Parameter on-line identification method suitable for PEM electrolytic cell electrochemical model

Abstract

The invention discloses a parameter online identification method suitable for a PEM electrolytic cell electrochemical model. According to the method, temperature sensitive parameters in an electrochemical model are decoupled into a reference parameter and a temperature correction coefficient, and a temperature decoupling model is constructed according to the reference parameter and the temperature correction coefficient; collecting current, temperature, pressure and voltage data under an actual operation condition to construct a measurable variable data set; performing parameter identification by using a particle swarm optimization algorithm, and substituting an identification result into the temperature decoupling model; and when a preset parameter identification period is reached or a fault of the electrolytic cell is detected, parameter identification is carried out again according to the updated measurable variable data set. According to the method, online parameter identification of the electrochemical model can be realized under the working condition of temperature and pressure fluctuation, and the voltage prediction precision of the PEM electrolytic cell under the conditions of working condition change, equipment aging and fault is remarkably improved.

Description

Technical Field

[0001] This invention belongs to the field of electrochemical model construction and parameter identification of PEM electrolyzers, and particularly relates to an online method for identifying electrochemical model parameters to improve the accuracy of voltage prediction under changing operating conditions, equipment aging and fault conditions of electrolyzers. Background Technology

[0002] Against the backdrop of large-scale application of renewable energy, the development of new energy bases in the western desert and deep-sea areas is accelerating, but centralized new energy sources face challenges in grid integration. Hydrogen energy, as a highly efficient secondary energy source, has become a core solution to the large-scale integration problem by converting intermittent renewable energy through electrolysis technology. Mainstream water electrolysis hydrogen production technologies cover four categories: alkaline, PEM, anion exchange membrane, and solid oxide electrolysis. Among them, PEM electrolyzers have become a research focus due to their compact structure, high current density, fast response, and wide power range.

[0003] Accurately constructing electrochemical models of electrolyzers is crucial for characterizing their operating properties and is of strategic significance for improving the reliability, durability, and economy of electrolysis systems. Optimizing electrolyzer performance prediction based on electrochemical models can provide theoretical guidance for engineering practice, thereby enhancing system efficiency and operational stability and promoting its coordinated operation with new energy power generation equipment.

[0004] In the study of electrochemical models for PEM electrolyzers, researchers initially constructed zero-dimensional empirical models to characterize polarization properties using a simplified functional relationship between voltage and current. Furthermore, some early modeling strategies relied on curve fitting techniques to approximate steady-state voltage-current characteristics. However, these models suffered from limitations due to a lack of internal mechanism characterization, resulting in simplistic structures, limited accuracy, and applicability restricted to specific electrolyzers. Therefore, researchers established a static mechanism model based on the molar conservation law for anodes and cathodes, systematically integrating loss mechanisms such as open-circuit voltage, activation overpotential, and ohmic overpotential. Further, subsequent models not only covered key operating parameters such as temperature and current density but also deeply analyzed the influence mechanisms of operating pressure and PEM thickness on performance, thereby achieving quantitative evaluation and optimization of electrolyzer efficiency. Furthermore, to reveal the mechanism by which exchange current density is affected by the inherent characteristics of the electrolyzer and the synergistic effect of temperature, researchers introduced the Arrhenius equation to accurately characterize the dependence of exchange current density and PEM conductivity on temperature.

[0005] Current research on electrochemical model construction and parameter identification of PEM electrolyzers is mainly limited to operating conditions where temperature and pressure are stable. However, with the rapid development of hydrogen production technology based on renewable energy sources with strong random fluctuations under limited energy storage conditions, the changes in current, temperature, and pressure during electrolyzer operation will become more frequent. If the impact of dynamic temperature and pressure changes on electrochemical model construction and parameter identification is ignored, the simulation accuracy and practical application value of the model will be significantly reduced.

[0006] Research on voltage prediction in PEM electrolyzers is relatively scarce, while domestic and international scholars have established a relatively systematic research framework in areas such as voltage prediction and health management of PEM fuel cells. Based on the similarity in core component structure and internal mechanisms between PEM electrolyzers and fuel cells, the driving factors of performance degradation in both are significantly correlated. Fuel cell voltage prediction methods can be systematically categorized into two types: data-driven methods and model-driven methods. Data-driven methods mainly rely on large-scale experimental datasets, do not require analysis of the internal degradation mechanism of the battery, and achieve the minimization of prediction error through sufficient model training and parameter optimization. Their core lies in using statistical methods to improve prediction accuracy, with typical technical paths including grey models, support vector machines, and neural networks. Model-driven approaches construct physical models based on material properties, load conditions, and degradation mechanisms to characterize internal processes. Among them, degradation models focus on the dynamic evolution of aging parameters and often combine Bayesian estimation frameworks such as Kalman filtering or particle filtering to achieve aging trend prediction and voltage forecasting. Load condition models focus on typical operating modes such as start-stop cycles, high power output, and rapid load changes, and construct empirical models to extrapolate voltage. Component degradation models construct multiphysics models through key component attenuation mechanisms such as membrane degradation and catalyst layer failure to achieve long-term voltage prediction.

[0007] However, the evolution trend of aging parameters revealed by short-term operating data under the existing research framework may have significant deviations. More importantly, the current mainstream voltage prediction methods are difficult to effectively deal with sudden changes in system parameters caused by sudden failures, such as mechanical damage to PEM, poisoning of catalyst layer ionomers, or corrosion of bipolar plates. The fundamental reason is that their modeling is generally based on the assumption of gradual degradation, which ultimately leads to a significant decline in prediction performance under abnormal operating conditions.

[0008] Therefore, it is urgent to study an online parameter identification method suitable for the electrochemical model of PEM electrolyzer. By constructing a temperature-decoupled electrochemical model, the parameters of the electrolyzer under high current density conditions can be identified in response to temperature and pressure fluctuations. Furthermore, online parameter identification can be performed using current, temperature, pressure, and voltage data from actual operation, thereby improving the voltage prediction accuracy of the electrolyzer under conditions of operating fluctuations, equipment aging, and sudden failures. Summary of the Invention

[0009] To address the shortcomings of the aforementioned technologies, the present invention aims to provide an online parameter identification method suitable for electrochemical models of PEM electrolyzers, thereby improving the voltage prediction accuracy of electrolyzers under fluctuating operating conditions, equipment aging, and fault states. At the same time, it overcomes the limitation of traditional high-precision parameter identification methods that require both temperature and pressure to remain stable for online parameter identification.

[0010] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0011] A method for online parameter identification of an electrochemical model for a PEM electrolyzer includes the following steps: S1, decoupling temperature-sensitive parameters in the PEM electrolyzer electrochemical model into reference parameters at a reference temperature and their corresponding temperature correction coefficients based on the Arrhenius equation; S2, constructing a temperature-decoupled PEM electrolyzer electrochemical model, identifying the parameters to be identified, and setting their reasonable value ranges; S3, collecting current, temperature, pressure, and voltage data of the PEM electrolyzer under actual operating conditions, and constructing a dataset of measurable variables for online parameter identification after processing; S4, performing parameter identification using a particle swarm optimization algorithm, and substituting the optimal parameter set obtained from the identification into the temperature-decoupled PEM electrolyzer electrochemical model; S5, re-performing parameter identification based on the updated dataset of measurable variables when a preset parameter identification cycle is reached or when a fault is detected in the electrolyzer.

[0012] Furthermore, in step S1, the temperature-sensitive parameters in the electrochemical model of the PEM electrolyzer are decoupled into reference parameters at the reference temperature and their corresponding temperature correction coefficients, which can be specifically expressed as follows:

[0013] The PEM electrolytic cell consists of several electrolysis chambers, each with an operating voltage U. cell Including open circuit voltage U ocv Activation overpotential Ohmic overpotential Concentration overpotential Four parts;

[0014] Open circuit voltage U ocv It can be represented as:

[0015]

[0016] Where T is the operating temperature of the PEM electrolyzer in K; R is the ideal gas constant with a value of 8.314 J / (mol·K); and F is the Faraday constant with a value of 96485 C / mol. and These are the partial pressures of hydrogen at the cathode and oxygen at the anode, respectively, in atm; The activity of the water between the electrode and the PEM;

[0017] Activation overpotential It can be represented as:

[0018]

[0019] Where, α ano and α cat These represent the anode charge transfer coefficient and the cathode charge transfer coefficient, respectively; J is the operating current density of the PEM electrolyzer, in A / cm². 2 J 0,ano and J 0,cat These are the anode exchange current density and the cathode exchange current density, respectively, in A / cm². 2 Based on the Arrhenius equation, after decoupling it into reference parameters at the reference temperature and their corresponding temperature correction coefficients, it can be expressed as follows:

[0020]

[0021] Among them, J 0,ano,ref and J 0,cat,ref These are the anode exchange current density and cathode exchange current density at the reference temperature (303.15 K), respectively; E exc,ano and E exc,cat These are the activation energies required for anodic electron transport and cathode electron transport, respectively.

[0022] Ohmic overpotential It can be represented as:

[0023]

[0024] Where, δ mem σ represents the PEM thickness in cm. mem The conductivity of PEM, expressed in S / cm, can be decoupled based on the Arrhenius equation and expressed as:

[0025]

[0026] Where, λ mem E represents the water content of the membrane. pro This is the activation energy for proton transport in the membrane;

[0027] Concentration overpotential It can be represented as:

[0028]

[0029] Among them, J lim The limiting current density is expressed in A / cm². 2 Based on the Arrhenius equation, after decoupling, it can be expressed as:

[0030]

[0031] Among them, J lim,ref The limiting current density at the reference temperature (303.15 K) is expressed in A / cm². 2 E diff The activation energy required for effective diffusion. By constructing a temperature-decoupled electrochemical model for the PEM electrolyzer, the problem of online parameter identification under temperature fluctuation conditions can be solved, and the voltage prediction accuracy of the electrolyzer under operating condition switching can be significantly improved.

[0032] Furthermore, the temperature-decoupled electrochemical model of the PEM electrolyzer in step S2 is as follows:

[0033] The operating voltage of the PEM electrolytic cell can be expressed as:

[0034]

[0035] Where, N stack The number of electrolysis chambers connected in series in the PEM electrolyzer;

[0036] The parameters to be identified include the anodic charge transfer coefficient α. ano Cathode charge transfer coefficient α cat Anode exchange current density J at reference temperature 0,ano,ref Cathode exchange current density J at reference temperature 0,cat,ref The activation energy E required for anodic electron transport exc,ano The activation energy E required for cathode electron transport exc,cat The activation energy E for proton transport in the membrane pro Limiting current density J at reference temperature lim,ref The activation energy E required for effective diffusion diff The reasonable range of values ​​for the parameters to be identified includes the anode charge transfer coefficient α. ano With cathode charge transfer coefficient α cat All are limited to a reasonable reaction kinetic range of 0.01 to 0.99; anodic exchange current density J at the reference temperature. 0,ano,ref Crossing 10 -10 Up to 10 -5 A / cm 2 Order of magnitude; cathode exchange current density J at reference temperature 0,cat,ref Set between 0.1 and 0.4; the activation energy E required for anodic electron transport. exc,ano Activation energy E required for cathode electron transport exc,cat The activation energy E required for effective diffusion diff All are limited to 10 2 Up to 10 5Range, activation energy E for proton transport in the membrane pro Limiting current density J at the reference temperature, between 9000 and 30000. lim,ref Set to 2.2 to 3.3 A / cm 2 between.

[0037] Furthermore, the current of the PEM electrolyzer under actual operating conditions is dynamically generated based on power generation prediction data using a stepped curve with non-uniform time intervals. The current switching process follows a smooth transition principle to suppress the impact of the electric double-layer effect on the electrolyzer voltage. The current stabilization time must be maintained for at least 5 minutes to ensure reliable steady-state operating condition data is obtained across different current density ranges. This strategy can suppress electrolyzer voltage distortion caused by the electric double-layer effect and ensure the reliability of data acquisition over a wide current density range, providing a crucial dynamic operating condition data foundation for constructing an accurate prediction model adapted to the fluctuating characteristics of renewable energy.

[0038] Furthermore, the processing method for the current, temperature, pressure, and voltage data of the PEM electrolyzer under actual operating conditions in step S3 includes: implementing abnormal data screening based on multi-physics field coupling boundary constraints, and eliminating double-layer effect interference by extracting steady-state operating condition data. This method can ensure that the current, temperature, pressure, and voltage parameter sets strictly conform to the coupling laws of electrochemistry, thermodynamics, and fluid mechanics, and the steady-state data after removing the double-layer effect can significantly improve the accuracy of PEM electrolyzer parameter identification.

[0039] Furthermore, in step S4, the particle swarm optimization algorithm sets the particle swarm position vector and velocity vector during the initialization process, and configures the parameter constraint boundary, the feasible region of the dynamic inertia weight coefficient, the individual cognition coefficient, and the group cognition coefficient; in each iteration, the objective function value of each particle is calculated in parallel, and the individual historical optimal solution and the group global optimal solution are updated synchronously; when the change in the objective function is lower than the convergence threshold or the number of iterations reaches the preset upper limit, the optimal parameter set is output.

[0040] The beneficial effects of this invention are as follows:

[0041] (1) Improve the voltage prediction accuracy of PEM electrolyzer model under changing operating conditions: By decoupling the temperature-sensitive parameters in the electrochemical model of PEM electrolyzer, the parameter identification method can use actual operating data under different temperature conditions for parameter identification, which significantly enhances the adaptability of the electrolyzer voltage prediction model under changing temperature conditions.

[0042] (2) Improve the voltage prediction accuracy of PEM electrolyzer model under equipment aging and failure conditions: By dynamically updating the model parameters, the problem of parameter aging caused by long-term operation of PEM electrolyzer and sudden failure caused by system parameter mutation is solved, which significantly improves the voltage prediction accuracy of PEM electrolyzer under complex conditions such as equipment aging and operation failure, and ensures that the model continuously and accurately reflects the actual operating status of electrolyzer. Attached Figure Description

[0043] Figure 1 This is a flowchart illustrating an online parameter identification method for an electrochemical model of a PEM electrolyzer, as described in an embodiment of the present invention.

[0044] Figure 2 These are the parameters to be identified and their value ranges for the temperature-decoupled electrochemical model of the PEM electrolyzer in this embodiment of the invention.

[0045] Figure 3 This is a flowchart of the particle swarm optimization algorithm used in the embodiments of the present invention;

[0046] Figure 4 This invention presents a comparison of simulated voltage and experimental voltage after parameter identification and voltage prediction for a traditional electrochemical model / temperature-decoupled electrochemical model under single / multi-temperature conditions in this embodiment of the invention. Among them, (a) is the result of parameter identification and voltage prediction for a traditional electrochemical model under a single temperature condition, (b) is the result of parameter identification and voltage prediction for a temperature-decoupled electrochemical model under a single temperature condition, (c) is the result of parameter identification and voltage prediction for a temperature-decoupled electrochemical model under different temperature conditions, and (d) is the parameter identification result.

[0047] Figure 5 This invention presents a comparison and analysis of the model's predicted voltage and the experimental voltage under actual operating conditions in this embodiment; wherein, (a) is a comparison of the predicted voltage and the experimental voltage, and (b) is the predicted voltage error. Detailed Implementation

[0048] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0049] Figure 1This is a flowchart illustrating an online parameter identification method for an electrochemical model of a PEM electrolyzer, as described in an embodiment of the present invention. The specific steps of this method include: S1, decoupling temperature-sensitive parameters in the PEM electrolyzer electrochemical model into reference parameters at a reference temperature and their corresponding temperature correction coefficients based on the Arrhenius equation; S2, constructing a temperature-decoupled PEM electrolyzer electrochemical model, identifying the parameters to be identified, and setting their reasonable value ranges; S3, collecting current, temperature, pressure, and voltage data of the PEM electrolyzer under actual operating conditions, and constructing a dataset of measurable variables for online parameter identification after processing; S4, performing parameter identification using a particle swarm optimization algorithm, and substituting the identified optimal parameter set into the temperature-decoupled PEM electrolyzer electrochemical model; S5, re-performing parameter identification based on the updated dataset of measurable variables when the preset parameter identification cycle is reached or when a fault is detected in the electrolyzer.

[0050] Specifically, based on the actual operating data of a PEM electrolyzer with a rated hydrogen production capacity of 100 standard cubic meters per hour at the Urumqi Hydrogen-Oil-Gas-Electricity Integrated Energy Station in Xinjiang on May 9, 2025 (this PEM electrolyzer was put into operation in July 2024), and the polarization curve data of five sets of different operating temperatures (60℃, 65℃, 70℃, 75℃ and 80℃) provided by the electrolyzer manufacturer, the implementation process of the online parameter identification method applicable to the electrochemical model of the PEM electrolyzer is systematically described.

[0051] S1. Based on the Arrhenius equation, the temperature-sensitive parameters in the electrochemical model of the PEM electrolyzer are decoupled into reference parameters at the reference temperature and their corresponding temperature correction coefficients.

[0052] The activation overpotential of the PEM electrolyzer can be expressed as:

[0053]

[0054] Among them, J 0,ano and J 0,cat These are the anode exchange current density and the cathode exchange current density, respectively, in A / cm². 2 This parameter is a temperature-sensitive parameter in the electrochemical model of the PEM electrolyzer. Therefore, it needs to be decoupled from the parameter at the reference temperature and its corresponding temperature correction coefficient based on the Arrhenius equation, which can be expressed as follows:

[0055]

[0056] The ohmic overpotential of a PEM electrolytic cell can be expressed as:

[0057]

[0058] Where, σ mem Let be the conductivity of PEM, expressed in S / cm. This is a temperature-sensitive parameter in the electrochemical model of the PEM electrolyzer; therefore, it needs to be decoupled based on the Arrhenius equation, and can be expressed as:

[0059]

[0060] The concentration overpotential of a PEM electrolyzer can be expressed as:

[0061]

[0062] Among them, J lim The limiting current density is expressed in A / cm². 2 This parameter is a temperature-sensitive parameter in the electrochemical model of the PEM electrolyzer. Therefore, it needs to be decoupled based on the Arrhenius equation, which can be expressed as:

[0063]

[0064] S2. Construct a temperature-decoupled electrochemical model of the PEM electrolyzer, identify the parameters to be identified, and set their reasonable value ranges.

[0065] The PEM electrolytic cell consists of several electrolysis chambers, each with an operating voltage U. cell Including open circuit voltage U ocv Activation overpotential Ohmic overpotential Concentration overpotential Therefore, the operating voltage of the PEM electrolyzer can be expressed as: (Four parts)

[0066]

[0067] Where, N stack The number of electrolysis chambers connected in series in the PEM electrolyzer;

[0068] Open circuit voltage U ocv It can be represented as:

[0069]

[0070] Where T is the operating temperature of the PEM electrolyzer in K; R is the ideal gas constant with a value of 8.314 J / (mol·K); and F is the Faraday constant with a value of 96485 C / mol. and These are the partial pressures of hydrogen at the cathode and oxygen at the anode, respectively, in atm; This represents the activity of the water between the electrode and the PEM.

[0071] The activation overpotential can be expressed as:

[0072]

[0073] Where, α ano and α cat These represent the anode charge transfer coefficient and the cathode charge transfer coefficient, respectively; J is the operating current density of the PEM electrolyzer, in A / cm². 2 J 0,ano and J 0,cat These are the anode exchange current density and the cathode exchange current density, respectively, in A / cm². 2 Based on the Arrhenius equation, after decoupling it into reference parameters at the reference temperature and their corresponding temperature correction coefficients, it can be expressed as follows:

[0074]

[0075] Among them, J 0,ano,ref and J 0,cat,ref These are the anode exchange current density and cathode exchange current density at the reference temperature (303.15 K), respectively; E exc,ano and E exc,cat These are the activation energies required for anodic electron transport and cathode electron transport, respectively.

[0076] Ohmic overpotential can be expressed as:

[0077]

[0078] Where, δ mem σ represents the PEM thickness in cm. mem The conductivity of PEM, expressed in S / cm, can be decoupled based on the Arrhenius equation and expressed as:

[0079]

[0080] Where, λ mem E represents the water content of the membrane. pro This is the activation energy for proton transport in the membrane;

[0081] Concentration overpotential can be expressed as:

[0082]

[0083] Among them, J lim The limiting current density is expressed in A / cm². 2 Based on the Arrhenius equation, after decoupling, it can be expressed as:

[0084]

[0085] Among them, J lim,ref The limiting current density at the reference temperature (303.15 K) is expressed in A / cm². 2 E diff The activation energy required for effective diffusion.

[0086] Figure 2 The parameters to be identified and their value ranges for the temperature-decoupled PEM electrolyzer electrochemical model are presented. As can be seen from the figure, the parameters to be identified include the anolyte charge transfer coefficient α. ano Cathode charge transfer coefficient α cat Anode exchange current density J at reference temperature 0,ano,ref Cathode exchange current density J at reference temperature 0,cat,ref The activation energy E required for anodic electron transport exc,ano The activation energy E required for cathode electron transport exc,cat The activation energy E for proton transport in the membrane pro Limiting current density J at reference temperature lim,ref The activation energy E required for effective diffusion diff The temperature-decoupled PEM electrolytic cell electrochemical model can adapt to fluctuations in external environments such as temperature and pressure, achieving high-precision prediction of electrolytic cell voltage.

[0087] Based on experimental and theoretical boundary constraints, the identification boundary for key parameters of the PEM electrolyzer is set as follows: the parameter to be identified is the anode charge transfer coefficient α. ano With cathode charge transfer coefficient α cat All are limited to a reasonable reaction kinetic range of 0.01 to 0.99; anodic exchange current density J at the reference temperature. 0,ano,ref Crossing 10 -10 Up to 10 -5 A / cm 2 Order of magnitude; cathode exchange current density J at reference temperature 0,cat,ref Set between 0.1 and 0.4; the activation energy E required for anodic electron transport. exc,ano Activation energy E required for cathode electron transport exc,cat The activation energy E required for effective diffusion diff All are limited to 10 2 Up to 10 5 Range, activation energy E for proton transport in the membrane pro Limiting current density J at the reference temperature, between 9000 and 30000. lim,ref Set to 2.2 to 3.3 A / cm 2 between.

[0088] S3. Collect current, temperature, pressure and voltage data of PEM electrolyzer under actual operating conditions, and construct a dataset of measurable variables for online parameter identification after processing.

[0089] In actual operation, this method synchronously collects current, temperature, hydrogen partial pressure, oxygen partial pressure, and voltage data of a PEM electrolyzer under various operating conditions using sensors. The collected data undergoes a rigorous preprocessing procedure: First, based on the inherent multi-physics coupling boundary constraints of the electrolyzer, including physical limitations such as the safe current density range, the system identifies and filters out all abnormal data points that violate these constraints in real time. Second, through steady-state condition determination, only data samples in a stable state are extracted from the collected data. This step effectively eliminates the interference of the electric double-layer effect caused by dynamic current changes on voltage measurement. Finally, the data after dual processing is constructed into a dynamically updated dataset of measurable variables for subsequent online parameter identification, laying a solid foundation for parameter identification under fluctuating operating conditions. This patent collected and processed actual operating data from a PEM electrolyzer with a rated hydrogen production capacity of 100 standard cubic meters per hour at the Urumqi Hydrogen-Oil-Gas-Electricity Integrated Energy Station in Xinjiang on May 9, 2025.

[0090] S4. The particle swarm optimization algorithm is used to perform parameter identification, and the optimal parameter set obtained by identification is substituted into the temperature-decoupled PEM electrolyzer electrochemical model.

[0091] This invention employs the particle swarm optimization algorithm to perform the key parameter identification process. With its excellent global search capability, efficient parallel computing characteristics, and low sensitivity to initial values, this algorithm is particularly suitable for solving online identification problems of multi-parameter strongly nonlinear models and can achieve efficient optimization in high-dimensional complex solution spaces. Figure 3 A flowchart of the particle swarm optimization algorithm used in this embodiment of the invention is provided. The specific implementation steps are as follows: Initialize the computational environment and import measured data of current density, temperature, pressure, and voltage; define unidentified physical constants and electrochemical model functions based on four components: open-circuit voltage, activation overpotential, ohmic overpotential, and concentration overpotential; set boundary constraints for nine parameters to be identified, and construct a weighted objective function that integrates root mean square error and maximum absolute error; configure optimization parameters such as particle swarm size, dynamic inertia weight coefficient, and individual and group recognition coefficients; perform iterative search to obtain the globally optimal parameter solution set; finally, feed the identified parameters back to the temperature-decoupled PEM electrolyzer electrochemical model in real time to update the model parameters. This mechanism ensures that the model accurately characterizes the electrochemical properties of the electrolyzer, significantly enhancing its voltage prediction accuracy and operational status response capability under renewable energy fluctuation conditions.

[0092] Based on polarization curve data at five specific operating temperatures provided by the equipment manufacturer, this invention first conducts initial parameter identification for the baseline state before the electrolyzer is put into operation. After substituting the optimal parameter set obtained from the identification into the temperature-decoupled PEM electrolyzer electrochemical model, the output voltage of the PEM electrolyzer under different temperature conditions can be predicted. To quantitatively verify the contribution of temperature decoupled modeling and parameter identification under multiple temperature conditions to improving the prediction accuracy of the PEM electrolyzer's dynamoya effect, three identification schemes are systematically compared: parameter identification of a traditional electrochemical model under a single temperature condition, parameter identification of a temperature-decoupled electrochemical model under a single temperature condition, and parameter identification of a temperature-decoupled electrochemical model under multiple temperature conditions. Figure 4 The comparison between voltage simulation and experimental data shows that using a temperature-decoupled electrochemical model improves the voltage fitting accuracy from 98.283% to 99.274%, and reduces the root mean square error from 13.950mV to 9.166mV. Furthermore, by applying data from different temperature conditions for parameter identification, the fitting accuracy of the PEM electrolyzer voltage improves from 99.274% to 99.774%, and reduces the root mean square error from 9.166mV to 5.101mV.

[0093] S5. When the preset parameter identification cycle is reached or a fault is detected in the electrolytic cell, parameter identification is re-executed based on the updated measurable variable dataset.

[0094] The parameter identification process of this invention possesses periodic dynamic update capabilities, aiming to continuously adapt to changes in the operating status of the electrolyzer. When any of the following preset trigger conditions are met, the system automatically initiates a new round of online parameter identification. One trigger condition is reaching a preset parameter update cycle: the system sets a fixed parameter identification interval, such as daily or dynamically adjusted according to the rate of change in operating conditions, and restarts the identification process upon reaching a predetermined time node. The second trigger condition is detecting an electrolyzer fault: when the status monitoring module identifies abnormal operating conditions or potential faults, typical signs including abnormal voltage fluctuations, sudden efficiency drops, and performance degradation of key components, the system will utilize the latest collected and updated measurable variable dataset to re-execute the particle swarm optimization parameter identification and model update process. This dynamic identification mechanism ensures that the system can adapt to changes in the external environment, quantify the trend of equipment performance degradation, and quickly respond to abnormal operating states, thereby significantly improving the prediction accuracy under fault conditions and providing a high-precision status assessment basis for the equipment health management system.

[0095] This patent is based on the actual operating data of a PEM electrolyzer with a rated hydrogen production capacity of 100 standard cubic meters per hour in the Urumqi Hydrogen Oil Gas Electricity Integrated Energy Station in Xinjiang on May 9, 2025, to verify the voltage prediction accuracy of online identification of electrochemical model parameters after the performance degradation caused by equipment aging. Figure 5A comparison and analysis of the model-predicted voltage and experimental voltage under actual operating conditions are presented. It can be seen that although the actual operating conditions are still affected by the electric double-layer effect, the maximum absolute error of the predicted voltage of the PEM electrolyzer is 3.259V, and the root mean square error is 556mV. Since the PEM electrolyzer consists of 62 electrolysis cells connected in series, the average root mean square error of the predicted voltage per electrolysis cell is only 8.968mV. This level of accuracy fully verifies the predictive reliability of the parameter identification method under aging conditions, providing effective technical support for performance prediction and condition assessment of electrolyzers in practical engineering applications.

[0096] Although preferred embodiments of this patent have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this patent.

[0097] Obviously, those skilled in the art can make various modifications and variations to this patent without departing from its spirit and scope. Therefore, if such modifications and variations fall within the scope of the patent claims and their equivalents, this patent also intends to include such modifications and variations.

Claims

1. A method for online parameter identification of an electrochemical model suitable for a PEM electrolyzer, characterized in that, The method includes: S1. Based on the Arrhenius equation, the temperature-sensitive parameters in the electrochemical model of the PEM electrolyzer are decoupled into reference parameters at the reference temperature and their corresponding temperature correction coefficients. S2. Construct a temperature-decoupled electrochemical model of the PEM electrolyzer, identify the parameters to be identified, and set their reasonable value ranges. S3. Collect current, temperature, pressure and voltage data of PEM electrolyzer under actual operating conditions, and construct a dataset of measurable variables for online parameter identification after processing. S4. The particle swarm optimization algorithm is used to perform parameter identification, and the optimal parameter set obtained by identification is substituted into the temperature-decoupled PEM electrolyzer electrochemical model. S5. When the preset parameter identification cycle is reached or a fault is detected in the electrolytic cell, parameter identification is re-executed based on the updated measurable variable dataset.

2. The method as described in claim 1, characterized in that: In step S1, the temperature-sensitive parameters in the electrochemical model of the PEM electrolyzer are decoupled into reference parameters at the reference temperature and their corresponding temperature correction coefficients, specifically as follows: The PEM electrolytic cell consists of several electrolysis chambers, each with an operating voltage U. cell Including open circuit voltage U ocv Activation overpotential Ohmic overpotential Concentration overpotential Four parts; Open circuit voltage U ocv Represented as: Where T is the operating temperature of the PEM electrolyzer in K; R is the ideal gas constant with a value of 8.314 J / (mol·K); and F is the Faraday constant with a value of 96485 C / mol. and These are the partial pressures of hydrogen at the cathode and oxygen at the anode, respectively, in atm; The activity of the water between the electrode and the PEM; Activation overpotential Represented as: Where, α ano and α cat These represent the anode charge transfer coefficient and the cathode charge transfer coefficient, respectively; J is the operating current density of the PEM electrolyzer, in A / cm². 2 J 0,ano and J 0,cat These are the anode exchange current density and the cathode exchange current density, respectively, in A / cm². 2 This is a temperature-sensitive parameter in the electrochemical model of the PEM electrolyzer. Based on the Arrhenius equation, it is decoupled into a reference parameter at the reference temperature and its corresponding temperature correction coefficient, which are expressed as follows: Among them, J 0,ano,ref and J 0,cat,ref These are the anode exchange current density and cathode exchange current density at a reference temperature of 303.15 K, respectively; E exc,ano and E exc,cat These are the activation energies required for anodic electron transport and cathodic electron transport, respectively. Ohmic overpotential Represented as: Where, δ mem σ represents the PEM thickness in cm. mem The conductivity of PEM, expressed in S / cm, is a temperature-sensitive parameter in the electrochemical model of the PEM electrolyzer. Decoupled from the PEM using the Arrhenius equation, it is expressed as: Where, λ mem E represents the water content of the membrane. pro This is the activation energy for proton transport in the membrane; Concentration overpotential Represented as: Among them, J lim The limiting current density is expressed in A / cm². 2 This is a temperature-sensitive parameter in the electrochemical model of the PEM electrolyzer, and after decoupling based on the Arrhenius equation, it is expressed as: Among them, J lim,ref The limiting current density at a reference temperature of 303.15 K, in A / cm². 2 E diff The activation energy required for effective diffusion.

3. The method as described in claim 2, characterized in that: The construction of the temperature-decoupled electrochemical model of the PEM electrolyzer in step S2 is specifically as follows: The operating voltage of the PEM electrolytic cell is expressed as follows: Where, N stack The number of electrolysis chambers connected in series in the PEM electrolyzer; The parameters to be identified include the anodic charge transfer coefficient α. ano Cathode charge transfer coefficient α cat Anode exchange current density J at reference temperature 0,ano,ref Cathode exchange current density J at reference temperature 0,cat,ref The activation energy E required for anodic electron transport exc,ano The activation energy E required for cathode electron transport exc,cat The activation energy E for proton transport in the membrane pro Limiting current density J at reference temperature lim,ref The activation energy E required for effective diffusion diff ; The reasonable range of values ​​for the parameters to be identified includes the anode charge transfer coefficient α. ano With cathode charge transfer coefficient α cat All are limited to a reasonable reaction kinetic range of 0.01 to 0.99; anodic exchange current density J at the reference temperature. 0,ano,ref Crossing 10 -10 Up to 10 -5 A / cm 2 Order of magnitude; cathode exchange current density J at reference temperature 0,cat,ref Set between 0.1 and 0.4; the activation energy E required for anodic electron transport. exc,ano Activation energy E required for cathode electron transport exc,cat The activation energy E required for effective diffusion diff All are limited to 10 2 Up to 10 5 Range, activation energy E for proton transport in the membrane pro Limiting current density J at the reference temperature, between 9000 and 30000. lim,ref Set to 2.2 to 3.3 A / cm 2 between.

4. The method as described in claim 1, characterized in that: The current of the PEM electrolyzer under the actual operating conditions is dynamically generated based on the predicted power generation data using a stepped curve with non-uniform time intervals. The current switching process follows the principle of smooth transition to suppress the influence of the double-layer effect on the electrolyzer voltage. The current stabilization time needs to be maintained for at least 5 minutes to ensure that reliable steady-state operating condition data are obtained in different current density ranges.

5. The method as described in claim 1, characterized in that: The processing method for PEM electrolyzer current, temperature, pressure and voltage data under actual operating conditions in step S3 includes: implementing abnormal data screening based on multi-physics field coupling boundary constraints, and eliminating double-layer effect interference by extracting steady-state operating condition data.

6. The method as described in claim 1, characterized in that: In step S4, the particle swarm optimization algorithm sets the particle swarm position vector and velocity vector during the initialization process, and configures the parameter constraint boundary, the feasible region of dynamic inertia weight coefficient, the individual cognition coefficient, and the group cognition coefficient. In each iteration, the objective function value of each particle is calculated in parallel, and the individual historical optimal solution and the group global optimal solution are updated synchronously. When the change in the objective function is lower than the convergence threshold or the number of iterations reaches the preset upper limit, the optimal parameter set is output.

Images