Electrolytic cell parallel connection method and system based on current-sharing impedance matching coupling regulation

Abstract

This application relates to the field of electrolytic hydrogen production control technology, and discloses a parallel electrolyzer method and system based on current sharing impedance matching coupling control. The method includes: acquiring the line impedance, converter output impedance, and equivalent impedance of each electrolyzer branch; establishing a branch impedance network model by series superposition; calculating impedance distribution data under different operating conditions; constructing a passive current sharing network based on impedance differences using a multi-winding coupled inductor topology; tuning the coupled inductor parameters with current sharing error and additional losses as optimization objectives; injecting a sinusoidal disturbance signal and acquiring the response waveform; calculating the source-end equivalent output impedance and the input impedance of each branch using Fourier transform; adjusting the virtual impedance compensation parameter according to the source-carrier impedance ratio; establishing an impedance coupling matrix to calculate the equivalent impedance coupling correction; and adding the two to obtain the converter control parameters. This application achieves both reduced hardware costs and improved current sharing accuracy and system stability through the coupling control of passive current sharing and active impedance matching.

Description

Technical Field

[0001] This application relates to the field of electrolytic hydrogen production control technology, and more specifically to a method and system for parallel electrolyzers based on current sharing impedance matching coupling control. Background Technology

[0002] In a wind-solar co-DC bus hydrogen production system, multiple electrolyzers of various types are typically connected in parallel on the same DC bus to increase hydrogen production scale and renewable energy consumption capacity. However, due to differences in cable length, interface converter parameters, and the impedance characteristics of each parallel branch, it is difficult to maintain consistent electrical parameters. This results in severely uneven current distribution during parallel operation, causing some electrolyzers to operate under overload conditions, experience lifespan reduction, and ultimately decrease the overall hydrogen production efficiency of the system.

[0003] Existing technologies suffer from the following problems: When using an active current sharing scheme, each branch requires an independent current sensor and a high-speed communication link. Current from each branch is collected in real-time and fed back to the centralized controller for current sharing adjustment. The system hardware cost and control complexity increase exponentially with the number of parallel branches, and communication delays can easily lead to current sharing failure during periods of drastic fluctuations in wind and solar power output. When using a fixed impedance matching scheme, the output impedance of the interface converter is set to a fixed value to maintain a stable source-carrier impedance ratio. However, the equivalent impedance of the electrolyzer drifts significantly with the operating point and cumulative operating time, making it difficult for the fixed matching parameters to meet stability criteria across the entire operating range, resulting in DC bus voltage oscillations. Furthermore, there is a lack of a unified coupling and control mechanism between the current sharing network and impedance matching. The integration of the current sharing network alters the equivalent impedance distribution of each branch, while the adjustment of impedance matching, in turn, affects the current distribution of each branch. Independent design of these two systems can easily lead to conflicting control effects. To address at least one of these problems, this application proposes a parallel electrolyzer method and system based on current sharing and impedance matching coupling control. Summary of the Invention

[0004] To address the shortcomings of existing technologies, the purpose of this application is to provide a method and system for parallel operation of electrolytic cells based on current-sharing impedance matching coupling control, which can effectively solve the problems in the background technology. The specific technical solution of this application is as follows:

[0005] A parallel electrolytic cell method based on current-sharing impedance matching coupling control, specifically including:

[0006] The line impedance, converter output impedance, and equivalent impedance of each electrolyzer branch in the hydrogen production system are obtained. The three are then connected in series and superimposed to obtain the total impedance of each branch. A branch impedance network model is established by connecting the branches in parallel and calculating the impedance distribution data of each branch under different operating conditions.

[0007] Based on the impedance difference between each branch, a multi-winding coupled inductor topology is selected. The primary winding of each branch of the multi-winding coupled inductor is connected in series with the positive bus of the corresponding branch, and the secondary winding of each branch is connected in series between the converter output terminal and the electrolytic cell input terminal of the corresponding branch. The turns ratio of the coupled inductor and the air gap length of the magnetic core are tuned with the current sharing error and additional loss as optimization objectives to construct a passive current sharing network.

[0008] Inject sinusoidal disturbance current signals into the DC bus and each branch, collect voltage and current response waveforms, extract the amplitude and phase corresponding to the disturbance frequency through Fourier transform, and calculate the equivalent output impedance of the source end and the input impedance of each branch.

[0009] The source-carrier impedance ratio is calculated based on the equivalent output impedance of the source and the input impedance of each branch. The virtual impedance compensation parameters of each branch interface converter are adjusted with the preset stability criterion as a constraint. An impedance coupling matrix is ​​established based on the self-inductance and mutual inductance parameters of the multi-winding coupled inductor. The equivalent impedance coupling correction amount of each branch is calculated. The equivalent impedance coupling correction amount is added to the virtual impedance compensation parameters to obtain the converter control parameters.

[0010] Specifically, establishing the branch impedance network model includes:

[0011] For each parallel branch of the electrolytic cell, the cable resistance and cable inductance from the DC bus to the input end of the electrolytic cell are equivalent to the line impedance, and the filter capacitor reactance and parasitic resistance at the output end of the interface converter are equivalent to the converter output impedance.

[0012] A working point is selected on the polarization characteristic curve of the electrolyzer. The slope of the tangent at the working point is taken as the equivalent resistance of the electrolyzer at the working point. The slope of the tangent is multiplied by a preset correction factor to obtain the equivalent impedance of the electrolyzer. The preset correction factors of the alkaline electrolyzer and the proton exchange membrane electrolyzer are determined according to the slope change rate of their respective polarization curves near the working point.

[0013] The total impedance of a branch is obtained by adding the line impedance, converter output impedance, and electrolytic cell equivalent impedance of the same branch. The total impedances of all branches are arranged in parallel to establish a branch impedance network model.

[0014] Specifically, the calculation of impedance distribution data for each branch under different operating conditions includes:

[0015] Based on the fluctuation range of the DC bus voltage, select multiple steady-state operating points, recalculate the equivalent impedance of the electrolytic cell of each branch at each steady-state operating point, remeasure the output impedance of the converter, and obtain the total impedance of each branch corresponding to that operating point.

[0016] The total impedance of each branch corresponding to all operating points is summarized to form impedance distribution data under different operating conditions.

[0017] When the cumulative operating time of the electrolytic cell exceeds the preset aging threshold, the polarization curve of the electrolytic cell is remeasured, the tangent slope at each working point is updated, the updated tangent slope is multiplied by the preset correction coefficient to obtain the updated equivalent impedance of the electrolytic cell, and the impedance distribution data is recalculated.

[0018] Specifically, the tuning of the coupling inductor turns ratio and core air gap length with current sharing error and additional loss as optimization targets includes:

[0019] The current sharing error is defined as the maximum deviation between the current in each branch and the average current of all branches divided by the average current. The additional loss is defined as the sum of the copper loss of each winding of the coupled inductor and the iron loss of the magnetic core.

[0020] The ratio of the number of turns in the secondary winding to the number of turns in the primary winding of each branch of the coupled inductor and the length of the air gap in the magnetic core are used as decision variables. With the minimization of current sharing error and the minimization of additional loss as optimization objectives, a multi-objective optimization algorithm is used to optimize the decision variables and obtain the Pareto front.

[0021] The solution with a current sharing error less than or equal to a preset error threshold and an additional loss less than or equal to a preset loss threshold is selected from the Pareto front and used as the tuning result of the coupled inductor parameters.

[0022] If there is no solution in the Pareto front that simultaneously satisfies both threshold conditions, then the solution with the smallest flow-averaging error is selected as the tuning result.

[0023] Specifically, the process of injecting sinusoidal disturbance current signals into the DC bus and each branch, acquiring voltage and current response waveforms, extracting the amplitude and phase corresponding to the disturbance frequency through Fourier transform, and calculating the equivalent output impedance at the source and the input impedance of each branch includes:

[0024] At the DC bus end, by adjusting the modulation wave of the interface converter and superimposing a sinusoidal disturbance component with a preset frequency and preset amplitude, a sinusoidal disturbance current signal with a corresponding frequency is generated in the DC bus current, and the voltage and current response waveforms of the DC bus are acquired simultaneously.

[0025] Fast Fourier transform is performed on the acquired voltage and current response waveforms respectively, and the voltage amplitude, current amplitude, and phase difference between voltage and current at the corresponding frequency points of the sinusoidal disturbance current signal are extracted from the transformation results.

[0026] Divide the voltage amplitude by the current amplitude to obtain the impedance amplitude, and use the phase difference as the impedance phase angle. The source-side equivalent output impedance is formed by the impedance amplitude and the impedance phase angle.

[0027] A sinusoidal disturbance current signal with the same frequency as the source is applied to the input terminal of each branch interface converter. The voltage and current response waveforms of each branch input terminal are collected synchronously, and the input impedance of each branch is calculated in the same way.

[0028] Specifically, adjusting the virtual impedance compensation parameters of each branch interface converter based on a preset stability criterion includes:

[0029] Divide the magnitude of the equivalent output impedance at the source end by the magnitude of the input impedance of each branch to obtain the source-carrier impedance ratio of each branch.

[0030] The constraint condition is that the source-carrier impedance ratio is less than a preset proportional threshold throughout the entire frequency band. The constraint condition is used to determine whether each branch meets the constraint condition.

[0031] For branches that do not meet the constraints, the virtual resistance and virtual reactance values ​​of the interface converter of the branch are adjusted by a preset step size. After each adjustment, the source-carrier impedance ratio of the branch is recalculated until the source-carrier impedance ratio is less than the preset ratio threshold in the entire frequency band. The virtual resistance and virtual reactance values ​​at this time are used as the virtual impedance compensation parameters of the branch.

[0032] Specifically, establishing the impedance coupling matrix based on the mutual inductance parameters of the current sharing network includes:

[0033] Obtain the self-inductance value of the secondary winding of each branch of the multi-winding coupled inductor, as well as the mutual inductance value between any two branches of the secondary winding;

[0034] Multiply the self-inductance value of the secondary winding of the i-th branch by the angular frequency and place it in the i-th row and i-th column of the impedance coupling matrix;

[0035] Multiply the mutual inductance values ​​of the secondary windings of the i-th branch and the j-th branch by the angular frequency, and place them in the i-th row and j-th column of the impedance coupling matrix, where i is not equal to j;

[0036] The complete impedance coupling matrix is ​​obtained by filling in rows and columns one by one. The diagonal elements of the impedance coupling matrix reflect the self-impedance of the current-sharing inductors of each branch, and the off-diagonal elements reflect the mutual impedance between the current-sharing inductors of different branches.

[0037] Specifically, the calculation of the equivalent impedance coupling correction for each branch includes:

[0038] Multiply the impedance coupling matrix by the current column vector of each branch to obtain the coupling voltage correction column vector of each branch. Each element in the coupling voltage correction column vector represents the additional voltage generated in the corresponding branch due to the mutual inductance coupling of the current sharing inductance of other branches.

[0039] Divide each element in the coupling voltage correction column vector by the current value of the branch corresponding to that element to obtain the equivalent impedance coupling correction amount of that branch. The equivalent impedance coupling correction amount represents the correction value of the equivalent impedance of the branch by the mutual inductance coupling of the current sharing network.

[0040] Specifically, after generating and issuing converter control parameters, multi-mode switching control is also included:

[0041] Real-time detection of DC bus voltage fluctuation amplitude and current imbalance of each branch, wherein the current imbalance is the maximum deviation of each branch current from the average current of all branches divided by the average current.

[0042] When the bus voltage fluctuation is lower than the preset fluctuation threshold and the current imbalance is lower than the preset balance threshold, each branch interface converter maintains the current control parameters and the current is shared only by the passive current sharing network.

[0043] When the bus voltage fluctuation exceeds the preset fluctuation threshold or the current imbalance exceeds the preset balancing threshold, the impedance online identification is restarted, the source-carrier impedance ratio and virtual impedance compensation parameters are recalculated, the equivalent impedance coupling correction is recalculated, the updated equivalent impedance coupling correction is added to the updated virtual impedance compensation parameters, the updated converter control parameters are generated and sent for execution.

[0044] A parallel electrolytic cell system based on current-sharing impedance matching coupling control, specifically including:

[0045] The impedance network analysis module is used to obtain the line impedance, converter output impedance and electrolytic cell equivalent impedance of each electrolytic cell branch. The three are connected in series and superimposed to obtain the total impedance of each branch. The branch impedance network model is established according to the parallel relationship, and the impedance distribution data of each branch under different operating conditions is calculated.

[0046] The passive current sharing network module is used to select the topology of the multi-winding coupled inductor based on the impedance difference between each branch. The primary winding of each branch of the multi-winding coupled inductor is connected in series with the positive bus of the corresponding branch, and the secondary winding of each branch is connected in series between the converter output terminal and the electrolytic cell input terminal of the corresponding branch. The turns ratio of the coupled inductor and the air gap length of the magnetic core are tuned with the current sharing error and additional loss as optimization targets.

[0047] The online impedance identification module is used to inject sinusoidal disturbance current signals into the DC bus and each branch, collect voltage and current response waveforms, extract the amplitude and phase corresponding to the disturbance frequency through Fourier transform, and calculate the equivalent output impedance of the source end and the input impedance of each branch.

[0048] The coupling control module is used to calculate the source-carrier impedance ratio based on the equivalent output impedance of the source end and the input impedance of each branch, adjust the virtual impedance compensation parameters of each branch interface converter with a preset stability criterion as a constraint, establish an impedance coupling matrix based on the self-inductance and mutual inductance parameters of the multi-winding coupled inductor and calculate the equivalent impedance coupling correction amount of each branch, add the equivalent impedance coupling correction amount to the virtual impedance compensation parameters to obtain the converter control parameters and send them down for execution.

[0049] The beneficial effects of this application are as follows: By establishing a branch impedance network model to analyze the impedance differences of each branch, a precise basis is provided for the parameter tuning of the passive current sharing network; passive current sharing is achieved by using a multi-winding coupled inductor topology, eliminating the need for independent current sensors and high-speed communication links in each branch, thus reducing system hardware costs; the source end and branch impedance are identified online by injecting sinusoidal disturbance signals, adapting to the effects of changes in the electrolytic cell's operating point and long-term aging; the coupling correction amount of the current sharing network to the equivalent impedance of each branch is calculated through the impedance coupling matrix, and the coupling correction amount is jointly controlled with the virtual impedance compensation parameters, solving the control conflict problem caused by the independent design of the current sharing network and impedance matching, and achieving a balance between current sharing accuracy and system stability. Attached Figure Description

[0050] Figure 1 This is a flowchart illustrating the process of the parallel electrolytic cell method based on current-sharing impedance matching coupling control in the embodiments of this application.

[0051] Figure 2 This is a schematic diagram of the branch impedance network model and the passive current sharing network structure in the embodiments of this application;

[0052] Figure 3 This is a schematic diagram of the online impedance identification and coupling control process in the embodiments of this application;

[0053] Figure 4 This is a schematic diagram of the structure of the parallel electrolytic cell system based on current sharing impedance matching coupling control in the embodiments of this application. Detailed Implementation

[0054] The present application will be further described in detail below with reference to the accompanying drawings and embodiments.

[0055] In the embodiments of this application, the terms "exemplary" or "for example" are used to indicate that something is an example, illustration, or description. Any embodiment or design that is described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design. Specifically, the use of the terms "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.

[0056] Hereinafter, the terms "first," "second," and other generic terms are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of the embodiments of this application, unless otherwise stated, "multiple" means two or more.

[0057] like Figure 1 The specific implementation of the parallel electrolytic cell method based on current sharing impedance matching coupling control in this application includes:

[0058] S101. Obtain the line impedance, converter output impedance, and equivalent impedance of each electrolyzer branch in the hydrogen production system. Connect these three in series to obtain the total impedance of each branch. Establish a branch impedance network model by connecting them in parallel and calculate the impedance distribution data of each branch under different operating conditions.

[0059] S102. Based on the impedance difference between each branch, a multi-winding coupled inductor topology is selected. The primary winding of the coupled inductor is connected in series with the positive bus of each branch, and the secondary winding is connected in series between the output of the converter and the input of the electrolytic cell. The turns ratio of the coupled inductor and the air gap length of the magnetic core are tuned with the current sharing error and additional loss as optimization objectives to construct a passive current sharing network.

[0060] S103. Inject sinusoidal disturbance current signals into the DC bus and each branch, collect voltage and current response waveforms, extract the amplitude and phase corresponding to the disturbance frequency through Fourier transform, and calculate the equivalent output impedance of the source end and the input impedance of each branch.

[0061] S104. Calculate the source-carrier impedance ratio based on the equivalent output impedance of the source end and the input impedance of each branch. Adjust the virtual impedance compensation parameters of each branch interface converter with the preset stability criterion as a constraint. Establish the impedance coupling matrix based on the mutual inductance parameters of the current sharing network. Calculate the equivalent impedance coupling correction amount of each branch. Add the equivalent impedance coupling correction amount to the virtual impedance compensation parameters to obtain the converter control parameters.

[0062] Currently, in wind-solar shared DC bus hydrogen production systems, multiple electrolyzers typically operate in parallel on the same DC bus. Differences in cable length, interface converter parameters, and the electrolyzer's own impedance characteristics among the parallel branches lead to uneven current distribution. Existing active current sharing solutions require independent current sensors and high-speed communication links for each branch, causing system hardware costs to increase exponentially with the number of parallel branches. Furthermore, communication delays can easily lead to current sharing failures during periods of significant fluctuation in wind and solar power output. This solution constructs a passive current sharing network using multi-winding coupled inductors, automatically balancing current among branches using electromagnetic coupling principles, eliminating the need for current sensors and communication links. Simultaneously, it uses a sinusoidal perturbation injection method to identify the impedance of each branch online, establishing an impedance coupling matrix to calculate the coupling correction amount of the current sharing network on the equivalent impedance of each branch. This coupling correction amount is then jointly controlled with virtual impedance compensation parameters, resolving the problem of asynchronous impedance matching updates due to coupling changes in the equivalent impedance of each branch after the current sharing network is connected.

[0063] In this embodiment, step S101 requires establishing a branch impedance network model and quantitatively analyzing the impedance distribution of each branch. The basic structure of the wind-solar shared DC bus hydrogen production system is as follows: the wind turbine generator and the photovoltaic array are connected to the DC bus via AC / DC converters and DC / DC converters, respectively. The DC bus voltage is maintained near a preset rated value. Multiple electrolyzer branches are connected in parallel to the DC bus. Each branch draws power from the DC bus, and after being stepped down or stepped up by the interface DC / DC converter, it supplies power to the electrolyzers. The type of electrolyzers can be all alkaline electrolyzers, all proton exchange membrane electrolyzers, or a mixture of alkaline and proton exchange membrane electrolyzers connected in parallel. Different types of electrolyzers have different polarization and impedance characteristics, which impose different requirements on current sharing and impedance matching.

[0064] Specifically, when modeling the impedance of each parallel branch of an electrolytic cell, three parts need to be considered. The first part is the line impedance. The connecting cables from the DC bus to the input end of the electrolytic cell have a certain resistance and inductance. The cable resistance depends on the cable material, cross-sectional area, and length, while the cable inductance depends on the cable wiring method and geometry. In actual engineering, due to the different installation positions of each electrolytic cell, the cable lengths of each branch may vary significantly. For example, the cables of electrolytic cell branches closer to the DC bus output end are shorter and have lower line impedance; the cables of electrolytic cell branches farther from the output end are longer and have higher line impedance. The difference in cable resistance directly leads to different voltage drops in each branch under the same bus voltage, thus affecting the current distribution. The cable resistance can be measured branch by branch using a DC resistance tester under power-off conditions, and the cable inductance can be measured using an LCR tester under AC excitation.

[0065] Specifically, the second part is the output impedance of the interface converter. Interface DC / DC converters typically employ buck or buck-boost topologies, with filter capacitors and filter inductors connected to their output terminals. Under steady-state DC conditions, the capacitive reactance of the filter capacitor tends to infinity, and its contribution to the output impedance is negligible. However, during dynamic processes, the equivalent series resistance of the filter capacitor and the DC resistance of the filter inductor constitute the main components of the converter's output impedance. Furthermore, the closed-loop output impedance introduced by the interface converter's control loop is also an important component of the total output impedance. The closed-loop output impedance is related to the converter's control strategy and controller parameters and can be obtained through theoretical derivation or experimental measurement. For converters with current closed-loop control, their output impedance typically exhibits a large inductive characteristic; for converters with voltage closed-loop control, their output impedance typically exhibits a small resistive characteristic.

[0066] The third part is the equivalent impedance of the electrolyzer. The current-voltage characteristics of the electrolyzer are described by the polarization curve, which reflects the terminal voltage of the electrolyzer at different current densities. In the low current density region, the voltage rises steeply with increasing current, mainly dominated by electrode activation polarization. In the medium current density region, the voltage rises linearly with current, mainly dominated by ohmic polarization. The slope of this segment is mainly determined by the electrolyte resistance, diaphragm resistance, and contact resistance. In the high current density region, the voltage rise accelerates again, mainly dominated by concentration polarization. Within the normal operating range, the electrolyzer typically operates in the linear segment dominated by ohmic polarization. Therefore, the polarization curve can be linearized near the operating point, and the tangent slope at the operating point can be used as the equivalent resistance of the electrolyzer at that operating point.

[0067] Specifically, the linearization process involves selecting the current operating point on the polarization curve and taking the tangent line at that point. The slope of the tangent line represents the rate of change of the terminal voltage when the electrolyzer current undergoes a small change near the operating point. This slope has the dimension of resistance and is defined as the equivalent resistance of the electrolyzer at that operating point. For alkaline electrolyzers, due to the high concentration of potassium hydroxide solution in the electrolyte, the ohmic internal resistance is relatively large, and the polarization curve is steeper in the ohmic polarization section. The equivalent resistance is typically in the range of tens to hundreds of milliohms, depending on the electrode area, electrode spacing, and electrolyte temperature of the electrolyzer. For proton exchange membrane electrolyzers, since perfluorosulfonic acid proton exchange membranes are used as the electrolyte, the membrane resistance is the main component of the ohmic internal resistance, and the equivalent resistance is typically in the range of several milliohms to tens of milliohms.

[0068] Furthermore, considering the nonlinearity of the polarization curve, the tangent slope is only effective within a small range near the operating point. When the DC bus voltage deviates significantly due to fluctuations in wind and solar power output, the operating point of the electrolyzer shifts accordingly, and the tangent slope also changes. Therefore, a preset correction factor needs to be multiplied when selecting the operating point to compensate for the deviation caused by nonlinearity. The value of the correction factor is determined based on the type of electrolyzer and the specific shape of the polarization curve. Alkaline electrolyzers exhibit better linearity in the ohmic polarization range, and the correction factor is typically taken as close to one. For proton exchange membrane electrolyzers, due to the nonlinear relationship between membrane resistance and current density, the correction factor can be appropriately relaxed.

[0069] The total impedance of a branch is obtained by adding the line impedance, converter output impedance, and electrolytic cell equivalent impedance together. Note that "addition" here refers to impedance series connection; that is, the three are connected in series sequentially within the same branch. Current flows sequentially through the line impedance, converter output impedance, and electrolytic cell equivalent impedance, therefore the total branch impedance equals the sum of the three. All branch total impedances are arranged in parallel to establish a branch impedance network model. This model describes the impedance coupling relationships between branches and forms the basis for subsequent current sharing network design and impedance matching control.

[0070] Specifically, calculating the impedance distribution data of each branch under different operating conditions is a crucial step after establishing the impedance network model. Due to the randomness and intermittency of wind and solar power output, the DC bus voltage fluctuates around its rated value during actual operation. This fluctuation changes the operating point of the electrolyzer, thus altering its equivalent impedance. Therefore, it is necessary to select multiple steady-state operating points based on the fluctuation range of the DC bus voltage. At each steady-state operating point, the equivalent impedance of each branch is recalculated, and the converter output impedance is remeasured to obtain the total impedance of each branch corresponding to that operating point. The total impedances of all branches corresponding to all operating points are then summarized to form the impedance distribution data under different operating conditions. The impedance distribution data is stored in a two-dimensional table, with row indices representing operating condition numbers and column indices representing branch numbers. The data in the table represents the total impedance value of the corresponding branch under the corresponding operating condition. The impedance distribution data can be used to analyze the variation patterns of impedance differences among branches under different operating conditions, providing constraints for the parameter design of the passive current sharing network.

[0071] Specifically, electrolyzers experience an aging effect during long-term operation. In alkaline electrolyzers, aging is primarily manifested as decreased electrode catalyst activity and membrane performance degradation, leading to an overall upward shift in the polarization curve and an increase in equivalent resistance. In proton exchange membrane electrolyzers, aging is mainly characterized by membrane chemical degradation and mechanical damage, resulting in increased membrane resistance and hydrogen permeation. This aging effect renders the polarization curve calibrated at the factory inaccurate after a certain period of operation, causing a deviation between the calculated equivalent impedance of the electrolyzer based on the factory polarization curve and the actual value. Therefore, when the cumulative operating time of the electrolyzer exceeds the preset aging threshold, it is necessary to remeasure the polarization curve, update the tangent slope at each operating point, multiply the updated tangent slope by a preset correction factor to obtain the updated equivalent impedance of the electrolyzer, and recalculate the impedance distribution data. The remeasurement of the polarization curve can be performed during planned downtime maintenance or through online scanning measurements using the electrolyzer's built-in voltage and current sensors.

[0072] In step S102, a passive current sharing network needs to be designed based on the impedance differences between the branches. The basic principle of passive current sharing is to establish magnetic coupling channels between the branches using electromagnetic coupling devices. When the current in a certain branch is larger due to its smaller impedance, the magnetic flux generated in the coupling device in that branch is stronger. Through mutual inductance, voltages are induced in other branches, automatically adjusting the equivalent impedance of each branch and making the current in each branch tend to be balanced. Multi-winding coupled inductors are an effective topology for realizing passive current sharing across multiple branches.

[0073] Specifically, a multi-winding coupled inductor consists of a common magnetic core and multiple windings. The primary winding is connected in series with the positive busbar of each branch, meaning that the positive busbars of all branches share the same conductor passing through the primary winding of the multi-winding coupled inductor. More precisely, each branch has a corresponding primary winding in the coupled inductor, and all primary windings are wound on the same magnetic core. The secondary winding is connected in series between the converter output terminal and the electrolytic cell input terminal of the corresponding branch. That is, one end of the secondary winding of each branch is connected to the positive output terminal of the interface converter, and the other end is connected to the positive input terminal of the electrolytic cell. The equivalent circuit of this connection method is: the primary and secondary windings of each branch form a transformer, with the primary side connected in series with the busbar side and the secondary side connected in series with the load side. When the currents in each branch are inconsistent, the magnetomotive force of the primary winding of each branch is unbalanced, generating alternating magnetic flux in the magnetic core. The alternating magnetic flux induces voltage in the secondary winding of each branch. The direction of this voltage opposes the change in current imbalance. That is, the branch with a larger current induces a reverse voltage, which reduces the equivalent voltage of the branch, while the branch with a smaller current induces a positive voltage, which increases the equivalent voltage of the branch. This achieves automatic balance of current in each branch.

[0074] Specifically, the current-sharing effect of multi-winding coupled inductors depends on the turns ratio and core parameters. The turns ratio determines the coupling strength between branches; a larger turns ratio results in stronger coupling and better current-sharing capability, but also increases additional losses. The air gap length of the core determines the magnetization characteristics and saturation current of the core; a larger air gap makes the core less prone to saturation, but results in a smaller magnetizing inductance and poorer coupling effect. Therefore, it is necessary to optimize the turns ratio and air gap length of the coupled inductor to achieve a balance between current-sharing accuracy and additional losses.

[0075] Specifically, the optimization tuning targets current sharing error and additional losses. Current sharing error is defined as the maximum deviation between the current in each branch and the average current of all branches, divided by the average current. This indicator reflects the degree of non-uniformity of the current in each branch. The smaller the current sharing error, the better the current sharing effect. Additional losses are defined as the sum of copper losses and core iron losses in each winding of the coupled inductor. Copper losses are Joule heat losses caused by the resistance of the winding conductors, and are proportional to the square of the current in the winding and the winding resistance. Iron losses are the hysteresis loss and eddy current loss of the core in the alternating magnetic field, and are related to the magnetization frequency, magnetic flux density amplitude, and loss characteristics of the core material. The greater the additional losses, the lower the energy conversion efficiency of the system.

[0076] Specifically, the ratio of the number of turns in the secondary winding to the number of turns in the primary winding of each branch of the coupled inductor, along with the air gap length of the magnetic core, are used as decision variables. The turns ratio determines the coupling coefficient between each branch. Different branches have different impedance differences, and their turns ratios can be set independently. Branches with higher impedance can have a larger turns ratio to achieve stronger compensation. The air gap length is a parameter shared by all branches because all windings share the same magnetic core. With the optimization objectives of minimizing current sharing error and minimizing additional losses, a multi-objective optimization algorithm is used to optimize the decision variables. The multi-objective optimization algorithm can be based on NSGA-II (genetic algorithm) or particle swarm optimization. The result of multi-objective optimization is not a single optimal solution, but a set of non-dominated solutions distributed on the Pareto front. Each solution on the Pareto front is better on one objective and worse on another. Decision-makers can select the most suitable compromise solution based on actual engineering needs.

[0077] Specifically, the principle for selecting the coupling inductor parameter tuning result from the Pareto front is as follows: Prioritize solutions where the current sharing error is less than or equal to a preset error threshold and the additional loss is less than or equal to a preset loss threshold. The preset error threshold is determined based on the system's requirements for current sharing accuracy, and the preset loss threshold is determined based on the system's upper tolerance limit for additional loss. If a solution exists in the Pareto front that simultaneously satisfies both threshold conditions, the solution with the minimum additional loss is selected as the tuning result, minimizing energy loss while meeting the current sharing accuracy requirements. If no solution exists in the Pareto front that simultaneously satisfies both threshold conditions, the solution with the minimum current sharing error is prioritized as the tuning result, because current sharing is the system's primary task, and additional losses can be compensated for by adjusting other parameters during subsequent operation.

[0078] like Figure 2 As shown in the diagram, the branch impedance network model and the passive current sharing network structure are illustrated. The left side represents the branch impedance network model, with parallel branches arranged from top to bottom. Each branch consists of the line impedance, converter output impedance, and electrolytic cell equivalent impedance connected in series. The input terminals of all branches are connected in parallel to the DC bus, and the output terminals are connected to their respective electrolytic cells. The diagram uses resistor symbols of different lengths to represent the differences in impedance between branches. The right side represents the passive current sharing network structure. The primary winding of the multi-winding coupled inductor is connected in series to the positive bus of each branch, and the secondary winding is connected in series between the converter output terminal and the electrolytic cell input terminal. All windings share a single toroidal core. The number of turns and winding direction of each winding are marked on the core, and the terminals with the same name are clearly marked.

[0079] In step S103, it is necessary to identify the equivalent output impedance of the source-end DC bus and the input impedance of each electrolytic cell branch online. The purpose of online impedance identification is to obtain the current impedance state of the system in real time, providing accurate parameter basis for impedance matching control. This scheme uses the sinusoidal disturbance injection method for online impedance identification. The basic principle of the sinusoidal disturbance injection method is: inject a sinusoidal disturbance current signal with a known frequency and amplitude into the system under test, simultaneously acquire the voltage and current response waveforms at the ports of the system under test, extract the voltage amplitude, current amplitude, and phase difference between them corresponding to the disturbance frequency from the response waveform through signal processing algorithms, and calculate the equivalent impedance of the system under test at the disturbance frequency according to the complex form of Ohm's law. By repeatedly injecting disturbance signals at multiple frequency points, the impedance characteristics of the system under test across the entire frequency band can be obtained.

[0080] Specifically, at the DC bus end, a sinusoidal disturbance current signal is generated by adjusting the modulation wave of the interface converter. The interface converter typically uses PWM modulation, and the modulation wave determines the magnitude of the output voltage or output current. A sinusoidal disturbance component with a preset frequency and amplitude is superimposed on the modulation wave. After PWM modulation, the converter's output current will contain this sinusoidal disturbance component at that frequency. The disturbance frequency needs to be selected to avoid the system's inherent resonant frequency and control bandwidth, and is usually chosen between tens of hertz and hundreds of hertz. The disturbance amplitude needs to be selected by balancing the signal-to-noise ratio and the disturbance's impact on the system. If the amplitude is too small, the signal is easily overwhelmed by noise, leading to a decrease in identification accuracy; if the amplitude is too large, it will cause non-negligible interference to the normal operation of the system. The disturbance amplitude is usually taken as a few percent of the rated current.

[0081] Specifically, synchronously acquiring the voltage and current response waveforms of the DC bus is a crucial step in impedance identification. The sampling frequency of voltage and current should be much higher than the disturbance frequency to meet the requirements of the Nyquist sampling theorem; typically, the sampling frequency is tens of times higher than the disturbance frequency. The sampling duration should cover multiple complete disturbance cycles to ensure the frequency resolution of the spectrum analysis. The acquired voltage and current response waveforms contain DC components, disturbance frequency components, inherent system harmonic components, and random noise. The time-domain waveform is transformed to the frequency domain using a Fast Fourier Transform (FFT), and the voltage amplitude, current amplitude, and phase information corresponding to the disturbance frequency are extracted from the spectrum. A Hanning window or Hamming window is chosen for the FFT to reduce spectral leakage. After the transformation, the frequency point where the disturbance frequency is located is found, and the complex voltage amplitude and complex current amplitude at that frequency point are obtained. The impedance amplitude is obtained by dividing the magnitude of the voltage complex amplitude by the magnitude of the current complex amplitude, and the impedance phase angle is obtained by subtracting the phase angle of the current complex amplitude from the phase angle of the voltage complex amplitude. The source-side equivalent output impedance is then represented by a complex number using the impedance amplitude and the impedance phase angle. The impedance magnitude reflects the source's ability to impede changes in current, while the impedance phase angle reflects the inductive and capacitive characteristics of the source.

[0082] Specifically, sinusoidal perturbation current signals are injected into the input terminals of each branch interface converter in the same manner as at the source terminal, with the perturbation frequency consistent with the injection frequency at the source terminal to ensure the comparability of the impedance identification results for each branch. The voltage and current response waveforms at the input terminals of each branch are simultaneously acquired, and the input impedance of each branch is obtained by performing the same Fast Fourier Transform and extraction operations as at the source terminal. The input impedance of each branch reflects its response characteristics to changes in external current.

[0083] like Figure 3As shown in the schematic diagram of the online impedance identification and coupling control process, the left side represents the online impedance identification process. Starting with the superposition of sinusoidal disturbance components, the process proceeds sequentially as follows: synchronous acquisition of voltage and current waveforms, fast Fourier transform spectrum analysis, extraction of disturbance frequency amplitude and phase, calculation of impedance amplitude and phase angle, and finally output of the equivalent output impedance at the source and the input impedance of each branch. The right side represents the coupling control process. Starting with reading the impedance at the source and each branch, the process proceeds sequentially as follows: calculation of the source-carrier impedance ratio, determination of whether the stability criterion is met, adjustment of virtual impedance compensation parameters, establishment of the impedance coupling matrix, calculation of the equivalent impedance coupling correction, addition of the correction to the compensation parameters, and finally output of the converter control parameters. The data transmission relationship between the two processes is indicated by dashed lines and arrows.

[0084] In step S104, impedance matching and coupling control are performed based on the identified impedance data. The purpose of impedance matching is to ensure the matching between the source output impedance and the load input impedance, preventing DC bus voltage oscillations caused by impedance mismatch. In a DC microgrid, if the source output impedance does not match the load input impedance, the DC bus voltage will ring or oscillate when the load undergoes a step change, affecting system stability and power quality. The Middlebrook stability criterion is a widely used stability criterion in DC microgrids. Its core condition is that the magnitude of the source output impedance is less than the magnitude of the load input impedance across the entire frequency band, i.e., the source-to-load impedance ratio is less than one. In practical engineering, to allow sufficient safety margin, the source-to-load impedance ratio is usually required to be less than a preset proportional threshold across the entire frequency band.

[0085] Specifically, the source-carrier impedance ratio (RCR) of each branch is obtained by dividing the magnitude of the equivalent output impedance at the source by the magnitude of the input impedance of each branch. The RCR is a function of frequency and needs to be verified across the entire frequency band. The full frequency band covers a range from DC to near the switching frequency, and the impedance characteristics across the entire frequency band can be approximated by impedance identification at multiple disturbance frequency points. Using the constraint that the RCR is less than a preset threshold across the entire frequency band as a constraint, each branch is checked to see if it meets the constraint. For branches that meet the constraint, it indicates that the source and load impedances of that branch are well matched and no adjustment is needed. For branches that do not meet the constraint, it indicates that there is a risk of impedance mismatch, and the virtual impedance compensation parameters of the interface converter need to be adjusted to improve the matching.

[0086] Specifically, virtual impedance compensation involves virtually connecting impedance elements in series or parallel at the converter output through the interface converter's control algorithm, thereby changing the converter's equivalent output impedance or the load's equivalent input impedance. Virtual impedance compensation does not require actual resistors, inductors, or capacitors; instead, it achieves this by changing the reference values ​​or feedback coefficients of the converter's control loop, offering advantages such as flexibility, adjustability, and no increase in hardware cost. When the source-carrier impedance ratio does not meet the constraints, the virtual resistance and virtual reactance values ​​of the interface converter are adjusted by a preset step size. The adjustment direction is to decrease the source-carrier impedance ratio. After each adjustment, the source-carrier impedance ratio of that branch is recalculated until it is less than a preset proportional threshold across the entire frequency band. The virtual resistance and virtual reactance values ​​that meet the constraints are used as the virtual impedance compensation parameters for that branch. This iterative adjustment process is essentially an online search for the optimal virtual impedance parameters, and the selection of the search step size requires a trade-off between convergence speed and adjustment accuracy.

[0087] Specifically, the introduction of a passive current-sharing network alters the equivalent impedance distribution of each branch. In multi-winding coupled inductors, mutual inductance exists between the secondary windings of each branch. A change in current in one branch induces voltage in other branches through mutual inductance, thus affecting the equivalent impedance of those branches. If the coupling effect of the current-sharing network is ignored during impedance matching, and the virtual impedance is adjusted only based on the uncoupled branch input impedance, the matching results will deviate in actual operation, potentially leading to matching failure. Therefore, the impedance coupling effect of the current-sharing network needs to be incorporated into impedance matching considerations, establishing an impedance coupling matrix and calculating the coupling correction amount of the current-sharing network on the equivalent impedance of each branch.

[0088] Specifically, the impedance coupling matrix is ​​established based on the self-inductance and mutual inductance parameters of a multi-winding coupled inductor. Each branch of the secondary winding of the multi-winding coupled inductor has a self-inductance value, and there is a mutual inductance value between any two branch secondary windings. The magnitude of the mutual inductance depends on the turns ratio of the two windings and the coupling coefficient of the magnetic core; the larger the mutual inductance, the tighter the coupling between the two windings. The self-inductance and mutual inductance parameters can be obtained through finite element electromagnetic field simulation or by measuring each winding of the actual coupled inductor using an LCR meter. After obtaining all the self-inductance and mutual inductance parameters, the impedance coupling matrix is ​​filled row by row and column by column.

[0089] Specifically, in an AC circuit, the impedance of an inductor is its self-inductance multiplied by its angular frequency, and the impedance of a mutual inductance is its mutual inductance multiplied by its angular frequency. Although this system is a DC system, the rate of change of current is not zero during dynamic processes, and inductors and mutual inductances will generate induced voltages, which are equivalent to AC impedance. Multiplying the self-inductance of the secondary winding of the i-th branch by its angular frequency and placing it in the i-th row and i-th column of the impedance coupling matrix yields diagonal elements, which reflect the self-impedance of the current-sharing inductors in each branch. Multiplying the mutual inductance of the secondary windings of the i-th and j-th branches by their angular frequencies and placing it in the i-th row and j-th column of the impedance coupling matrix (where i is not equal to j) yields off-diagonal elements, which reflect the mutual impedance between the current-sharing inductors of different branches. Filling the matrix row by row and column by column yields the complete impedance coupling matrix, where the number of rows and columns equals the number of parallel branches.

[0090] Multiplying the impedance coupling matrix by the current column vectors of each branch yields the coupling voltage correction column vector for each branch. The current column vectors consist of the current values ​​of each branch. Each element in the coupling voltage correction column vector represents the additional voltage induced in the corresponding branch by the mutual inductance coupling of current-sharing networks with other branches. The magnitude of the additional voltage depends on the current and mutual inductance parameters of each branch; branches with larger currents induce larger voltages in other branches through mutual inductance. Dividing each element in the coupling voltage correction column vector by the current value of the corresponding branch yields the equivalent impedance coupling correction for that branch. The equivalent impedance coupling correction represents the correction value of the equivalent impedance of the branch by the mutual inductance coupling of the current-sharing network. Adding this correction to the virtual impedance compensation parameters compensates for the impact of the current-sharing network connection on impedance matching.

[0091] The equivalent impedance coupling correction is added to the virtual impedance compensation parameter to obtain the converter control parameters after current sharing and matching coupling regulation. After the converter control parameters are sent to each branch interface converter, the converter control loop operates according to the new virtual impedance parameters. At the same time, the passive current sharing network automatically balances the current in each branch through electromagnetic coupling. The two work together to achieve joint regulation of passive current sharing and active impedance matching.

[0092] In actual system operation, the power output conditions of wind and solar power change frequently, sometimes with slow power increases and decreases, and sometimes with sudden power drops or surges. For different operating conditions, the system adopts a multi-mode switching control strategy to balance control accuracy and operational efficiency. The system monitors the DC bus voltage fluctuation amplitude and the current imbalance of each branch in real time. The bus voltage fluctuation amplitude reflects the severity of changes in wind and solar power output, while the current imbalance reflects the uniformity of current in each branch. When the bus voltage fluctuation amplitude is below a preset fluctuation threshold and the current imbalance is below a preset balancing threshold, it indicates that the current operating condition is stable, the current sharing effect is good, and the interface converters of each branch maintain the current control parameters. Current sharing is performed only by the passive current sharing network, without initiating online impedance identification and coupling control, to reduce disturbance to the system and save computational resources for online identification. When the bus voltage fluctuation exceeds the preset fluctuation threshold or the current imbalance exceeds the preset equalization threshold, it indicates that the current operating conditions have changed drastically or the current sharing accuracy has decreased. It is necessary to restart the online impedance identification, recalculate the source-carrier impedance ratio and virtual impedance compensation parameters, recalculate the equivalent impedance coupling correction, add the updated equivalent impedance coupling correction to the updated virtual impedance compensation parameters, generate updated converter control parameters, and send them out for execution to adapt to the new operating conditions and correct the current sharing deviation.

[0093] This embodiment uses an engineering application scenario of a wind-solar shared DC bus hydrogen production system as an example to illustrate the complete implementation process and actual effects of the method proposed in this application. A wind-solar hydrogen production station is located in Northwest China, an area rich in wind energy resources with high average annual wind speeds and annual photovoltaic irradiance classified as a Class I region in China. The station has a total installed capacity of several megawatts each for wind turbine generators and photovoltaic arrays. Power generated is boosted and rectified before being fed into a DC bus, with the DC bus rated at a certain nominal voltage. Several alkaline electrolyzers and proton exchange membrane electrolyzers are configured and connected in parallel to the DC bus.

[0094] The rated power and models of the various electrolyzers differ. Alkaline electrolyzers have a larger single-unit capacity but slower start-up and load-changing response, making them suitable for base-load hydrogen production. Proton exchange membrane electrolyzers have a relatively smaller single-unit capacity but faster response, making them suitable for operating in response to wind and solar load fluctuations. Each electrolyzer is connected to the DC bus via an interface DC / DC converter, with the converter's rated power matched to the corresponding electrolyzer. The site design fully considers the economics of alkaline electrolyzers and the flexibility of proton exchange membrane electrolyzers, achieving a significant expansion of the hydrogen production load range through hybrid operation.

[0095] During the commissioning phase before system operation, the line impedance of each branch was measured branch by branch. The measurements showed that the cable lengths of each branch ranged from tens to nearly a hundred meters, with significant differences in cable resistance; the ratio of the maximum to the minimum line resistance was several times. If this impedance difference is not addressed, it will lead to severe uneven current distribution. Polarization curve measurements were performed on each electrolyzer. The alkaline electrolyzer exhibited a steeper tangent slope near its rated operating point, indicating a higher equivalent resistance; the proton exchange membrane electrolyzer showed a shallower tangent slope, indicating a lower equivalent resistance. The difference in equivalent resistance between the different electrolyzers further exacerbated the unevenness of the parallel current distribution.

[0096] Based on measured branch impedance data, a branch impedance network model was established. According to the variation of bus voltage within the fluctuation range of wind and solar power output, multiple steady-state operating points were selected, and the impedance distribution data at each operating point were calculated. The impedance distribution data shows that the impedance differences between branches are relatively small under low-power conditions, but significantly increase under high-power conditions. This is because the electrolyzer current density is higher under high-power conditions, leading to a larger ohmic voltage drop. However, the ohmic internal resistance of the alkaline electrolyzer and the proton exchange membrane electrolyzer exhibits different trends with current density, causing the difference in their equivalent impedance to further widen in the high-power region.

[0097] Based on the impedance difference characteristics of each branch in the impedance distribution data, a passive current-sharing network with multi-winding coupled inductors was designed. The coupled inductor core is made of amorphous nanocrystalline material, which features high saturation flux density and low loss, making it suitable for use on high-current DC bus sides. The core air gap length and winding turns ratio were tuned using a multi-objective optimization algorithm. During optimization, the objectives were to minimize the maximum current-sharing error under each operating condition and the additional loss under rated power. The NSGA-II multi-objective genetic algorithm was used for optimization. The algorithm population size was set to an appropriate value, and the number of generations was set to the number required for the solution to converge to the Pareto front. The obtained Pareto front showed a clear trade-off between current-sharing error and additional loss. The final selected tuning result ensured that the current-sharing error was within a preset threshold, while the additional loss was within the acceptable range for the system.

[0098] The coupled inductor was wound and installed according to the adjusted parameters. Both the primary and secondary windings of the coupled inductor used multi-strand enameled copper wire to reduce additional copper losses caused by the high-frequency skin effect. The magnetic core adopted a toroidal structure, with the windings evenly distributed on the circumference of the core to reduce leakage inductance. During installation, attention was paid to the connection polarity and the correspondence of the same-name terminals between the coupled inductor and each branch interface converter and electrolytic cell to ensure the correct direction of the induced voltage in the secondary winding, thus suppressing current deviation.

[0099] After the system is put into operation, the online impedance identification module periodically injects sinusoidal disturbance signals into the DC bus for impedance monitoring. During normal operation, the disturbance injection frequency is selected to avoid the converter switching frequency and the system resonant point, and the disturbance amplitude is set to an appropriate proportion of the rated current. During special periods of drastic fluctuations in wind and solar power output, the system automatically increases the frequency of disturbance injection to ensure that the impedance identification data can reflect rapid changes in the system status in a timely manner.

[0100] During a typical summer day's operation, photovoltaic (PV) output gradually increased in the morning, while wind power output remained relatively stable. The DC bus voltage fluctuated slightly around its rated value, and the current imbalance in each branch remained at a low level. The system operated in a stable mode, with current sharing solely handled by a passive current sharing network, and the interface converters maintained fixed control parameters. In the afternoon, a sudden severe convective weather event occurred, causing a rapid increase in wind speed. PV output decreased rapidly due to cloud cover, the DC bus voltage fluctuated significantly, and the current imbalance in each branch increased rapidly, exceeding the preset balancing threshold. Upon detecting the change in operating conditions, the system automatically switched to a disturbance mode, restarted online impedance identification, updated the input impedance data for each branch, recalculated the source-carrier impedance ratio and virtual impedance compensation parameters, and updated the impedance coupling matrix and equivalent impedance coupling correction. After the updated converter control parameters were executed, the current in each branch gradually returned to equilibrium, the bus voltage fluctuation was effectively suppressed, and the system returned to a stable operating state.

[0101] One autumn night, wind power output remained high, while photovoltaic output was zero. The alkaline electrolyzer handled the main hydrogen production load, and the proton exchange membrane electrolyzer was in low-power hot standby mode. At this time, the current distribution in each branch was relatively uniform, and the system operated in a stable mode. In the early morning, wind speed suddenly decreased, causing a sharp drop in wind power output and a subsequent decrease in DC bus voltage. This caused the operating point of each electrolyzer to shift downwards, resulting in a change in the equivalent impedance of the electrolyzers. The system promptly detected the change in operating conditions, initiated online impedance identification and coupling control procedures, updated control parameters, and completed the readjustment of current sharing and matching within seconds.

[0102] During long-term system operation, the electrolyzers gradually age, and their polarization characteristics shift. The aging threshold is set when the cumulative operating time reaches a certain value. When the cumulative operating time of a particular electrolyzer reaches the aging threshold, its polarization curve is retested during a planned system shutdown for maintenance. The retest results show that the polarization curve of the alkaline electrolyzer shifts upward overall, and the equivalent resistance increases by about ten percent compared to the factory value. The membrane resistance of the proton exchange membrane electrolyzer also increases, but the increase is small. The retested polarization curves are updated in the impedance network model, and the impedance distribution data is recalculated. Based on the updated impedance data, the coupling inductance parameters of the passive current sharing network are fine-tuned to adapt to the new impedance characteristics after the electrolyzers have aged.

[0103] This application also provides a parallel electrolytic cell system based on current sharing impedance matching coupling control, including an impedance network analysis module, a passive current sharing network module, an online impedance identification module, and a coupling control module. Each module is implemented in software on a central controller, which is connected to the interface converters and sensors of each branch via analog signal acquisition and communication interfaces.

[0104] The impedance network analysis module incorporates branch impedance modeling and impedance distribution calculation algorithms. The branch impedance modeling algorithm reads preset parameters for the resistance and inductance of each branch cable, the equivalent series resistance of the converter output filter capacitor, and the piecewise linearization parameters of the electrolytic cell polarization curve. It then establishes a branch impedance network model using series superposition and parallel networking logic. The impedance distribution calculation algorithm calculates the total impedance of each branch based on the preset DC bus voltage fluctuation range and operating condition sampling step size, traversing each steady-state operating point and outputting an impedance distribution data table. This module also receives trigger signals for updating the electrolytic cell polarization curve and automatically recalculates the impedance distribution data after the electrolytic cell's cumulative operating time exceeds a preset aging threshold.

[0105] The passive current sharing network module incorporates a multi-objective optimization solver and parameter tuning logic. The multi-objective optimizer uses the current sharing error function and the additional loss function as two independent objective functions, and the turns ratio of each branch and the core air gap length as decision variables. It searches for a Pareto front solution set within a preset range of decision variable values. The parameter tuning logic selects the optimal tuning result from the Pareto front according to preset threshold conditions, outputting the number of turns in the secondary winding, the number of turns in the primary winding, and the core air gap length for each branch, generating a structural parameter table for the passive current sharing network for reference during hardware construction and maintenance.

[0106] The online impedance identification module incorporates a sinusoidal disturbance signal generator, a data acquisition buffer, and a spectrum analysis algorithm. The sinusoidal disturbance signal generator produces sinusoidal disturbance components at preset frequencies and amplitudes, which are then superimposed onto the modulation wave of the interface converter. The data acquisition buffer synchronously stores sampled data of bus voltage, bus current, and the input voltage and current of each branch. The spectrum analysis algorithm performs a Fast Fourier Transform on the buffered data, extracting the amplitude and phase of the disturbance frequency points, and outputs the equivalent output impedance at the source end and the magnitude and phase angle of the input impedance of each branch according to the impedance calculation formula.

[0107] The coupling control module incorporates an impedance matching controller and a coupling correction calculator. The impedance matching controller reads the equivalent output impedance of the source and the input impedance of each branch, calculates the source-carrier impedance ratio for each branch, compares it with a preset proportional threshold, and iteratively adjusts the virtual resistance and virtual reactance values ​​of branches that do not meet the condition by a preset step size, outputting the virtual impedance compensation parameters for each branch. The coupling correction calculator reads the structural parameters of the passive current sharing network and the current data of each branch, establishes an impedance coupling matrix according to the self-inductance and mutual inductance formulas, multiplies it by the current column vector to obtain the coupling voltage correction column vector, and then divides it by the current value to obtain the equivalent impedance coupling correction amount. Finally, the equivalent impedance coupling correction amount is added to the virtual impedance compensation parameters, encoded as control parameters, and sent to the interface converters of each branch for execution via the communication interface.

[0108] The above description is merely a preferred embodiment of this application. The scope of protection of this application is not limited to the above embodiments. All technical solutions falling within the scope of this application's concept are within the scope of protection of this application. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principles of this application should also be considered within the scope of protection of this application.

Claims

1. A method for parallel operation of electrolytic cells based on current-equalizing impedance matching coupling control, characterized in that, include: The line impedance, converter output impedance, and equivalent impedance of each electrolyzer branch in the hydrogen production system are obtained. The three are then connected in series and superimposed to obtain the total impedance of each branch. A branch impedance network model is established by connecting the branches in parallel and calculating the impedance distribution data of each branch under different operating conditions. Based on the impedance difference between each branch, a multi-winding coupled inductor topology is selected. The primary winding of each branch of the multi-winding coupled inductor is connected in series with the positive bus of the corresponding branch, and the secondary winding of each branch is connected in series between the converter output terminal and the electrolytic cell input terminal of the corresponding branch. The turns ratio of the coupled inductor and the air gap length of the magnetic core are tuned with the current sharing error and additional loss as optimization objectives to construct a passive current sharing network. Inject sinusoidal disturbance current signals into the DC bus and each branch, collect voltage and current response waveforms, extract the amplitude and phase corresponding to the disturbance frequency through Fourier transform, and calculate the equivalent output impedance of the source end and the input impedance of each branch. The source-carrier impedance ratio is calculated based on the equivalent output impedance of the source and the input impedance of each branch. The virtual impedance compensation parameters of each branch interface converter are adjusted with the preset stability criterion as a constraint. An impedance coupling matrix is ​​established based on the self-inductance and mutual inductance parameters of the multi-winding coupled inductor. The equivalent impedance coupling correction amount of each branch is calculated. The equivalent impedance coupling correction amount is added to the virtual impedance compensation parameters to obtain the converter control parameters.

2. The method for parallel connection of electrolytic cells based on current-sharing impedance matching coupling control according to claim 1, characterized in that, The establishment of the branch impedance network model includes: For each parallel branch of the electrolytic cell, the cable resistance and cable inductance from the DC bus to the input end of the electrolytic cell are equivalent to the line impedance, and the filter capacitor reactance and parasitic resistance at the output end of the interface converter are equivalent to the converter output impedance. A working point is selected on the polarization characteristic curve of the electrolyzer. The slope of the tangent at the working point is taken as the equivalent resistance of the electrolyzer at the working point. The slope of the tangent is multiplied by a preset correction factor to obtain the equivalent impedance of the electrolyzer. The preset correction factors of the alkaline electrolyzer and the proton exchange membrane electrolyzer are determined according to the slope change rate of their respective polarization curves near the working point. The total impedance of a branch is obtained by adding the line impedance, converter output impedance, and electrolytic cell equivalent impedance of the same branch. The total impedances of all branches are arranged in parallel to establish a branch impedance network model.

3. The method for parallel connection of electrolytic cells based on current-sharing impedance matching coupling control according to claim 2, characterized in that, The calculation of impedance distribution data for each branch under different operating conditions includes: Based on the fluctuation range of the DC bus voltage, select multiple steady-state operating points, recalculate the equivalent impedance of the electrolytic cell of each branch at each steady-state operating point, remeasure the output impedance of the converter, and obtain the total impedance of each branch corresponding to that operating point. The total impedance of each branch corresponding to all operating points is summarized to form impedance distribution data under different operating conditions. When the cumulative operating time of the electrolytic cell exceeds the preset aging threshold, the polarization curve of the electrolytic cell is remeasured, the tangent slope at each working point is updated, the updated tangent slope is multiplied by the preset correction coefficient to obtain the updated equivalent impedance of the electrolytic cell, and the impedance distribution data is recalculated.

4. The method for parallel connection of electrolytic cells based on current-sharing impedance matching coupling control according to claim 1, characterized in that, The tuning of the coupled inductor turns ratio and core air gap length with current sharing error and additional loss as optimization objectives includes: The current sharing error is defined as the maximum deviation between the current in each branch and the average current of all branches divided by the average current. The additional loss is defined as the sum of the copper loss of each winding of the coupled inductor and the iron loss of the magnetic core. The ratio of the number of turns in the secondary winding to the number of turns in the primary winding of each branch of the coupled inductor and the length of the air gap in the magnetic core are used as decision variables. With the minimization of current sharing error and the minimization of additional loss as optimization objectives, a multi-objective optimization algorithm is used to optimize the decision variables and obtain the Pareto front. The solution with a current sharing error less than or equal to a preset error threshold and an additional loss less than or equal to a preset loss threshold is selected from the Pareto front and used as the tuning result of the coupled inductor parameters. If there is no solution in the Pareto front that simultaneously satisfies both threshold conditions, then the solution with the smallest flow-averaging error is selected as the tuning result.

5. The method for parallel connection of electrolytic cells based on current-sharing impedance matching coupling control according to claim 1, characterized in that, The process involves injecting sinusoidal disturbance current signals into the DC bus and each branch, acquiring voltage and current response waveforms, extracting the amplitude and phase corresponding to the disturbance frequency through Fourier transform, and calculating the equivalent output impedance at the source and the input impedance of each branch, including: At the DC bus end, by adjusting the modulation wave of the interface converter and superimposing a sinusoidal disturbance component with a preset frequency and preset amplitude, a sinusoidal disturbance current signal with a corresponding frequency is generated in the DC bus current, and the voltage and current response waveforms of the DC bus are acquired simultaneously. Fast Fourier transform is performed on the acquired voltage and current response waveforms respectively, and the voltage amplitude, current amplitude, and phase difference between voltage and current at the corresponding frequency points of the sinusoidal disturbance current signal are extracted from the transformation results. Divide the voltage amplitude by the current amplitude to obtain the impedance amplitude, and use the phase difference as the impedance phase angle. The source-side equivalent output impedance is formed by the impedance amplitude and the impedance phase angle. A sinusoidal disturbance current signal with the same frequency as the source is applied to the input terminal of each branch interface converter. The voltage and current response waveforms of each branch input terminal are collected synchronously, and the input impedance of each branch is calculated in the same way.

6. The method for parallel connection of electrolytic cells based on current-sharing impedance matching coupling control according to claim 1, characterized in that, The adjustment of the virtual impedance compensation parameters of each branch interface converter based on a preset stability criterion includes: Divide the magnitude of the equivalent output impedance at the source end by the magnitude of the input impedance of each branch to obtain the source-carrier impedance ratio of each branch. The constraint condition is that the source-carrier impedance ratio is less than a preset proportional threshold throughout the entire frequency band. The constraint condition is used to determine whether each branch meets the constraint condition. For branches that do not meet the constraints, the virtual resistance and virtual reactance values ​​of the interface converter of the branch are adjusted by a preset step size. After each adjustment, the source-carrier impedance ratio of the branch is recalculated until the source-carrier impedance ratio is less than the preset ratio threshold in the entire frequency band. The virtual resistance and virtual reactance values ​​at this time are used as the virtual impedance compensation parameters of the branch.

7. The method for parallel connection of electrolytic cells based on current-sharing impedance matching coupling control according to claim 1, characterized in that, The step of establishing the impedance coupling matrix based on the mutual inductance parameters of the current sharing network includes: Obtain the self-inductance value of the secondary winding of each branch of the multi-winding coupled inductor, as well as the mutual inductance value between any two branches of the secondary winding; Multiply the self-inductance value of the secondary winding of the i-th branch by the angular frequency and place it in the i-th row and i-th column of the impedance coupling matrix; Multiply the mutual inductance values ​​of the secondary windings of the i-th branch and the j-th branch by the angular frequency, and place them in the i-th row and j-th column of the impedance coupling matrix, where i is not equal to j; The complete impedance coupling matrix is ​​obtained by filling in rows and columns one by one. The diagonal elements of the impedance coupling matrix reflect the self-impedance of the current-sharing inductors of each branch, and the off-diagonal elements reflect the mutual impedance between the current-sharing inductors of different branches.

8. The method for parallel connection of electrolytic cells based on current-sharing impedance matching coupling control according to claim 7, characterized in that, The calculation of the equivalent impedance coupling correction for each branch includes: Multiply the impedance coupling matrix by the current column vector of each branch to obtain the coupling voltage correction column vector of each branch. Each element in the coupling voltage correction column vector represents the additional voltage generated in the corresponding branch due to the mutual inductance coupling of the current sharing inductance of other branches. Divide each element in the coupling voltage correction column vector by the current value of the branch corresponding to that element to obtain the equivalent impedance coupling correction amount of that branch. The equivalent impedance coupling correction amount represents the correction value of the equivalent impedance of the branch by the mutual inductance coupling of the current sharing network.

9. The method for parallel connection of electrolytic cells based on current-sharing impedance matching coupling control according to claim 1, characterized in that, After generating and issuing the converter control parameters, multi-mode switching control is also included: Real-time detection of DC bus voltage fluctuation amplitude and current imbalance of each branch, wherein the current imbalance is the maximum deviation of each branch current from the average current of all branches divided by the average current. When the bus voltage fluctuation is lower than the preset fluctuation threshold and the current imbalance is lower than the preset balance threshold, each branch interface converter maintains the current control parameters and the current is shared only by the passive current sharing network. When the bus voltage fluctuation exceeds the preset fluctuation threshold or the current imbalance exceeds the preset balancing threshold, the impedance online identification is restarted, the source-carrier impedance ratio and virtual impedance compensation parameters are recalculated, the equivalent impedance coupling correction is recalculated, the updated equivalent impedance coupling correction is added to the updated virtual impedance compensation parameters, the updated converter control parameters are generated and sent for execution.

10. A parallel electrolytic cell system based on current-equalizing impedance matching coupling control, characterized in that, A method for implementing a parallel electrolytic cell connection based on current-sharing impedance matching coupling control as described in any one of claims 1 to 9 includes: The impedance network analysis module is used to obtain the line impedance, converter output impedance and electrolytic cell equivalent impedance of each electrolytic cell branch. The three are connected in series and superimposed to obtain the total impedance of each branch. The branch impedance network model is established according to the parallel relationship, and the impedance distribution data of each branch under different operating conditions is calculated. The passive current sharing network module is used to select the topology of the multi-winding coupled inductor based on the impedance difference between each branch. The primary winding of each branch of the multi-winding coupled inductor is connected in series with the positive bus of the corresponding branch, and the secondary winding of each branch is connected in series between the converter output terminal and the electrolytic cell input terminal of the corresponding branch. The turns ratio of the coupled inductor and the air gap length of the magnetic core are tuned with the current sharing error and additional loss as optimization targets. The online impedance identification module is used to inject sinusoidal disturbance current signals into the DC bus and each branch, collect voltage and current response waveforms, extract the amplitude and phase corresponding to the disturbance frequency through Fourier transform, and calculate the equivalent output impedance of the source end and the input impedance of each branch. The coupling control module is used to calculate the source-carrier impedance ratio based on the equivalent output impedance of the source end and the input impedance of each branch, adjust the virtual impedance compensation parameters of each branch interface converter with a preset stability criterion as a constraint, establish an impedance coupling matrix based on the self-inductance and mutual inductance parameters of the multi-winding coupled inductor and calculate the equivalent impedance coupling correction amount of each branch, add the equivalent impedance coupling correction amount to the virtual impedance compensation parameters to obtain the converter control parameters and send them down for execution.

Images