Optimization method for preventing freezing of fuel cell during cold start

Abstract

The invention discloses an optimization method for preventing freezing in cold start of a fuel cell, and relates to the technical field of fuel cells, the optimization method comprises the following steps: S1, collecting environment temperature and stack internal freezing condition data, and analyzing water production rate and temperature rise ice melting capability; s2, constructing a prediction model capable of reflecting the relationship among the environment temperature, the internal icing condition of the electric pile and the ideal current density; s3, inputting the environment temperature and the internal icing condition of the electric pile into the prediction model, and adjusting the parameters of the fuel cell; the air temperature is indirectly controlled through a current density curve predicted by a model, stack water production-ice melting balance is converted into a flow distribution optimization problem, and voltage fluctuation feed-forward compensation is combined, so that cooperative control of the temperature and the electrochemical state is realized, the performance of the fuel cell in the cold start process is optimized, and the performance of the fuel cell in the cold start process is improved. And the stability and the output efficiency of the system are enhanced, and powerful technical support is provided for application of the fuel cell in a low-temperature environment.

Description

Technical Field

[0001] The present invention relates to the technical field of fuel cells, especially heat exchange, and specifically provides an optimized method for preventing fuel cell cold start icing. Background Art

[0002] Hydrogen fuel cells (PEMFCs), as an efficient and environmentally friendly energy conversion device, have shown great application potential in recent years in the fields of new energy vehicles, distributed power generation systems, and backup power supplies. The working principle of PEMFCs is based on the electrochemical reaction of hydrogen and oxygen under the action of a catalyst, directly generating electric energy and water. Its only by-product - water, is not only harmless to the environment but also reflects the high cleanliness of the energy conversion process. Therefore, hydrogen energy, as a clean energy source, has received extensive attention and application.

[0003] In practical applications, especially under extreme climate conditions, PEMFCs face many technical challenges. Especially in low-temperature environments, the cold start problem of PEMFCs is particularly prominent. In an environment below zero degrees Celsius, ice and frost are likely to accumulate inside and outside the stack of PEMFCs (such as pipelines, valves, etc.). In particular, the water vapor generated by the electrochemical reaction will quickly condense and may freeze after contacting the cold components. This will not only block the gas flow channels, reduce the effective supply of reaction gases, but also increase the internal resistance of the stack, resulting in a decline in battery performance and even inability to start.

[0004] Currently, the industry mainly adopts two strategies to improve the start-up ability of PEMFCs in low-temperature environments: one is to use a PTC heating system to preheat key components, and the other is to use the heat generated by the chemical reaction of the stack itself to achieve self-heating. However, the cold start response time of domestic systems is currently too long. During this process, the water generated by the stack will freeze due to the low-temperature environment, affecting the start-up time and performance of the stack. Summary of the Invention

[0005] The present invention overcomes the deficiencies of the prior art and provides an optimized method for preventing fuel cell cold start icing.

[0006] To achieve the above object, the technical solution adopted by the present invention is: a device for preventing fuel cell cold start icing, including: a stack, a first electronic throttle valve, an intercooler, and an air compressor sequentially installed on the air inlet pipeline of the stack, and a back pressure valve installed on the air outlet pipeline of the stack; A three-way valve is installed between the first electronic throttle valve and the intercooler. One end of the three-way valve is connected to a bypass pipeline, and one end of the bypass pipeline is connected to the air inlet pipeline between the intercooler and the air compressor; A communication pipeline is connected between the air outlet pipeline and the air inlet pipeline. One end of the communication pipeline is located between the three-way valve and the first electronic throttle valve, and the other end is located on one side of the back-pressure valve.

[0007] In a preferred embodiment of the present invention, an air filter for ensuring the purity of the sucked gas is installed at one end of the air inlet pipeline; the air filter is located on one side of the air compressor.

[0008] In a preferred embodiment of the present invention, a pressure sensor for detecting gas pressure is installed between the three-way valve and the first electronic throttle valve.

[0009] In a preferred embodiment of the present invention, a second electronic throttle valve for adjusting the pressure value of the sucked gas is installed on the communication pipeline.

[0010] In a preferred embodiment of the present invention, a silencer for reducing the noise of gas discharge is installed at one end of the air outlet pipeline.

[0011] The present invention provides an optimized method for preventing fuel cell cold start icing, including the following steps: S1. Collect the ambient temperature and data on the internal icing condition of the fuel cell stack under various working conditions, and analyze the water production rate and temperature rise and ice melting ability under each working condition; S2. According to the water production rate and temperature rise and ice melting ability, construct a prediction model that can reflect the relationship between the ambient temperature, the internal icing condition of the fuel cell stack, and the ideal current density; S3. Input the ambient temperature and data on the internal icing condition of the fuel cell stack during cold start into the prediction model, output the current density curve, and the flow distribution ratio between the intercooler and the bypass pipeline that matches, and adjust the fuel cell parameters in combination with the control algorithm.

[0012] In a preferred embodiment of the present invention, in the step of S1, it includes the following sub-steps: S11. Collect the ambient temperature and data on the internal icing condition of the fuel cell stack under various working conditions; S12. Preprocess the collected data; including removing duplicate, incomplete or abnormal data items involved, and performing normalization processing; S13. Analyze the ambient temperature and the internal icing condition of the fuel cell stack under each working condition, the corresponding water production rate and temperature rise and ice melting ability.

[0013] In a preferred embodiment of the present invention, in the step of S2, it includes the following sub-steps: S21. Analyze the correlation between the ambient temperature, the internal icing condition of the fuel cell stack, and the current density; S22. Based on the Transformer architecture, using the ambient temperature and the internal icing condition of the fuel cell stack as input features and the ideal current density as the output target, construct a prediction model that can predict the current density curve and the flow distribution ratio between the matched intercooler and the bypass pipeline. S23. Divide the data collected and processed in step S1 into a training set and a validation set. Use the training set to train the prediction model and use the validation set to evaluate the performance of the prediction model.

[0014] In a preferred embodiment of the present invention, in step S3, the following sub-steps are included: S31. During the cold start of the fuel cell, collect the ambient temperature and the internal icing condition data of the fuel cell stack in real time and input them into the prediction model. S32. Obtain the output current density curve and the flow distribution ratio between the matched intercooler and the bypass pipeline through the prediction model. S33. According to the current density curve and the flow distribution ratio, combine the control algorithm to adjust the parameters of the fuel cell, and dynamically adjust the parameters of the control algorithm according to the real-time feedback data.

[0015] In a preferred embodiment of the present invention, in step S33, for the parameter adjustment of the fuel cell: taking the internal temperature of the fuel cell stack and the water production-ice melting balance as the control targets, where the internal temperature of the fuel cell stack directly affects the ice melting ability and the water production rate. By comparing the actually measured internal temperature of the fuel cell stack with , calculate the temperature error: ; By comparing the actual water production rate with the ice melting rate , calculate the water production-ice melting balance error: ; Through proportional (P), integral (I), and derivative (D) control, adjust the control parameters to reduce the error. The output of the PID controller is: ; where , and are the proportional, integral, and derivative gains respectively. For the strategy of adjusting the control algorithm parameters: Proportional gain Adjustment: In the initial stage, if the temperature or the water production-ice melting balance deviates from the target value greatly, increase To improve the response speed and make the stack temperature or the water production - ice melting state quickly approach the target value; when approaching the target value, reduce To avoid oscillations caused by over - adjustment; Integral gain Adjustment: When the temperature or the water production - ice melting balance error persists, appropriately increase Accumulate the error signal to prompt the system to eliminate long - term deviations; when the error changes rapidly, appropriately reduce To avoid integral saturation; Differential gain Adjustment: When the temperature or the water production - ice melting state changes sharply, increase To offset the inertial influence and improve the system stability; when the system tends to be stable, reduce To avoid over - inhibition; Input the predicted current density curve and the flow distribution ratio As a feed - forward signal into the control system to guide the operating state of the stack to change in the target direction in advance, then the final control parameter Is: ; Wherein, Is the output of the PID controller; Allocate the final control quantity To the current density adjustment and flow distribution ratio adjustment mechanisms to achieve precise control of the operating parameters of the stack; Adjustment based on real - time feedback data: According to the real - time collected feedback data of the stack voltage fluctuation, temperature change, and water production - ice melting state, periodically evaluate the current control effect and dynamically adjust the parameters of the PID controller: ; ; ; Wherein, 、 And Are the adjustment steps of the proportional, integral, and differential gains; 、 And Are the adjustment terms for the water production - ice melting balance error.

[0016] The present invention solves the defects existing in the background technology, and the present invention has the following beneficial effects: (1) The present invention provides a device for preventing fuel cell cold start icing. By means of the first electronic throttle valve, intercooler and air compressor on the air inlet pipeline, and the back pressure valve on the air outlet pipeline, when the fuel cell system starts in a sub-zero low temperature environment, through the cooperation of the back pressure valve, the outlet pressure of the air compressor can be increased, that is, the pressure ratio of the air compressor is increased. And with the cooperation of the three-way valve and the bypass pipeline, the flow distribution of air between the intercooler and the bypass pipeline can be reasonably regulated, and the air temperature entering the stack can be effectively adjusted, thus effectively solving the problem that the water generated by the stack during cold start directly freezes, and avoiding the start-up time and performance of the stack.

[0017] (2) The present invention provides an optimization method for preventing fuel cell cold start icing. When the fuel cell is cold started, based on the prediction model of the Transformer architecture and combined with the adjustment strategy, the problem of multi-field coupling imbalance of heat-electric-fluid during the cold start of the fuel cell in an extremely low temperature environment is effectively solved. The cold start success rate is significantly increased to more than 90%, and at the same time, the voltage fluctuation is controlled within 5%. By indirectly controlling the air temperature through the current density curve predicted by the model, the water production - ice melting balance of the stack is transformed into a flow distribution optimization problem, and combined with voltage fluctuation feedforward compensation, the coordinated control of temperature and electrochemical state is achieved. This not only optimizes the performance of the fuel cell during cold start, but also enhances the stability and output efficiency of the system, providing strong technical support for the application of fuel cells in low temperature environments.

[0018] (3) In the present invention, by combining a control algorithm to adjust the parameters of the fuel cell, the water production - ice melting balance of the stack is transformed into a flow distribution optimization problem. According to the flow distribution ratio output by the prediction model, the flow between the intercooler and the bypass pipeline is adjusted in real time, and the air temperature is indirectly controlled, which can accurately control the thermal field and reaction environment inside the stack, make the water production rate match the temperature rise and ice melting ability, avoid ice accumulation caused by too fast water production rate, and then prevent the situation that ice blocks the catalyst layer and affects the diffusion of reactants, ensuring the smooth progress of the reaction, reducing voltage fluctuation, and improving the output performance of the system.

[0019] (4) In the present invention, by installing an air filter at one end of the air inlet pipeline, when the air compressor draws in external gas, through the cooperation of the air filter, particulate matter, dust, grease or other pollutants in the drawn gas can be filtered, ensuring the purity of the drawn gas, and avoiding the adverse effects of impurities or pollutants in the drawn gas on the operating performance and life of the air compressor, thereby reducing the pollution and damage to the fuel cell. Description of the Drawings

[0020] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the accompanying drawings required for the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are only some embodiments recorded in the present invention. For those of ordinary skill in the art, without creative efforts, other accompanying drawings can also be obtained based on these drawings; Figure 1 is a schematic diagram of the overall structure of a device for preventing fuel cell cold start icing according to a preferred embodiment of the present invention; Figure 2 is a schematic diagram of the flow of an optimization method for preventing fuel cell cold start icing according to a preferred embodiment of the present invention; In the figure: 1. First electronic throttle valve; 2. Intercooler; 3. Air compressor; 4. Back pressure valve; 5. Three-way valve; 6. Bypass pipeline; 7. Connecting pipeline; 8. Air filter; 9. Pressure sensor; 10. Second electronic throttle valve; 11. Muffler. Detailed implementation manners

[0021] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of them. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0022] In the following description, many specific details are set forth to fully understand the present invention. However, the present invention can also be implemented in other ways different from those described herein. Therefore, the scope of protection of the present invention is not limited by the specific embodiments disclosed below.

[0023] Application overview: In an extremely low temperature environment, the cold start process of a fuel cell faces multiple challenges such as drastic temperature changes and imbalance between water production rate and ice melting ability. These factors can easily lead to voltage fluctuations, seriously affecting the stability and output performance of the system. Traditional control strategies, such as constant current or constant voltage startup, often fail to start or cause large voltage fluctuations because they cannot accurately match the heat generation and ice formation rates.

[0024] In practical applications, it is found that when a fuel cell is cold started, if the water production rate exceeds the temperature rise ice melting ability, ice accumulation will occur, which not only hinders the progress of chemical reactions, but also increases the impedance of the system, reducing the voltage and current output. Especially in the catalyst layer, ice blockage will seriously affect the diffusion of reactants, further deteriorating the system performance. Therefore, how to achieve a dynamic balance between water production and ice melting during the cold start process has become a key problem to be solved urgently.

[0025] In view of the above technical problems, the concept of the present invention is to propose an optimized method for preventing fuel cell cold start icing. Through an intelligent prediction model and a dynamic control strategy, the problem of multi-field coupling imbalance during the cold start process of fuel cells in extremely low temperature environments is solved, significantly improving the cold start success rate and system stability, and providing strong technical support for the wide application of fuel cells in low temperature environments.

[0026] As Figure 1 shown, a device for preventing fuel cell cold start icing includes: a fuel cell stack, a first electronic throttle valve 1, an intercooler 2, and an air compressor 3 sequentially installed on the air inlet pipeline of the fuel cell stack, and a back pressure valve 4 installed on the air outlet pipeline of the fuel cell stack; a three-way valve 5 is installed between the first electronic throttle valve 1 and the intercooler 2, one end of the three-way valve 5 is connected to a bypass pipeline 6, and one end of the bypass pipeline 6 is connected to the air inlet pipeline between the intercooler 2 and the air compressor 3; a communication pipeline 7 is connected between the air outlet pipeline and the air inlet pipeline, one end of the communication pipeline 7 is located between the three-way valve 5 and the first electronic throttle valve 1, and the other end is located on one side of the back pressure valve 4.

[0027] It should be noted that when the fuel cell system starts in a sub-zero low temperature environment, the air compressor 3 sucks in low temperature air through one end of the air inlet pipeline, fully opens the first electronic throttle valve 1, and regulates the back pressure valve 4 on the air outlet pipeline to increase the outlet pressure of the air compressor 3, that is, increase the pressure ratio of the air compressor 3. At this time, the air temperature at the outlet of the air compressor 3 will rise above zero degrees Celsius. By adjusting the opening of the three-way valve 5, the flow distribution of air between the intercooler 2 and the bypass pipeline 6 can be reasonably regulated, effectively adjusting the air temperature entering the fuel cell stack, thereby effectively solving the problem of direct freezing of water generated by the fuel cell stack during the cold start process, and avoiding the start-up time and performance of the fuel cell stack.

[0028] In some embodiments, an air filter 8 for ensuring the purity of the sucked gas is installed at one end of the air inlet pipeline.

[0029] It should be noted that the air filter 8 is located on one side of the air compressor 3; by installing the air filter 8 at one end of the air inlet pipeline, when the air compressor 3 sucks in external gas, through the cooperation of the air filter 8, particulate matter, dust, grease or other pollutants in the sucked gas can be filtered to ensure the purity of the sucked gas, avoiding adverse effects of impurities or pollutants in the sucked gas on the operating performance and life of the air compressor 3, thereby reducing pollution and damage to the fuel cell.

[0030] In some embodiments, a pressure sensor 9 for detecting gas pressure is installed between the three-way valve 5 and the first electronic throttle valve 1; with the arrangement of the pressure sensor 9, since the pressure sensor 9 is located between the three-way valve 5 and the first electronic throttle valve 1, it is convenient to detect the pressure value of the incoming air, thereby facilitating the regulation operation of the back pressure valve 4 according to the feedback value.

[0031] In some embodiments, a second electronic throttle valve 10 for regulating the pressure value of the pumped-in gas is installed on the connecting pipeline 7; with the arrangement of the second electronic throttle valve 10, since the second electronic throttle valve 10 is installed on the connecting pipeline 7, by regulating the second electronic throttle valve 10, it is possible to conveniently adjust the gas pressure of the air entering the fuel cell stack.

[0032] In some embodiments, a muffler 11 for reducing the noise of the discharged gas is installed at one end of the air outlet pipeline; with the arrangement of the muffler 11, it is possible to achieve the effect of noise reduction for the gas discharged from the air outlet pipeline of the fuel cell stack.

[0033] When the present invention is in use, when the fuel cell system starts in a low-temperature environment below zero, the low-temperature air enters the air compressor 3 through the air filter 8. The first electronic throttle valve 1 is fully opened, and through the feedback of the pressure value detected by the pressure sensor 9, the back pressure valve 4 on the air outlet pipeline is regulated to increase the outlet pressure of the air compressor 3, that is, to increase the pressure ratio of the air compressor 3. At this time, the air temperature at the outlet of the air compressor 3 will rise above zero degrees Celsius. By adjusting the opening degree of the three-way valve 5, the flow rate distribution of the air between the intercooler 2 and the bypass pipeline 6 can be reasonably regulated, effectively regulating the air temperature entering the fuel cell stack, thereby effectively solving the problem of direct freezing of the water generated in the fuel cell stack during the cold start process and avoiding the start-up time and performance of the fuel cell stack.

[0034] As Figure 2 shown, the present invention provides an optimized method for preventing fuel cell cold start icing, including the following steps: S1. Collect the ambient temperature and fuel cell stack internal icing condition data under various working conditions of the fuel cell, and analyze the water production rate and temperature rise and ice melting ability under each working condition; S2. According to the water production rate and temperature rise and ice melting ability, construct a prediction model that can reflect the relationship between the ambient temperature, the fuel cell stack internal icing condition and the ideal current density; S3. Input the ambient temperature and fuel cell stack internal icing condition data during the cold start of the fuel cell into the prediction model, output the current density curve, and the flow rate distribution ratio between the intercooler 2 and the bypass pipeline 6 that matches, and combine the control algorithm to adjust the fuel cell parameters.

[0035] It should be noted that during the cold start of the fuel cell, based on the prediction model of the Transformer architecture and combined with the adjustment strategy, the problem of multi-field coupling imbalance of heat-electricity-fluid during the cold start of the fuel cell in extremely low temperature environments is effectively solved. The cold start success rate is significantly increased to over 90%, and at the same time, the voltage fluctuation is controlled within 5%. The air temperature is indirectly controlled through the current density curve predicted by the model, converting the water production-ice melting balance of the stack into a flow distribution optimization problem, and combined with voltage fluctuation feedforward compensation, the coordinated control of temperature and electrochemical state is achieved. This not only optimizes the performance of the fuel cell during cold start but also enhances the stability and output efficiency of the system, providing strong technical support for the application of fuel cells in low temperature environments.

[0036] In some specific implementation manners, in the step of S1, the following sub-steps are included: S11. Collect the ambient temperature and the data of the ice formation condition inside the fuel cell stack under various working conditions; S12. Preprocess the collected data; S13. Analyze the ambient temperature and the ice formation condition inside the fuel cell stack under each working condition, and the corresponding water production rate and temperature rise and ice melting capacity.

[0037] In this embodiment, in the step of S11, the various working conditions include but are not limited to low-temperature cold start, normal temperature operation, and operation under different load conditions; the ambient temperature is collected by arranging multiple temperature sensors at key positions of the fuel cell system, such as at the inlet of the air filter 8, around the air compressor 3, and on the outer surface of the fuel cell stack, etc., to ensure that the ambient temperature information can be comprehensively and accurately obtained; the ice formation condition inside the fuel cell stack is collected by using fiber optic sensors, ultrasonic sensors, or electrical impedance tomography technology to monitor the ice formation conditions of various key components inside the fuel cell stack, such as the catalyst layer, gas diffusion layer, and proton exchange membrane, etc., and perform operations such as filtering, amplifying, and analog-to-digital conversion on the original signals output by the sensors to convert them into digital signals that are convenient for analysis and processing.

[0038] In this embodiment, in the step of S12, the preprocessing includes removing duplicate, incomplete, or abnormal data items and performing normalization processing; removing duplicate data from the collected operation data and identifying and processing outliers: outliers may be caused by data entry errors or special patient conditions and need to be processed according to the actual situation, such as replacing them with the mean, median, or deleting them; the normalization processing is specifically: scaling the data proportionally so that it falls into a small specific interval, such as [0,1] or [-1,1], which helps to eliminate the influence of the dimension on the result, and the calculation formula is: ; where is the original data; is the normalized data.

[0039] In this embodiment, in step S13, for the water production rate: according to the current density of the fuel cell and the effective area of the stack , the total current is calculated . The relationship between the number of moles of water produced and the current is: ; wherein, is time; is the number of electrons transferred in the reaction (for a proton exchange membrane fuel cell, ); is the Faraday constant (about 96485 C / mol); The water production rate is affected by the actual working conditions (such as the supply of reactants, temperature, pressure, etc.), and the actual water production rate is obtained by correction : ; wherein, is the molar mass of water (18×10 -3 kg / mol); is the water production efficiency, which reflects the ratio of the actual water production rate to the theoretical water production rate; For the ice melting ability: calculate the heat generation power of the stack : ; wherein, is the reaction heat generation power, which is calculated according to the reaction heat and the water production rate: ; is the ohmic loss power, which is calculated according to the current and the internal resistance of the stack : ; According to the temperature distribution inside the stack and the physical properties of ice, evaluate the temperature rise ice melting ability. When the temperature on the surface or inside of the stack rises above the melting point of ice, the ice begins to melt, and the ice melting rate is: ; wherein, is the latent heat of fusion of ice (about 3.34×10 5 J / kg); is the heat used for ice melting, which is calculated by integrating the heat transferred to the ice surface by the stack over a period of time: , is the heat transfer efficiency, which reflects the ratio of the heat transferred to the ice surface to the total heat generated by the stack.

[0040] In some specific embodiments, in the step of S2, the following sub-steps are included: S21. Analyze the correlation between the ambient temperature, the ice formation condition inside the stack, and the current density; S22. Based on the Transformer architecture, use the ambient temperature and the ice formation condition inside the stack as input features, and the ideal current density as the output target to construct a prediction model that can predict the current density curve and the flow distribution ratio between the matching intercooler 2 and the bypass pipeline 6; S23. Divide the data collected and processed in the S1 step into a training set and a validation set, use the training set to train the prediction model, and use the validation set to evaluate the performance of the prediction model.

[0041] In this embodiment, in the step of S21, the ambient temperature and the ice formation condition data inside the stack are used as influencing factors for establishing the prediction model. For each influencing factor and the current density calculate the Pearson correlation coefficient : ; wherein, is the ambient temperature or the ice formation condition data value of the th sample; is the th sample corresponding to and the current density; is the mean value of the ambient temperature or the ice formation condition data, that is, ; is the mean value of the current density, that is, ; is the number of samples, that is, the total number of observations; When the Pearson correlation coefficient it indicates that the influencing factor and the current density are completely positively correlated, that is, an increase in one variable is always accompanied by an increase in the other variable; when it indicates that the influencing factor and the current density are completely negatively correlated, that is, an increase in one variable is always accompanied by a decrease in the other variable; when it indicates that there is no linear correlation between the influencing factor and the current density ; The closer the absolute value of is to 1, the stronger the linear relationship between the influencing factor and the current density, and the closer it is to 0, the weaker the linear relationship.

[0042] In this embodiment, in step S22, for the input layer: Let the time series of the environmental temperature be , where is the length of the time series; the relevant features (icing thickness, icing area) of the internal icing condition of the stack are represented as a feature vector ; For the encoder: It is set to multiple layers (6 layers or 8 layers), and the multi-head self-attention mechanism is adopted. The dimension of each head is , where is the number of heads; for the input sequence , the query , key and value tensors are obtained through linear projection: , , ; Among them, ; Calculate the self-attention output: ; Among them, is the dimension of the key; Each encoder layer contains a feed-forward neural network: ; Among them, is the output of the multi-head attention layer; and are weight matrices; and are bias vectors; is the hidden layer dimension of the feed-forward neural network; the input dimension is , and the intermediate layer dimension is a larger value (such as ); Add residual connections after each sub-layer (self-attention and feed-forward network) and apply layer normalization to stabilize the training process; For the decoder: The number of layers is the same as that of the encoder, and the multi-head self-attention mechanism is also adopted. At the same time, in addition to self-attention, each layer of the decoder also contains a cross-attention mechanism for paying attention to the encoder output and the input of the decoder itself. The feed-forward neural network structure is the same as that in the encoder, and residual connections and layer normalization are also applied after each sub-layer; For the output layer: Output the time series, representing the ideal current density at each time step, which is ; Predict the flow distribution ratio between the intercooler 2 and the bypass pipeline 6, which is represented as a vector , where is the flow distribution ratio at each time step.

[0043] In this embodiment, in step S23, during the prediction model training, the error between the predicted value and the true value is quantified: Current density prediction loss: ; where, is the length of the time series; is the time step; is the current density value predicted by the model at time ; is the current density value at time in the true data; Flow distribution ratio prediction loss: ; where, is the true flow distribution ratio value at time ; is the flow distribution ratio value predicted by the model at time ; The weight parameters of the model are continuously adjusted through the backpropagation algorithm, and the gradient of the loss function with respect to each parameter is calculated: ; where, is the loss function; is the model output; is the intermediate variable; is the model parameter; The model parameters are updated using an optimizer: ; where, is the parameter at the th iteration; is the learning rate; is the first moment estimate; is the second moment estimate; is a small constant used to prevent the denominator from being zero; After each training cycle ends, calculate the loss on the validation set. If the loss value on the validation set does not decrease for consecutive training cycles, stop training: .

[0044] In some specific implementation schemes, in step S3, the following sub-steps are included: S31. When the fuel cell is cold-started, the ambient temperature and the data of the icing condition inside the stack are collected in real time and input into the prediction model; S32. Obtain the output current density curve and the matching flow distribution ratio between the intercooler 2 and the bypass pipeline 6 through the prediction model; S33. Adjust the parameters of the fuel cell according to the current density curve and the flow distribution ratio, and dynamically adjust the parameters of the control algorithm based on the real-time feedback data.

[0045] It should be noted that by adjusting the parameters of the fuel cell in combination with the control algorithm, the water production - ice melting balance of the stack is transformed into a flow distribution optimization problem. According to the flow distribution ratio output by the prediction model, the flow between the intercooler 2 and the bypass pipeline 6 is adjusted in real time, and the air temperature is indirectly controlled, which can accurately control the thermal field and reaction environment inside the stack, make the water production rate match the temperature rise and ice melting ability, avoid ice accumulation caused by too fast water production rate, and then prevent the ice blockage of the catalyst layer from affecting the diffusion of reactants, ensure the smooth progress of the reaction, reduce the voltage fluctuation, and improve the output performance of the system.

[0046] In this embodiment, in the step of S31, the collected data input to the prediction model is preprocessed by the same steps as in S12, which will not be elaborated here in detail.

[0047] In this embodiment, in the step of S33, for the parameter adjustment of the fuel cell: taking the internal temperature of the stack and the water production - ice melting balance as the control objectives, where the internal temperature of the stack directly affects the ice melting ability and the water production rate; By comparing the actually measured internal temperature of the stack with , calculate the temperature error: ; By comparing the actual water production rate with the ice melting rate , calculate the water production - ice melting balance error: ; Through proportional (P), integral (I), and derivative (D) control, adjust the control parameters to reduce the error, and the output of the PID controller is: ; Among them, , and are the proportional, integral, and derivative gains respectively; For the strategy of adjusting the control algorithm parameters: Proportional gain Adjustment: In the initial stage, if the temperature or the water production - ice melting balance deviates greatly from the target value, increase to improve the response speed and make the stack temperature or the water production - ice melting state quickly approach the target value; when approaching the target value, reduce To avoid oscillations caused by excessive adjustment; Integral gain Adjustment: When the temperature or the water production - ice melting balance error persists, appropriately increase the accumulated error signal to prompt the system to eliminate long - term deviations; when the error changes rapidly, appropriately decrease to avoid integral saturation; Derivative gain Adjustment: When the temperature or the water production - ice melting state changes sharply, increase to counteract the inertial effect and improve the system stability; when the system tends to be stable, decrease to avoid over - inhibition; Take the predicted current density curve and the flow distribution ratio as feed - forward signals and input them into the control system to guide the operating state of the stack to change in the target direction in advance. Then the final control parameters are: ; where is the output of the PID controller; Allocate the final control quantity to the current density regulation (such as controlling the speed of the air compressor 3, hydrogen flow rate, etc.) and the flow distribution ratio regulation mechanism (such as adjusting the opening of the three - way valve 5) to achieve precise control of the operating parameters of the stack; Adjustment based on real - time feedback data: According to the real - time collected feedback data of the stack voltage fluctuation, temperature change, and water production - ice melting state, periodically evaluate the current control effect and dynamically adjust the parameters of the PID controller: ; ; ; where , and are the adjustment steps of the proportional, integral, and derivative gains; , and are the adjustment terms for the water production - ice melting balance error.

[0048] Based on the inspiration of the ideal embodiments of the present invention, through the above description, for those skilled in the art, it is obvious that the present invention is not limited to the details of the above exemplary embodiments, and can be implemented in other specific forms without departing from the spirit or basic characteristics of the present invention. Therefore, from any point of view, the embodiments should be regarded as exemplary and non-limiting. The scope of the present invention is defined by the appended claims rather than the above description. Therefore, all changes falling within the meaning and scope of the equivalent elements of the claims are intended to be embraced within the present invention. Any reference signs in the claims should not be construed as limiting the claims involved.

[0049] In addition, it should be understood that although this specification is described according to embodiments, not every embodiment only contains an independent technical solution. This narrative way of the specification is only for clarity. Those skilled in the art should regard the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. An optimized method for preventing ice formation during cold start of a fuel cell, characterized in that, It includes the following steps: S1. Collect the ambient temperature and data on the icing condition inside the fuel cell stack under various operating conditions, and analyze the water production rate and temperature rise ice melting capacity under each operating condition; S2. Based on the water production rate and temperature rise ice melting capacity, construct a prediction model that can reflect the relationship between the ambient temperature, the icing condition inside the fuel cell stack, and the ideal current density; S3. Input the ambient temperature and data on the icing condition inside the fuel cell stack during cold start into the prediction model, output the current density curve, as well as the flow distribution ratio between the matched intercooler and the bypass pipeline, and adjust the fuel cell parameters in combination with the control algorithm; The device based on the optimization method includes: a fuel cell stack, a first electronic throttle valve, an intercooler, and an air compressor sequentially installed on the air inlet pipeline of the fuel cell stack, and a back pressure valve installed on the air outlet pipeline of the fuel cell stack; A three-way valve is installed between the first electronic throttle valve and the intercooler. One end of the three-way valve is connected to a bypass pipeline, and one end of the bypass pipeline is connected to the air inlet pipeline between the intercooler and the air compressor; A connecting pipeline is connected between the air outlet pipeline and the air inlet pipeline. One end of the connecting pipeline is located between the three-way valve and the first electronic throttle valve, and the other end is located on one side of the back pressure valve.

2. The optimized method for preventing fuel cell cold start icing according to claim 1, characterized in that: In the step of S1, it includes the following sub-steps: S11. Collect the ambient temperature and data on the icing condition inside the fuel cell stack under various operating conditions; S12. Preprocess the collected data, including removing duplicate, incomplete, or abnormal data items and performing normalization processing; S13. Analyze the ambient temperature and the icing condition inside the fuel cell stack, the corresponding water production rate, and the temperature rise ice melting capacity under each operating condition.

3. An optimized method for preventing ice formation during cold start of a fuel cell according to claim 1, characterized in that: In the step of S2, it includes the following sub-steps: S21. Analyze the correlation between the ambient temperature, the icing condition inside the fuel cell stack, and the current density; S22. Based on the Transformer architecture, use the ambient temperature and the icing condition inside the fuel cell stack as input features and the ideal current density as the output target to construct a prediction model that can predict the current density curve and the flow distribution ratio between the matched intercooler and the bypass pipeline; S23. Divide the data collected and processed in step S1 into a training set and a validation set, use the training set to train the prediction model, and use the validation set to evaluate the performance of the prediction model.

4. An optimized method for preventing fuel cell cold start icing according to claim 1, characterized in that: In the step of S3, it includes the following sub-steps: S31. During the cold start of the fuel cell, collect the ambient temperature and data on the icing condition inside the fuel cell stack in real time and input them into the prediction model; S32. Obtain the output current density curve and the flow distribution ratio between the matched intercooler and the bypass pipeline through the prediction model; S33. According to the current density curve and the flow distribution ratio, adjust the parameters of the fuel cell in combination with the control algorithm, and dynamically adjust the parameters of the control algorithm according to the real-time feedback data.

5. An optimization method for preventing fuel cell cold start icing according to claim 4, characterized in that: In the step of S33, for the parameter adjustment of the fuel cell: taking the internal temperature of the stack and the balance of water production and ice melting as the control objectives, wherein the internal temperature of the stack directly affects the ice melting ability and the water production rate; By comparing the actually measured internal temperature of the stack with , calculate the temperature error: ； By comparing the actual water production rate with the ice melting rate , calculate the water production-ice melting balance error: ； By proportional, integral, and derivative control, the control parameters are adjusted to reduce the error, and the output of the PID controller is as follows: ； Among them, , and are the proportional, integral, and derivative gains respectively; Regarding the strategy for adjusting the parameters of the control algorithm: Proportional gain Adjustment: In the initial stage, if the temperature or the water production - ice melting balance deviates significantly from the target value, increase to improve the response speed and make the stack temperature or the water production - ice melting state quickly approach the target value; when approaching the target value, decrease to avoid oscillations caused by over - adjustment; Integral gain Adjustment: When the temperature or the water production - ice melting balance error persists, appropriately increase the accumulated error signal to prompt the system to eliminate long - term deviations; when the error changes rapidly, appropriately decrease to avoid integral saturation; Differential gain Adjustment: When the temperature or the water production - ice melting state changes sharply, increase to counteract the inertial influence and improve the system stability; when the system tends to be stable, decrease to avoid over - suppression; The predicted current density curve and flow distribution ratio are input into the control system as feedforward signals to guide the operating state of the stack to change in the target direction in advance, and the final control parameters are as follows: ； Among them, is the output of the PID controller; Allocate the final control quantity to the current density adjustment and flow distribution ratio adjustment mechanisms to achieve precise control of the operating parameters of the stack; Adjustment based on real-time feedback data: According to the real-time collected feedback data of stack voltage fluctuation, temperature change, and water production - ice melting state, periodically evaluate the current control effect, and dynamically adjust the parameters of the PID controller: ； ； ； Among them, , and are the adjustment steps of the proportional, integral, and derivative gains; , and are the adjustment terms for the produced water - ice melting balance error.

6. An optimized method for preventing fuel cell cold start icing, characterized in that: One end of the air inlet pipeline is equipped with an air filter for ensuring the purity of the sucked-in gas; the air filter is located on one side of the air compressor.

7. An optimized method for preventing fuel cell cold start icing, characterized in that: A pressure sensor for detecting gas pressure is installed between the three-way valve and the first electronic throttle valve.

8. An optimization method for preventing fuel cell cold start icing according to claim 1, characterized in that: A second electronic throttle valve for adjusting the pressure value of the sucked-in gas is installed on the communication pipeline.

9. An optimized method for preventing fuel cell cold start icing, characterized in that: One end of the air outlet pipeline is equipped with a silencer for reducing the noise of the discharged gas.

Images