Vanadium redox flow battery stack optimization and performance prediction method

Abstract

The invention relates to the field of electrochemical energy storage, and discloses a vanadium redox flow battery stack optimization and performance prediction method, which comprises the following steps: constructing a high-precision reaction kinetic model, and accurately describing the kinetic behavior of an electrode reaction based on a multi-step reaction mechanism, a surface adsorption-desorption process and heterogeneous surface reaction distribution; based on the output of the reaction kinetic model, constructing a multi-physical field coupling model, including ion mass transfer, fluid dynamics and heat transfer models, to cooperatively simulate the ion concentration, fluid flow and temperature field distribution in the galvanic pile; and based on the state variables of the multi-physics field coupling model, constructing a dynamic prediction model of chemical reaction-heat-fluid coupling, and predicting the future performance of the galvanic pile in real time through a long and short term memory network model. According to the method, the energy efficiency, the power density and the service life of the vanadium redox flow battery can be improved, and support is provided for intelligent management and optimization of the vanadium redox flow battery.

Description

Technical Field

[0001] The invention relates to the technical field of electrochemical energy storage, and in particular to a method for optimizing and predicting the performance of a vanadium liquid flow battery stack. Background Art

[0002] Vanadium Redox Flow Battery (VRFB) is an energy storage technology with broad application prospects. Due to its high charge and discharge efficiency, long cycle life, and good safety, it has been widely used in large-scale grid energy storage systems. However, the performance of vanadium redox flow batteries is largely affected by multiple physical and chemical processes inside the battery stack, including complex phenomena such as electrode reaction kinetics, ion mass transfer, fluid flow, and heat transfer. Therefore, how to accurately describe and optimize these processes to improve the energy efficiency, power density, and service life of vanadium redox flow batteries is an important challenge facing current technology.

[0003] In the prior art, a single physical field or a separate multi-physical field model is usually used to describe the internal behavior of the vanadium flow battery stack, such as constructing an electrode reaction kinetic model or an ion mass transfer model separately. However, traditional reaction kinetic models are mostly based on the single-step reaction assumption, and fail to fully consider the multi-step electron transfer, surface adsorption-desorption process, and heterogeneous reaction distribution on the electrode surface in the actual redox reaction, which results in limited accuracy of the model and cannot accurately reflect the effect of the electrode microstructure on the reaction rate. In addition, the existing ion mass transfer, fluid flow, and heat transfer models are often independent of each other, lacking an effective coupling solution method, and it is difficult to capture the interaction between different physical fields. The separability of this model limits the overall optimization of the battery stack performance, resulting in problems such as concentration polarization and local overheating in the actual operation of the battery, reducing the energy efficiency and cycle life of the vanadium flow battery.

[0004] At the same time, in terms of performance prediction, most existing studies rely on static models or empirical formulas, which makes it difficult to achieve dynamic predictions of future battery performance, especially under complex working conditions or long-term operation conditions. The prediction accuracy of the model is often low, which makes it difficult to meet the needs of intelligent battery management and preventive maintenance in practical applications. Therefore, the current optimization and performance prediction methods of vanadium flow battery stacks have major deficiencies in accuracy, coupling, and real-time performance. A high-precision prediction method based on multi-physical field coupling and data-driven is urgently needed to improve the overall performance of vanadium flow batteries. Summary of the invention

[0005] In view of the shortcomings of the prior art, the present invention provides a method for optimizing and predicting the performance of a vanadium liquid flow battery stack, which accurately simulates and optimizes the complex reaction kinetics, mass transfer, fluid flow and heat transfer processes within the vanadium liquid flow battery stack to improve its energy efficiency, power density and service life, and realize dynamic prediction of its future performance.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: A method for optimizing and predicting the performance of a vanadium liquid flow battery stack, comprising the following steps:

[0007] Construct a high-precision reaction kinetic model to describe the kinetic behavior of the electrode reaction based on the multi-step reaction mechanism, surface adsorption-desorption process and heterogeneous surface reaction distribution, and obtain the electrode surface reaction rate and local current density distribution;

[0008] Based on the high-precision reaction kinetics model, a multi-physics field coupling model is constructed, including an ion mass transfer model, a fluid dynamics model, and a heat transfer model. The current density distribution output by the reaction kinetics model is used as the boundary condition of the ion mass transfer model to perform a collaborative simulation of the ion mass transfer, fluid flow, and thermal effects inside the vanadium liquid flow battery stack;

[0009] Based on the state variables of the multi-physics field coupling model, a dynamic prediction model of chemical reaction-heat-fluid coupling is constructed, and based on the real-time monitored fuel cell state variables, the future performance of the fuel cell is predicted through a data-driven model.

[0010] Preferably, the high-precision reaction kinetics model comprises the following steps:

[0011] The redox reaction on the electrode surface is decomposed into multiple electron transfer steps through a multi-step reaction mechanism, and the charge transfer coefficient and reaction rate of each step are described respectively;

[0012] The Langmuir isotherm equation is used to describe the adsorption and desorption equilibrium process of vanadium ions on the electrode surface to reflect the effect of adsorption sites on the reaction rate.

[0013] Based on the heterogeneous activity distribution function of the electrode surface, the reaction activity differences in different areas of the electrode surface are described, and the overall current density is calculated by surface integration.

[0014] Preferably, the heterogeneous activity distribution function is used to characterize the distribution of different active sites on the electrode surface, and the reaction current density in different areas of the electrode surface is calculated based on the function. The heterogeneous activity distribution function is the distribution ratio of active sites per unit area of ​​the surface.

[0015] Preferably, in the high-precision reaction kinetics model, the electrode surface reaction current density is related to the electrode surface roughness, and the electrode surface is described by a fractal geometry model, and its fractal dimension is proportional to the surface current density.

[0016] Preferably, the multi-physics field coupling model comprises the following steps:

[0017] An ion mass transfer model was established based on the Nernst-Planck equation to describe the diffusion, migration and convection processes of ions;

[0018] A fluid dynamics model is established based on the Navier-Stokes equation to simulate the flow state and pressure distribution of the electrolyte;

[0019] A heat transfer model is established based on the heat transfer equation to describe the temperature distribution and thermal effect of the fuel cell stack during operation;

[0020] The ion mass transfer model, fluid dynamics model and heat transfer model are coupled and solved through the finite element method to optimize the flow channel design, flow velocity and temperature distribution of the fuel cell stack.

[0021] Preferably, the ion mass transfer model calculates the ion concentration distribution according to the concentration, diffusion coefficient, charge number and mobility of the vanadium ions to optimize the flow rate and flow channel design of the electrolyte.

[0022] Preferably, the heat source term in the heat transfer model includes reaction heat generated by chemical reactions and ohmic heat inside the battery stack to ensure that the battery operating temperature is controlled within an optimal range.

[0023] Preferably, the chemical reaction-heat-fluid coupling dynamic prediction model comprises the following steps:

[0024] The dependence of reaction rate on temperature is described based on the Arrhenius equation, and the effect of temperature field on reaction rate is reflected by the change of reaction rate constant;

[0025] Add the reaction rate coupling term to the fluid dynamics equation to describe the feedback effect of chemical reactions in the battery stack on fluid flow;

[0026] A data-driven model is constructed based on the long short-term memory network, and the state variables of the multi-physics field coupling model are used as input to predict the future performance of the fuel cell stack.

[0027] Preferably, the long short-term memory network model is used to predict the voltage attenuation, power density and energy efficiency of the fuel cell stack in the future, and the long short-term memory network model adapts to the changes in the actual working conditions of the fuel cell stack through incremental updates.

[0028] The present invention also provides a vanadium liquid flow battery stack optimization and performance prediction device, comprising:

[0029] Reaction kinetics modeling module, used to build multi-step reaction kinetics model of vanadium flow battery stack and heterogeneous surface reaction distribution model;

[0030] Multi-physics coupling module, which is used to couple ion mass transfer, fluid dynamics and heat transfer models to perform numerical simulation of the multi-field coupling state inside the fuel cell stack;

[0031] The dynamic prediction module, based on the LSTM data-driven model of chemical reaction-heat-fluid coupling, realizes real-time prediction of the future performance of the battery stack.

[0032] The present invention provides a method for optimizing and predicting the performance of a vanadium liquid flow battery stack. It has the following beneficial effects:

[0033] 1. Through high-precision modeling of multi-step reaction mechanisms, adsorption-desorption processes, and heterogeneous surface reaction distribution, the present invention can more accurately describe the microscopic reaction behavior of the electrode surface of vanadium flow batteries. This detailed kinetic modeling makes the prediction of the local reaction rate and current density of the battery stack more accurate, thereby providing reliable boundary conditions for subsequent multi-physics field coupling models and helping to optimize the overall performance of the battery stack.

[0034] 2. The multi-physics field coupling model constructed by the present invention solves the problem of complex physical field interactions that are difficult to describe with traditional separation models through the collaborative simulation of ion mass transfer, fluid flow and heat transfer. By accurately obtaining the concentration field, flow velocity field and temperature field distribution inside the battery stack, the electrolyte flow rate, flow channel geometry and cooling system design can be optimized, thereby significantly improving the mass transfer efficiency and power density of the battery.

[0035] 3. The data-driven prediction model based on the long short-term memory network (LSTM) combined with the coupled state variables of multiple physical fields realizes the dynamic prediction of the future performance of the battery stack. During the battery operation, the model can predict the voltage attenuation, power density and energy efficiency changes based on the real-time monitored battery stack state variables, ensuring that the battery operates under the best working conditions, reducing the risk of failure and extending the battery life.

[0036] 4. By optimizing the reaction rate, fluid distribution and temperature control inside the battery stack, the present invention effectively suppresses local overheating and concentration polarization of the battery while achieving efficient mass transfer and energy conversion. This method helps to maintain the battery working in an efficient state, prolong the service life, and meet the long-term operation requirements of vanadium liquid flow batteries in the field of energy storage. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 It is a schematic diagram of the method flow of the present invention;

[0038] Figure 2 It is a schematic diagram of the device structure of the present invention.

[0039] Among them, 10. Reaction kinetics modeling module; 20. Multi-physics field coupling module; 30. Dynamic prediction module. DETAILED DESCRIPTION

[0040] The following will be combined with the drawings in the specification of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0041] Please refer to the attached Figure 1 The embodiment of the present invention provides a method for optimizing and predicting the performance of a vanadium flow battery stack, which realizes efficient optimization and performance prediction of a vanadium flow battery stack by constructing a high-precision reaction kinetics model, a multi-physics field coupling model, and a chemical reaction-heat-fluid dynamic prediction model. The various steps of the method of the present invention are described in detail below.

[0042] S1. Constructing a high-precision reaction kinetics model

[0043] First, the performance optimization of vanadium flow batteries depends largely on the accurate modeling of electrode reaction kinetics. Existing electrode reaction kinetic models usually only consider single-step reaction processes, ignoring the multi-step electron transfer, surface adsorption-desorption process and heterogeneous reaction distribution on the electrode surface in actual redox reactions. It is difficult to accurately describe the complex reaction behavior of vanadium ions on the electrode surface, thus affecting the optimization effect of the overall performance of the battery stack.

[0044] Based on this, the present invention proposes a high-precision reaction kinetic model for a refined description of the reaction process on the electrode surface of the vanadium flow battery stack. The model combines the multi-step reaction mechanism, the surface adsorption-desorption model, and the heterogeneous surface reaction distribution, and provides accurate reaction rates and local current density distributions, providing precise boundary conditions for subsequent mass transfer models and fluid models, thereby achieving performance optimization and prediction of the vanadium flow battery stack.

[0045] In this embodiment, the construction of a high-precision reaction kinetics model mainly includes three parts: multi-step reaction mechanism, surface adsorption-desorption process and heterogeneous surface reaction distribution.

[0046] In the multi-step reaction mechanism, it is considered that the redox reaction of the vanadium flow battery is usually not a single-step process, but is completed step by step by multiple electron transfer reactions. Therefore, in general, the present invention decomposes the redox reaction on the electrode surface into multiple independent electron transfer steps, each of which is affected by the charge transfer coefficient and the reaction overpotential. Specifically, assuming that the anode reaction is the reduction of oxidized vanadium ions to reduced vanadium ions, this process includes the stepwise transfer of multiple electrons and can be expressed as follows:

[0047]

[0048] Each step in this reaction uses the extended Tafel equation to describe the reaction rate to meet the characteristics of stepwise electron transfer. The basic expression of the extended Tafel equation is:

[0049]

[0050] Where i represents the local current density, i0 is the exchange current density, α aj and α cj are the anode and cathode charge transfer coefficients of each step of electron transfer, η is the overpotential, F is the Faraday constant, R is the gas constant, T is the temperature, and n is the total number of steps in the reaction. The above formula significantly improves the accuracy of the reaction kinetic model by describing each electron transfer process step by step.

[0051] In addition, considering that the adsorption of vanadium ions on the electrode surface will affect the reaction rate, the present invention further introduces a surface adsorption-desorption model. Specifically, in some embodiments, vanadium ions form a temporary adsorption state on the electrode surface, that is, the vanadium ions are first adsorbed on the electrode surface, and then released into the solution through desorption to complete the redox reaction. Therefore, the Langmuir isotherm adsorption equation is used to describe the adsorption-desorption equilibrium process of vanadium ions on the electrode surface. In general, the Langmuir isotherm adsorption equation can be expressed as:

[0052]

[0053] Among them, θ is the coverage of adsorption sites on the electrode surface, K is the adsorption equilibrium constant, and c is the concentration of vanadium ions in the solution. Through this equation, the amount of vanadium ions adsorbed on the electrode surface can be calculated, thereby further adjusting the reaction rate in the reaction kinetic model.

[0054] In order to further improve the accuracy of the reaction kinetics model, the present invention also considers the problem of heterogeneous reaction distribution on the electrode surface. In practical applications, due to the differences in the microstructure of the electrode material, the distribution of active sites on the electrode surface is usually not uniform. Therefore, the present invention introduces a heterogeneous surface activity distribution function f(x, y) to describe the activity differences in different areas of the electrode surface. In general, the heterogeneous surface activity distribution function represents the distribution ratio of active sites per unit area, which can be obtained by actual measurement or numerical simulation. According to this function, the local reaction current density in different areas of the electrode surface can be expressed as:

[0055]

[0056] By integrating over the entire electrode surface, the total current density distribution is obtained, namely:

[0057] I=∫ 表面 i(x,y)dxdy

[0058] In some embodiments, the heterogeneous surface activity distribution function can be correlated with the roughness and pore distribution of the electrode surface to further describe the effect of the electrode surface microstructure on the reaction rate.

[0059] As an option, in order to describe the microstructure of the electrode surface in more detail, the present invention uses a fractal geometry model to model the roughness of the electrode surface. Specifically, the fractal dimension D of the electrode surface f It can be used to describe the distribution of surface roughness and active sites, thereby affecting the local reaction rate. In this case, the electrode surface reaction current density is related to the fractal dimension D f The following relationships can be established:

[0060]

[0061] This model further improves the applicability of the reaction kinetics model by describing the roughness and pore structure of the electrode surface.

[0062] In summary, the high-precision reaction kinetics model in the present invention can more accurately describe the reaction behavior of the electrode surface in the vanadium liquid flow battery stack by combining the multi-step reaction mechanism, the surface adsorption-desorption process and the heterogeneous surface reaction distribution model, and provide accurate boundary conditions for subsequent mass transfer and fluid dynamics models.

[0063] S2. Build a multi-physics coupling model

[0064] After constructing a high-precision reaction kinetics model, the present invention further proposes a multi-physics field coupling model to collaboratively simulate the ion mass transfer, fluid dynamics and heat transfer processes in the vanadium liquid flow battery stack. In actual operation, the internal stack of the vanadium liquid flow battery contains complex mass transfer, fluid flow and heat transfer processes, and these processes are interrelated and affect the overall performance of the stack. In traditional methods, these physical processes are often simulated separately, making it difficult to achieve accurate optimization of the vanadium liquid flow battery stack at the system level. To overcome this shortcoming, the multi-physics field coupling model of the present invention, based on the high-precision reaction kinetics model, uses the finite element method to couple and solve each physical field, and uses the current density distribution output by the reaction kinetics model as the boundary condition of the ion mass transfer model, so as to comprehensively analyze and optimize the ion concentration field, fluid flow velocity field and temperature field inside the stack.

[0065] In this embodiment, the multi-physics field coupling model mainly includes an ion mass transfer model, a fluid dynamics model and a heat transfer model, and the finite element method is used to achieve collaborative calculation in the coupling solution process.

[0066] In the ion mass transfer model, the present invention describes the mass transfer process of vanadium ions based on the Nernst-Planck equation. In general, the ion mass transfer process is affected by diffusion, migration and convection, so the model should consider the influence of diffusion coefficient, charge number, mobility, electric field gradient and fluid flow rate. Specifically, assuming that the vanadium ion concentration is c i , the diffusion coefficient is D i , the charge number is z i , the mobility is μ i , the electrolyte flow rate is u, and the potential is φ, then the basic form of the model is as follows:

[0067]

[0068] Among them, R i Represents the reaction rate, which is usually provided by the current density and reaction rate in the high-precision reaction kinetics model. In some embodiments, the initial conditions and boundary conditions of the mass transfer model can be dynamically adjusted according to the actual operating conditions of the fuel cell stack to ensure the applicability of the model.

[0069] As an option, in the actual solution process, the local current density distribution on the electrode surface can be used as the boundary condition of the ion mass transfer model to realize the coupled calculation of the kinetic model and the mass transfer model, so as to accurately obtain information such as the ion concentration distribution and concentration polarization effect inside the fuel cell stack, providing a theoretical basis for the performance optimization of the fuel cell stack.

[0070] In the fluid dynamics model part, the present invention simulates the flow state and pressure distribution of the electrolyte in the battery stack based on the Navier-Stokes equation. In general, the flow of the electrolyte is affected by density, flow velocity and external pressure. Specifically, assuming that the density of the electrolyte is ρ, the flow velocity is u, and the pressure is p, the fluid dynamics model can be expressed as:

[0071]

[0072] In the formula, μ is the viscosity of the electrolyte and F is the body force term. The solution of the fluid model directly affects the convection term in the ion mass transfer process, so the two need to be calculated synchronously during the solution. In addition, as a specific implementation scheme, the initial parameters of the flow velocity and flow channel structure in the fluid dynamics model can be appropriately adjusted according to the working environment of the battery stack, thereby effectively improving the flow efficiency and uniformity of the fluid.

[0073] In some embodiments, the fluid dynamics model can be further combined with the actual flow channel geometry and fluid properties of the battery stack to optimize the flow channel design and electrolyte flow rate. During implementation, the geometric design of the flow channel of the battery stack is usually input into the model as a boundary condition to ensure the practical applicability of the simulation results.

[0074] In the heat transfer model part, the present invention uses the heat transfer equation to model the temperature distribution inside the battery stack. Generally, vanadium liquid flow batteries will generate a lot of heat due to the reaction during operation. If the heat cannot be effectively dissipated, the long-term stability of the battery stack will be affected. Therefore, in this embodiment, the energy conservation equation is used to describe the heat transfer process. Specifically, it is assumed that the specific heat capacity of the electrolyte is cpc_pcp, the temperature is T, the thermal conductivity is k, and the heat source term is q gen , then the basic expression of the model is:

[0075]

[0076] Among them, the heat source term q gen The heat transfer model solution reflects the temperature distribution of the battery stack, including the reaction heat generated by chemical reactions in the battery stack and the ohmic heat generated by current flow. In some embodiments, the local temperature rise of the battery can be reduced by adjusting the parameters of the cooling system or optimizing the flow channel design to ensure that the battery stack operates in the optimal temperature range.

[0077] As an optimization option, temperature-sensitive parameters can be introduced into the heat transfer model so that the model can dynamically reflect the impact of temperature changes on reaction rate and heat transfer performance, thereby improving the prediction accuracy and reliability of the model.

[0078] In the multi-physics field coupling solution part, the present invention adopts the finite element method to collaboratively solve the ion mass transfer model, fluid dynamics model and heat transfer model. In the actual solution, the ion mass transfer model, fluid dynamics model and heat transfer model are coupled with each other, and real-time data interaction is required on the model boundary conditions and state variables. In general, the reaction rate data in the ion mass transfer model is used as the feedback input of the fluid dynamics model, and the flow rate and pressure distribution data in the fluid dynamics model are used as the input conditions of the heat transfer model. In the coupled solution, an iterative solution strategy is adopted to ensure that each physical field reaches numerical convergence, so the coupled solution result has high stability and accuracy.

[0079] In some embodiments, the multi-physics field coupling model of the present invention can dynamically adjust the boundary conditions through real-time monitoring data. For example, in the actual operation of vanadium liquid flow batteries, the flow rate, temperature and pressure distribution may deviate due to changes in external conditions. At this time, the model boundary conditions can be corrected based on the monitoring data to make the solution more in line with the actual situation and improve the effectiveness of stack performance optimization.

[0080] S3. Construct a dynamic prediction model for chemical reaction-heat-fluid coupling

[0081] After completing the construction of a high-precision reaction kinetics model and a multi-physics field coupling model, the present invention further proposes a dynamic prediction model of chemical reaction-heat-fluid coupling to achieve accurate prediction of the future performance of the vanadium liquid flow battery stack. In actual operation, the vanadium liquid flow battery stack is dynamically affected by multiple factors such as temperature change, reaction rate, fluid flow, etc. The physical fields are coupled with each other, and large performance fluctuations may occur under different operating conditions. In general, a single physical model is difficult to accurately predict the future changes in the performance of the stack. Therefore, the present invention is based on the existing multi-physics field coupling model, combined with a data-driven dynamic prediction model, and uses the coupled state variables of chemical reaction, heat transfer and fluid flow as input to construct a long short-term memory network (LSTM) model, which can predict the future performance of the stack based on the real-time monitored stack state variables. Through the combination of the multi-field dynamic coupling model of chemical reaction-heat-fluid and the data-driven deep learning prediction model, real-time monitoring and prediction of the future state of the stack can be achieved.

[0082] In this embodiment, the dynamic prediction model of chemical reaction-heat-fluid coupling includes three parts: a chemical reaction-heat coupling model, a fluid-reaction coupling model, and a dynamic performance prediction model based on LSTM.

[0083] In the chemical reaction-thermal coupling model part, the present invention describes the relationship between reaction rate and temperature based on the Arrhenius equation to reflect the dependence of reaction rate on temperature. In general, the chemical reaction rate k(T) in the battery stack is exponentially related to temperature, so when the temperature rises, the reaction rate will increase significantly. Specifically, the Arrhenius equation can be expressed as:

[0084]

[0085] Among them, A is the frequency factor, E a is the activation energy of the reaction, R is the gas constant, and T is the reaction temperature. By applying this equation, the changing trend of the chemical reaction rate can be predicted based on the current temperature. In addition, the heat (reaction heat) generated during the chemical reaction will further affect the temperature field distribution. In general, the heat source term q gen It can be expressed as the product of the reaction rate k(T) and the reaction enthalpy change ΔH:

[0086] q gen =k(T)·ΔH

[0087] Among them, ΔH represents the enthalpy change of the reaction. Through this model, the dynamic coupling of reaction rate and temperature field can be realized, so that the temperature change can be fed back to the reaction rate in real time, thus accurately describing the thermal-reaction coupling relationship of the fuel cell during operation.

[0088] In the fluid-reaction coupling model part, the present invention introduces the reaction rate of the chemical reaction into the fluid dynamics model to describe the dynamic coupling of fluid flow and chemical reaction. In general, changes in the chemical reaction rate will have a certain impact on the fluid flow, especially in systems where mass transfer and reaction processes are closely related, changes in fluid flow rate will further feed back into the reaction rate. Specifically, the reaction rate distribution R(x, y, z) is added to the body term of the fluid dynamics equation to reflect the interaction between fluid flow and the reaction process. Assuming that the electrolyte fluid density is ρ, the flow rate is u, and the viscosity is μ, the fluid dynamics equation can be expressed as:

[0089]

[0090] Wherein, γ is the fluid-reaction coupling coefficient, which is used to control the contribution of the reaction rate in the fluid dynamics equation. In some embodiments, the value of γ can be calibrated according to experimental data to enhance the applicability of the fluid-reaction coupling model. By adding the reaction rate to the fluid dynamics equation, the effect of the chemical reaction on the fluid flow rate can be dynamically simulated to achieve the coupling optimization of the reaction and the fluid.

[0091] In the dynamic performance prediction model based on LSTM, the present invention builds a data-driven performance prediction model based on the long short-term memory network (LSTM) to achieve dynamic prediction of the future state of the battery stack. In general, the LSTM model can use time series data to capture the changing rules of the system under different states. Specifically, the input of the LSTM model includes the state variables of the chemical reaction-heat-fluid coupling model, such as the real-time temperature, current density, flow rate, etc. of the battery stack. Assume that the input state variable is x t , the hidden layer state of the LSTM model is h t , then the output of the LSTM model represents the performance prediction value at the next moment, which can be calculated by the following recursive relationship:

[0092] h t =f(W h x t +U h h t-1 +b h )

[0093] Among them, W h and U h is the weight matrix of the model, b h is a bias term, and f represents an activation function. Through the combination of multiple layers of LSTM units, the present invention can construct a deep time series model to achieve dynamic prediction of the future state of the battery stack (such as voltage attenuation, power density, and energy efficiency, etc.).

[0094] As an optimization option, in some embodiments, the training data of the LSTM model can be incrementally updated according to the actual operating conditions of the vanadium liquid flow battery, so that the model can adapt to changes in the battery operating conditions. For example, during the operation of the battery, changes in state variables such as current density, temperature and flow rate may produce certain deviations. At this time, the weights of the LSTM model can be adjusted through incremental learning to improve the prediction accuracy and adaptability of the model. In general, the incremental learning method can improve the real-time performance of the model, so for battery systems that require long-term online monitoring, this method has high practicality.

[0095] In summary, the chemical reaction-heat-fluid coupling dynamic prediction model of the present invention realizes the dynamic prediction of the future performance of the vanadium liquid flow battery stack by combining the coupled state variables of multiple physical fields and the LSTM model, and can provide data support for battery performance monitoring and intelligent management.

[0096] The present invention realizes accurate simulation and coordinated optimization of complex physical and chemical processes inside the battery stack by constructing a high-precision reaction kinetic model, a multi-physics field coupling model and a chemical reaction-heat-fluid dynamic prediction model. The method first describes the reaction kinetic behavior of the electrode surface through a multi-step reaction mechanism, an adsorption-desorption process and a heterogeneous reaction distribution, and then combines the multi-physics field coupling model of ion mass transfer, fluid dynamics and heat transfer to obtain the ion concentration, flow rate and temperature field distribution inside the battery stack. On this basis, the present invention constructs a data-driven prediction model based on a long short-term memory network to realize dynamic prediction of the future performance of the battery stack. This method can be used to improve the energy efficiency and power density of vanadium liquid flow batteries and provide support for their intelligent management.

[0097] The vanadium liquid flow battery stack optimization and performance prediction device described below and the vanadium liquid flow battery stack optimization and performance prediction method described above can be referred to each other.

[0098] Please refer to the attached Figure 2 The present invention also provides a vanadium liquid flow battery stack optimization and performance prediction device, comprising:

[0099] Reaction kinetics modeling module 10, used to construct a multi-step reaction kinetics model of a vanadium liquid flow battery stack and a heterogeneous surface reaction distribution model;

[0100] The multi-physics coupling module 20 is used to couple the ion mass transfer, fluid dynamics and heat transfer models to perform numerical simulation on the multi-field coupling state inside the fuel cell stack;

[0101] The dynamic prediction module 30 realizes real-time prediction of the future performance of the fuel cell stack based on the LSTM data-driven model of chemical reaction-heat-fluid coupling.

[0102] The device of this embodiment can be used to execute the above method embodiment, and its principles and technical effects are similar, which will not be repeated here.

[0103] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for optimizing and predicting the performance of a vanadium liquid flow battery stack, characterized in that: The following steps are involved: Construct a high-precision reaction kinetic model to describe the kinetic behavior of the electrode reaction based on the multi-step reaction mechanism, surface adsorption-desorption process and heterogeneous surface reaction distribution, and obtain the electrode surface reaction rate and local current density distribution; Based on the high-precision reaction kinetics model, a multi-physics field coupling model is constructed, including an ion mass transfer model, a fluid dynamics model, and a heat transfer model. The current density distribution output by the reaction kinetics model is used as the boundary condition of the ion mass transfer model to perform a collaborative simulation of the ion mass transfer, fluid flow, and thermal effects inside the vanadium liquid flow battery stack; Based on the state variables of the multi-physics field coupling model, a dynamic prediction model of chemical reaction-heat-fluid coupling is constructed, and based on the real-time monitored fuel cell state variables, the future performance of the fuel cell is predicted through a data-driven model.

2. A vanadium liquid flow battery stack optimization and performance prediction method according to claim 1, characterized in that: The high-precision reaction kinetics model comprises the following steps: The redox reaction on the electrode surface is decomposed into multiple electron transfer steps through a multi-step reaction mechanism, and the charge transfer coefficient and reaction rate of each step are described respectively; The Langmuir isotherm equation is used to describe the adsorption and desorption equilibrium process of vanadium ions on the electrode surface to reflect the effect of adsorption sites on the reaction rate. Based on the heterogeneous activity distribution function of the electrode surface, the reaction activity differences in different areas of the electrode surface are described, and the overall current density is calculated by surface integration.

3. A vanadium liquid flow battery stack optimization and performance prediction method according to claim 2, characterized in that: The heterogeneous activity distribution function is used to characterize the distribution of different active sites on the electrode surface. The reaction current density in different areas of the electrode surface is calculated based on the function. The heterogeneous activity distribution function is the distribution ratio of active sites per unit area of ​​the surface.

4. The method for optimizing and predicting the performance of a vanadium liquid flow battery stack according to claim 1, characterized in that: In the high-precision reaction kinetics model, the electrode surface reaction current density is related to the electrode surface roughness, and the electrode surface is described by a fractal geometry model, and its fractal dimension is proportional to the surface current density.

5. The method for optimizing and predicting the performance of a vanadium liquid flow battery stack according to claim 1, characterized in that: The multi-physics coupling model includes the following steps: An ion mass transfer model was established based on the Nernst-Planck equation to describe the diffusion, migration and convection processes of ions; A fluid dynamics model is established based on the Navier-Stokes equation to simulate the flow state and pressure distribution of the electrolyte; A heat transfer model is established based on the heat transfer equation to describe the temperature distribution and thermal effect of the fuel cell stack during operation; The ion mass transfer model, fluid dynamics model and heat transfer model are coupled and solved through the finite element method to optimize the flow channel design, flow velocity and temperature distribution of the fuel cell stack.

6. A method for optimizing and predicting the performance of a vanadium liquid flow battery stack according to claim 5, characterized in that: The ion mass transfer model calculates the ion concentration distribution according to the concentration, diffusion coefficient, charge number and mobility of the vanadium ions to optimize the flow rate and flow channel design of the electrolyte.

7. A method for optimizing and predicting the performance of a vanadium liquid flow battery stack according to claim 5, characterized in that: The heat source term in the heat transfer model includes the reaction heat generated by the chemical reaction and the ohmic heat inside the battery stack to ensure that the battery operating temperature is controlled within the optimal range.

8. The method for optimizing and predicting the performance of a vanadium liquid flow battery stack according to claim 1, characterized in that: The chemical reaction-heat-fluid coupling dynamic prediction model comprises the following steps: The dependence of reaction rate on temperature is described based on the Arrhenius equation, and the effect of temperature field on reaction rate is reflected by the change of reaction rate constant; Add the reaction rate coupling term to the fluid dynamics equation to describe the feedback effect of chemical reactions in the battery stack on fluid flow; A data-driven model is constructed based on the long short-term memory network, and the state variables of the multi-physics field coupling model are used as input to predict the future performance of the fuel cell stack.

9. A method for optimizing and predicting the performance of a vanadium liquid flow battery stack according to claim 8, characterized in that: The long short-term memory network model is used to predict the voltage attenuation, power density and energy efficiency of the fuel cell stack in the future. The long short-term memory network model adapts to the changes in the actual working conditions of the fuel cell stack through incremental updates.

10. A vanadium liquid flow battery stack optimization and performance prediction device, used to execute the method according to any one of claims 1 to 9, characterized in that: include: Reaction kinetics modeling module, used to build multi-step reaction kinetics model of vanadium liquid flow battery stack and heterogeneous surface reaction distribution model; Multi-physics coupling module, which is used to couple ion mass transfer, fluid dynamics and heat transfer models to perform numerical simulation of the multi-field coupling state inside the fuel cell stack; The dynamic prediction module, based on the LSTM data-driven model of chemical reaction-heat-fluid coupling, realizes real-time prediction of the future performance of the fuel cell stack.

Images