An optimization method to prevent fuel cell cold start icing

Abstract

The present invention discloses an optimization method for preventing freezing of a fuel cell during cold start, which relates to the technical field of fuel cells. The optimization method comprises the following steps: S1, collecting data on ambient temperature and freezing conditions inside a fuel cell stack, and analyzing the water production rate and temperature rise ice-melting capacity; S2, constructing a prediction model that can reflect the relationship between ambient temperature, freezing conditions inside the fuel cell stack, and ideal current density; S3, inputting the ambient temperature and freezing conditions inside the fuel cell stack into the prediction model to adjust the fuel cell parameters; indirectly controlling the air temperature through the current density curve predicted by the model, converting the water production-ice melting balance of the fuel cell stack into a flow distribution optimization problem, and combining with voltage fluctuation feedforward compensation to achieve coordinated control of temperature and electrochemical state, which 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.

Description

Technical Field

[0001] The present invention relates to the technical field of fuel cells, in particular to heat exchange, and specifically to an optimization method for preventing freezing of a fuel cell during cold start. Background Art

[0002] As a highly efficient and environmentally friendly energy conversion device, the PEMFC (hydrogen-embedded electromagnetic fuel cell) has demonstrated tremendous potential in recent years in areas such as new energy vehicles, distributed power generation systems, and backup power. The PEMFC operates on the electrochemical reaction of hydrogen and oxygen in the presence of a catalyst, directly generating electricity and water. Water, its sole byproduct, is not only environmentally friendly but also demonstrates the highly clean nature of the energy conversion process. Consequently, hydrogen energy, as a clean energy source, has attracted widespread attention and application.

[0003] PEMFCs face numerous technical challenges in practical applications, especially under extreme climate conditions. Cold start issues are particularly prominent in low-temperature environments. In subzero temperatures, frost easily accumulates inside the PEMFC stack and on external components (such as pipes and valves). Water vapor generated by electrochemical reactions, in particular, rapidly condenses and potentially freezes upon contact with cold components. This not only blocks gas flow channels, reducing the effective supply of reactant gases, but also increases internal resistance in the stack, leading to decreased cell performance and even failure to start.

[0004] Currently, the industry primarily employs two strategies to improve PEMFC startup capabilities in low-temperature environments: one is to preheat key components using a PTC heating system, and the other is to utilize the chemical reactions within the stack to generate heat for self-heating. However, the response time for cold starts in domestic systems is excessively long. During this process, water generated by the stack can freeze due to the low temperature, impacting startup time and performance. Summary of the Invention

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

[0006] To achieve the above-mentioned object, the technical solution adopted by the present invention is as follows: a device for preventing icing of a fuel cell during cold start, comprising: a fuel cell stack, a first electronic throttle valve, an intercooler, and an air compressor sequentially installed on the air inlet pipe of the fuel cell stack, and a back pressure valve installed on the air outlet pipe of the fuel cell stack;

[0007] 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;

[0008] 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.

[0009] In a preferred embodiment of the present invention, an air filter for ensuring the purity of the drawn-in gas is installed at one end of the air inlet pipe; the air filter is located on one side of the air compressor.

[0010] 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.

[0011] In a preferred embodiment of the present invention, a second electronic throttle valve for adjusting the pressure value of the drawn-in gas is installed on the communicating pipeline.

[0012] In a preferred embodiment of the present invention, a muffler for reducing gas exhaust noise is installed at one end of the air outlet pipe.

[0013] The present invention provides an optimization method for preventing icing of a fuel cell during cold start, comprising the following steps:

[0014] S1. Collect data on the ambient temperature and internal icing conditions of the fuel cell stack under various operating conditions, and analyze the water production rate and temperature rise ice melting capacity under each operating condition;

[0015] S2. Based on the water production rate and temperature rise ice melting capacity, a prediction model is constructed that reflects the relationship between ambient temperature, internal ice formation conditions of the fuel cell stack, and ideal current density.

[0016] S3. Input the ambient temperature of the fuel cell during cold start and the icing condition data inside the stack into the prediction model, output the current density curve, and the flow distribution ratio between the matching intercooler and the bypass line, and adjust the fuel cell parameters in combination with the control algorithm.

[0017] In a preferred embodiment of the present invention, the step S1 includes the following sub-steps:

[0018] S11. Collecting data on the ambient temperature and internal icing conditions of the fuel cell stack under various operating conditions;

[0019] S12. Preprocessing the collected data, including removing duplicate, incomplete, or abnormal data items and performing normalization processing;

[0020] S13. Analyze the ambient temperature and internal icing conditions of the fuel cell stack under each operating condition, and the corresponding water production rate and temperature rise ice melting capacity.

[0021] In a preferred embodiment of the present invention, the step S2 includes the following sub-steps:

[0022] S21. Analyze the correlation between ambient temperature, internal icing conditions of the fuel cell stack, and current density.

[0023] S22. Based on the Transformer architecture, using ambient temperature and internal stack icing conditions as input features and ideal current density as the output target, a prediction model was constructed that can predict the current density curve and the matching flow distribution ratio between the intercooler and bypass line.

[0024] 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.

[0025] In a preferred embodiment of the present invention, the step S3 includes the following sub-steps:

[0026] S31. During a cold start of the fuel cell, real-time data on ambient temperature and icing conditions inside the fuel cell stack are collected and input into a prediction model;

[0027] S32, obtaining an output current density curve and a matching flow distribution ratio between the intercooler and the bypass line through a prediction model;

[0028] S33. According to the current density curve and the flow distribution ratio, the parameters of the fuel cell are adjusted in combination with the control algorithm, and the parameters of the control algorithm are dynamically adjusted according to the real-time feedback data.

[0029] In a preferred embodiment of the present invention, in the step S33, the parameters of the fuel cell are adjusted: the internal temperature of the stack The control objectives are water production-ice melting balance, where the internal temperature of the stack directly affects the ice melting capacity and water production rate;

[0030] By comparing the actual measured internal temperature of the battery stack and , calculate the temperature error:

[0031] ;

[0032] By comparing the actual water production rate and ice melting rate , calculate the water production-ice melting balance error:

[0033] ;

[0034] Through proportional (P), integral (I) and differential (D) control, the control parameters are adjusted to reduce the error. The output of the PID controller for:

[0035] ;

[0036] in, 、 and are proportional, integral, and derivative gains respectively;

[0037] Strategy for adjusting the control algorithm parameters:

[0038] Proportional gain Adjustment: In the initial stage, if the temperature or water production-melt ice balance deviates greatly from the target value, increase To improve the response speed, the stack temperature or water production-melting state quickly approaches the target value; when approaching the target value, reduce To avoid shocks caused by over-adjustment;

[0039] Integral gain Adjustment: When the temperature or water production-melt ice balance error persists, increase the Accumulate error signals to prompt the system to eliminate long-term deviations; when the error changes rapidly, appropriately reduce To avoid integral saturation;

[0040] Differential gain Adjustment: When the temperature or water production-melting state changes rapidly, increase Offset the inertia effect and improve system stability; when the system tends to be flat, reduce Avoid over-inhibition;

[0041] The predicted current density curve and flow distribution ratio As the feedforward signal input control system, it guides the operation state of the battery stack to change in the target direction in advance, so the final control parameter for:

[0042] ;

[0043] in, is the output of the PID controller;

[0044] The final control amount Allocate to the current density adjustment and flow distribution ratio adjustment mechanisms to achieve precise control of the stack operating parameters;

[0045] Adjustment based on real-time feedback data: Based on the real-time collected data on stack voltage fluctuations, temperature changes, and water production and ice melting status feedback, the current control effect is periodically evaluated and the parameters of the PID controller are dynamically adjusted:

[0046] ;

[0047] ;

[0048] ;

[0049] in, 、 and is the adjustment step size of the proportional, integral and derivative gains; 、 and It is an adjustment item for the water production-melt ice balance error.

[0050] The present invention solves the defects existing in the background technology and has the following beneficial effects:

[0051] (1) The present invention provides a device for preventing fuel cell cold start from freezing. The device comprises a first electronic throttle valve, an intercooler and an air compressor on the air inlet pipe, and a back pressure valve on the air outlet pipe. When the fuel cell system is started in a sub-zero low temperature environment, the back pressure valve cooperates to increase the outlet pressure of the air compressor, that is, to increase the pressure ratio of the air compressor. In addition, with the cooperation of the three-way valve and the bypass pipe, the flow distribution of air between the intercooler and the bypass pipe can be reasonably regulated, and the temperature of the air entering the fuel cell stack can be effectively adjusted, thereby effectively solving the problem of direct freezing of 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.

[0052] (2) The present invention provides an optimization method for preventing freezing during cold start of a fuel cell. By combining a prediction model based on a Transformer architecture with an adjustment strategy during cold start of the fuel cell, the method effectively solves the problem of thermal-electric-fluid multi-field coupling imbalance during cold start of the fuel cell in an extremely low temperature environment, significantly improves the cold start success rate to more than 90%, and controls the voltage fluctuation within 5%. The air temperature is indirectly controlled through the current density curve predicted by the model, and the water production-ice melting balance of the stack is converted into a flow distribution optimization problem. Combined with voltage fluctuation feedforward compensation, the coordinated control of temperature and electrochemical state is achieved, which 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.

[0053] (3) In the present invention, the parameters of the fuel cell are adjusted by combining a control algorithm, and the water production-ice melting balance of the fuel cell stack is converted 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 pipe is adjusted in real time. The air temperature is indirectly controlled in a way that can accurately control the thermal field and reaction environment inside the fuel cell stack, so that the water production rate matches the temperature rise and ice melting capacity, avoiding ice accumulation caused by excessive water production rate, and further preventing ice from clogging the catalyst layer and affecting the diffusion of reactants, thereby ensuring the smooth progress of the reaction, reducing voltage fluctuations, and improving the output performance of the system.

[0054] (4) In the present invention, by installing an air filter at one end of the air inlet pipe, when the air compressor draws in external air, the air filter can filter out particulate matter, dust, grease or other pollutants in the drawn-in gas, thereby ensuring the purity of the drawn-in gas and preventing impurities or pollutants in the drawn-in gas from adversely affecting the operating performance and life of the air compressor, thereby reducing pollution and damage to the fuel cell. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments described in the present invention. Those skilled in the art can also derive other drawings based on these drawings without inventive efforts.

[0056] Figure 1 1 is a schematic diagram of the overall structure of a device for preventing icing during cold start of a fuel cell according to a preferred embodiment of the present invention;

[0057] Figure 2 This is a flow chart of an optimization method for preventing icing during cold start of a fuel cell according to a preferred embodiment of the present invention;

[0058] In the figure: 1. First electronic throttle valve; 2. Intercooler; 3. Air compressor; 4. Back pressure valve; 5. Three-way valve; 6. Bypass line; 7. Connecting line; 8. Air filter; 9. Pressure sensor; 10. Second electronic throttle valve; 11. Muffler. DETAILED DESCRIPTION

[0059] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0060] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the scope of protection of the present invention is not limited to the specific embodiments disclosed below.

[0061] Application Overview:

[0062] In extremely low-temperature environments, the cold start process of fuel cells faces multiple challenges, including drastic temperature fluctuations and an imbalance between water production rate and ice melting capacity. These factors can easily lead to voltage fluctuations, seriously affecting system stability and output performance. Traditional control strategies, such as constant current or constant voltage starting, often fail to accurately match heat production and ice formation rates, resulting in startup failures or large voltage fluctuations.

[0063] In practical applications, it has been found that when a fuel cell is cold-started, if the water production rate exceeds the temperature rise and ice melting capacity, ice accumulation will occur. This not only hinders the chemical reaction but also increases the system impedance, reducing the voltage and current output. Ice blockage, particularly in the catalyst layer, can severely affect the diffusion of reactants, further deteriorating system performance. Therefore, achieving a dynamic balance between water production and ice melting during the cold-start process has become a key issue that needs to be addressed.

[0064] In view of the above technical problems, the concept of the present invention is to propose an optimization method to prevent the freezing of fuel cells during cold start. Through intelligent prediction models and dynamic control strategies, the problem of multi-field coupling imbalance during the cold start of fuel cells in extremely low temperature environments is solved, which significantly improves the success rate of cold start and the stability of the system, and provides strong technical support for the widespread application of fuel cells in low temperature environments.

[0065] like Figure 1 As shown, a device for preventing fuel cell cold start from icing includes: a fuel cell stack, a first electronic throttle valve 1, an intercooler 2, and an air compressor 3 installed in sequence 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, one end of the bypass pipeline 6 is connected to the air inlet pipeline between the intercooler 2 and the air compressor 3; a connecting pipeline 7 is connected between the air outlet pipeline and the air inlet pipeline, one end of the connecting 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.

[0066] It should be noted that when the fuel cell system is started in a sub-zero low-temperature environment, the air compressor 3 draws low-temperature air through one end of the air inlet pipe, fully opens the first electronic throttle valve 1, and regulates the back pressure valve 4 on the air outlet pipe 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 to above zero degrees Celsius. By adjusting the opening of the three-way valve 5 and reasonably regulating the flow distribution of air between the intercooler 2 and the bypass pipe 6, the temperature of the air entering the fuel cell stack can be effectively adjusted, thereby effectively solving the problem of direct freezing of water generated in the fuel cell stack during cold start, and avoiding the startup time and performance of the fuel cell stack.

[0067] In some embodiments, an air filter 8 is installed at one end of the air inlet pipe to ensure the purity of the drawn-in gas.

[0068] 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 pipe, when the air compressor 3 draws in external air, the air filter 8 can filter out particulate matter, dust, grease or other pollutants in the drawn-in gas, thereby ensuring the purity of the drawn-in gas and preventing impurities or pollutants in the drawn-in gas from adversely affecting the operating performance and life of the air compressor 3, thereby reducing pollution and damage to the fuel cell.

[0069] 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; through the setting 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 air entering, thereby facilitating the regulation operation of the back pressure valve 4 according to the feedback value.

[0070] In some embodiments, a second electronic throttle valve 10 for adjusting the pressure value of the drawn-in gas is installed on the connecting pipe 7; through the setting of the second electronic throttle valve 10, since the second electronic throttle valve 10 is installed on the connecting pipe 7, the gas pressure of the air entering the fuel cell stack can be easily adjusted by regulating the second electronic throttle valve 10.

[0071] In some embodiments, a muffler 11 for reducing gas exhaust noise is installed at one end of the air outlet pipe; through the setting of the muffler 11, the noise of the gas discharged from the air outlet pipe of the fuel cell stack can be reduced.

[0072] When the present invention is used, when the fuel cell system is started in a low-temperature environment below zero, low-temperature air enters the air compressor 3 through the air filter 8, the first electronic throttle valve 1 is fully opened, and the back pressure valve 4 on the air outlet pipeline is regulated by feedback of the pressure value detected by the pressure sensor 9, so that the outlet pressure of the air compressor 3 is increased, that is, the pressure ratio of the air compressor 3 is increased. At this time, the air temperature at the outlet of the air compressor 3 will rise to 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 line 6 can be reasonably regulated, and the temperature of the air entering the fuel cell stack can be effectively adjusted, thereby effectively solving the problem of direct freezing of water generated in the fuel cell stack during cold start, and avoiding the startup time and performance of the fuel cell stack.

[0073] like Figure 2 As shown, the present invention provides an optimization method for preventing icing of a fuel cell during cold start, comprising the following steps:

[0074] S1. Collect data on the ambient temperature and internal icing conditions of the fuel cell stack under various operating conditions, and analyze the water production rate and temperature rise ice melting capacity under each operating condition;

[0075] S2. Based on the water production rate and temperature rise ice melting capacity, a prediction model is constructed that reflects the relationship between ambient temperature, internal ice formation conditions of the fuel cell stack, and ideal current density.

[0076] S3. Input the ambient temperature of the fuel cell during cold start and the icing condition data inside the stack into the prediction model, output the current density curve, and the matching flow distribution ratio between the intercooler 2 and the bypass line 6, and adjust the fuel cell parameters in combination with the control algorithm.

[0077] It should be noted that by using a Transformer-based prediction model combined with an adjustment strategy during the cold start of the fuel cell, the problem of thermal-electrical-fluid multi-field coupling imbalance during the cold start of the fuel cell in an extremely low temperature environment is effectively solved, significantly improving the cold start success rate to more than 90%. At the same time, the voltage fluctuation is controlled within 5%, and the air temperature is indirectly controlled through the current density curve predicted by the model. The water production-ice melting balance of the fuel cell stack is converted 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 the cold start process, 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.

[0078] In some specific embodiments, the step S1 includes the following sub-steps:

[0079] S11. Collecting data on the ambient temperature and internal icing conditions of the fuel cell stack under various operating conditions;

[0080] S12, preprocessing the collected data;

[0081] S13. Analyze the ambient temperature and internal icing conditions of the fuel cell stack under each operating condition, and the corresponding water production rate and temperature rise ice melting capacity.

[0082] In this embodiment, in step S11, the various operating 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 the external surface of the fuel cell stack, so as to ensure that the ambient temperature information can be fully and accurately obtained; the icing condition inside the fuel cell stack is collected by using optical fiber sensors, ultrasonic sensors or electrical impedance tomography technology to monitor the icing conditions of various key components inside the fuel cell stack, such as the catalyst layer, gas diffusion layer and proton exchange membrane, and the original signal output by the sensor is filtered, amplified and analog-to-digital converted to convert it into a digital signal that is easy to analyze and process.

[0083] In this embodiment, in step S12, preprocessing includes removing duplicate, incomplete or abnormal data items and performing normalization processing; removing duplicate data from the collected operating data, identifying and processing abnormal values: abnormal values ​​may be caused by data entry errors or special patient conditions, and need to be processed according to actual conditions, such as replacing them with mean, median or deletion; normalization processing specifically includes: scaling the data so that it falls into a small specific interval, such as [0,1] or [-1,1], which helps to eliminate the impact of dimension on the results. The calculation formula is: ;in, It is the original data; is the normalized data.

[0084] In this embodiment, in step S13, the water production rate is: according to the current density of the fuel cell and the effective area of ​​the stack , calculate the total current , the number of moles of water produced The relationship with current is:

[0085] ;

[0086] in, It’s time; is the number of electrons transferred in the reaction (for PEMFCs, ); is the Faraday constant (about 96485C / mol);

[0087] The water production rate will be affected by the actual working conditions (such as the supply of reactants, temperature, pressure, etc.), and the actual water production rate can be corrected. :

[0088] ;

[0089] in, is the molar mass of water (18×10 -3 kg / mol); It is the water production efficiency, which reflects the ratio of the actual water production rate to the theoretical water production rate;

[0090] For ice melting capacity: Calculate the heat generation power of the fuel cell :

[0091] ;

[0092] in, is the heat generation power of the reaction, according to the reaction heat And water production rate calculation: ; is the ohmic power loss, which is determined by the current and the internal resistance of the battery stack. calculate: ;

[0093] The ice melting capacity is evaluated based on the temperature distribution inside the stack and the physical properties of ice. When the temperature on the surface or inside the stack rises above the melting point of ice, the ice begins to melt and the melting rate is for:

[0094] ;

[0095] in, is the latent heat of melting of ice (about 3.34×10 5 J / kg); is the heat used to melt the ice, calculated by integrating the heat transferred to the ice surface by the stack over a period of time: , It 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.

[0096] In some specific embodiments, the step S2 includes the following sub-steps:

[0097] S21. Analyze the correlation between ambient temperature, internal icing conditions of the fuel cell stack, and current density.

[0098] S22. Based on the Transformer architecture, using ambient temperature and internal stack icing conditions as input features and ideal current density as the output target, a prediction model is constructed that can predict the current density curve and the matching flow distribution ratio between intercooler 2 and bypass line 6.

[0099] 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.

[0100] In this embodiment, in step S21, the ambient temperature and the icing condition data inside the stack are used as influencing factors for establishing the prediction model. and current density Calculate the Pearson correlation coefficient :

[0101] ;

[0102] in, It is The ambient temperature or icing condition data value of each sample; is with The corresponding The current density of each sample; is the mean value of the ambient temperature or the icing condition data inside the stack, that is, ; is the mean current density, that is ; is the sample size, that is, the total number of observations;

[0103] When the Pearson correlation coefficient , indicating the influencing factors and current density Perfect positive correlation means that an increase in one variable is always accompanied by an increase in the other variable; , indicating the influencing factors and current density Perfect negative correlation, where an increase in one variable is always accompanied by a decrease in the other; , indicating the influencing factors and current density There is no linear correlation between them; The closer the absolute value is to 1, the more influencing factors and current density The closer the linear relationship is to 0, the weaker the linear relationship is.

[0104] In this embodiment, in step S22, for the input layer: let the time series of the ambient temperature be ,in is the length of the time series; the relevant characteristics of the icing condition inside the stack (ice thickness, ice area) are represented as a feature vector ;

[0105] For the encoder: set to multiple layers (6 or 8 layers), using a multi-head self-attention mechanism, the dimension of each head is ,in is the number of heads; for the input sequence , query is obtained by linear projection ,key Sum Tensor:

[0106] , , ;

[0107] in, ;

[0108] Calculate the self-attention output:

[0109] ;

[0110] in, is the dimension of the key;

[0111] Each encoder layer contains a feed-forward neural network:

[0112] ;

[0113] in, is the output of the multi-head attention layer; and is the weight matrix; and is the bias vector; is the hidden layer dimension of the feedforward neural network; the input dimension is , the middle dimension is a larger value (such as );

[0114] Add residual connections after each sub-layer (self-attention and feed-forward network) and apply layer normalization to stabilize the training process;

[0115] For the decoder: The number of layers is consistent with that of the encoder, and the multi-head self-attention mechanism is also used. In addition to self-attention, each layer of the decoder also includes a cross-attention mechanism to focus on the encoder output and the decoder's own input. The feedforward neural network structure is the same as the feedforward neural network in the encoder, and residual connections and layer normalization are also applied after each sublayer.

[0116] For the output layer: the output time series represents the ideal current density at each time step, which is ; Predict the flow distribution ratio between intercooler 2 and bypass line 6, expressed as a vector ,in, is the flow distribution ratio at each time step.

[0117] In this embodiment, in step S23, during the prediction model training, the error between the predicted value and the true value is quantified:

[0118] Current density predicts losses:

[0119] ;

[0120] in, is the length of the time series; is the time step; The model is in time Predicted current density value; Is the real data in time The current density value;

[0121] Traffic distribution ratio prediction loss:

[0122] ;

[0123] in, It's in time The actual traffic distribution ratio; The model is in time Predicted traffic distribution ratio;

[0124] The weight parameters of the model are continuously adjusted through the back-propagation algorithm, and the gradient of the loss function for each parameter is calculated: ;in, is the loss function; is the model output; is an intermediate variable; are model parameters;

[0125] Use the optimizer to update the model parameters: ;in, It is The parameters of the iteration; is the learning rate; is the first moment estimate; is the second-order moment estimate; is a small constant used to prevent the denominator from being zero;

[0126] After each training cycle, the loss on the validation set is calculated. If the loss value on the validation set is continuous If the number of training cycles does not decrease, stop training: .

[0127] In some specific embodiments, the step of S3 includes the following sub-steps:

[0128] S31. During a cold start of the fuel cell, real-time data on ambient temperature and icing conditions inside the fuel cell stack are collected and input into a prediction model;

[0129] S32, obtaining an output current density curve and a matching flow distribution ratio between the intercooler 2 and the bypass line 6 through the prediction model;

[0130] S33. According to the current density curve and the flow distribution ratio, the parameters of the fuel cell are adjusted in combination with the control algorithm, and the parameters of the control algorithm are dynamically adjusted according to the real-time feedback data.

[0131] 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 fuel cell stack is converted 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 pipe 6 is adjusted in real time. The indirect control of the air temperature can accurately control the thermal field and reaction environment inside the fuel cell stack, so that the water production rate matches the temperature rise and ice melting capacity, avoiding ice accumulation caused by excessive water production rate, and thus preventing ice from clogging the catalyst layer and affecting the diffusion of reactants, ensuring the smooth progress of the reaction, reducing voltage fluctuations, and improving the output performance of the system.

[0132] In this embodiment, in step S31, the collected data input into the prediction model is preprocessed using the same steps as S12, which will not be described in detail here.

[0133] In this embodiment, in step S33, the fuel cell parameters are adjusted: the internal temperature of the stack is The control objectives are water production-ice melting balance, where the internal temperature of the stack directly affects the ice melting capacity and water production rate;

[0134] By comparing the actual measured internal temperature of the battery stack and , calculate the temperature error:

[0135] ;

[0136] By comparing the actual water production rate and ice melting rate , calculate the water production-ice melting balance error:

[0137] ;

[0138] Through proportional (P), integral (I) and differential (D) control, the control parameters are adjusted to reduce the error. The output of the PID controller for:

[0139] ;

[0140] in, 、 and are proportional, integral, and derivative gains respectively;

[0141] Strategy for adjusting the control algorithm parameters:

[0142] Proportional gain Adjustment: In the initial stage, if the temperature or water production-melt ice balance deviates greatly from the target value, increase To improve the response speed, the stack temperature or water production-melting state quickly approaches the target value; when approaching the target value, reduce To avoid shocks caused by over-adjustment;

[0143] Integral gain Adjustment: When the temperature or water production-melt ice balance error persists, increase the Accumulate error signals to prompt the system to eliminate long-term deviations; when the error changes rapidly, appropriately reduce To avoid integral saturation;

[0144] Differential gain Adjustment: When the temperature or water production-melting state changes rapidly, increase Offset the inertia effect and improve system stability; when the system tends to be flat, reduce Avoid over-inhibition;

[0145] The predicted current density curve and flow distribution ratio As the feedforward signal input control system, it guides the operation state of the battery stack to change in the target direction in advance, so the final control parameter for:

[0146] ;

[0147] in, is the output of the PID controller;

[0148] The final control amount Allocate to current density regulation (such as controlling the speed of air compressor 3, hydrogen flow, etc.) and flow distribution ratio adjustment mechanism (such as adjusting the opening of three-way valve 5) to achieve precise control of fuel cell stack operating parameters;

[0149] Adjustment based on real-time feedback data: Based on the real-time collected data on stack voltage fluctuations, temperature changes, and water production and ice melting status feedback, the current control effect is periodically evaluated and the parameters of the PID controller are dynamically adjusted:

[0150] ;

[0151] ;

[0152] ;

[0153] in, 、 and is the adjustment step size of the proportional, integral and derivative gains; 、 and It is an adjustment item for the water production-melt ice balance error.

[0154] The above description is based on the ideal embodiment of the present invention. It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention. Therefore, the embodiments should be considered as illustrative and non-restrictive in all respects. The scope of the present invention is defined by the appended claims, not the foregoing description, and it is intended that all variations within the meaning and range of equivalents of the claims be encompassed within the present invention. Any reference signs in the claims should not be construed as limiting the claim to which they relate.

[0155] In addition, it should be understood that although this specification is described in terms of implementation methods, not every implementation method contains only one independent technical solution. This narrative method of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment can also be appropriately combined to form other implementation methods that can be understood by those skilled in the art.

Claims

1. An optimization method for preventing cold start icing of a fuel cell, characterized in that: The following steps are involved: S1. Collect data on the ambient temperature and internal icing conditions of 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, a prediction model is constructed that reflects the relationship between ambient temperature, internal ice formation conditions of the fuel cell stack, and ideal current density. S3. Inputting the ambient temperature during cold start of the fuel cell and the icing condition data inside the fuel cell stack into the prediction model, outputting the current density curve and the matching flow distribution ratio between the intercooler and the bypass line, and combining it with the control algorithm to adjust the fuel cell parameters; The device based on the optimization method includes: a fuel cell stack, a first electronic throttle valve, an intercooler, and an air compressor installed in sequence on an air inlet pipe of the fuel cell stack, and a back pressure valve installed on an air outlet pipe 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 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; The step S3 includes the following sub-steps: S31. During a cold start of the fuel cell, real-time data on ambient temperature and icing conditions inside the fuel cell stack are collected and input into a prediction model; S32, obtaining an output current density curve and a matching flow distribution ratio between the intercooler and the bypass line through a prediction model; S33, adjusting the parameters of the fuel cell in combination with the control algorithm according to the current density curve and the flow distribution ratio, and dynamically adjusting the parameters of the control algorithm according to the real-time feedback data; In the step S33, the fuel cell parameters are adjusted by adjusting the internal temperature of the stack. The control objectives are water production-ice melting balance, where the internal temperature of the stack directly affects the ice melting capacity and water production rate; By comparing the actual measured internal temperature of the battery stack and , calculate the temperature error: ； By comparing the actual water production rate and ice melting rate , calculate the water production-ice melting balance error: ； Through proportional, integral and differential control, the control parameters are adjusted to reduce the error. The output of the PID controller for: ； in, ,and are proportional, integral, and derivative gains respectively; Strategy for adjusting the control algorithm parameters: Proportional gain Adjustment: In the initial stage, if the temperature or water production-melt ice balance deviates greatly from the target value, increase the response speed to make the stack temperature or water production-melt ice state quickly approach the target value; when approaching the target value, reduce To avoid shocks caused by over-adjustment; Integral gain Adjustment: When the temperature or water production-melt ice balance error persists, increase the Accumulate error signals 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 water production-melting state changes rapidly, increase Offset the inertia effect and improve system stability; when the system tends to be flat, reduce Avoid over-inhibition; The predicted current density curve and flow distribution ratio As the feedforward signal input control system, it guides the operation state of the battery stack to change in the target direction in advance, so the final control parameter for: ； in, is the output of the PID controller; The final control amount Allocate to the current density adjustment and flow distribution ratio adjustment mechanisms to achieve precise control of the stack operating parameters; Adjustment based on real-time feedback data: Based on the real-time collected data on stack voltage fluctuations, temperature changes, and water production and ice melting status feedback, the current control effect is periodically evaluated and the parameters of the PID controller are dynamically adjusted: ； ； ； in, 、 and is the adjustment step size of the proportional, integral and derivative gains; 、 and It is an adjustment item for the water production-ice melting balance error.

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

3. The optimization method for preventing cold start icing of a fuel cell according to claim 1, characterized in that: The step S2 includes the following sub-steps: S21. Analyze the correlation between ambient temperature, internal icing conditions of the fuel cell stack, and current density. S22. Based on the Transformer architecture, using ambient temperature and internal stack icing conditions as input features and ideal current density as the output target, a prediction model was constructed that can predict the current density curve and the matching flow distribution ratio between the intercooler and bypass line. 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. The optimization method for preventing cold start icing of a fuel cell according to claim 1, characterized in that: An air filter for ensuring the purity of the drawn-in gas is installed at one end of the air inlet pipeline; the air filter is located on one side of the air compressor.

5. The optimization method for preventing cold start icing of a fuel cell according to claim 1, characterized in that: A pressure sensor for detecting gas pressure is installed between the three-way valve and the first electronic throttle valve.

6. The optimization method for preventing cold start icing of a fuel cell according to claim 1, characterized in that: The communicating pipeline is provided with a second electronic throttle valve for adjusting the pressure value of the drawn-in gas.

7. The optimization method for preventing cold start icing of a fuel cell according to claim 1, characterized in that: A muffler is installed at one end of the air outlet pipeline to reduce the noise of gas discharge.

Images