Method and device for estimating water content of fuel cell system

Abstract

The invention provides a water content estimation method and device for a fuel cell system, and relates to the technical field of fuel cells, and the method comprises the steps: obtaining a first parameter and a second parameter of the fuel cell system; on the basis of the first parameter, predicting the particle state of the state transition model at the current moment to obtain the prediction state of each particle; calculating a theoretical observation value corresponding to the prediction state of each particle by using the observation model based on the second parameter, and updating the weight of each particle based on the difference between the theoretical observation value and the actual measurement value; and resampling each particle based on the updated particle weight to generate a new particle set, and calculating a weighted average value of each state variable of the state transition model based on the new particle set to obtain an estimated value of the water content index at the current moment. The method and the device for estimating the water content of the fuel cell system provided by the invention are used for accurately estimating the water content of a proton exchange membrane on line in the actual operation process of the fuel cell system.

Description

Technical Field

[0001] This application relates to the field of fuel cell technology, and in particular to a method and apparatus for estimating the water content of a fuel cell system. Background Technology

[0002] Proton exchange membrane fuel cells (PEMFCs) directly convert hydrogen and oxygen into electrical energy through an electrochemical reaction. They typically use a perfluorosulfonic acid proton exchange membrane (such as a Nafion membrane) to block electrons and gas molecules, thus isolating protons from the reactant gases. Proton transfer during the reaction process depends on a transition mechanism (hydrogen ions migrate through the hydrogen bond network of water molecules) and an on-board mechanism (hydrogen ions move by combining with sulfonic acid groups). It is necessary to maintain a suitable water content inside the membrane.

[0003] When the water content inside the membrane is insufficient, the sulfonic acid groups cannot be fully hydrated, the proton conductivity decreases, leading to increased ohmic polarization and damage to the membrane structure. Excessive water content inside the membrane will cause flooding, block the pores of the gas diffusion layer, hinder oxygen transport, cause concentration gradient polarization, and accelerate the corrosion of the electrode material.

[0004] However, the water content estimation methods in related technologies are not only inaccurate, but also difficult to provide water content information in a timely manner during the actual operation of fuel cells. Summary of the Invention

[0005] The purpose of this application is to provide a method and apparatus for estimating the water content of a fuel cell system, which is used to accurately estimate the water content of the proton exchange membrane during the actual operation of the fuel cell system.

[0006] This application provides a method for estimating the water content of a fuel cell system, including: The system acquires the current operating state parameters of the fuel cell system. These parameters include a first parameter and a second parameter. The first parameter characterizes the operating control variables of the fuel cell system, while the second parameter characterizes external parameters related to water content. Based on the first parameter, the state of each particle in the state transition model is predicted to obtain a predicted state for each particle. The state transition model describes the hydrothermal state of the fuel cell system. Based on the second parameter, the theoretical observation value corresponding to the predicted state of each particle is calculated using an observation model. The weight of each particle is updated based on the difference between the theoretical observation value and the actual measurement value. Based on the updated particle weights, each particle is resampled to generate a new particle set. Based on the new particle set, the weighted average of each state variable in the state transition model is calculated to obtain an estimated value of the water content index at the current moment.

[0007] Optionally, the initialization process of the state transition model includes the following steps: based on a pre-defined prior distribution of state variables; generating an initial particle set of the state transition model, and assigning an initial weight to each initial particle in the initial particle set; wherein the prior distribution of state variables is set based on the startup characteristics of the fuel cell system.

[0008] Optionally, process noise is added during the prediction of the particle state at the current moment in the state transition model; the process noise includes: process noise acting on the hydration degree of the proton exchange membrane and process noise acting on the liquid water saturation of the cathode gas diffusion layer.

[0009] Optionally, the step of resampling each particle based on the updated particle weights to generate a new particle set includes: calculating the cumulative weight of each particle; generating a random number that is the same as the number of particles and is uniformly distributed in the interval [0,1]; and extracting particles with replacement from the original particle set according to the random number and the cumulative weights to form a new particle set; wherein the weights of all particles in the new particle set are set to be uniformly distributed.

[0010] Optionally, the state variables of the state transition model include at least one of the following: proton exchange membrane hydration degree, cathode gas diffusion layer liquid water saturation degree, battery average temperature, and mole fraction of water vapor in the cathode flow channel; the water content index includes: proton exchange membrane hydration degree and cathode gas diffusion layer liquid water saturation degree; the first parameter includes at least one of the following: cathode inlet air flow rate, anode inlet hydrogen flow rate, coolant inlet flow rate, coolant inlet temperature, and load current density.

[0011] This application also provides a water content estimation device for a fuel cell system, comprising: The system includes a parameter acquisition module for acquiring the current operating state parameters of the fuel cell system. These parameters include a first parameter and a second parameter. The first parameter characterizes the operating control variables of the fuel cell system, and the second parameter characterizes external parameters related to water content. A state prediction module predicts the particle state of the state transition model based on the first parameter, obtaining the predicted state of each particle. The state transition model describes the hydrothermal state of the fuel cell system. A weight update module calculates the theoretical observation value corresponding to the predicted state of each particle using the observation model based on the second parameter, and updates the weight of each particle based on the difference between the theoretical observation value and the actual measurement value. A water content estimation module resamples each particle based on the updated particle weights to generate a new particle set, and calculates the weighted average of the state variables of the state transition model based on the new particle set, obtaining an estimated value of the water content index at the current moment.

[0012] Optionally, the device further includes: an initialization module; the initialization module is configured to generate an initial particle set of the state transition model based on a pre-set prior distribution of state variables; the initialization module is further configured to assign an initial weight to each initial particle in the initial particle set; wherein the prior distribution of state variables is set based on the start-up characteristics of the fuel cell system.

[0013] Optionally, the state prediction module is specifically used to add process noise during the prediction of the particle state at the current moment of the state transition model; wherein, the process noise includes: process noise acting on the hydration degree of the proton exchange membrane and process noise acting on the liquid water saturation of the cathode gas diffusion layer.

[0014] Optionally, the water content estimation module is specifically used to calculate the cumulative weight of each particle; the water content estimation module is also specifically used to generate random numbers that are the same as the number of particles and are uniformly distributed in the interval [0,1], and to extract particles with replacement from the original particle set according to the random numbers and the cumulative weights to form a new particle set; wherein, the weights of all particles in the new particle set are set to be uniformly distributed.

[0015] Optionally, the state variables of the state transition model include at least one of the following: proton exchange membrane hydration degree, cathode gas diffusion layer liquid water saturation degree, battery average temperature, and mole fraction of water vapor in the cathode flow channel; the water content index includes: proton exchange membrane hydration degree and cathode gas diffusion layer liquid water saturation degree; the first parameter includes at least one of the following: cathode inlet air flow rate, anode inlet hydrogen flow rate, coolant inlet flow rate, coolant inlet temperature, and load current density.

[0016] Optionally, the prior probability distribution of each state variable in the state transition model is set as follows: the initial distribution of the proton exchange membrane hydration degree is as follows: The initial distribution of liquid water saturation in the cathode gas diffusion layer is as follows: The average temperature of the battery The initial distribution is The initial distribution of water vapor mole fraction within the cathode channel is as follows: .

[0017] This application also provides a computer program product, including a computer program / instructions that, when executed by a processor, implement the steps of the water content estimation method for a fuel cell system as described above.

[0018] This application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of any of the above-described fuel cell system water content estimation methods.

[0019] This application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the water content estimation method for a fuel cell system as described above.

[0020] The method and apparatus for estimating the water content of a fuel cell system provided in this application first obtain the current operating state parameters of the fuel cell system. These operating state parameters include a first parameter and a second parameter. The first parameter characterizes the operating control variables of the fuel cell system, and the second parameter characterizes external parameters related to water content. Then, based on the first parameter, the particle states of the state transition model at the current moment are predicted to obtain the predicted state of each particle. The state transition model describes the hydrothermal state of the fuel cell system. Based on the second parameter, the theoretical observation value corresponding to the predicted state of each particle is calculated using an observation model, and the weight of each particle is updated based on the difference between the theoretical observation value and the actual measurement value. Finally, based on the updated particle weights, each particle is resampled to generate a new particle set, and based on the new particle set, the weighted average of each state variable of the state transition model is calculated to obtain the estimated value of the water content index at the current moment. Thus, the water content of the proton exchange membrane can be accurately estimated online during the actual operation of the fuel cell system. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 This is one of the flowcharts illustrating the water content estimation method for the fuel cell system provided in this application; Figure 2 This is the second flowchart illustrating the water content estimation method for the fuel cell system provided in this application; Figure 3 This is a schematic diagram of the structure of the water content estimation device for the fuel cell system provided in this application; Figure 4 This is a schematic diagram of the structure of the electronic device provided in this application. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0024] The terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such use of data can be interchanged where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class and the number of objects is not limited; for example, a first object can be one or more. Furthermore, "and / or" in the specification and claims indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship. All actions involving the acquisition of signal information or data in this application are performed in accordance with the relevant data protection laws and policies of the country where the application is located and with authorization from the owner of the relevant device.

[0025] The aforementioned technical solutions in related technologies mainly suffer from the following shortcomings: Offline estimation methods, such as those based on electrochemical impedance spectroscopy, relaxation time distribution techniques, and electron microscopy, typically rely on relatively precise and expensive measurement equipment. The large size of this equipment increases measurement costs and limits practical applications, and it cannot meet real-time monitoring requirements, making it difficult to provide timely water content information during actual fuel cell operation. Electrochemical signal-based methods require modifications to the fuel cell stack structure and adaptation to online dynamic operating conditions, offering strong engineering applicability, but they are susceptible to polarization phenomena and membrane electrode aging, making it difficult to pinpoint local water content. Data-driven methods rely heavily on large amounts of data for prediction; the size of the data determines its prediction accuracy, and most data requires invasive methods for acquisition, increasing measurement complexity and cost.

[0026] To address the aforementioned technical problems in related technologies, this application provides a method for estimating the water content of a fuel cell system, which mainly includes the following: Particle filtering is a filtering method based on probability theory and stochastic process theory, particularly suitable for solving state estimation problems of nonlinear and non-Gaussian systems. In the estimation of water content in a fuel cell system, particle filtering approximates the probability distribution of the system state by generating and updating a set of random particles, thus achieving an effective estimate of water content. Its core principle is based on Monte Carlo simulation; the basic idea is to use a set of weighted particles to represent the posterior probability distribution of the system state. Assume the system's state space is... The observation space is At any moment The system state is The observed value is Particle filtering approximates the probability distribution of the true state through a series of steps.

[0027] The method for estimating the water content of a fuel cell system provided in this application embodiment, such as Figure 1 As shown, the first step is the initialization process, which randomly generates states in the state space based on prior knowledge. One particle: And assign initial weights to each particle: These particles and weights constitute the initial particle set, representing an estimate of the initial state of the system.

[0028] Next, we move into the prediction phase, based on the system's state transition model: The particles from the previous moment Propagation to the current moment yields the predicted particle. State transition models describe how the system state changes over time. For example, in a fuel cell system, it includes the influence of factors such as the physicochemical reactions inside the cell, gas flow, and heat transfer on the water content state. By applying the state transition model to each particle and adding a certain amount of random noise to simulate the uncertainty of the system, a predicted particle set is obtained.

[0029] Then comes the update step, based on the observation model. and the actual observed value at the current time Calculate the weight of each predicted particle. The observation model establishes the relationship between the system state and observed values. In a fuel cell, these observed values ​​may come from measurements by sensors such as voltage, current, and temperature. The weights of particles are adjusted by comparing the predicted observed values ​​with the actual observed values. The smaller the difference, the closer the state represented by the particle is to the true state, and the larger its weight; conversely, the larger the difference, the smaller the weight. Through this step, the weight distribution of particles can more accurately reflect the posterior probability distribution of the system state.

[0030] Finally, there's the resampling step. During the update process, the weights of some particles may become very small, while a few particles have very large weights. This leads to a reduction in the effective sample size of the particle set, a problem known as "particle degradation." To address this, the resampling process samples particles with replacement based on their weights. Particles with larger weights are selected more frequently, resulting in a new particle set. The new set contains the same number of particles as the original, but their distribution is more concentrated in high-probability regions, effectively improving the representativeness of the particle set to the system state and enhancing the accuracy of the estimation. By continuously repeating the prediction, update, and resampling process, particle filtering can gradually approximate the true state of the system, achieving accurate online estimation of the water content in a fuel cell system.

[0031] In this embodiment of the application, a noise-resistant online estimation system for water content in a fuel cell system is also provided. The system includes: a fuel cell body, a sensor unit, a data acquisition unit, a particle filter processing unit, and an estimation result output unit, and each unit is connected in communication with the others in sequence.

[0032] Fuel cell body: It adopts a proton exchange membrane fuel cell (PEMFC), and the core components include proton exchange membrane, cathode gas diffusion layer (GDL), anode structure, flow channel and cooling system, providing the test object for water content estimation.

[0033] Sensor unit: includes input parameter sensor and observation parameter sensor, wherein the input parameter sensor is used to collect fuel cell operation control variables, and the observation parameter sensor is used to collect external characterization parameters that are strongly correlated with water content.

[0034] Input parameter sensors: Cathode inlet air flow sensor (measuring range 0-5 kg / s, accuracy ±0.01 kg / s), Anode inlet hydrogen flow sensor (measuring range 0-0.5 kg / s, accuracy ±0.001 kg / s), Coolant inlet flow sensor (measuring range 0-2 kg / s, accuracy ±0.01 kg / s), Coolant inlet temperature sensor (measuring range 293-353 K, accuracy ±0.1 K), Load current density sensor (measuring range 0-2 A / cm²). 2 Accuracy ±0.01A / cm 2).

[0035] Observation parameter sensors: battery terminal voltage sensor (measurement range 0.6-1.0V, accuracy ±1mV), coolant outlet temperature sensor (measurement range 293-353K, accuracy ±0.1K), cathode outlet air relative humidity sensor (measurement range 0-100%, accuracy ±2%).

[0036] Data acquisition unit: It adopts a serial communication unit (communication rate 9600bps) to receive the analog signals output by the sensor unit, convert them into digital signals and transmit them to the particle filter processing unit. At the same time, it performs filtering preprocessing on the data to remove high-frequency interference.

[0037] Particle filter processing unit: Based on an embedded processor (main frequency ≥ 1GHz), it incorporates a fuel cell water-thermal coupling model and particle filter algorithm to process digital signals and achieve water content state estimation.

[0038] Estimation result output unit: includes a display screen and a communication interface. The display screen is used to display the core water content index in real time, and the communication interface is used to transmit the estimation results to the fuel cell controller (FCU) to provide data support for water management control.

[0039] The method for estimating the water content of a fuel cell system provided in this application will be described in detail below with reference to the accompanying drawings, through specific embodiments and application scenarios.

[0040] like Figure 2 As shown in the embodiment of this application, a method for estimating the water content of a fuel cell system is provided. This method may include the following steps 201 to 204: Step 201: Obtain the current operating status parameters of the fuel cell system.

[0041] The operating status parameters include: a first parameter (i.e., the input parameters collected by the aforementioned input parameter sensor) and a second parameter (i.e., the observation output parameters collected by the aforementioned observation parameter sensor); the first parameter is used to characterize the operating control variables of the fuel cell system; the second parameter is used to characterize the external parameters related to water content.

[0042] For example, after obtaining the current operating status parameters of the fuel cell system, the water content at the current moment can be estimated.

[0043] Step 202: Based on the first parameter, predict the particle state of the state transition model at the current moment to obtain the predicted state of each particle.

[0044] The state transition model is used to describe the hydrothermal state of the fuel cell system.

[0045] For example, the state variables of the above state transition model include at least one of the following: proton exchange membrane hydration degree, cathode gas diffusion layer liquid water saturation, battery average temperature, and mole fraction of water vapor in the cathode flow channel; the first parameter includes at least one of the following: cathode inlet air flow rate, anode inlet hydrogen flow rate, coolant inlet flow rate, coolant inlet temperature, and load current density.

[0046] For example, the state variables in the above state transition model include: proton exchange membrane hydration degree. (Value range 1-14) Liquid water saturation in the cathode gas diffusion layer (Value range 0-1), Average battery temperature (Value range 293-353K), Mole fraction of water vapor in the cathode channel (Value range 0-0.5), where proton exchange membrane hydration and the saturation of liquid water in the cathode gas diffusion layer The core indicator is moisture content.

[0047] For example, the input parameters of the above state transition model include: cathode inlet air flow rate. Hydrogen flow rate at the anode inlet Coolant inlet flow rate Coolant inlet temperature External load current density The above input parameters are collected in real time by an input parameter sensor.

[0048] For example, before estimating the water content, the state transition model needs to be initialized, and this initialization process includes the following steps 301 and 302: Step 301: Based on the pre-defined prior distribution of state variables.

[0049] Step 302: Generate the initial particle set of the state transition model and assign initial weights to each initial particle in the initial particle set.

[0050] The prior distribution of the state variables is set based on the startup characteristics of the fuel cell system.

[0051] For example, the prior probability distribution of each state variable in the state transition model is as follows: the initial distribution of the proton exchange membrane hydration degree is as follows: The initial distribution of liquid water saturation in the cathode gas diffusion layer is as follows: The average temperature of the battery The initial distribution is The initial distribution of water vapor mole fraction within the cathode channel is as follows: .

[0052] For example, in this embodiment of the application, a priori distribution of state variables is set based on the start-up characteristics of the fuel cell, an initial particle set is generated and weights are assigned: membrane hydration degree The initial distribution is (The membrane is typically in a semi-wet state at startup; the mean is a typical value of 8, and the variance reflects initial uncertainty); GDL liquid water saturation The initial distribution is (GDL moisture content is low during initial startup, averaging 15%); average battery temperature The initial distribution is (Approximately ambient temperature 298K, with allowable fluctuations of ±2K); Mole fraction of water vapor in the cathode channel The initial distribution is (Matching ambient humidity). Based on the above distribution, Latin hypercube sampling was used to generate... An initial number of particles (the optimal number balancing estimation accuracy and computational efficiency), each particle being a four-dimensional state vector. and assign uniform initial weights. To prevent particles from exceeding physical boundaries, constraints are imposed on the sampling results: , , (Battery safe operating temperature range). After generating the initial particles, an initial weight is assigned to each particle. It should be noted that, without additional information, the initial weights are usually set to a uniform distribution.

[0053] For example, the core equations of the state transition model include: the membrane hydration dynamic equation, the liquid water saturation equation of the cathode gas diffusion layer, and the battery temperature equation. The dynamic equation for membrane hydration degree is: in, The degree of hydration of the proton exchange membrane; For follow Variation in battery drag coefficient (( hour ), The water diffusion coefficient inside the membrane, For membrane density, For membrane volume, Let I be the water vapor permeation rate across the membrane, I be the load current density, and A be the effective reaction area of ​​the fuel cell. This equation quantifies the effects of three mechanisms—electroosmotic drag, diffusion, and permeation—on membrane hydration.

[0054] The equation for the saturation of liquid water is: in, The liquid water saturation level in the cathode gas diffusion layer. The porosity of the cathode gas diffusion layer (typically 0.4-0.6). The amount of liquid water generated by the electrochemical reaction ( , (where Faraday's constant is used) This refers to the amount of liquid water that evaporates. This refers to the amount of liquid water transported between the cathode gas diffusion layer and the flow channel. The density of liquid water, This represents the volume of the cathode gas diffusion layer.

[0055] The battery temperature equation is as follows: in, The equivalent heat capacity of the battery. It is a reversible voltage (approximately 1.23V). Heat carried away by the coolant ( ), The latent heat of vaporization of water ( ), This refers to the battery terminal voltage. This represents the average battery temperature.

[0056] For example, based on the above state transition model, the particle state at the current moment can be predicted. In this embodiment, a hydrothermal coupling model is used as an example of the state transition model. During the prediction phase, the particle is substituted into the state transition equation of the hydrothermal coupling model, and the state is updated by combining real-time input parameters. Taking time k as an example, the particle state transition formula is: in, (Control cycle to meet real-time requirements) , This represents process noise, the variance of which is identified experimentally (positively correlated with load fluctuation amplitude). Physical constraints need to be embedded in the state transition: when The time limit is set to 1 (when the membrane is completely dry). Set the value to 0.8 (GDL saturation).

[0057] Specifically, step 202 above, which involves predicting the particle state of the state transition model at the current moment, may further include the following step 202a: Step 202a: In the process of predicting the particle state at the current moment of the state transition model, process noise is added.

[0058] The process noise includes: process noise affecting the hydration degree of the proton exchange membrane. Process noise affecting the liquid water saturation of the cathode gas diffusion layer .

[0059] For example, after the particle state prediction at the current moment is completed, the weights can be updated based on the second parameter mentioned above.

[0060] Step 203: Based on the second parameter, calculate the theoretical observation value corresponding to the predicted state of each particle using the observation model, and update the weight of each particle based on the difference between the theoretical observation value and the actual measurement value.

[0061] For example, the second parameter mentioned above includes: battery terminal voltage, coolant inlet and outlet temperature difference, and cathode outlet air relative humidity; the observation model includes: terminal voltage observation equation; the terminal voltage observation equation is: in, To activate the overpotential, This is an ohmic overpotential. This is the concentration overpotential. Activation overpotential. ( (Tafel slope, approximately 0.06 V / dec), ohmic overpotential ( Film resistor , (membrane conductivity), concentration overpotential ( (where the limiting current density is).

[0062] For example, in the embodiments of this application, the water content and operating status of the fuel cell are described by the above-mentioned membrane hydration dynamic equation, liquid water saturation equation, battery temperature equation, and terminal voltage observation equation.

[0063] For example, the observed output parameter (i.e., the second parameter mentioned above) in the embodiments of this application includes: battery terminal voltage. Temperature difference between coolant inlet and outlet Cathode outlet air relative humidity The above-mentioned observation output parameters are acquired in real time by the observation parameter sensor.

[0064] For example, during the update phase, particle weights are calculated based on terminal voltage, coolant temperature difference, and outlet humidity using a pre-built fuel cell-specific observation model. This specifically includes the following steps: 1. First, theoretical observations are calculated by predicting particle states. ①by Calculate membrane conductivity Substituting into the voltage equation, we get ; ②by Calculate the theoretical outlet temperature of the coolant ③By Convert to theoretical relative humidity .

[0065] 2. Subsequently, a weight calculation method based on multi-observation fusion is adopted. The weight update formula is: in, , , The standard deviation of the observed noise is determined by the sensor's accuracy level.

[0066] Step 204: Based on the updated particle weights, resample each particle to generate a new particle set, and based on the new particle set, calculate the weighted average of each state variable in the state transition model to obtain the estimated value of the water content index at the current moment.

[0067] The water content indicators include: proton exchange membrane hydration degree and cathode gas diffusion layer liquid water saturation degree.

[0068] For example, after the weights are updated, resampling can be performed, and the water content can be estimated based on the resampled particles.

[0069] Specifically, step 204 above, which involves resampling each particle to generate a new set of particles, may further include steps 204a1 and 204a2: Step 204a1: Calculate the cumulative weight of each particle.

[0070] Step 204a2: Generate a random number that is the same as the number of particles and is uniformly distributed in the interval [0,1]. Based on the random number and the cumulative weight, extract particles with replacement from the original particle set to form a new particle set.

[0071] In this new particle set, the weights of all particles are set to a uniform distribution.

[0072] For example, in an embodiment of this application, the cumulative weight of each particle is first calculated: , After that, generate A uniformly distributed in random numbers in an interval For each random number Find satisfaction index Then the first One particle is selected to enter a new particle set. This process is repeated. Next, a new set of particles is obtained after resampling: After resampling, the weights of all new particles are typically reset to a uniform distribution, i.e. This allows the weights to be readjusted in subsequent iterations based on new observation data.

[0073] For example, a weighted average is performed on the resampled particles to output an estimated value of the core water content index, which can be calculated using the following formula: ; For example, based on the above formula, the water content index at the current moment can be estimated to obtain the estimated value of the water content index at the current moment. By repeating the above steps, the water content index can be estimated in real time based on the acquired operating status parameters.

[0074] The method for estimating the water content of a fuel cell system provided in this application has the following advantages: 1. Adaptability to nonlinear and non-Gaussian systems The unique advantage of particle filtering lies in its ability to effectively handle non-Gaussian noise without requiring the assumptions of linear system and observation models, which aligns perfectly with the characteristics of fuel cell systems. Significant nonlinearity exists in the hydrothermal coupling process of fuel cells: membrane hydration degree... With conductivity It exhibits a piecewise nonlinear relationship. hour Rising sharply (Then it tends to stabilize); GDL saturation Beyond 50%, the resistance to liquid water transport increases exponentially; the electrochemical reaction rate and temperature satisfy the Arrhenius equation (exponential relationship). Simultaneously, system noise exhibits non-Gaussian characteristics: load current fluctuations (such as in car start-stop scenarios) are impulse noise, and sensor measurement errors are subject to environmental interference, exhibiting a mixed Gaussian distribution. Traditional Kalman filtering requires linearization of voltage equations, etc. (e.g., ...). The approximation of a linear function leads to increased estimation error (experiments show an error exceeding 15%); while particle filtering directly approximates the state distribution of a nonlinear system through a swarm of particles, without requiring any linearization assumptions. Experiments show that in Under dynamic operating conditions where the value changes abruptly from 3 to 12, the estimation error of this method remains stable within ±0.3, while the linearized Kalman filter error reaches ±1.2, fully demonstrating its adaptability to the nonlinear characteristics of fuel cells.

[0075] 2. High-precision estimation By deeply fusing the physical mechanism model with particle weight updates, this method achieves high-precision estimation of water content. Its core advantages lie in two aspects: first, the particle weights can accurately reflect the physical constraints of the fuel cell; for example, when a particle predicts... (Membrane drying) However, when the actual observed voltage is high, this particle will be assigned a very low weight due to the large voltage deviation, to avoid its interference with the estimation results; secondly, a stratified weight allocation strategy was designed for key water content indicators, for and The corresponding observation errors are assigned higher weighting coefficients (voltage error coefficient set to 1.5, humidity error coefficient set to 1.0) to enhance the estimation accuracy of the core state. Experimental verification was conducted using a 10kW PEMFC test bench under dynamic loads (0.2-1.0 A / cm²). 2 Under cyclic operating conditions, this method is applicable to... The estimated root mean square error (RMSE) is 0.28. The RMSE is 0.023, compared to the traditional impedance method ( RMSE=1.1) and sliding mode observer ( The accuracy is significantly improved (RMSE=0.06). Meanwhile, under steady-state conditions, the estimation error can be stabilized within ±0.1, meeting the accuracy requirements of fuel cell water management control (control accuracy must be ≤0.5).

[0076] 3. It has strong real-time performance. Based on the real-time control requirements of fuel cells (control cycle is typically ≤10ms), this method ensures real-time performance from two aspects: algorithm optimization and model simplification. First, an adaptive particle number strategy is adopted, reducing the particle number from 200 to 100 under steady-state conditions and automatically increasing it back to 200 under dynamic conditions, reducing computation time by 40% compared to a fixed particle number scheme. Second, the hydrothermal coupling model is simplified in an engineering manner, such as removing the intramembrane water diffusion term. Replace with empirical coefficients (based on the fit of experimental data) This method avoids the high computational cost of solving partial differential equations. Tested on an Intel Core i5 processor (2.5GHz), the single-iteration calculation time is 3.2ms, fully meeting the 10ms control cycle requirement. Compared to the finite element model (requiring over 500ms per calculation) and the complex mechanism model (requiring 20ms per calculation), this method significantly improves computational efficiency while maintaining accuracy. In practical automotive applications, the calculation time can be further compressed to less than 1ms through FPGA hardware acceleration, adapting to the real-time estimation requirements under extreme dynamic conditions. Furthermore, the algorithm code is written in C and integrated into the fuel cell controller (FCU), exhibiting good compatibility with existing control logic without requiring additional hardware costs.

[0077] 4. High robustness To address the uncertainties in the actual operation of fuel cells, this method demonstrates strong robustness, as verified through three typical operating conditions: 1. Sensor data missing scenario: When the cathode outlet humidity sensor fails, the algorithm automatically removes the humidity observation item and updates the weights only based on voltage and coolant temperature difference. The estimated RMSE increased from 0.28 to 0.45, still outperforming traditional methods; 2. Model parameter drift scenario: when membrane aging leads to a decrease in diffusion coefficient... Even with a 30% decrease, particle filtering can still capture the true value through multi-observation fusion. Changes occurred, and the Kalman filter's RMSE increased to 1.8 due to model mismatch; 3. Extreme operating condition interference scenarios: startup at -10℃ (membrane water freezing caused initial...) Even under conditions of large measurement deviations and high altitude, low air pressure (which increases airflow measurement error by 20%), the estimation error of this method can still be controlled within ±0.6, meeting the requirements for safe operation. This robustness stems from the diversity of the particle ensemble—even if some particles deviate from the true state due to model or data errors, high-weight particles can still maintain tracking of the true state, preventing the estimation results from diverging. In contrast, observers based on a single model all exhibit estimation failures or significant deviations in the aforementioned scenarios.

[0078] In summary, compared with estimation methods in related technologies, the water content estimation method for fuel cell systems provided in this application overcomes the limitations of linear and Gaussian assumptions, and can well adapt to the complex nonlinear physicochemical processes and non-Gaussian noise environment within the fuel cell system, thereby achieving high-precision estimation of water content. Its efficient computation process ensures real-time performance, providing accurate state information for the timely operation and control of the fuel cell; its strong robustness allows it to maintain reliable estimation performance even when faced with abnormal sensor data or system model deviations. The particle filter-based online water content state estimation method is expected to play an important role in applications such as new energy vehicles and distributed generation, providing solid data support for the optimized control and water management strategies of fuel cell systems, further improving the energy conversion efficiency and service life of fuel cells.

[0079] The water content estimation method for a fuel cell system provided in this application first obtains the current operating state parameters of the fuel cell system. These operating state parameters include a first parameter and a second parameter. The first parameter characterizes the operating control variables of the fuel cell system, and the second parameter characterizes external parameters related to water content. Then, based on the first parameter, the particle states of the state transition model at the current moment are predicted to obtain the predicted state of each particle. The state transition model describes the hydrothermal state of the fuel cell system. Based on the second parameter, the theoretical observation value corresponding to the predicted state of each particle is calculated using an observation model, and the weight of each particle is updated based on the difference between the theoretical observation value and the actual measurement value. Finally, based on the updated particle weights, each particle is resampled to generate a new particle set, and based on the new particle set, the weighted average of each state variable in the state transition model is calculated to obtain the estimated value of the water content index at the current moment. Thus, the water content of the proton exchange membrane can be accurately estimated online during the actual operation of the fuel cell system.

[0080] It should be noted that the fuel cell system water content estimation method provided in this application embodiment can be executed by a fuel cell system water content estimation device, or a control module within that device for executing the method. This application embodiment uses the fuel cell system water content estimation device executing the method as an example to illustrate the fuel cell system water content estimation device provided in this application embodiment.

[0081] It should be noted that, in the embodiments of this application, the water content estimation methods for fuel cell systems shown in the accompanying drawings are all illustrated by way of example with reference to one of the accompanying drawings in the embodiments of this application. In specific implementation, the water content estimation methods for fuel cell systems shown in the accompanying drawings of the above methods can also be implemented in conjunction with any other accompanying drawings that can be combined with the above embodiments, which will not be elaborated here.

[0082] The water content estimation device for a fuel cell system provided in this application is described below. The water content estimation method for a fuel cell system described below can be referred to in correspondence with the water content estimation method described above.

[0083] Figure 3 This is a schematic diagram of the structure of the water content estimation device for a fuel cell system provided in an embodiment of this application, as shown below. Figure 3 As shown, it specifically includes: The parameter acquisition module 301 is used to acquire the current operating state parameters of the fuel cell system; the operating state parameters include: a first parameter and a second parameter; the first parameter is used to characterize the operating control variables of the fuel cell system; the second parameter is used to characterize the external parameters related to water content; the state prediction module 302 is used to predict the particle state of the state transition model at the current moment based on the first parameter, and obtain the predicted state of each particle; the state transition model is used to describe the hydrothermal state of the fuel cell system; the weight update module 303 is used to calculate the theoretical observation value corresponding to the predicted state of each particle using the observation model based on the second parameter, and update the weight of each particle based on the difference between the theoretical observation value and the actual measurement value; the water content estimation module 304 is used to resample each particle based on the updated particle weight, generate a new particle set, and calculate the weighted average of each state variable of the state transition model based on the new particle set, and obtain the estimated value of the water content index at the current moment.

[0084] Optionally, the device further includes: an initialization module; the initialization module is configured to generate an initial particle set of the state transition model based on a pre-set prior distribution of state variables; the initialization module is further configured to assign an initial weight to each initial particle in the initial particle set; wherein the prior distribution of state variables is set based on the start-up characteristics of the fuel cell system.

[0085] Optionally, the state prediction module 302 is specifically used to add process noise during the prediction of the particle state at the current moment of the state transition model; wherein, the process noise includes: process noise acting on the hydration degree of the proton exchange membrane and process noise acting on the liquid water saturation of the cathode gas diffusion layer.

[0086] Optionally, the water content estimation module 304 is specifically used to calculate the cumulative weight of each particle; the water content estimation module 304 is also specifically used to generate random numbers that are the same as the number of particles and are uniformly distributed in the interval [0,1], and to extract particles with replacement from the original particle set according to the random numbers and the cumulative weights to form a new particle set; wherein, the weights of all particles in the new particle set are set to be uniformly distributed.

[0087] Optionally, the state variables of the state transition model include at least one of the following: proton exchange membrane hydration degree, cathode gas diffusion layer liquid water saturation degree, battery average temperature, and mole fraction of water vapor in the cathode flow channel; the water content index includes: proton exchange membrane hydration degree and cathode gas diffusion layer liquid water saturation degree; the first parameter includes at least one of the following: cathode inlet air flow rate, anode inlet hydrogen flow rate, coolant inlet flow rate, coolant inlet temperature, and load current density.

[0088] Optionally, the prior probability distribution of each state variable in the state transition model is set as follows: the initial distribution of the proton exchange membrane hydration degree is as follows: The initial distribution of liquid water saturation in the cathode gas diffusion layer is as follows: The average temperature of the battery The initial distribution is The initial distribution of water vapor mole fraction within the cathode channel is as follows: .

[0089] The fuel cell system water content estimation device provided in this application first acquires the current operating state parameters of the fuel cell system. These operating state parameters include a first parameter and a second parameter. The first parameter characterizes the operating control variables of the fuel cell system, and the second parameter characterizes external parameters related to water content. Then, based on the first parameter, the particle states of the state transition model at the current moment are predicted to obtain the predicted state of each particle. The state transition model describes the hydrothermal state of the fuel cell system. Based on the second parameter, the theoretical observation value corresponding to the predicted state of each particle is calculated using an observation model, and the weight of each particle is updated based on the difference between the theoretical observation value and the actual measurement value. Finally, based on the updated particle weights, each particle is resampled to generate a new particle set, and based on the new particle set, the weighted average of each state variable of the state transition model is calculated to obtain the estimated value of the water content index at the current moment. Thus, the water content of the proton exchange membrane can be accurately estimated online during the actual operation of the fuel cell system.

[0090] Figure 4 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 4As shown, the electronic device may include: a processor 410, a communications interface 420, a memory 430, and a communications bus 440, wherein the processor 410, the communications interface 420, and the memory 430 communicate with each other through the communications bus 440. The processor 410 can call logic instructions in the memory 430 to execute a method for estimating the water content of a fuel cell system. This method includes: first, acquiring the current operating state parameters of the fuel cell system; the operating state parameters include a first parameter and a second parameter; the first parameter characterizes the operating control variables of the fuel cell system; the second parameter characterizes external parameters related to water content; then, based on the first parameter, predicting the particle states of the state transition model at the current moment to obtain the predicted state of each particle; the state transition model describes the hydrothermal state of the fuel cell system; based on the second parameter, calculating the theoretical observation value corresponding to the predicted state of each particle using an observation model, and updating the weight of each particle based on the difference between the theoretical observation value and the actual measurement value; finally, based on the updated particle weights, resampling each particle to generate a new particle set, and calculating the weighted average of each state variable of the state transition model based on the new particle set to obtain the estimated value of the water content index at the current moment. Thus, the water content of the proton exchange membrane can be accurately estimated online during the actual operation of the fuel cell system.

[0091] Furthermore, the logical instructions in the aforementioned memory 430 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0092] On the other hand, this application also provides a computer program product, which includes a computer program stored on a computer-readable storage medium. The computer program includes program instructions, and when the program instructions are executed by a computer, the computer can execute the fuel cell system water content estimation method provided by the above methods. The method includes: first, obtaining the current operating state parameters of the fuel cell system; the operating state parameters include: a first parameter and a second parameter; the first parameter is used to characterize the operating control variables of the fuel cell system; the second parameter is used to characterize the external parameters related to water content; then, based on the first parameter, predicting the particle state of the state transition model at the current moment to obtain the predicted state of each particle; the state transition model is used to describe the hydrothermal state of the fuel cell system; based on the second parameter, using an observation model to calculate the theoretical observation value corresponding to the predicted state of each particle, and updating the weight of each particle based on the difference between the theoretical observation value and the actual measurement value; finally, based on the updated particle weights, resampling each particle to generate a new particle set, and based on the new particle set, calculating the weighted average of each state variable of the state transition model to obtain the estimated value of the water content index at the current moment. In this way, the water content of the proton exchange membrane can be accurately estimated online during the actual operation of the fuel cell system.

[0093] Furthermore, this application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, performs the aforementioned methods for estimating the water content of a fuel cell system. This method includes: first, acquiring the current operating state parameters of the fuel cell system; the operating state parameters include a first parameter and a second parameter; the first parameter characterizes the operating control variables of the fuel cell system; the second parameter characterizes external parameters related to water content; then, based on the first parameter, predicting the particle states of the state transition model at the current moment to obtain a predicted state for each particle; the state transition model describes the hydrothermal state of the fuel cell system; based on the second parameter, calculating the theoretical observation value corresponding to the predicted state of each particle using an observation model, and updating the weight of each particle based on the difference between the theoretical observation value and the actual measurement value; finally, based on the updated particle weights, resampling each particle to generate a new particle set, and calculating the weighted average of the state variables of the state transition model based on the new particle set to obtain an estimated value of the water content index at the current moment. Thus, the water content of the proton exchange membrane can be accurately estimated online during the actual operation of the fuel cell system.

[0094] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0095] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0096] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for estimating the water content of a fuel cell system, characterized in that, include: Obtain the current operating status parameters of the fuel cell system; The operating status parameters include: a first parameter and a second parameter; the first parameter is used to characterize the operating control variables of the fuel cell system; the second parameter is used to characterize external parameters related to water content. Based on the first parameter, the particle state at the current moment of the state transition model is predicted to obtain the predicted state of each particle; the state transition model is used to describe the hydrothermal state of the fuel cell system. Based on the second parameter, the theoretical observation value corresponding to the predicted state of each particle is calculated using the observation model, and the weight of each particle is updated based on the difference between the theoretical observation value and the actual measurement value. Based on the updated particle weights, each particle is resampled to generate a new particle set. Based on the new particle set, the weighted average of each state variable in the state transition model is calculated to obtain the estimated value of the water content index at the current moment.

2. The method according to claim 1, characterized in that, The prediction of the particle state at the current moment in the state transition model includes: Process noise is added during the prediction of the particle state at the current moment in the state transition model. The process noise includes: process noise affecting the hydration degree of the proton exchange membrane and process noise affecting the liquid water saturation degree of the cathode gas diffusion layer.

3. The method according to claim 1, characterized in that, The process of resampling each particle based on the updated particle weights to generate a new particle set includes: Calculate the cumulative weight for each particle; Generate a random number that is the same as the number of particles and is uniformly distributed in the interval [0,1]. Based on the random number and the cumulative weight, draw particles with replacement from the original particle set to form a new particle set. In this new particle set, the weights of all particles are set to a uniform distribution.

4. The method according to claim 1, characterized in that, The initialization process of the state transition model includes the following steps: Based on a pre-defined prior distribution of state variables; Generate an initial particle set for the state transition model and assign initial weights to each initial particle in the initial particle set; The prior distribution of the state variables is set based on the startup characteristics of the fuel cell system.

5. The method according to claim 4, characterized in that, The state variables of the state transition model include at least one of the following: proton exchange membrane hydration degree, cathode gas diffusion layer liquid water saturation degree, battery average temperature, and mole fraction of water vapor in the cathode flow channel; the water content index includes: proton exchange membrane hydration degree and cathode gas diffusion layer liquid water saturation degree; the first parameter includes at least one of the following: cathode inlet air flow rate, anode inlet hydrogen flow rate, coolant inlet flow rate, coolant inlet temperature, and load current density.

6. The method according to claim 5, characterized in that, The prior probability distribution of each state variable in the state transition model is as follows: The initial distribution of hydration degree of the proton exchange membrane is as follows: The initial distribution of liquid water saturation in the cathode gas diffusion layer is as follows: The average temperature of the battery The initial distribution is The initial distribution of water vapor mole fraction within the cathode channel is as follows: .

7. The method according to claim 5, characterized in that, The core equations of the state transition model include: the membrane hydration dynamic equation, the liquid water saturation equation of the cathode gas diffusion layer, and the battery temperature equation. The dynamic equation for membrane hydration degree is: in, The degree of hydration of the proton exchange membrane; For follow The varying battery drag coefficient The water diffusion coefficient inside the membrane. For membrane density, For membrane volume, This represents the water vapor permeation rate across the membrane. I For load current density, A This refers to the effective reaction area of ​​the fuel cell; The equation for the saturation of liquid water is: in, The liquid water saturation level in the cathode gas diffusion layer. The porosity of the cathode gas diffusion layer. The amount of liquid water produced by the electrochemical reaction, This refers to the amount of liquid water that evaporates. This refers to the amount of liquid water transported between the cathode gas diffusion layer and the flow channel. The density of liquid water, The volume of the cathode gas diffusion layer; The battery temperature equation is as follows: in, The equivalent heat capacity of the battery. It is a reversible voltage. The heat carried away by the coolant. The latent heat of vaporization of water, This refers to the battery terminal voltage. This represents the average battery temperature.

8. The method according to any one of claims 7, characterized in that, The second parameter includes: battery terminal voltage, coolant inlet and outlet temperature difference, and relative humidity of the air at the cathode outlet; the observation model includes: terminal voltage observation equation; The terminal voltage observation equation is as follows: in, To activate the overpotential, This is an ohmic overpotential. This is concentration overpotential.

9. A water content estimation device for a fuel cell system, characterized in that, The device includes: The parameter acquisition module is used to acquire the operating status parameters of the fuel cell system at the current moment; the operating status parameters include: a first parameter and a second parameter; the first parameter is used to characterize the operating control variables of the fuel cell system; the second parameter is used to characterize external parameters related to water content; The state prediction module is used to predict the particle state of the state transition model at the current moment based on the first parameter, so as to obtain the predicted state of each particle; the state transition model is used to describe the hydrothermal state of the fuel cell system. The weight update module is used to calculate the theoretical observation value corresponding to the predicted state of each particle based on the second parameter using the observation model, and update the weight of each particle based on the difference between the theoretical observation value and the actual measurement value. The water content estimation module is used to resample each particle based on the updated particle weights to generate a new particle set, and calculate the weighted average of each state variable of the state transition model based on the new particle set to obtain the estimated value of the water content index at the current time.

10. An electronic device, characterized in that, It includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the water content estimation method for a fuel cell system as described in any one of claims 1 to 8.

Images