Performance prediction method and prediction system of proton exchange fuel cell system

Abstract

The invention relates to the technical field of fuel cells, and particularly discloses a proton exchange fuel cell system performance prediction method and system, and the method comprises the steps: S1, collecting parameters related to the performance of a proton exchange fuel cell system PEMFC, and obtaining real-time operation data; s2, performing preprocessing and feature extraction on the real-time operation data in the step S1 based on an anomaly detection method of deep learning to obtain standardized data; s3, constructing a machine learning model, training the model by using the standardized data, and performing multi-objective optimization by using a particle swarm optimization algorithm to obtain a sequence prediction model; s4, performing data prediction and verification by using the sequence prediction model in the step S3, and predicting to obtain the service life and performance data of the fuel cell; by optimizing the operation parameters, the performance fluctuation of the fuel cell in a high-power output or high-temperature environment can be reduced, and a plurality of targets are incorporated into the optimization process, so that the method is suitable for various application scenes.

Description

Technical Field

[0001] The present invention belongs to the technical field of fuel cells, and in particular relates to a method and system for predicting the performance of a proton exchange fuel cell system. Background Art

[0002] As a new generation of clean energy technology, proton exchange fuel cells (PEMFC) have shown great application potential in transportation, electricity, military and industry with their advantages of high power density, low emissions and fast response. In order to give full play to the performance advantages of PEMFC and ensure its stable and efficient operation under various working conditions, it is particularly important to accurately predict and optimize the performance of PEMFC systems.

[0003] At present, the performance prediction of PEMFC system mainly relies on mathematical modeling, simulation analysis and experimental testing. Mathematical modeling simulates the electrochemical reaction, material transfer and heat transfer process inside PEMFC by constructing complex physical and chemical models, so as to predict the performance of the system. Simulation analysis uses computer simulation software to solve the mathematical model and obtain the performance parameters of the system under different working conditions. Experimental testing directly measures the actual performance of the system by building a PEMFC system test platform to verify the accuracy of the prediction results.

[0004] However, with the increasing complexity of PEMFC systems and the diversification of application scenarios, existing prediction methods have gradually exposed some limitations.

[0005] Traditional PEMFC system optimization often focuses on a single performance indicator, such as power density or energy efficiency. However, in practical applications, PEMFC systems need to meet the requirements of multiple performance indicators at the same time, such as CO2 emissions, durability, cost, etc. Single-target optimization methods are difficult to balance these performance indicators, resulting in limited improvement in the overall performance of the system.

[0006] In order to solve the above problems, it is particularly important to develop a PEMFC system performance prediction method and system that can comprehensively consider multiple performance indicators and has adaptive capabilities. In this regard, the inventor proposes a proton exchange fuel cell system performance prediction method and system to solve the above problems. Summary of the invention

[0007] The present invention is intended to provide a method and system for predicting the performance of a proton exchange fuel cell system to solve the problems raised in the above background technology.

[0008] A method for predicting performance of a proton exchange fuel cell system, comprising:

[0009] S1. Collect parameters related to the performance of the proton exchange fuel cell system PEMFC and obtain real-time operation data;

[0010] S2. Preprocessing and feature extraction of the real-time operation data in step S1 using an anomaly detection method based on deep learning to obtain standardized data;

[0011] S3, build a machine learning model, use standardized data to train the model, use the particle swarm optimization algorithm to perform multi-objective optimization, and obtain a sequence prediction model;

[0012] S4. Use the sequence prediction model in step S3 to perform data prediction and verification, and predict the fuel cell life and performance data.

[0013] Preferably, the collection of parameters related to the performance of the proton exchange fuel cell system PEMFC is achieved by installing a sensor network in the fuel cell system.

[0014] Preferably, the sensor network includes a voltage sensor, a temperature sensor and a humidity sensor.

[0015] Preferably, the multi-objective optimization includes power density optimization, energy efficiency optimization and CO2 emission optimization.

[0016] Preferably, the power density optimization is an indicator for evaluating the performance of PEMFC, and its formula is as follows:

[0017] P d =V·I / A

[0018] V: battery cell output voltage V;

[0019] I: current A;

[0020] A: Active surface area of ​​battery cell cm 2 .

[0021] Preferably, the energy efficiency optimization represents the efficiency of the fuel cell in utilizing the input fuel, and the formula is as follows:

[0022]

[0023] Vcell: fuel cell monomer output voltage V;

[0024] ΔG: reaction Gibbs free energy J / mol;

[0025] n: the number of electrons in the reaction, which is 2;

[0026] F: Faraday constant, taken as 96485C / mol.

[0027] Preferably, the CO2 emission optimization represents the indirect carbon emission optimization of the fuel, and the formula is as follows:

[0028] E CO2 =Q·f

[0029] E CO2 : CO2 emissions, kg;

[0030] Q: amount of hydrogen consumed, mol;

[0031] f: Carbon emission coefficient of hydrogen production kg CO2 / mol H2.

[0032] Preferably, the speed update formula of the particle swarm optimization algorithm in step S3 is:

[0033] v i (t+1)=ωv i (t)+c1r1(p i -x i (t))+c2r2(gx i (t))

[0034] v i (t+1): velocity of particle i at step t+1;

[0035] ω: Inertia weight, which controls how much the particle depends on its previous velocity;

[0036] c1, c2: acceleration constants, adjusting the particle towards the individual optimal solution p i and the moving speed of the global optimal solution g;

[0037] r1, r2: random numbers in the range of [0,1], used to increase randomness;

[0038] x i (t): the position of particle i at step t;

[0039] p i : The optimal position found by particle i so far;

[0040] g: global optimal position.

[0041] Preferably, the sequence prediction model is a variational autoencoder VAE, including an encoder and a decoder, wherein the encoder maps the input sequence x to a distribution z in a latent space, and the formula is:

[0042] z~σ φ (z|x)=N(μ φ (x),σ φ 2 (x)

[0043] Among them, μ φ (x) and σ φ (x) is the output of the encoder network, representing the mean and standard deviation of the latent space, respectively.

[0044] The decoder reconstructs the input sequence hatx from the sampling point z in the latent space, and its formula is:

[0045] hatx=p θ (x|z)

[0046] Among them, p θ (x|z) is the output of the decoder network and represents the distribution of the input sequence given the latent variable z.

[0047] The present invention also provides a proton exchange fuel cell system performance prediction system, comprising a processor and a computer-readable storage medium; the processor loads and executes any one of the above-mentioned proton exchange fuel cell system performance prediction methods, the computer-readable storage medium is used to store multiple instructions, the processor is used to implement the instructions, and the instructions are suitable for the prediction method.

[0048] Compared with the prior art, the present invention has the following beneficial effects:

[0049] The present invention can reduce the performance fluctuation of the fuel cell under high power output or high temperature environment by optimizing the operating parameters, and incorporates multiple objectives into the optimization process, including power density, energy efficiency, CO2 emissions, etc., making it suitable for a variety of application scenarios. It integrates advanced optimization algorithms and can further develop an adaptive optimization system to monitor and adjust the operating parameters of the fuel cell in real time to cope with complex and changeable operating conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 The present invention is a flow chart of a method for predicting performance of a proton exchange fuel cell system. DETAILED DESCRIPTION

[0051] The following is further described in detail through specific implementation methods:

[0052] Embodiment 1:

[0053] See also Figure 1 As shown, a method for predicting the performance of a proton exchange fuel cell system comprises:

[0054] S1. By installing voltage sensors, temperature sensors, humidity sensors, etc. in the fuel cell system, parameters related to the performance of the proton exchange fuel cell system PEMFC are collected to obtain real-time operation data;

[0055] S2. Preprocessing and feature extraction of the real-time operation data in step S1 using an anomaly detection method based on deep learning to obtain standardized data;

[0056] S3. Build a machine learning model, use standardized data to train the model, and use the particle swarm optimization algorithm to perform multi-objective optimization, including power density optimization, energy efficiency optimization, and CO2 emission optimization, to obtain a sequence prediction model;

[0057] S4. Use the sequence prediction model in step S3 to perform data prediction and verification, and predict the fuel cell life and performance data.

[0058] Specifically, the power density optimization is an indicator for evaluating the performance of PEMFC, and its formula is as follows:

[0059] P d =V·I / A

[0060] V: battery cell output voltage V;

[0061] I: current A;

[0062] A: Active surface area of ​​battery cell cm 2 ;

[0063] Power density reflects the output power per unit area and is a key parameter for evaluating the energy output of fuel cells. By increasing the power density, fuel cells can provide higher power output in a relatively small volume.

[0064] In multi-objective optimization, this formula is used to maximize the power density to improve the output efficiency of the fuel cell.

[0065] Specifically, the energy efficiency optimization represents the efficiency of the fuel cell in utilizing the input fuel, and the formula is as follows:

[0066]

[0067] Vcell: fuel cell monomer output voltage V;

[0068] ΔG: reaction Gibbs free energy J / mol;

[0069] n: the number of electrons in the reaction, which is 2;

[0070] F: Faraday constant, taken as 96485C / mol;

[0071] Energy efficiency measures the proportion of electrical energy that a fuel cell obtains from fuel. By maximizing energy efficiency, fuel consumption is kept as low as possible while output power is kept as high as possible.

[0072] In multi-objective optimization, energy efficiency is one of the key optimization objectives. Maximizing energy efficiency can not only reduce fuel waste, but also reduce system operating costs.

[0073] Specifically, the CO2 emission optimization represents the indirect carbon emission optimization of fuel, and its formula is as follows:

[0074] E CO2 =Q·f

[0075] E CO2 : CO2 emissions, kg;

[0076] Q: amount of hydrogen consumed, mol;

[0077] f: Carbon emission coefficient of hydrogen production kg CO2 / mol H2;

[0078] Calculating indirect carbon emissions can help us evaluate the environmental impact of the system, which is important for evaluating the environmental benefits of fuel cells throughout their life cycle;

[0079] A key metric in multi-objective optimization is environmental impact. Minimizing CO2 emissions can make fuel cell systems more environmentally friendly. The formula helps the optimization algorithm identify the best operating conditions to reduce carbon emissions.

[0080] Specifically, the speed update formula of the particle swarm optimization algorithm in step S3 is:

[0081] v i (t+1)=ωv i (t)+c1r1(p i -x i (t))+c2r2(gx i (t))

[0082] v i (t+1): velocity of particle i at step t+1;

[0083] ω: Inertia weight, which controls how much the particle depends on its previous velocity;

[0084] c1, c2: acceleration constants, adjusting the particle towards the individual optimal solution p i and the moving speed of the global optimal solution g;

[0085] r1, r2: random numbers in the range of [0,1], used to increase randomness;

[0086] x i (t): the position of particle i at step t;

[0087] p i : The optimal position found by particle i so far;

[0088] g: global optimal position;

[0089] The particle velocity update formula helps the particle swarm algorithm find the optimal solution in the search space; by continuously adjusting the velocity and position, the particles tend to the optimal solution, thus solving the multi-objective optimization problem;

[0090] The PSO algorithm uses the speed update formula to enable particles to achieve a dynamic balance between the global optimum and the local optimum, thereby accelerating the convergence speed and achieving the goal of balancing power density, energy efficiency and environmental impact.

[0091] Specifically, the sequence prediction model is a variational autoencoder VAE, which includes an encoder and a decoder. The encoder maps the input sequence x to the distribution z in the latent space, and the formula is:

[0092] z~σ φ (z|x)=N(μ φ (x),σ φ 2 (x)

[0093] Among them, μ φ (x) and σ φ (x) is the output of the encoder network, representing the mean and standard deviation of the latent space, respectively.

[0094] Specifically, the decoder reconstructs the input sequence hatx from the sampling point z in the latent space, and the formula is:

[0095] hatx=p θ (x|z)

[0096] Among them, p θ (x|z) is the output of the decoder network, which represents the distribution of the input sequence given the latent variable z;

[0097] In sequence prediction, VAE can first learn a latent representation of the input sequence and then generate future sequences based on this latent representation, using RNN or LSTM to capture the time series dynamics in the latent space.

[0098] From the above, it can be seen that by optimizing the operating parameters, the performance fluctuations of fuel cells under high power output or high temperature environments can be reduced, and multiple objectives can be incorporated into the optimization process, including power density, energy efficiency, CO2 emissions, etc., making it suitable for a variety of application scenarios. By integrating advanced optimization algorithms, adaptive optimization systems can be further developed to monitor and adjust the operating parameters of fuel cells in real time to cope with complex and changeable operating conditions.

[0099] The present invention is specifically applied to the performance prediction of proton exchange fuel cell systems. As a power source for vehicles such as automobiles, ships, and airplanes, proton exchange fuel cell systems can achieve ultra-low emission operation and reduce environmental pollution. With the continuous advancement of technology and the reduction of costs, proton exchange fuel cell vehicles are expected to achieve large-scale commercial applications in the next few years. As a distributed power source, the proton exchange fuel cell system can provide power guarantee for remote areas or emergency situations. In smart grids and distributed energy systems, proton exchange fuel cell systems can be used as important energy storage and power generation equipment to improve the stability and reliability of the power system.

[0100] Application examples:

[0101] Referring to the method of Example 1, there is a 100W small PEMFC system, the goal is to optimize its power density and energy efficiency while reducing CO2 emissions, and balance power, efficiency and environmental impact through genetic algorithms or particle swarm optimization algorithms.

[0102] System parameters:

[0103] Unit area A = 50 cm 2

[0104] Hydrogen consumption Q = 0.05 mol / min

[0105] Carbon emission coefficient for hydrogen production f = 0.8 kg CO2 / mol H2

[0106] Working voltage V range: 0.6–0.9V (adjustable)

[0107] Working temperature T range: 60℃–80℃ (adjustable)

[0108] Working humidity H range: 50%–100% (adjustable)

[0109] Target optimization:

[0110] Maximize power density Pd

[0111] Maximize energy efficiency η

[0112] Minimize CO2 emissions CO2

[0113] 1. Data Collection

[0114] Assume that the system initial conditions and collected data are as follows:

[0115] Voltage V = 0.7V

[0116] Current I = 0.2A

[0117] Hydrogen consumption Q = 0.05 mol / min

[0118] 2. Calculate power density

[0119] According to the formula:

[0120] P d =V·I / A

[0121] Substitute the data into:

[0122] Pd = 0.7 × 0.250 = 0.0028 W / cm 2

[0123] 3. Calculate energy efficiency

[0124] Given the Gibbs free energy ΔG = 237000 J / mol (i.e., the free energy change per mole of hydrogen) and the number of reaction electrons n = 2, the energy efficiency η is calculated as:

[0125]

[0126] Substitute the data and calculate:

[0127] η=0.7237000 / (2×96485)≈0.57

[0128] 4. Calculate CO2 emissions

[0129] CO2 emissions CO2 The calculation is:

[0130] E CO2 =Q·f

[0131] Substitute the data and calculate:

[0132] ECO2=0.05×0.8=0.04kg CO2 / min

[0133] 5: Define the optimization objective function

[0134] For the multi-objective optimization problem, the power density Pd, energy efficiency η and CO2 emissions E are set. CO2 The objective function f(x) is:

[0135] f(x)=w1·Pd+w2·η+w3·E CO2

[0136] The weight coefficients w1 = 0.4, w2 = 0.4, and w3 = 0.2 are used to focus on balancing power density and efficiency while taking into account environmental impact.

[0137] Substitute the above data and calculate:

[0138] f(x)=0.4·1 / 0.0028+0.4·1 / 0.57+0.2·0.04

[0139] ≈142.86+0.7+0.008

[0140] =143.568

[0141] 6: Apply particle swarm optimization algorithm

[0142] The particle swarm optimization algorithm (PSO) is used to optimize the objective function f(x) as the fitness function. Assume that the following optimal parameters are obtained after iteration:

[0143] Optimum operating voltage V = 0.75V

[0144] Optimum working temperature T = 75 °C

[0145] Optimum humidity H = 90%

[0146] Substitute these optimization parameters and recalculate:

[0147] Update data collection

[0148] New voltage V = 0.75V

[0149] Assume that the current I = 0.25A

[0150] Recalculate power density

[0151] Pd=0.75×0.25 / 50=0.00375W / cm 2

[0152] Recalculate energy efficiency

[0153] η=0.75 / [237000 / (2×96485)]≈0.61

[0154] Recalculating CO2 emissions

[0155] E CO2 =0.04kg CO2 / min

[0156] Recalculate the objective function

[0157] f(x)=0.4·10.00375+0.4·10.61+0.2·0.04

[0158] ≈106.67+0.655+0.008

[0159] =107.333

[0160] Application Effect

[0161] After optimization, the objective function value dropped from 143.568 to 107.333, and the power density of the system increased to 0.00375W / cm 2 , the energy efficiency is also slightly improved to 0.61, meeting the multiple goals of optimizing performance and reducing environmental impact.

[0162] As can be seen from the above, by optimizing operating parameters such as voltage, temperature, and humidity, this method can significantly improve power density and energy efficiency. This means that under the same conditions, the fuel cell system can output higher power, thereby increasing the effective output of the device. In practical applications, improving energy efficiency helps reduce fuel consumption, so that more electricity is generated per unit of hydrogen, which helps reduce energy usage costs and system operating expenses.

[0163] The optimization method takes into account CO2 emissions as one of the optimization targets, which is essential for sustainable development and environmental protection. By reducing the carbon emission coefficient and reducing carbon emissions, fuel cells can meet environmental standards in a wider range of applications. In addition, combined with multi-objective optimization methods, fuel cell systems can find a better balance between performance and environmental impact, meeting the requirements of low-carbon economy and environmentally friendly energy.

[0164] By optimizing the operating parameters, the performance fluctuation of the fuel cell under high power output or high temperature environment can be reduced. For example, controlling the temperature and humidity helps to maintain the hydration state of the membrane and avoid membrane degradation and system failure caused by dryness or over-humidity. This enables the fuel cell to maintain higher stability and reliability during long-term operation, reducing the occurrence of failures and extending the system life.

[0165] The present invention supports incorporating multiple objectives into the optimization process, including power density, energy efficiency, CO2 emissions, etc. This flexibility makes it suitable for a variety of application scenarios. For example, in mobile devices that require high power density, power output can be optimized first; in scenarios with high environmental friendliness requirements, emissions reduction can be prioritized. This multi-objective optimization feature makes the method more versatile and can meet the diverse needs of different industries.

[0166] By improving efficiency and reducing fuel consumption, this method can directly reduce the operating cost of the system. In addition, due to the improvement of system stability and optimization of operating parameters, it can also reduce maintenance requirements and related costs, and extend the service life of key components such as membrane electrodes, which has great economic value for fuel cell applications that require long-term and continuous operation, such as transportation, backup power, etc.

[0167] Embodiment 2:

[0168] A proton exchange fuel cell system performance prediction system, the system comprising a processor and a computer-readable storage medium; the computer-readable storage medium is used to store multiple instructions, the processor is used to implement each instruction, and the instructions are suitable for being loaded by the processor and executing the above-mentioned proton exchange fuel cell system performance prediction method.

[0169] The beneficial effects and technical effects of the same proton exchange fuel cell system performance prediction method embodiment are not repeated here.

[0170] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "examples", "specific examples" or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine different embodiments or examples described in this specification and the features of different embodiments or examples, unless they are contradictory.

[0171] Although the embodiments of the present invention have been shown and described, it is understood by those skilled in the art that various changes, modifications, substitutions and variations may be made to these embodiments without departing from the principles and spirit of the present invention, and the scope of the present invention is defined by the appended claims and their equivalents. The embodiments described above are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

Claims

1. A method for predicting the performance of a proton exchange fuel cell system, characterized in that: include: S1. Collect parameters related to the performance of the proton exchange fuel cell system PEMFC and obtain real-time operation data; S2. Preprocessing and feature extraction of the real-time operation data in step S1 using an anomaly detection method based on deep learning to obtain standardized data; S3, build a machine learning model, use standardized data to train the model, use the particle swarm optimization algorithm to perform multi-objective optimization, and obtain a sequence prediction model; S4. Use the sequence prediction model in step S3 to perform data prediction and verification, and predict the fuel cell life and performance data.

2. A proton exchange fuel cell system performance prediction method according to claim 1, characterized in that: The collection of parameters related to the performance of the proton exchange fuel cell system PEMFC is achieved by installing a sensor network in the fuel cell system.

3. A proton exchange fuel cell system performance prediction method according to claim 2, characterized in that: The sensor network includes a voltage sensor, a temperature sensor, and a humidity sensor.

4. A proton exchange fuel cell system performance prediction method according to claim 1, characterized in that: The multi-objective optimization includes power density optimization, energy efficiency optimization and CO2 emission optimization.

5. A proton exchange fuel cell system performance prediction method according to claim 4, characterized in that: The power density optimization is an indicator for evaluating the performance of PEMFC, and its formula is as follows: P d =V·I / A V: battery cell output voltage V; I: current A; A: Active surface area of ​​battery cell cm 2 .

6. A proton exchange fuel cell system performance prediction method according to claim 4, characterized in that: The energy efficiency optimization represents the efficiency of the fuel cell in utilizing the input fuel, and its formula is as follows: Vcell: fuel cell monomer output voltage V; ΔG: reaction Gibbs free energy J / mol; n: the number of electrons in the reaction, which is 2; F: Faraday constant, taken as 96485C / mol.

7. A proton exchange fuel cell system performance prediction method according to claim 4, characterized in that: The CO2 emission optimization represents the indirect carbon emission optimization of the fuel, and its formula is as follows: E CO2 =Q·f E CO2 : CO2 emissions, kg; Q: amount of hydrogen consumed, mol; f: Carbon emission coefficient of hydrogen production kg CO2 / mol H2.

8. A proton exchange fuel cell system performance prediction method according to claim 1, characterized in that: The speed update formula of the particle swarm optimization algorithm in step S3 is: v i (t+1)=ωv i (t)+c1r1(p i -x i (t))+c2r2(g-x i (t)) v i (t+1): velocity of particle i at step t+1; ω: Inertia weight, which controls how much the particle depends on its previous velocity; c1, c2: acceleration constants, adjusting the particle towards the individual optimal solution p i and the moving speed of the global optimal solution g; r1, r2: random numbers in the range of [0,1], used to increase randomness; x i (t): the position of particle i at step t; p i : The optimal position found by particle i so far; g: global optimal position.

9. A proton exchange fuel cell system performance prediction method according to claim 1, characterized in that: The sequence prediction model is a variational autoencoder VAE, which includes an encoder and a decoder. The encoder maps the input sequence x to the distribution z in the latent space, and the formula is: z~σ φ (z|x)=N(μ φ (x),s φ 2 (x)) Among them, μ φ (x) and σ φ (x) is the output of the encoder network, representing the mean and standard deviation of the latent space; The decoder reconstructs the input sequence hatx from the sampling point z in the latent space, and its formula is: hatx=p θ (x|z) Among them, p θ (x|z) is the output of the decoder network and represents the distribution of the input sequence given the latent variable z.

10. A proton exchange fuel cell system performance prediction system, characterized in that: It comprises a processor and a computer-readable storage medium; the processor loads and executes a proton exchange fuel cell system performance prediction method according to any one of claims 1 to 9, the computer-readable storage medium is used to store multiple instructions, the processor is used to implement the instructions, and the instructions are suitable for the prediction method.

Images