Temperature control method, device and equipment of solid oxide electrolytic cell system, medium and product

Abstract

The invention discloses a temperature control method, device and equipment of a solid oxide electrolytic cell system, a medium and a product, and relates to the technical field of electrolytic pile temperature control, and the method comprises the following steps: obtaining a reference output temperature of an electric pile in the solid oxide electrolytic cell system at the current moment, and current density and air flow input into the electric pile; an LM-BP NN model is adopted to determine the initial fuel flow rate; conveying fuel to the stack according to the initial fuel flow rate; acquiring a temperature output value of the electric pile corresponding to the initial fuel flow rate; according to the temperature output value and the reference output temperature, a PID control algorithm is adopted to obtain fuel flow velocity correction; determining a corrected fuel flow rate for conveying fuel to the solid oxide electrolytic cell system according to the fuel flow rate correction and the initial fuel flow rate; and conveying fuel to the solid oxide electrolytic cell system according to the corrected fuel flow rate, so that the real-time temperature output value, corresponding to the corrected fuel flow rate, of the electric pile and the reference output temperature meet preset conditions. The durability of the galvanic pile can be improved.

Description

Technical Field

[0001] The present application relates to the technical field of electrolytic stack temperature control, and in particular to a temperature control method, device, equipment, medium and product for a solid oxide electrolytic cell system. Background Art

[0002] Currently, temperature control methods for solid oxide electrolyzer (SOEC) systems primarily rely on air flow control (AFC). However, this approach is unable to cope with complex operating conditions such as large current fluctuations and has significant limitations. Therefore, recent research has focused on the role of fuel flow control (FFC) in temperature regulation.

[0003] Existing research involving fuel flow control mainly focuses on maintaining a constant gas conversion rate or preventing fuel starvation, and rarely conducts in-depth analysis of fuel flow as a core parameter for active temperature control. In fact, as a reactant gas, fuel not only affects the reaction efficiency of the stack, but also plays a key role in the temperature distribution and thermal management of the stack through its flow regulation. Reasonable regulation of fuel flow is expected to more effectively suppress temperature overshoot and local overheating, thereby slowing down the degradation of the electrolyte and electrode interface caused by high temperature or temperature fluctuations, and improving the long-term operating stability and durability of the stack. Therefore, in-depth exploration of the role of fuel flow in active temperature control and durability improvement of SOEC stacks is the key to achieving an efficient and reliable SOEC system. Summary of the Invention

[0004] The purpose of this application is to provide a temperature control method, device, equipment, medium and product for a solid oxide electrolyzer system, which can improve the durability of the battery stack.

[0005] To achieve the above objectives, this application provides the following solutions:

[0006] In a first aspect, the present application provides a temperature control method for a solid oxide electrolyzer system, the method comprising:

[0007] Obtaining a reference output temperature of a stack in a solid oxide electrolyzer system at a current moment, a current density input to the stack, and an air flow rate;

[0008] Determining a preliminary fuel flow rate using a LM-BP NN model based on the reference output temperature, the current density, and the air flow rate as a feedforward fuel flow rate input for the fuel cell stack; the LM-BP NN model is used to predict the preliminary fuel flow rate based on the reference output temperature, the current density, and the air flow rate;

[0009] delivering fuel to the stack according to the preliminary fuel flow rate;

[0010] obtaining a temperature output value of the fuel cell stack corresponding to the preliminary fuel flow rate;

[0011] According to the temperature output value and the reference output temperature, a PID control algorithm is used to obtain a fuel flow rate correction value as a feedback fuel flow rate input of the fuel cell stack;

[0012] determining a corrected fuel flow rate for delivering fuel to the solid oxide electrolyzer system based on the fuel flow rate correction and the preliminary fuel flow rate;

[0013] Fuel is delivered to the solid oxide electrolyzer system according to the corrected fuel flow rate, so that the real-time temperature output value of the stack corresponding to the corrected fuel flow rate and the reference output temperature meet preset conditions.

[0014] In a second aspect, the present application provides a temperature control device for a solid oxide electrolyzer system, the device comprising:

[0015] A reference output temperature and current density acquisition module is used to obtain the reference output temperature of the stack in the solid oxide electrolyzer system at the current moment, the current density input to the stack, and the air flow rate;

[0016] a preliminary fuel flow rate determination module, configured to determine a preliminary fuel flow rate using a LM-BPNN model based on the reference output temperature, the current density, and the air flow rate, as a feedforward fuel flow rate input for the stack; the LM-BPNN model being configured to predict the preliminary fuel flow rate based on the reference output temperature, the current density, and the air flow rate;

[0017] a fuel delivery module, configured to deliver fuel to the fuel stack according to the preliminary fuel flow rate;

[0018] a temperature output value acquisition module, configured to acquire a temperature output value of the fuel cell stack corresponding to the preliminary fuel flow rate;

[0019] a fuel flow rate correction module, configured to obtain a fuel flow rate correction value using a PID control algorithm based on the temperature output value and the reference output temperature, as a feedback fuel flow rate input of the fuel cell stack;

[0020] a corrected fuel flow rate module for determining a corrected fuel flow rate for delivering fuel to the solid oxide electrolyzer system based on the fuel flow rate correction amount and the preliminary fuel flow rate;

[0021] The fuel delivery module is further used to deliver fuel to the solid oxide electrolyzer system according to the corrected fuel flow rate, so that the real-time temperature output value of the stack corresponding to the corrected fuel flow rate and the reference output temperature meet preset conditions.

[0022] In a third aspect, the present application provides a computer device comprising: a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the temperature control method for a solid oxide electrolysis cell system described in any one of the above.

[0023] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the temperature control method for a solid oxide electrolysis cell system described in any one of the above.

[0024] In a fifth aspect, the present application provides a computer program product, comprising a computer program, which, when executed by a processor, implements the steps of the temperature control method for a solid oxide electrolysis cell system described in any one of the above.

[0025] According to the specific embodiments provided in this application, this application discloses the following technical effects:

[0026] The present application provides a temperature control method, device, equipment, medium and product for a solid oxide electrolyzer system, and specifically proposes a feedforward-feedback control method based on LM-BP NN. In feedforward control, the LM-BP NN model, with its rapid response capability, comprehensively considers multiple parameters such as reference output temperature, current density, and air flow to accurately predict the initial fuel flow rate, regulate the system in advance, and reduce the risk of temperature overshoot due to changes in operating conditions. Feedback control is based on the feedforward control output, and uses a PID control algorithm to calculate the fuel flow rate correction value in real time based on the error between the temperature output value of the stack corresponding to the initial fuel flow rate and the reference output temperature, and dynamically corrects the initial fuel flow rate. When the system shows a temperature overshoot trend, the fuel flow rate is adjusted in time to restore the temperature to stability. Finally, the outputs of the feedforward and feedback controls are combined to form the final fuel flow rate of the system. Through this two-pronged approach, the output temperature overshoot problem of the solid oxide electrolyzer system is effectively suppressed, thereby improving the durability of the stack and providing guidance for the future coupling of SOEC with renewable energy. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0028] Figure 1 is a schematic diagram of a SOEC system according to an exemplary embodiment;

[0029] Figure 2 is a diagram showing a verification result of a SOEC model according to an exemplary embodiment;

[0030] Figure 3 is a flow chart showing a temperature control method for a solid oxide electrolysis cell system according to an exemplary embodiment;

[0031] Figure 4 is a schematic diagram showing changes in the MSE of the training set during the training of the LM-BP NN model according to an exemplary embodiment;

[0032] Figure 5 is a schematic diagram showing a comparison between a predicted value and an actual value using a LM-BP NN model according to an exemplary embodiment;

[0033] Figure 6 is a control logic diagram of a feedforward-feedback control method based on an LM-BP NN model according to an exemplary embodiment;

[0034] Figure 7 The performance comparison diagram of the feedforward-feedback control method based on LM-BP NN model and the basic fuel flow control method based on PID control is shown in accordance with an exemplary embodiment. Figure 1 ;

[0035] Figure 8 The performance comparison diagram of the feedforward-feedback control method based on LM-BP NN model and the basic fuel flow control method based on PID control is shown in accordance with an exemplary embodiment. Figure 2 ;

[0036] Figure 9 The performance comparison diagram of the feedforward-feedback control method based on LM-BP NN model and the basic fuel flow control method based on PID control is shown in accordance with an exemplary embodiment. Figure 3 ;

[0037] Figure 10 The performance comparison diagram of the feedforward-feedback control method based on LM-BP NN model and the basic fuel flow control method based on PID control is shown in accordance with an exemplary embodiment. Figure 4 ;

[0038] Figure 11 The performance comparison diagram of the feedforward-feedback control method based on LM-BP NN model and the basic fuel flow control method based on PID control is shown in accordance with an exemplary embodiment. Figure 5 ;

[0039] Figure 12The performance comparison diagram of the feedforward-feedback control method based on LM-BP NN model and the basic fuel flow control method based on PID control is shown in accordance with an exemplary embodiment. Figure 6 ;

[0040] Figure 13 The control effect of the feedforward-feedback control method based on the LM-BP NN model under actual fluctuating current conditions is shown according to an exemplary embodiment;

[0041] Figure 14 is a block diagram of a temperature control device for a solid oxide electrolysis cell system according to an exemplary embodiment;

[0042] Figure 15 A schematic diagram of the structure of a computer device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0043] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0044] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0045] The SOEC system is an advanced energy conversion device, mainly consisting of two parts: the SOEC stack and BoP (Balance of Plant).

[0046] like Figure 1 As shown in the figure, the BoP includes auxiliary equipment such as steam generator, gas mixer, solenoid valve, heat exchanger and electric heater.

[0047] The main function of the steam generator is to generate the steam required by the system. The gas mixer is used to mix different types of gases in a certain proportion. The solenoid valve is a valve used to control the flow of gas or liquid, and the valve is opened and closed by electromagnetic force. In the SOEC system, the solenoid valve is mainly used to accurately control the on-off and flow of gas. The function of the heat exchanger is to achieve heat exchange. In the SOEC system, the heat exchanger is used to recover the waste heat generated in the system and transfer it to the fluid or gas that needs to be heated to improve the energy utilization efficiency of the system. The electric heater is mainly used to heat the gas or liquid to meet the temperature requirements of the system in different operating stages. When the SOEC system is started, the electric heater can quickly heat the gas or liquid to the operating temperature required by the fuel cell stack; during operation, when the system heat is insufficient or the temperature needs to be adjusted, the electric heater can provide additional heat.

[0048] The SOEC stack includes, from top to bottom, a channel at the top, an air electrode, an electrolyte, a fuel electrode, a channel at the bottom, and a power supply.

[0049] During the electrolysis process, water vapor and carbon dioxide gain electrons at the fuel electrode (cathode) and are reduced to generate hydrogen and carbon monoxide, respectively; oxygen ions migrate from the fuel electrode through the electrolyte to the air electrode (anode), where they lose electrons to generate oxygen.

[0050] The power supply provides an external current for the electrolysis reaction of the SOEC stack. The external voltage applied by the power supply forms an electric field between the air electrode (anode) and the fuel electrode (cathode), thereby driving the electrons in the external circuit to flow from the air electrode to the fuel electrode (i.e., the direction of the external circuit). Inside the stack, the electrons enter the reaction zone from the fuel electrode. At the same time, under the action of the electric field, oxygen ions (O 2- ) migrates from the fuel electrode side to the air electrode side in the electrolyte.

[0051] On the fuel side, a steam generator produces a certain amount of water vapor, which is thoroughly mixed with H2 and CO2 in a gas mixer. The mixed gas is preheated in a heat exchanger, where the gas temperature is raised by waste heat from the reaction products at the SOEC stack outlet. The hot mixed gas is fed into the SOEC stack for a reaction. The water vapor and CO2 gain electrons and are reduced to a synthesis gas composed of H2 and CO, producing oxygen ions. On the air side, air is introduced as a purge gas. The oxygen ions generated by the fuel electrode pass through the electrolyte layer and enter the air electrode, where they are oxidized to oxygen and carried away by the air.

[0052] In the above-mentioned SOEC system, the main input parameters of the SOEC system include current density, fuel flow rate and air flow rate. Based on these input parameters, the system outputs temperature and stack voltage. The main input parameters of the SOEC system include current density, fuel flow rate and air flow rate. Based on these input parameters, the system outputs temperature and stack voltage. This process contains a series of physical and chemical phenomena, including electrochemical reactions, chemical reactions, multi-component mass transfer and heat transfer, and the flow process of the working fluid. The system outputs the stack voltage V cell It consists of two parts, one of which is the Nernst voltage E Nernst , and the other part includes the ohmic overpotential η ohm and activation overpotential η act Overpotential inside:

[0053] V cell =E Nernst +η ohm +η act ;

[0054] Among them E Nernst It can be determined by the following formula:

[0055]

[0056] Where, and E CO Respectively represent the Nernst potential corresponding to the electrolysis generation of H2 and CO, R represents the universal gas constant (8.3145 J / mol·K), T represents the temperature of the battery, and F represents the Faraday constant (96485 C / mol). P CO and are the partial pressures of hydrogen, oxygen, water, carbon monoxide, and carbon dioxide, respectively. and are the mole fractions of water and carbon dioxide.

[0057] The magnitude of the ohmic overpotential is linearly related to the current density of the electrolytic cell and can be determined by the following formula:

[0058]

[0059] Where i represents the stack current, L e Indicates the thickness of the electrolyte layer.

[0060] The activation overpotential is the voltage loss caused by the limitation of electrochemical reaction kinetics, including the activation overpotential loss of hydrogen, oxygen and carbon monoxide:

[0061]

[0062] in, represents the activation overpotential of hydrogen, η act,CO represents the activation overpotential of carbon monoxide, represents the activation overpotential of oxygen, represents the current density of the water vapor reaction, represents the exchange current density of the water vapor reaction, represents the number of electrons transferred by the water vapor reaction, F represents the Faraday constant, represents the current density of the carbon dioxide reaction, represents the exchange current density of the carbon dioxide reaction, represents the number of electrons transferred by the carbon dioxide reaction, represents the current density of the oxygen reaction, represents the number of electrons transferred by the oxygen reaction, represents the current density of the oxygen reaction.

[0063] The exchange current density can be obtained by the following equation:

[0064]

[0065] Where, represents the mole fraction of water vapor in the mixed gas, represents the partial pressure of water vapor, represents the reference pressure of water vapor, represents the activation energy of the water vapor reaction, Indicates the mole fraction of CO2 in the mixed gas, represents the partial pressure of CO2, represents the reference pressure of CO2, represents the activation energy of the CO2 reaction, represents the mole fraction of O2 in the mixed gas, represents the oxygen partial pressure, represents the reference pressure of oxygen, It represents the activation energy of oxygen reaction. m, n, and k are empirical exponents that reflect the nonlinear effect of pressure on reaction rate.

[0066] The SOEC stack is a complex multi-physics coupling system. In addition to the electrochemical process, multiple factors such as chemical reaction heat and conduction heat need to be considered when modeling. - 3 s -1 ) can be determined by the Haberman model:

[0067]

[0068] k ps=exp(-0.2935Z 3 +0.6351Z 2 +4.1799Z+0.3169);

[0069]

[0070] Among them, k sf is the rate constant, k ps is the equilibrium constant, and Z is the temperature correlation coefficient.

[0071] The energy conservation equation of the battery stack can be established to calculate the temperature T of the battery stack:

[0072]

[0073] Among them, C v Indicates the specific heat capacity of the stack, h i represents the enthalpy of the i-th gas, P s Represents the externally supplied electrical energy (i.e. the electrical energy of the power supply), represents the molar flow rate of the i-th gas at the inlet of the fuel electrode, represents the molar flow rate of the i-th gas at the inlet of the air electrode, represents the molar flow rate of the i-th gas at the outlet of the fuel electrode, represents the molar flow rate of the i-th gas at the outlet of the air electrode, represents the mole fraction of the i-th gas at the fuel electrode inlet, represents the mole fraction of the i-th gas at the air electrode inlet, represents the mole fraction of the i-th gas at the outlet of the fuel electrode, represents the mole fraction of the i-th gas at the outlet of the air electrode.

[0074] Figure 1 The heat exchanger in the heat exchanger can be a countercurrent heat exchanger for gas heat recovery. The outlet temperature of the hot and cold streams can be determined by the following equations:

[0075]

[0076] Among them, C v,hot Specific heat capacity, T, which represents heat flow hot The temperature of heat flow, W hot.in Indicates the intake volume of heat flow, h hot,in represents the specific enthalpy of the heat flow inlet, W hot,out Indicates the intake volume of heat flow, h hot,out represents the specific enthalpy of the heat flow outlet, k hot The convection heat transfer coefficient, A, represents the heat flow hot Indicates the heat flow heat transfer area, Tplate Indicates the instantaneous temperature of the middle plate of the heat exchanger, C p,plate Indicates the specific heat capacity of the middle plate of the heat exchanger, k cold represents the convective heat transfer coefficient of the cold flow, A cold represents the cold flow heat transfer area, T cold Indicates the temperature of the cold stream.

[0077] When the inlet gas of the SOEC stack does not reach the preset temperature after passing through the heat exchanger, an electric heater is required for further heating. The electric heater model can be determined by the following equation:

[0078]

[0079] Among them, C V represents the specific heat at constant volume, W in Indicates the mass flow rate of the inlet gas, h in represents the specific enthalpy of the inlet gas, W out Indicates the mass flow rate of the outlet gas, h out represents the specific enthalpy of the outlet gas, and q represents the power of the electric heater.

[0080] In the present disclosure, a dynamic SOEC system simulation model can be established based on the above structure.

[0081] The above complete modeling process is implemented through the Matlab / Simulink software platform, and the dynamic nonlinear operation of the fuel cell stack is simulated. In order to verify the accuracy and effectiveness of the model, the simulation model is compared with the experimental data, such as Figure 2 As shown, the intake fuel composition is 10% H2, 45% CO2 and 45% H2O, with a flow rate of 25L / min; the air intake is oxygen, with a flow rate of 20L / min. Figure 2 It can be seen that at different temperatures (1023K, 1073K and 1123K), the simulation results of the model's polarization curves are in good agreement with the experimental results, with the maximum error not exceeding 3.9%, which indicates that the model has high accuracy.

[0082] Current research on temperature control in SOEC systems primarily focuses on the SOEC stack, with limited focus on the entire SOEC control system. While the SOEC stack is the core of the SOEC system, it also includes important thermal management components such as heat exchangers and electric heaters. This involves complex device interactions and cross-timescale responses. Relying solely on SOEC stack control is insufficient to ensure overall performance, requiring system-level temperature control to guarantee overall energy efficiency and reliability. Therefore, expanding research to the SOEC system level remains essential.

[0083] Even when the current control methods for SOEC stacks or SOEC systems involve the regulation of fuel flow rate, most studies focus on ensuring the conversion rate remains unchanged or preventing fuel shortages, and rarely conduct relevant research on the main control parameter of the stack temperature.

[0084] However, this type of solution is prone to temperature overshoot during actual control. For SOEC co-electrolysis systems, temperature overshoot not only affects the kinetics of the coordinated electrolysis of H2O and CO2, but can also cause abnormal fluctuations in the reaction gas conversion rate and product ratio, further exacerbating thermal stress and material degradation within the cell, affecting the stability of the electrolyte and electrode interface, accelerating the degradation of the stack performance, and thus reducing the overall syngas efficiency of the system.

[0085] In order to solve the above technical problems, the present disclosure provides a temperature control method, device, equipment, medium and product for a solid oxide electrolytic cell system.

[0086] Figure 3 is a flow chart showing a method for controlling the temperature of a solid oxide electrolytic cell system according to an exemplary embodiment. Figure 3 As shown, the method includes the following steps S101-S107:

[0087] In S101, the reference output temperature of the stack, the current density of the input stack, and the air flow rate in the solid oxide electrolysis cell system at the current moment are obtained.

[0088] The reference output temperature is the target temperature that the solid oxide electrolyzer (SOEC) system is expected to achieve. Its setting is based on a comprehensive range of factors, such as the material properties of the stack, the optimal conditions for the chemical reaction, and the efficiency requirements of the entire system. Different solid oxide materials have optimal ion conductivity at specific temperatures. To ensure efficient and stable electrolysis within the stack, the stack temperature must be controlled within an appropriate range. The ideal value within this range serves as the reference output temperature. Typically, this reference output temperature can be determined through preliminary experimental measurements, theoretical calculations, or based on the system's historical operating data. It can be set manually through the system's control interface or automatically by the system based on different operating modes.

[0089] Current density refers to the amount of current flowing through a fuel cell stack per unit area. It is one of the key factors affecting the electrolytic reaction rate within the stack. The higher the current, the more intense the electrolytic reaction, and the corresponding increase in heat generation. Current density is also closely related to the performance and lifespan of the fuel cell stack. Excessively high current density can cause localized overheating in the stack, accelerating material aging and damage; while too low a current density can slow the electrolytic reaction rate and reduce system efficiency. Current sensors and other devices can measure the current input to the stack in real time. Combined with the stack's effective area, the current density can be calculated.

[0090] In S102, the LM-BP NN model is used to determine the preliminary fuel flow rate based on the reference output temperature, current density and air flow, which serves as the feedforward fuel flow rate input of the fuel cell stack; the LM-BP NN model is used to predict the preliminary fuel flow rate based on the reference output temperature, current density and air flow.

[0091] The LM-BP NN model is a Levenberg-Marquardt back-propagation neural network model. To a certain extent, the LM-BP NN model overcomes the shortcomings of the traditional BP model, such as slow convergence speed and easy falling into local optimum.

[0092] The currently acquired reference output temperature, current density, and air flow rate are fed into the trained LM-BP NN model as input data. The model processes and calculates the input data based on its internally learned mapping relationships, and outputs a preliminary fuel flow rate value. This value is the fuel flow rate that the model predicts will enable the fuel stack to achieve optimal operating conditions under the reference output temperature, current density, and air flow rate, based on historical data and learned patterns.

[0093] In one example, the training process of the LM-BP NN model includes the following steps S1021-S1023:

[0094] S1021. Obtain a training set, where the training set includes samples and corresponding labels. The samples include sample temperature, sample current density, and sample air flow rate. The labels include sample fuel flow rate.

[0095] The labels used in this disclosure for training are sample fuel flow rates. These are the fuel flow rates required by the system under the specified operating conditions, corresponding to the sample temperature, sample current density, and sample air flow rate. The exact sample fuel flow rate corresponding to each sample can be obtained through experimental measurements or actual system operation records. Combining the samples and their corresponding labels forms the training set used to train the model.

[0096] It is worth noting that during the training process, the sample air flow rate is a fixed value and remains unchanged.

[0097] S1022: Input the sample into the initial LM-BP NN model to obtain the predicted fuel flow rate.

[0098] Before training begins, an initial LM-BP NN model must be constructed. This model has a specific network structure, consisting of an input layer, hidden layers, and an output layer. The number of neurons in the input layer is typically the same as the number of sample features—in this case, three (sample temperature, sample current density, and sample air flow). The output layer has one neuron, corresponding to the predicted fuel flow rate. The number of neurons and the number of layers in the hidden layer need to be adjusted and optimized based on the specific problem. Furthermore, model parameters such as weights and thresholds are randomly initialized.

[0099] Each sample in the training set (sample temperature, sample current density, and sample air flow) is input into the initial LM-BP NN model. The model processes the input sample according to its internal calculation rules. Specifically, the input signal passes from the input layer through the neurons in the hidden layer for weighted summation and nonlinear transformation, and is ultimately passed to the output layer to obtain a predicted fuel flow rate value. This predicted value is the model's initial estimate of the fuel flow rate corresponding to the sample, based on the current parameter settings.

[0100] S1023. Adjust model parameters in the initial LM-BP NN model according to the predicted fuel flow rate and the corresponding sample fuel flow rate to obtain a LM-BP NN model.

[0101] A loss function can be calculated based on the predicted fuel flow rate and the corresponding sample fuel flow rate. Based on minimizing the loss function, an iterative algorithm is used to adjust model parameters of the initial LM-BP NN model until the loss function is minimized or a preset number of iterations is reached, thereby obtaining the LM-BP NN model. The iterative algorithm includes a gradient descent algorithm.

[0102] The loss function measures the degree of discrepancy between the model's predictions and the actual labels. Common loss functions include mean squared error (MSE). Taking MSE as an example, for a given sample, the loss function can be expressed as the average of the squares of the difference between the predicted fuel flow rate and the sampled fuel flow rate. By calculating the loss function values ​​for all samples in the training set, an overall loss value can be obtained, which reflects the overall prediction error of the model under the current parameters.

[0103] After each update of the model parameters, the value of the loss function needs to be recalculated. This is to evaluate whether the prediction performance of the model has improved under the new parameters. An iterative algorithm (such as the gradient descent algorithm) is used to continuously adjust the model parameters. In each iteration, the gradient of the loss function is calculated, and the model parameters are updated based on the gradient. As the iteration proceeds, the value of the loss function will gradually decrease, and the prediction performance of the model will continue to improve. The iterative process will continue until the termination condition is met. There are usually two types of termination conditions: one is that the loss function is minimized, that is, the value of the loss function no longer decreases significantly, indicating that the model has converged to a better state; the other is that the preset number of iterations is reached, which is to avoid the algorithm falling into an infinite loop or the calculation time being too long. When the termination condition is met, the iterative process stops, and the model obtained at this time is the trained LM-BP NN model, which can be used to predict the fuel flow rate of new samples.

[0104] Because current density directly affects the electrochemical reaction rate and heat generation within the stack, it is a key parameter for temperature control. Fuel flow directly affects the supply of reactants and heat transfer within the stack, and is more efficient for temperature regulation than controlling the cooling air flow. Therefore, fuel flow rate control is chosen in this algorithm to achieve stack output temperature control. These input and output parameters are used to train a neural network model to establish a nonlinear mapping relationship between current density, fuel flow, air flow, and temperature.

[0105] In this disclosure, the basic form of the LM algorithm weight update parameter is:

[0106]

[0107] Among them, w k is the current weight parameter, which indicates the influence of input data on output results. represents the first-order derivative of the objective function F(w), w k+1 Represents the weight parameter for the next iteration.

[0108] During the training process of the neural network, an error function is used to quantify the difference between the predicted output and the actual output. Assume that the error function used in the BP neural network is:

[0109]

[0110] Among them, y k is the actual output, f(x k ,w k ) The predicted output of the neural network, x k is the input data, then the total error of the entire network (objective function, F(w)) can be expressed as:

[0111]

[0112] Where E(w) is the error vector, each element of which corresponds to the error of a sample. i (w) represents the square of the error value of the BP neural network corresponding to the i-th sample; then the j-th gradient component of F(w) is:

[0113]

[0114] Where w j represents the weight value corresponding to the jth training sample, and N represents the total number of samples;

[0115] In order to minimize the error function, it is necessary to calculate the gradient of the objective function F(w) with respect to the weight w, that is:

[0116]

[0117] Among them, J T (w) is the transpose of the Jacobian matrix of the error, which contains the partial derivatives of each output of the network with respect to the weights. In order to accelerate convergence, the LM algorithm also introduces the second-order derivative of the objective function, namely:

[0118]

[0119] Ignore the quadratic partial derivatives, only consider the dominant terms and write them in matrix form:

[0120]

[0121] Combining the first-order derivative and second-order derivative information and introducing the damping factor λ, we have:

[0122] w k+1 =w k -[J T (w k )J(w k )+λ k I] -1 J T (w k )E(w k );

[0123] Among them, λ k represents the damping factor of the kth iteration, J(w k ) represents the Jacobian matrix at the kth iteration, J T (w k ) represents its transpose, and I represents the identity matrix.

[0124] During training, a test set can also be included. After the weight update is completed, the test data (input state variables) in the test set are input into the LM-BP NN model after the weight update, and the predicted output temperature is obtained through forward propagation.

[0125] This disclosure also provides training and verification of neural network algorithms:

[0126] In this study, a LM-BP NN model with an input layer of 3 nodes, 5 hidden layers and an output layer of 1 node was established. In order to ensure the accuracy of the LM-BP NN model, this paper used 1500 sample data sets for model training, of which 70% of the data was used for training and 30% for testing and verification. Through the training and debugging of the network model, the network can predict the output temperature of the fuel cell according to the input parameters (current density and fuel flow rate), and automatically adjust the input parameters such as fuel flow rate based on the prediction results to achieve precise temperature control. Figure 4 The training results shown show that the mean squared error (MSE) of both the training set and the validation set converged quickly after 60 iterations. After about 150 iterations, the MSE of the training set and the validation set reached 0.1198 and 0.1003 respectively. Figure 5 The comparison between the true values ​​in the validation set and the predictions of the LM-BP NN model is shown, with a correlation coefficient (R) of 0.99985. This high correlation coefficient emphasizes the strong predictive ability of the LM-BP NN model.

[0127] In this disclosure, a feedforward-feedback control method based on the LM-BP NN model is proposed, such as Figure 6 As shown, the SOEC stack temperature can be adjusted by precisely controlling the fuel flow rate, and the fluctuation of key indicators such as voltage can be reduced.

[0128] Feedforward control logic: Feedforward control is a control method based on the prediction of the LM-BP NN model. The system uses the influence of known input variables to predict the output without relying on real-time feedback. This method predicts the control quantity in advance based on the current system state, so that interference can be responded to in advance. The LM-BP NN model in the feedforward loop can preliminarily predict the temperature change trend based on the current operating state of the SOEC system (including parameters such as current density, fuel flow and air flow), clarify the approximate range of control parameters, and adjust the fuel flow in advance, so that the network has basic responsiveness to input disturbances, thereby minimizing temperature fluctuations. There is no feedback information involved in this process. It is completely based on the feedforward hierarchy, which directly maps the input variables to the predicted output.

[0129] In S103 , fuel is delivered to the fuel cell stack according to the preliminary fuel flow rate.

[0130] By controlling Figure 2 The opening of the solenoid valve in the fuel cell is adjusted to accurately deliver the fuel to the fuel cell stack according to the set value of the initial fuel flow rate.

[0131] In S104 , a temperature output value of the fuel cell stack corresponding to a preliminary fuel flow rate is obtained.

[0132] After inputting a preliminary fuel flow rate to the stack, a temperature sensor can be used to measure the stack's current output temperature in real time, obtaining the stack's temperature output value corresponding to the preliminary fuel flow rate. Temperature sensors are typically installed in key locations on the stack to accurately reflect the stack's actual temperature.

[0133] In S105 , a PID control algorithm is used to obtain a fuel flow rate correction value based on the temperature output value and the reference output temperature, which is used as the feedback fuel flow rate input of the fuel cell stack.

[0134] The PID (Proportional-Integral-Derivative) control algorithm calculates the corresponding control variable (fuel flow rate correction) by comparing the deviation between the actual output of the system (current output temperature) and the desired output (reference output temperature).

[0135] In one embodiment, step S105 uses a PID control algorithm to obtain a fuel flow rate correction value based on the current output temperature and the reference output temperature, including the following sub-steps A1-A2:

[0136] A1. Obtain the error between the temperature output value and the reference output temperature.

[0137] In order to achieve precise control of the temperature of the fuel cell stack, it is necessary to calculate the difference between the two temperatures, that is, the error value.

[0138] A2. Use the PID control algorithm to obtain the fuel flow rate correction value based on the error value.

[0139] Specifically, the fuel flow rate correction amount is obtained according to the error value, the preset proportional coefficient, the preset integral coefficient, and the preset differential coefficient.

[0140] The PID control algorithm is a widely used feedback control algorithm consisting of three components: proportional (P), integral (I), and differential (D). These three components control the system from different perspectives, and their combined effect enables precise regulation of the system output. In this scenario, the required correction to the fuel flow rate is calculated based on the temperature error.

[0141] Proportional (P) control part:

[0142] Proportional control outputs a control quantity proportionally to the current error value. The calculation formula is:

[0143] up(t)=Kp×e(t);

[0144] Where up(t) is the output of the proportional control component, Kp is the preset proportional coefficient, and e(t) represents the error value. The proportional coefficient Kp determines the degree to which the error value affects the control variable. A larger Kp value results in a faster system response to the error, but may cause a larger overshoot. A smaller Kp value results in a slower system response, but may also improve stability.

[0145] Integral (I) control part:

[0146] Integral control is primarily used to eliminate steady-state errors in a system. It integrates the error value over a period of time. As time passes, the integral term accumulates the error, allowing adjustments to be made when the system experiences persistent errors. The calculation formula is:

[0147] u I (t) = K I ×∫0 t e(τ)dτ;

[0148] Among them, u I (t) is the output of the integral control part, K I is the preset integral coefficient; It is the integral of the error value from the initial time to the current time t. The integral coefficient K I The larger the value, the stronger the integral effect and the faster the system eliminates the steady-state error, but it may cause the system to experience integral saturation. I The smaller it is, the weaker the integral effect is and the slower the elimination of steady-state errors is.

[0149] Differential (D) control part:

[0150] Differential control predicts the future trend of the system based on the rate of change of the error value and makes adjustments in advance to suppress excessive oscillation of the system and improve the stability of the system. Its calculation formula is:

[0151]

[0152] Among them, u D (t) is the output of the differential control part, K D is the preset differential coefficient, Indicates the rate of change of the error value. Differential coefficient K D The larger the K is, the stronger the differential effect is, and the more sensitive the system is to the error change, but it may amplify the influence of noise. D The smaller it is, the weaker the differential effect is and the less obvious the response to error changes is.

[0153] Comprehensive calculation of fuel flow rate correction:

[0154] By adding the outputs of the proportional, integral and differential parts, we can get the final fuel flow rate correction u(t):

[0155] u(t)=up(t)+u I (t)+u D (t).

[0156] In S106, a corrected fuel flow rate for delivering fuel to the solid oxide electrolyzer system is determined based on the fuel flow rate correction amount and the preliminary fuel flow rate.

[0157] Combining the fuel flow rate correction with the preliminary fuel flow rate can yield the corrected fuel flow rate. Typically, the two are combined using a simple addition operation: corrected fuel flow rate = preliminary fuel flow rate + fuel flow rate correction. If the fuel flow rate correction is positive, it indicates that the fuel delivery flow rate needs to be increased; if it is negative, the fuel delivery flow rate needs to be reduced. In this way, the fuel delivery flow rate can be dynamically adjusted based on the actual temperature of the stack, allowing the current output temperature of the stack to gradually meet the preset conditions, ensuring stable and efficient operation of the stack.

[0158] Feedback control logic: Feedback control is a mechanism that relies on real-time feedback information and dynamically adjusts the control parameters according to the output results. In this method, it further accurately adjusts the fuel flow rate based on the temperature output value of the fuel cell stack corresponding to the preliminary fuel flow rate to ensure that the temperature can stabilize near the target value (reference output temperature). In the transient response stage, it compares the current output temperature with the reference output temperature, and the calculated error is corrected through the PID control algorithm to generate a fuel flow rate correction for fine-tuning. This correction is superimposed on the preliminary fuel flow rate output by the feedforward control to form the final fuel flow rate input to the fuel cell stack.

[0159] In S107, fuel is delivered to the solid oxide electrolyzer system according to the corrected fuel flow rate, so that the real-time temperature output value of the stack corresponding to the corrected fuel flow rate and the reference output temperature meet preset conditions.

[0160] This feedforward-feedback synergy can not only use the rapid response capability of the neural network to achieve preliminary predictions, but also improve the stability and accuracy of the predictions through more precise control algorithms. Feedforward control is to provide preliminary predictions so that the network has basic response capabilities to input disturbances. Through the feedforward structure, the network can quickly respond to changes in input variables and achieve rapid control. Feedback control is based on feedforward control and uses error feedback to continuously correct the prediction results to improve the accuracy and stability of the system. The outputs of the feedforward and feedback controllers are combined as the final fuel flow rate control signal of the system and applied to the SOEC simulation system to output the optimized real-time stable temperature.

[0161] Dynamic response verification:

[0162] The SOEC system using feedforward-feedback control based on LM-BP NN showed higher efficiency than the traditional basic fuel flow control based on PID control, and the accuracy and speed of temperature regulation were greatly improved. Figure 7-12 The performance of the LM-BP NN-based feedforward-feedback control method was compared with the basic fuel flow control method based on PID control. In Figure 7, it is clear that the LM-BP NN-based feedforward-feedback control significantly improves the overshoot and stabilization time of the stack temperature. Figure 8 The comparison of voltage regulation is shown, and the voltage fluctuation range is reduced from 1.12-2.07 V to 1.37-1.61 V, indicating that the feedforward-feedback control based on LM-BP NN effectively suppresses the voltage oscillation. Figure 9 The variation of fuel flow rate was demonstrated, and more precise regulation was observed, with the maximum fuel flow rate reduced from 0.0021 mol / s to 0.0012 mol / s, minimizing unnecessary fuel consumption and reducing system energy consumption. Figure 10-12 The effectiveness of feedforward-feedback control based on the LM-BP neural network on the syngas ratio, system conversion rate, and system efficiency was demonstrated. Due to precise fuel control, the LM-BP neural network-based feedforward-feedback approach effectively mitigated the impact of current fluctuations on the syngas ratio, conversion rate, and efficiency. The syngas ratio fluctuation range was reduced from 2.06-2.16 to 2.07-2.08, the system conversion rate range was narrowed from 40.5-82.8% to 68-76.1%, and the efficiency range was reduced from 56.2-77.8% to 62.4-72.7%. These results demonstrate that the LM-BP neural network-based feedforward-feedback approach not only improves temperature regulation but also effectively suppresses fluctuations in voltage and other performance indicators, addressing the shortcomings of basic fuel feedback control methods.

[0163] To further verify the effectiveness of the feedforward-feedback control strategy, the present invention uses the actual current data of a photovoltaic power station in a certain province and city as input and analyzes the changes in key system indicators when applying the feedforward-feedback method based on LM-BP NN under fluctuating conditions. It is found that the neural network control strategy can effectively suppress temperature overshoot when dealing with large nonlinear fluctuations in the system, and has higher response speed and stability. Figure 13 As shown, the stack temperature stabilized at approximately 1073K, demonstrating that the control algorithm effectively managed the system's thermal behavior. The stack voltage fluctuated around 1.5V, reaching a maximum of 1.51V and a minimum of 1.48V, with no significant deviations. The fuel flow rate pattern closely aligned with the current fluctuations, remaining low during periods of low current and increasing as current increased.

[0164] While traditional PID control methods can achieve temperature regulation in SOEC systems to a certain extent, and their algorithmic structure is simple and computationally fast, they lack robustness when dealing with complex operating conditions and large fluctuations. Especially when dealing with nonlinear systems and large current density fluctuations, PID control often suffers from temperature overshoot or regulation lag, making precise temperature control difficult. Furthermore, PID control requires frequent parameter tuning, making it difficult to adapt to rapid changes in dynamic environments.

[0165] In contrast, the neural network-based control strategy disclosed in this paper can respond to the system's nonlinear behavior in real time by adaptively adjusting control parameters. Especially under current steps or complex operating conditions, the neural network approach exhibits higher response speed and control accuracy, thus offering significant advantages in robustness and efficiency.

[0166] The method disclosed herein can improve control accuracy. Compared with the traditional PID control method, the feedforward-feedback method based on LM-BPNN can significantly reduce the temperature overshoot of the SOEC stack and reduce the temperature adjustment time. It can also reduce fluctuations. This method effectively suppresses the fluctuations of the key performance indicators of the system. The fluctuation range of the stack voltage is reduced from 1.12-2.07V to 1.37-1.61V, and the fluctuations of the synthesis gas ratio and system efficiency are also greatly reduced. It can also improve system efficiency. Due to the precise control of the fuel flow, the fuel consumption of the system is reduced and the overall efficiency is significantly improved.

[0167] Based on the same inventive concept, embodiments of the present application further provide a temperature control device for a solid oxide electrolytic cell system for implementing the temperature control method for a solid oxide electrolytic cell system. The solution to the problem provided by this device is similar to the solution described in the method described above. Therefore, the specific limitations of one or more embodiments of the temperature control device for a solid oxide electrolytic cell system provided below can be found in the above-mentioned limitations of the temperature control method for a solid oxide electrolytic cell system, and will not be further elaborated here.

[0168] In an exemplary embodiment, Figure 14 As shown, a temperature control device for a solid oxide electrolytic cell system is provided, the device comprising:

[0169] An input data acquisition module 11 is used to obtain a reference output temperature of the stack in the solid oxide electrolyzer system at the current moment, and input a current density and air flow rate of the stack;

[0170] a preliminary fuel flow rate determination module 12, configured to determine a preliminary fuel flow rate using a LM-BP NN model based on the reference output temperature, the current density, and the air flow rate, as a feedforward fuel flow rate input for the fuel cell stack; the LM-BP NN model is configured to predict the preliminary fuel flow rate based on the reference output temperature, the current density, and the air flow rate;

[0171] a fuel delivery module 13, configured to deliver fuel to the fuel stack according to the preliminary fuel flow rate;

[0172] a temperature output value acquiring module 14, configured to acquire a temperature output value of the fuel cell stack corresponding to the preliminary fuel flow rate;

[0173] A fuel flow rate correction module 15 is configured to obtain a fuel flow rate correction value based on the temperature output value and the reference output temperature using a PID control algorithm as a feedback fuel flow rate input of the fuel cell stack;

[0174] a corrected fuel flow rate module 16 for determining a corrected fuel flow rate for delivering fuel to the solid oxide electrolyzer system based on the fuel flow rate correction amount and the preliminary fuel flow rate;

[0175] The fuel delivery module 13 is further configured to deliver fuel to the solid oxide electrolyzer system according to the corrected fuel flow rate, so that the real-time temperature output value of the stack corresponding to the corrected fuel flow rate and the reference output temperature meet preset conditions.

[0176] As an optional implementation, in the aspect of obtaining the fuel flow rate correction value by using a PID control algorithm according to the temperature output value and the reference output temperature, the fuel flow rate correction value module is specifically configured to:

[0177] Obtaining an error value between the temperature output value and the reference output temperature;

[0178] The fuel flow rate correction amount is obtained by using the PID control algorithm according to the error value.

[0179] As an optional implementation manner, in the aspect of obtaining the fuel flow rate correction value by using the PID control algorithm according to the error value, the fuel flow rate correction value module is specifically configured to:

[0180] The fuel flow rate correction amount is obtained according to the error value, a preset proportional coefficient, a preset integral coefficient, and a preset differential coefficient.

[0181] As an optional implementation, the device further includes:

[0182] A training set acquisition module is used to acquire a training set, wherein the training set includes: samples and corresponding labels; the samples include: sample temperature, sample current density and sample air flow; the labels include: sample fuel flow rate;

[0183] a predicted fuel flow rate acquisition module, configured to input the sample into an initial LM-BP NN model to obtain a predicted fuel flow rate;

[0184] The LM-BP NN model acquisition module is configured to adjust model parameters in the initial LM-BP NN model according to the predicted fuel flow rate and the corresponding sample fuel flow rate to acquire the LM-BP NN model.

[0185] As an optional embodiment, in the aspect of adjusting the model parameters in the initial LM-BP NN model according to the predicted fuel flow rate and the corresponding sample fuel flow rate, the LM-BP NN model acquisition module is specifically configured to:

[0186] A loss function is calculated based on the predicted fuel flow rate and the corresponding sample fuel flow rate. According to the minimization of the loss function, the model parameters of the initial LM-BP NN model are adjusted using an iterative algorithm until the loss function is minimized or a preset number of iterations is reached, thereby obtaining the LM-BP NN model.

[0187] As an optional embodiment, in determining the corrected fuel flow rate for delivering fuel to the solid oxide electrolyzer system based on the fuel flow rate correction amount and the preliminary fuel flow rate, the LM-BP NN model acquisition module is specifically configured to:

[0188] The sum of the fuel flow rate correction amount and the preliminary fuel flow rate is taken as the corrected fuel flow rate.

[0189] In an exemplary embodiment, a computer device is provided. The computer device may be a server or a terminal. The internal structure diagram thereof may be as follows: Figure 15As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, abbreviated as I / O) and a communication interface. The processor, memory and input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a temperature control method for a solid oxide electrolyzer system is implemented.

[0190] Those skilled in the art will understand that Figure 15 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0191] In an exemplary embodiment, a computer device is further provided, including a memory and a processor. The memory stores a computer program, and the processor implements the steps in the above method embodiments when executing the computer program.

[0192] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0193] In an exemplary embodiment, a computer program product is provided, including a computer program. When the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0194] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0195] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, database or other media used in the embodiments provided in this application may include at least one of non-volatile and volatile memory. Non-volatile memory may include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory may include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0196] The databases involved in the various embodiments provided herein may include at least one of a relational database and a non-relational database. Non-relational databases may include, but are not limited to, distributed databases based on blockchains. The processors involved in the various embodiments provided herein may include, but are not limited to, general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic units, data processing logic units based on quantum computing, and the like.

[0197] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0198] This document uses specific examples to illustrate the principles and implementation methods of this application. The description of the above examples is only intended to help understand the method and core concept of this application. At the same time, for those skilled in the art, based on the concept of this application, there may be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A temperature control method for a solid oxide electrolyzer system, characterized in that: The method comprises: Obtaining a reference output temperature of a stack in a solid oxide electrolyzer system at a current moment, a current density input to the stack, and an air flow rate; Determining a preliminary fuel flow rate using a LM-BP NN model based on the reference output temperature, the current density, and the air flow rate as a feedforward fuel flow rate input for the fuel cell stack; the LM-BP NN model is used to predict the preliminary fuel flow rate based on the reference output temperature, the current density, and the air flow rate; delivering fuel to the stack according to the preliminary fuel flow rate; obtaining a temperature output value of the fuel cell stack corresponding to the preliminary fuel flow rate; According to the temperature output value and the reference output temperature, a PID control algorithm is used to obtain a fuel flow rate correction value as a feedback fuel flow rate input of the fuel cell stack; determining a corrected fuel flow rate for delivering fuel to the solid oxide electrolyzer system based on the fuel flow rate correction and the preliminary fuel flow rate; Fuel is delivered to the solid oxide electrolyzer system according to the corrected fuel flow rate, so that the real-time temperature output value of the stack corresponding to the corrected fuel flow rate and the reference output temperature meet preset conditions.

2. The temperature control method of the solid oxide electrolyzer system according to claim 1, characterized in that: The method of obtaining a fuel flow rate correction value by using a PID control algorithm according to the temperature output value and the reference output temperature includes: Obtaining an error value between the temperature output value and the reference output temperature; The fuel flow rate correction amount is obtained by using the PID control algorithm according to the error value.

3. The temperature control method of the solid oxide electrolyzer system according to claim 2, characterized in that: The step of obtaining the fuel flow rate correction value by using the PID control algorithm according to the error value includes: The fuel flow rate correction amount is obtained according to the error value, a preset proportional coefficient, a preset integral coefficient, and a preset differential coefficient.

4. The temperature control method of the solid oxide electrolyzer system according to claim 1, characterized in that: The method further comprises: Acquire a training set, the training set comprising: samples and corresponding labels; the samples comprising: sample temperature, sample current density, and sample air flow; the labels comprising: sample fuel flow rate; Inputting the sample into an initial LM-BP NN model to obtain a predicted fuel flow rate; The model parameters in the initial LM-BP NN model are adjusted according to the predicted fuel flow rate and the corresponding sample fuel flow rate to obtain the LM-BP NN model.

5. The temperature control method of the solid oxide electrolyzer system according to claim 4, characterized in that: The adjusting the model parameters in the initial LM-BP NN model according to the predicted fuel flow rate and the corresponding sample fuel flow rate to obtain the LM-BP NN model includes: A loss function is calculated based on the predicted fuel flow rate and the corresponding sample fuel flow rate. According to the minimization of the loss function, the model parameters of the initial LM-BP NN model are adjusted using an iterative algorithm until the loss function is minimized or a preset number of iterations is reached, thereby obtaining the LM-BP NN model.

6. The temperature control method of the solid oxide electrolytic cell system according to claim 5, characterized in that: The determining, based on the fuel flow rate correction amount and the preliminary fuel flow rate, of a corrected fuel flow rate for delivering fuel to the solid oxide electrolyzer system comprises: The sum of the fuel flow rate correction amount and the preliminary fuel flow rate is taken as the corrected fuel flow rate.

7. A temperature control device for a solid oxide electrolytic cell system, characterized in that: The device comprises: A reference output temperature and current density acquisition module is used to obtain the reference output temperature of the stack in the solid oxide electrolyzer system at the current moment, the current density input to the stack, and the air flow rate; a preliminary fuel flow rate determination module, configured to determine a preliminary fuel flow rate using a LM-BP NN model based on the reference output temperature, the current density, and the air flow rate, as a feedforward fuel flow rate input for the fuel cell stack; the LM-BP NN model is configured to predict the preliminary fuel flow rate based on the reference output temperature, the current density, and the air flow rate; a fuel delivery module, configured to deliver fuel to the fuel stack according to the preliminary fuel flow rate; a temperature output value acquisition module, configured to acquire a temperature output value of the fuel cell stack corresponding to the preliminary fuel flow rate; a fuel flow rate correction module, configured to obtain a fuel flow rate correction value using a PID control algorithm based on the temperature output value and the reference output temperature, as a feedback fuel flow rate input of the fuel cell stack; a corrected fuel flow rate module for determining a corrected fuel flow rate for delivering fuel to the solid oxide electrolyzer system based on the fuel flow rate correction amount and the preliminary fuel flow rate; The fuel delivery module is further used to deliver fuel to the solid oxide electrolyzer system according to the corrected fuel flow rate, so that the real-time temperature output value of the stack corresponding to the corrected fuel flow rate and the reference output temperature meet preset conditions.

8. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the temperature control method for a solid oxide electrolysis cell system according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the temperature control method for a solid oxide electrolysis cell system according to any one of claims 1 to 6 are implemented.

10. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the temperature control method for a solid oxide electrolysis cell system according to any one of claims 1 to 6 are implemented.

Images