapping relationship between operating parameters, tail gas characteristic parameters and a fourth fitting coefficient. The fourth fitting coefficient includes at least one fitting coefficient, and the fourth fitting coefficient is different from the second fitting coefficient and the third fitting coefficient.

5. A fuel cell performance prediction method, characterized in that: The method comprises: Obtaining actual operating parameters of the fuel cell; Inputting the actual operating parameters into a performance prediction model of the fuel cell according to any one of claims 1 to 4; Using the target neural network model in the performance prediction model, the fitting coefficients in the semi-empirical physical model are predicted based on the actual operating parameters to obtain target fitting coefficients; The actual performance parameters of the fuel cell output by the semi-empirical physical model are obtained by taking the actual operating parameters and the corresponding target fitting coefficients as inputs of the semi-empirical physical model in the performance prediction model.

6. A fuel cell performance prediction model determination device, characterized in that: The device comprises: a parameter acquisition module, configured to use preset operating parameters as input to a pre-built computational fluid dynamics model of the fuel cell, and acquire a first performance parameter of the fuel cell output by the computational fluid dynamics model corresponding to the input preset operating parameters; a first training data building module, configured to build a fuel cell performance parameter training database based on the preset operating parameters and the first performance parameters corresponding to the preset operating parameters; a coefficient acquisition module, configured to input the data in the performance parameter training database into a pre-built semi-empirical physical model of the fuel cell, and perform fitting calibration on the fitting coefficients in the semi-empirical physical model to obtain first fitting coefficients of the semi-empirical physical model corresponding to the preset operating parameters, wherein the semi-empirical physical model is configured to reflect a mapping relationship between the operating parameters, fitting coefficients, and performance parameters of the fuel cell; a second training data construction module, configured to construct a fitting coefficient training database based on the preset operating parameters in the performance parameter training database and the first fitting coefficients corresponding to the preset operating parameters; A model training module is used to train a preset neural network model based on the data in the fitting coefficient training database to obtain a target neural network model, wherein the target neural network model is used to output corresponding fitting coefficients based on the input operating parameters; A model fusion module is used to fuse the target neural network model and the semi-empirical physical model to obtain a performance prediction model of the fuel cell, and the output of the target neural network model in the performance prediction model is used as the fitting coefficient input of the semi-empirical physical model.

7. A fuel cell performance prediction device, characterized in that: The device comprises: A parameter acquisition module is used to obtain the actual operating parameters of the fuel cell; a model input module, configured to input the actual operating parameters into a performance prediction model of the fuel cell according to any one of claims 1 to 4; a coefficient determination module, configured to predict the fitting coefficients in the semi-empirical physical model based on the actual operating parameters using a target neural network model in the performance prediction model to obtain target fitting coefficients; The performance calculation module is used to obtain the actual performance parameters of the fuel cell output by the semi-empirical physical model by taking the actual operating parameters and the corresponding target fitting coefficients as inputs of the semi-empirical physical model in the performance prediction model.

8. An electronic device, characterized in that: The electronic device includes: a processor and a memory storing computer program instructions; When the processor executes the computer program instructions, the method for determining a fuel cell performance prediction model according to any one of claims 1 to 4 is implemented; Alternatively, the fuel cell performance prediction method as claimed in claim 5 is implemented.

9. A computer-readable storage medium, characterized in that The computer readable storage medium stores computer program instructions, and when the computer program instructions are executed by the processor, the method for determining a performance prediction model of a fuel cell according to any one of claims 1 to 4 is implemented; Alternatively, the fuel cell performance prediction method as claimed in claim 5 is implemented.

10. A computer program product, characterized in that When the instructions in the computer program product are executed by a processor of an electronic device, the electronic device executes the method for determining a performance prediction model of a fuel cell according to any one of claims 1 to 4; Alternatively, the fuel cell performance prediction method as claimed in claim 5 is implemented.

Images