A multi-time scale coordinated control method for an off-grid wind power hydrogen production system
By employing a multi-timescale coordinated control method in an off-grid wind power hydrogen production system, the problem of abnormal gas purity on the anode side when wind power is insufficient was solved, achieving coordination between electrical stability and gas-side safety, and improving the system's operational continuity and hydrogen production efficiency.
Patent Information
- Application Number
- CN202610241771.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-02-28
- Publication Date
- 2026-05-29
AI Technical Summary
Existing off-grid wind power hydrogen production systems suffer from problems such as rapid reduction of electrolysis load when wind power is insufficient, leading to transient abnormalities in the purity of gas on the anode side, which in turn causes frequent protection shutdowns and a decrease in the overall hydrogen production efficiency of the system.
A multi-timescale coordinated control method is adopted to calculate the negative ramp intensity by acquiring system operation data, construct the dissolved hydrogen inventory proxy state, generate the purity peak prediction, and perform robust normalization processing to generate target electrolysis current command and energy storage power command. This is to mitigate the mismatch between the sudden drop in oxygen generation rate and the lag in transmembrane hydrogen release, and ensure electrical stability and gas-side safety.
It effectively suppressed the formation of gas purity spikes exceeding the limits on the anode side, ensuring the power balance and absolute electrical stability of the off-grid microgrid, improving the overall operational continuity of the system and the level of wind energy absorption, and increasing the achievement rate of the long-term hydrogen production plan.
Smart Images

Figure CN122118843A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of wind power hydrogen production control technology, and in particular to a multi-timescale coordinated control method for off-grid wind power hydrogen production systems. Background Technology
[0002] Off-grid wind-powered hydrogen production systems typically consist of wind turbines, their converters, energy storage units, converters, electrolysis power supplies and electrolyzers, and system loads. In off-grid operation, the system lacks an external power grid for power support and must rely on internally connected parallel equipment for power sharing and regulation to maintain grid stability. Due to random fluctuations in wind power, when wind speed decreases and the available wind power is insufficient to meet the system's total demand, the control system is forced to frequently and rapidly reduce the electrolyzer's operating current to eliminate the power shortfall on the bus and ensure the power balance and electrical safety of the off-grid microgrid.
[0003] Existing off-grid coordinated control methods primarily focus on electrical stability, with conventional strategies involving rapid reduction of electrolysis load when power shortages occur. While this approach maintains electrical stability, it overlooks a serious hidden danger on the gas side: the negative ramp-up purity peak window. The physical reason for this is that hydrogen that can be temporarily retained inside the electrolyzer exists mainly in dissolved form, exhibiting short-term storage and mass transfer lag. When the electrolysis current is rapidly reduced, the oxygen generation rate drops almost instantaneously, leading to a rapid weakening of gas dilution capacity. However, the release of transmembrane hydrogen exhibits a significant lag. This temporal inconsistency causes a transient peak in the proportion of hydrogen mixed with oxygen on the anode side within a short time window. Existing control strategies, lacking quantification and constraints on the purity risk within this short window, directly translate the electrical stability-maintaining reduction action into a dangerous action on the gas side. This can easily trigger purity exceedance shutdowns or force the system to adopt an overly conservative operating strategy, ultimately resulting in reduced wind energy absorption and a significant decrease in the overall hydrogen production efficiency of the system. Summary of the Invention
[0004] The purpose of this invention is to address the shortcomings of existing technologies where rapid reduction of electrolysis load, primarily driven by electrical stability, easily leads to transient anomalies in the purity of the gas on the anode side, resulting in frequent protection shutdowns and a decrease in the overall hydrogen production efficiency of the system. Therefore, this invention proposes a multi-timescale coordinated control method for off-grid wind power hydrogen production systems.
[0005] To address the problems existing in the prior art, the present invention adopts the following technical solution: A multi-timescale coordinated control method for an off-grid wind power hydrogen production system includes: S1. Obtain the system's operating data and calculate the negative ramp intensity based on the rate of decrease of the electrolytic current in the operating data; S2. Construct the dissolved hydrogen inventory proxy state based on electrolysis current and negative ramp intensity, and generate purity peak prediction. S3. The predicted purity peak is robustly normalized within the rolling window to obtain the risk normalization index. S4. Determine the minimum feasible current based on the planned current, operating data and energy storage capacity of the current cycle, and shape the minimum feasible current by adjusting the lower edge according to the risk normalization index to generate the target electrolysis current command. S5. Evaluate the real-time power gap of the power supply network based on the target electrolysis current command, and generate energy storage power command.
[0006] Preferably, the operating data includes available wind power, upper limit of energy storage charging and discharging power, electrolysis current, electrolytic cell voltage, and total system load power excluding electrolysis load.
[0007] Preferably, the negative ramp intensity is calculated based on the rate of decrease of the electrolytic current in the operating data, including: Calculate the difference between the electrolytic current of the previous sampling period and the electrolytic current of the current sampling period. If the difference is less than zero, the negative ramp intensity is set to zero; otherwise, the ratio of the difference to the sampling period is taken as the negative ramp intensity.
[0008] Preferably, a dissolved hydrogen inventory proxy state is constructed based on the electrolysis current and the negative ramp intensity, and a purity peak prediction is generated based on the dissolved hydrogen inventory proxy state, including: The dissolved hydrogen inventory proxy status of the previous sampling period is recursively updated by using the online updated equivalent time constant and the electrolysis current of the current sampling period to obtain the dissolved hydrogen inventory proxy status of the current sampling period. The upside term is obtained based on the ratio of the dissolved hydrogen inventory proxy status to the electrolysis current during the current sampling period; The purity peak prediction is generated by linearly combining the uplift term and negative ramp intensity by identifying and updating parameters online.
[0009] Preferably, the equivalent time constant and the online identification and updating parameters are updated online using the recursive least squares method. The online update includes: fitting the online measured anode-side gas purity with the purity peak prediction to generate a prediction error; and updating the parameter vector using the gain calculated from the regression vector and the prediction error.
[0010] Preferably, the predicted purity peak is robustly normalized within the rolling window to obtain a risk normalization index, including: The median of the predicted purity spike within the set rolling window is used as the baseline value; Calculate the absolute deviation between the predicted purity peak and the baseline value, and use the median of this absolute deviation within the rolling window as the deviation baseline value; Calculate the difference between the purity peak prediction and the baseline value for the current sampling period. If the difference is less than zero, the risk normalization index is set to zero; otherwise, the ratio of the difference to the deviation benchmark value is used as the risk normalization index.
[0011] Preferably, the minimum feasible current is determined based on the planned current, operating data, and energy storage capacity for the current cycle, including: The initial power gap is obtained by subtracting the available wind power from the sum of the products of the total system load power and planned current in the operating data and the electrolyzer voltage in the operating data. Subtract the upper limit of energy storage charging and discharging power in the operating data from the initial power gap, and take the maximum value of the difference and zero as the reduced electrolytic power; The amount of current to be reduced is obtained by the ratio of the reduced electrolytic power to the electrolytic cell voltage. The difference between the planned current and the reduced current is taken as the minimum feasible current.
[0012] Preferably, the minimum feasible current is shaped by a downward adjustment based on the risk normalization index to generate the target electrolysis current command, including: Add 1 to the risk normalization index to obtain the risk weighting coefficient; Using the risk weighting coefficient as a weighting factor, the target electrolysis current command is obtained by weighted averaging the minimum feasible current and the electrolysis current of the previous sampling period.
[0013] Preferably, the real-time power gap of the power supply network is assessed based on the target electrolysis current command, and an energy storage power command is generated, including: The real-time power gap of the power supply network is obtained by subtracting the available wind power from the sum of the products of the total system load power, the target electrolysis current command and the electrolyzer voltage in the operating data. The energy storage power command is obtained by limiting the real-time power gap of the power supply network between the upper limit of energy storage charging and discharging power and the negative upper limit of energy storage charging and discharging power.
[0014] To address the aforementioned problems, this invention also provides a multi-timescale coordinated control system for an off-grid wind power hydrogen production system, the system comprising: The negative ramp calculation module is used to acquire the system's operating data and calculate the negative ramp intensity based on the rate of decrease of the electrolytic current in the operating data. The peak prediction module is used to construct the dissolved hydrogen inventory proxy state based on the electrolysis current and the negative ramp intensity, and generate the purity peak prediction. The risk normalization module is used to robustly normalize the predicted purity peak within a rolling window to obtain the risk normalization index. The current shaping module is used to determine the minimum feasible current based on the planned current, operating data and energy storage capacity of the current cycle, and to shape the minimum feasible current by a downward adjustment according to the risk normalization index to generate the target electrolysis current command. The energy storage scheduling module is used to assess the real-time power gap of the power supply network based on the target electrolysis current command and generate energy storage power commands.
[0015] Compared with the prior art, the beneficial effects of the present invention are: 1. This invention integrates the negative ramp intensity of the electrolysis current with the physical characteristics of the delayed release of dissolved hydrogen to construct a purity peak prediction quantity and extract a robust risk normalization index. Then, at the control command level, it directly performs adaptive downward edge shaping processing on the minimum feasible current, transforming the implicit transient anomaly of gas purity into a calculable and dynamically perceived risk constraint. This allows the electrolyzer current to automatically slow down its rate of decline under conditions such as a sudden drop in wind power and forced load reduction, thereby fundamentally mitigating the mismatch caused by the sudden drop in oxygen generation rate and the delayed release of transmembrane hydrogen, and effectively suppressing the formation of gas purity out-of-bounds peaks on the anode side.
[0016] 2. This invention accurately assesses the real-time power gap based on the target electrolysis current command after the formation, and utilizes a parallel power supply power sharing mechanism to prioritize the energy storage unit to bear the instantaneous gap caused by the gradual reduction of electrolysis load. At the same time, it dynamically corrects the planned current trajectory by combining the cyclic risk energy and hydrogen production deviation on a slow time scale. This multi-time scale coordination architecture not only ensures the power balance and absolute electrical stability of the off-grid microgrid bus, but also effectively avoids the system's protection shutdown and forced conservative strategy due to frequent triggering of purity anomalies. Under the premise of taking into account the physical safety boundary of the gas side, it significantly improves the continuity of the system's overall operation, the level of wind energy absorption, and the achievement rate of the long-term hydrogen production plan. Attached Figure Description
[0017] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings: Figure 1 This is a flowchart illustrating a multi-timescale coordinated control method for an off-grid wind power hydrogen production system according to the present invention. Figure 2 This is a functional block diagram of a multi-timescale coordinated control system for an off-grid wind power hydrogen production system according to the present invention. Figure 3 This is a schematic diagram of the topology of the off-grid wind power hydrogen production system of the present invention. Detailed Implementation
[0018] 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.
[0019] Example: This example provides a multi-timescale coordinated control method for an off-grid wind power hydrogen production system. See [link to relevant documentation]. Figure 1 Specifically, including: S1. Obtain the system's operating data and calculate the negative ramp intensity based on the rate of decrease of the electrolytic current in the operating data; S2. Construct the dissolved hydrogen inventory proxy state based on electrolysis current and negative ramp intensity, and generate purity peak prediction. S3. The predicted purity peak is robustly normalized within the rolling window to obtain the risk normalization index. S4. Determine the minimum feasible current based on the planned current, operating data and energy storage capacity of the current cycle, and shape the minimum feasible current by adjusting the lower edge according to the risk normalization index to generate the target electrolysis current command. S5. Evaluate the real-time power gap of the power supply network based on the target electrolysis current command, and generate energy storage power command.
[0020] In an embodiment of the present invention, obtaining the operating data of the hydrogen production system and calculating the negative ramp intensity based on the rate of decrease of the electrolysis current in the operating data includes: This embodiment is based on the actual operation scenario of an off-grid wind power hydrogen production system. Specifically, it implements the steps of acquiring the operating data of the hydrogen production system and calculating the negative ramp intensity based on the rate of decrease in electrolysis current within the operating data. The off-grid wind power hydrogen production system includes wind turbine generators and their converters, energy storage units and their converters, electrolysis power supplies and electrolyzers, system load, and online monitoring equipment. All equipment is connected to the main controller via an industrial bus. The main controller uses an embedded processor and is set with a fixed sampling period to achieve real-time acquisition of operating data and real-time calculation of the negative ramp intensity, ensuring that the real-time requirements of the system's fast-scale coordinated control are met. The specific implementation process is as follows: The main controller establishes communication connections with the corresponding monitoring modules of each device to acquire various operational data of the hydrogen production system in real time. Specifically, the available wind power is collected through the power monitoring port of the wind turbine converter, which integrates a power sensor to detect the actual available power output from the wind turbine to the off-grid power supply network after conversion by the converter. The upper limit of energy storage charging and discharging power is obtained through the control unit of the energy storage converter. This control unit calculates and outputs the maximum safe charging and discharging power output of the energy storage unit under the current operating conditions, i.e., the upper limit of energy storage charging and discharging power, based on the current SOC state of the energy storage unit, the battery cell voltage, and temperature parameters. The electrolysis current is collected through a current sensor connected in series in the output circuit of the electrolysis power supply. The current sensor uses a high-precision Hall current sensor, with a measurement range adapted to the rated operating current of the electrolytic cell and a measurement accuracy of no less than 0.5%. It can accurately capture instantaneous changes in the electrolytic current. The electrolytic cell voltage is acquired by a voltage sensor connected in parallel across the positive and negative terminals of the electrolytic cell. This voltage sensor is an isolated voltage sensor to avoid electromagnetic interference during the operation of the electrolytic cell affecting the measurement accuracy. The total system load power is acquired by the bus power monitoring module of the off-grid power supply network. This module can detect the total power consumed by all system loads other than the electrolytic load in real time, ensuring the accuracy of subsequent power deficit calculations. All the above operating data are synchronously transmitted to the storage unit of the main controller for subsequent calculations.
[0021] The main controller continuously collects electrolytic current data according to a set sampling period of Δt, which is 0.01 seconds. This value takes into account both the response speed of the electrolytic current change and the computing power of the main controller. It can capture the instantaneous decrease in electrolytic current without causing the main controller to be overloaded due to an excessively short sampling period. Δt is the sampling period, which represents the time interval between two consecutive collections of electrolytic current data, in seconds. k is set as the sequence number of the current sampling period, and k is a positive integer that increases sequentially as the sampling process continues. I(k) is the electrolytic current data collected in the current sampling period k. This data is the actual electrolytic current value measured in real time by the current sensor and transmitted to the main controller, which can truly reflect the current working current state of the electrolytic cell. I(k-1) is the electrolytic current data collected in the previous sampling period k-1. This data is retrieved from the main controller's storage unit and is the electrolytic current value captured at the previous sampling moment. It is used to compare with the electrolytic current value in the current sampling period to calculate the current change.
[0022] Based on the collected electrolysis current data, the rate of decrease in electrolysis current is calculated, and thus the negative ramp intensity is obtained. The specific calculation formula is as follows: ,in The negative ramp intensity of the current sampling period k is used to characterize the instantaneous rate of decrease of the electrolysis current. This parameter is a core input parameter for subsequently constructing the dissolved hydrogen inventory proxy state and predicting the purity peak; the max function in the formula takes values between 0 and 1. The calculation logic for the maximum value of the two is as follows: When I(k-1) is less than or equal to I(k), it means that the electrolytic current in the current sampling period has not decreased compared to the previous sampling period. At this time, there is no negative ramp phenomenon in the electrolytic current, and the corresponding negative ramp intensity should be taken as 0 to avoid generating invalid negative ramp intensity data when there is no downward trend, thus ensuring the rationality of the parameters. When I(k-1) is greater than I(k), it means that the electrolytic current in the current sampling period has decreased compared to the previous sampling period. At this time, the difference between I(k-1) and I(k) is positive. The ratio of this difference to the sampling period Δt is the instantaneous decrease rate of the electrolytic current. Using this decrease rate as the negative ramp intensity of the current period can accurately quantify the decrease speed of the electrolytic current. This calculation method has a small computational load, which can meet the real-time calculation needs of the main controller and accurately capture the negative ramp characteristics of the electrolytic current. It provides reliable basic parameters for the subsequent quantification of purity peak risk, ensuring the feasibility and control accuracy of the entire coordinated control method.
[0023] It should be noted that the available power of wind power refers to the effective electrical power that a wind turbine can continuously provide to the off-grid power supply network after the wind energy is converted by the rotor aerodynamically, transmitted by the transmission chain, and output by the converter under the current wind speed and air density conditions. Its magnitude is jointly limited by wind speed fluctuations, rotor speed constraints, converter rated capacity, and voltage and frequency stability requirements. The upper limit of energy storage charging and discharging power refers to the maximum charging power and maximum discharging power allowed by the energy storage unit under the current state of charge, terminal voltage, and temperature conditions, constrained by the battery electrochemical reaction rate, internal resistance heating, and the rated current capability of the power converter. It is used to characterize the instantaneous support capability of energy storage for power gaps and its ability to absorb surplus electrical energy. Electrolysis current refers to the DC current provided by the electrolytic power supply to the electrolytic cell stack. This current directly determines the electrochemical reaction rate occurring at the electrode interface. The charge throughput per unit time corresponds to the rate of hydrogen and oxygen production and affects the heating level and polarization state of the electrolyzer. Electrolyzer voltage refers to the potential difference across the electrolyzer stack, composed of the reversible reaction voltage, voltage loss due to polarization, and ohmic voltage drop. It reflects the driving voltage required for the conversion of electrical energy into chemical energy during electrolysis and varies with current, temperature, bubble coverage, and membrane resistance. The total system load power excluding electrolysis load refers to the total active power demand of all electrical equipment in the off-grid power supply network at the current moment, excluding the electrolysis power source and its electrolyzer. This typically includes control and communication equipment, cooling and circulating pumps, auxiliary equipment such as compressors and dryers, and other on-site electrical loads. This power characterizes the non-electrolysis power consumption that the system must prioritize and participates in the real-time power balance calculation of the power supply network. It should be noted that the negative ramp intensity is a quantity used to characterize the rate of decrease of the electrolytic current when it decreases between adjacent sampling periods. When the electrolytic current of the previous sampling period is greater than that of the current sampling period, the ratio of the difference between the two to the sampling period is used as the negative ramp intensity to characterize the steepness of the downward adjustment of the electrolytic load, thereby reflecting the electrolytic reaction rate and its abrupt change in the gas generation and dilution process. When the electrolytic current of the previous sampling period is less than or equal to that of the current sampling period, the negative ramp intensity is set to zero so that this quantity only responds to the decreasing process of the electrolytic current and avoids introducing interference to the increasing process.
[0024] In an embodiment of the present invention, a proxies state of dissolved hydrogen inventory is constructed based on electrolysis current and negative ramp intensity, and a purity peak prediction is generated, including: In the real-time control process of this off-grid wind power hydrogen production system, the main controller performs recursive updates of the dissolved hydrogen inventory proxy state and recursive least squares online updates of the equivalent time constant and online identification parameters at a fixed sampling period. During system power-on initialization, the initial dissolved hydrogen inventory proxy state is set to the initial electrolysis current, and the equivalent time constant parameter vector and covariance matrix are both set to initial values that conform to the initial operating characteristics of the electrolyzer. Within each sampling period, the main controller first performs a recursive calculation of the dissolved hydrogen inventory proxy state based on the dissolved hydrogen inventory proxy state of the previous period, the current period's electrolysis current, and the online updated equivalent time constant. The calculation formula used is as follows:
[0025] in This represents the proxies status of dissolved hydrogen inventory during the current sampling period, used to characterize the equivalent retention level of dissolved hydrogen inside the electrolyzer at the current moment. This represents the proxies status of the dissolved hydrogen inventory from the previous sampling period. The fixed sampling period set for the system This is the equivalent time constant updated online for the current sampling period. The formula, which uses the electrolysis current collected in the current sampling period, adopts a first-order lag recursive form to accurately match the physical laws of dissolved hydrogen storage and delayed release within the electrolyzer. Subsequently, the main controller constructs a regression vector and fits the online measured anode-side gas purity with the purity peak prediction, calculates the prediction error, and then calculates the update gain through the regression vector and covariance matrix. The update gain and prediction error are used to complete the iterative correction of the parameter vector and synchronously update the covariance matrix, realizing the synchronous adaptive update of the equivalent time constant and online identification parameters. The above online update process of recursive least squares method enables the model parameters to be corrected in real time according to the actual operating conditions of the electrolyzer and the gas purity response, ensuring the accuracy and reliability of the dissolved hydrogen inventory proxy state and the purity peak prediction. The overall calculation process only relies on the data of the previous period and the current period's collection, with low computational load and high real-time performance, fully meeting the embedded control requirements of off-grid wind power hydrogen production systems.
[0026] This embodiment provides a specific calculation example of the online parameter update process using the recursive least squares method. During system power-on initialization, the parameter vector is set to an initial vector of all zeros, and the covariance matrix is set to a unit diagonal matrix with large diagonal elements to ensure parameter identification accuracy in the initial stage. Simultaneously, extremely small values are set to avoid zero denominators. Within the current sampling period, the main controller first acquires the measured values of the negative ramp intensity of the electrolytic current and the purity of the anode-side gas under the dissolved hydrogen inventory proxy state. Then, a regression vector is constructed based on these values. The regression vector expression used is as follows: The first element of the vector is a constant term used to fit the fixed offset component of the purity peak prediction; the second element is the ratio of the dissolved hydrogen inventory proxy state to the electrolysis current, reflecting the purity rise characteristics caused by dissolved hydrogen hysteresis; and the third element is the negative ramp intensity, reflecting the influence of the current decrease rate on the purity peak. The prediction error between the predicted purity peak and the measured anode-side gas purity is calculated, and the prediction error expression is as follows: in To measure the purity of the gas on the anode side, Given the parameter vector updated in the previous sampling period, the update gain for the current period is calculated based on the regression vector and the covariance matrix of the previous period. The gain calculation formula is as follows: in Given the covariance matrix of the previous sampling period, the parameter vector is iteratively updated using the calculated update gain and prediction error. The parameter vector update formula is as follows: Simultaneously, the recursive update of the covariance matrix is performed. The formula for updating the covariance matrix is: in This is the parameter vector updated in the current period. The covariance matrix is updated for the current period, and the equivalent time constant is adaptively corrected synchronously based on the real-time update results of the parameter vector.
[0027] Here is a specific numerical example. The initial parameter vector is a vector consisting of three zero elements. The covariance matrix is a third-order diagonal matrix with all diagonal elements set to 10,000. The current sampling period's dissolved hydrogen inventory proxy status is set to 500, the electrolysis current to 480, the negative ramp intensity to 20, and the measured anode-side gas purity to 2.5. Substituting these values into the calculation, the second element of the regression vector is approximately 1.04, which is the sum of 500 divided by 480 and 1 / 100,000. The three elements of the regression vector are 1, 1.04, and 20, respectively. The product of the transpose of the parameter vector and the regression vector from the previous period is approximately zero. The prediction error is set to 2.5. Substituting these values into the gain formula, the current period's update gain is calculated. Then, the parameter vector is corrected using the gain and prediction error, updating it from a zero vector to a vector containing valid identification values. The covariance matrix iterates synchronously. Through the above specific numerical calculation process, real-time online identification and updating of the parameter vector and equivalent time constant can be achieved.
[0028] It should be noted that the purity of the anode-side gas refers to the proportion of oxygen, the target component, in the gas discharged from the anode side of the electrolyzer, or an equivalent characterization quantity related to impurities. It is used to reflect the degree of hydrogen contamination in the anode-side gas and corresponds to the gas safety and quality level. Its value fluctuates with changes in transmembrane permeation, dissolved hydrogen release, pressure difference, and transient operating conditions. The dissolved hydrogen inventory proxy state is a state quantity used to characterize the equivalent storage level of hydrogen that can be temporarily retained inside the electrolysis unit. This hydrogen mainly exists in the form of dissolved hydrogen in the membrane or electrode catalyst layer and its microporous structure and can be slowly released into the gas channel within a certain period of time. Therefore, this state quantity reflects the degree of short-term storage and mass transfer lag of hydrogen inside the material. The dissolved hydrogen inventory proxy state is derived recursively from the historical changes of the electrolysis current. It gradually accumulates as the electrolysis current increases and gradually decreases as the electrolysis current decreases. This allows it to characterize the short-term contamination effect caused by the rapid decrease in oxygen generation rate while hydrogen continues to be released when the current is reduced, thus providing an interpretable internal memory quantity for purity peak prediction and negative ramp-up risk assessment.
[0029] The calculation process retrieves the dissolved hydrogen inventory proxy state and the real-time collected electrolysis current, which have already been calculated for the current sampling period. Simultaneously, a preset minimum value is retrieved to prevent division by zero anomalies caused by excessively low electrolysis current, ensuring the stability and continuity of the calculation process. Then, the rise term is calculated by the ratio of the dissolved hydrogen inventory proxy state for the current sampling period to the sum of the electrolysis current and the minimum value. The calculation logic of the rise term aligns with the physical characteristics of the delayed release of dissolved hydrogen inside the electrolyzer. When the electrolysis current experiences a rapid decrease (i.e., a negative ramp-up), the electrolysis current will decrease instantaneously, while the dissolved hydrogen inventory proxy state, due to its lag characteristic, remains at a relatively high level. The ratio of the two will rise accordingly. This rise term can directly quantify the effect of delayed release of dissolved hydrogen on the purity of the anode-side gas, providing a core driving factor for subsequent purity peak prediction.
[0030] After calculating the lifting term, the parameters are identified and updated online using the recursive least squares method. These parameters include three real-time updated coefficients, which are used to adapt the different influence weights of the lifting term and negative ramp intensity on the purity peak. Simultaneously, the negative ramp intensity already calculated for the current sampling period is used. The lifting term and negative ramp intensity are linearly combined using the parameters identified and updated online to generate the purity peak prediction. The formula for calculating the purity peak prediction is as follows:
[0031] in This is the predicted purity peak for the current sampling period, used to characterize the transient peak level of gas purity that may occur on the anode side of the electrolyzer under the current operating conditions, and to provide direct input for subsequent risk normalization index calculation; The constant parameters are identified and updated online for the current sampling period, which are used to fit the fixed offset component of the gas purity on the anode side, and to offset the influence of the inherent error of the electrolyzer and the zero-point deviation of the sensor on the prediction results. The online identification and updating of the rise term coefficient parameters for the current sampling period are used to adjust the influence weight of the rise term on the predicted amount of purity peak, and to adapt to the degree of influence of the hysteresis release of dissolved hydrogen on the purity peak under different operating conditions. The negative ramp intensity coefficient parameter is identified and updated online for the current sampling period. It is used to adjust the influence weight of the negative ramp intensity on the purity peak prediction and adapt to the change law of purity peak under different current descent rates. This represents the proxies status of dissolved hydrogen inventory during the current sampling period, used to characterize the equivalent retention level of dissolved hydrogen inside the electrolyzer at the current moment; The electrolytic current collected during the current sampling period directly reflects the current operating current status of the electrolytic cell; This is a preset minimum value, taken as one ten-thousandth, used to avoid division by zero errors when the electrolysis current approaches zero, and to ensure the stability of the calculation process; The negative ramp intensity of the current sampling period is used to characterize the instantaneous rate of decrease of the electrolytic current. The formula is calculated using a linear combination. The core reason is that the gas purity peak on the anode side has a significant linear correlation with the rise term and the negative ramp intensity. By identifying and updating the three parameters online, the dynamic characteristics of the electrolyzer under different operating periods and working conditions can be adapted in real time, ensuring that the purity peak prediction can accurately track the actual purity change trend and accurately capture the negative ramp purity peak window.
[0032] For example, when the dissolved hydrogen inventory proxy status is in the current sampling period Electrolysis current extremely small amount At that time, the calculated result of the lift term was 500 divided by 480, which is approximately 1.04. The parameters were then identified and updated online. , , Current cycle negative ramp intensity Substituting into the formula, the predicted purity peak can be obtained. This prediction can accurately reflect the transient peak level of gas purity on the anode side under the current operating conditions, providing reliable data support for subsequent risk assessment.
[0033] It should be noted that the rise term refers to the characterization quantity formed by the ratio between the dissolved hydrogen inventory proxy state and the electrolysis current. It is used to characterize the relative amplification effect caused by the rapid decrease of the electrolysis current when the electrolysis current is reduced, while the dissolved hydrogen inventory proxy state remains at a high level due to mass transfer lag. This reflects the short-term inconsistency of the oxygen generation and dilution capacity decreasing first and the hydrogen release decreasing later. The purity peak prediction quantity is a characterization quantity for predicting the transient abnormal trend of gas purity on the anode side, constructed based on the rise term and the negative ramp intensity. It is obtained by linearly combining the rise term and the negative ramp intensity. It is used to characterize the probability and intensity of the appearance of the purity peak window during the negative ramp in advance and to provide input for the subsequent calculation of the risk normalization index and the shaping of the electrolysis current reduction edge.
[0034] In an embodiment of the present invention, the purity peak prediction is robustly normalized within a rolling window to obtain a risk normalization index, including: A fixed-length scrolling window is used to store purity peak prediction data for multiple consecutive sampling periods. The scrolling window adopts a first-in, first-out sliding method, filling it with the purity peak prediction data collected in the initial stage. In each sampling period, the latest purity peak prediction is written into the scrolling window and the oldest historical data is removed, keeping the scrolling window length constant. The main controller first extracts all purity peak prediction data within the scrolling window and calculates the median. This median is used as the baseline value, which is used to characterize the normal operating level of the purity peak prediction under the current operating conditions. It can eliminate abnormal data interference and form an adaptive operating benchmark. Then, the absolute deviation between each purity peak prediction and the baseline value is calculated within the scrolling window. The median of all absolute deviation data is then calculated and used as the deviation benchmark value. The deviation benchmark value is used to characterize the fluctuation range of the purity peak prediction under normal operating conditions. Using the median to calculate the baseline value and the deviation benchmark value can improve the robustness of the overall calculation and avoid the impact of extreme data or instantaneous interference on the risk assessment results.
[0035] The difference between the current sampling period's purity peak prediction and the baseline value is calculated. This difference is then compared to zero using a larger function to obtain the final risk normalization index for the current sampling period. The formula for calculating the risk normalization index is as follows:
[0036] in This is the risk normalization index for the current sampling period, used to quantify the risk level of transient gas purity spikes induced by negative ramping under the current operating conditions. A higher value indicates a higher risk of purity spikes. This is the purity peak prediction for the current sampling period, used to reflect the purity peak prediction level at the current moment. This is the set of all purity peak predictions stored within the scrolling window. The baseline value is the median of the predicted purity spike within the rolling window. The median of the absolute deviation between the purity peak prediction and the baseline value within the rolling window is the deviation benchmark value. The max function is used to ensure that the risk normalization index only takes a valid positive value when the purity peak prediction is higher than the baseline value, and returns to zero when the purity peak prediction is lower than or equal to the baseline value, thus eliminating invalid calculation results under risk-free conditions. This calculation method can adaptively adjust the risk benchmark according to the actual operating conditions of the system, and is suitable for different electrolytic cell models, different sensor calibration states, and different wind condition fluctuation scenarios.
[0037] It should be noted that the risk normalization index is a dimensionless quantity used to characterize the positive deviation of the current sampling period's purity peak prediction from the recent normal level. It is achieved by extracting the median of the purity peak prediction within the rolling window as the baseline level, and extracting the median of the absolute deviation of the purity peak prediction relative to this baseline as a robust scale. This allows for adaptive normalization of the fluctuation amplitude under different operating conditions without relying on a fixed threshold. When the purity peak prediction of the current sampling period is not higher than the baseline level, the risk normalization index is set to zero to suppress the response to fluctuations in non-risk directions. When the purity peak prediction of the current sampling period is higher than the baseline level, the ratio of its difference from the baseline to the robust scale is used as the risk normalization index. This index reflects the relative strength of the negative climbing purity peak window and serves as a risk driver for the formation of the down-edge.
[0038] In an embodiment of the present invention, a minimum feasible current is determined based on the planned current, operating data, and energy storage capacity of the current cycle. The minimum feasible current is then adjusted downwards according to a risk normalization index to generate a target electrolysis current command, including: After calculating the risk normalization index in each sampling period, the system retrieves various operational data for the current sampling period and the slow-scale planned current issued by the external dispatch system from the storage unit. The total system load power is the total power consumed by all system loads except the electrolysis load, representing the current power demand of the non-electrolysis portion of the system. The planned current is the slow-scale electrolysis current target value set by the external dispatch system or maintenance personnel, serving as a benchmark for electrolysis load adjustment. Subsequently, the total system power demand is calculated as the sum of the total system load power and the planned electrolysis power. The planned electrolysis power is obtained by multiplying the planned current by the electrolytic cell voltage. This product accurately reflects the power consumed by the electrolytic cell under the current planned operating conditions. Subtracting the currently available wind power from the total system power demand yields the initial power gap. The initial power gap represents the power difference when the available wind power cannot meet the total system power demand under the current operating conditions. The formula for calculating the initial power gap is as follows: ,in This represents the initial power gap for the current sampling period. If the value is positive, it indicates that the available wind power is insufficient and the gap needs to be made up by discharging energy storage or reducing the electrolytic load. If the value is negative, it indicates that the available wind power is excessive and can be used for energy storage charging or to maintain the normal operation of the electrolytic load. This represents the total system load power during the current sampling period. This represents the electrolytic cell voltage during the current sampling period. This represents the planned current for the current sampling period. This represents the available wind power during the current sampling period.
[0039] After completing the initial power gap calculation, the upper limit of the energy storage charging and discharging power for the current sampling period is retrieved. This parameter is the maximum charging and discharging power that the energy storage unit can safely output, calculated in real time by the energy storage converter control unit based on the current SOC state of the energy storage unit, the battery cell voltage, and temperature parameters. It is used to characterize the maximum power gap compensation capability that the energy storage system can bear. Subsequently, the upper limit of the energy storage charging and discharging power is subtracted from the initial power gap to obtain the power gap portion that the energy storage system cannot bear. Since the reduced electrolytic power cannot be negative (i.e., it is zero when no reduction in electrolytic power is required), the maximum value between this difference and zero is taken as the reduced electrolytic power. The reduced electrolytic power is used to characterize the power gap that must be compensated by reducing the electrolytic current. The formula for calculating the reduced electrolytic power is as follows: , in the formula This represents the electrolysis power reduction during the current sampling period. The maximum power limit for energy storage charging and discharging during the current sampling period is defined by the `max` function. This function ensures that when the initial power gap is less than or equal to the maximum power limit for energy storage charging and discharging, the energy storage system can fully handle the power gap without reducing the electrolysis power. In this case, the reduced electrolysis power is zero. When the initial power gap is greater than the maximum power limit for energy storage charging and discharging, the excess must be compensated by reducing the electrolysis power. In this case, the value is the power value of the excess portion. This calculation logic strictly follows the power balance principle of off-grid systems, prioritizing the use of energy storage systems to compensate for the power gap and only reducing the electrolysis load when the energy storage capacity is insufficient. This ensures the stability of the off-grid power supply network while minimizing the impact of electrolysis load reduction on hydrogen production efficiency.
[0040] The system retrieves the calculated reduction in electrolytic power and the real-time collected electrolytic cell voltage from the storage unit for the current sampling period. The reduced electrolytic power represents the power gap that must be compensated by lowering the electrolytic current, used to determine the magnitude of power reduction required for the electrolytic load. The electrolytic cell voltage is the voltage value collected in real-time by isolated voltage sensors connected in parallel across the positive and negative terminals of the electrolytic cell. Its value is stable and accurately reflects the current operating voltage state of the electrolytic cell. Since the electrolytic power, electrolytic current, and electrolytic cell voltage satisfy the power calculation formula (i.e., electrolytic power equals the product of electrolytic current and electrolytic cell voltage), it can be deduced that the electrolytic current equals the ratio of electrolytic power to electrolytic cell voltage. Based on this principle, the ratio of the reduced electrolytic power to the electrolytic cell voltage is calculated to obtain the amount of current to be reduced. This amount of current reduction is used to characterize the specific value by which the electrolytic current needs to be reduced to compensate for the corresponding power gap.
[0041] After calculating the reduced current, the planned current for the current sampling period is retrieved. This planned current is the slow-scale electrolysis current target value set by the external scheduling system or maintenance personnel, serving as the benchmark for electrolysis load adjustment. Then, the difference between the planned current and the calculated reduced current is calculated to obtain the minimum feasible current. The minimum feasible current characterizes the minimum current value that the electrolyzer can operate under current conditions to compensate for the power gap and maintain system power balance. The core of this calculation logic is to subtract the necessary reduced current from the planned current, ensuring that the reduced electrolysis current accurately compensates for the power gap that energy storage cannot handle, while also ensuring that the electrolyzer does not shut down or experience operational abnormalities due to excessively low current. The formula used for calculating the minimum feasible current is as follows: , in the formula The lowest feasible current for the current sampling period; The minimum feasible current is the planned current for the current sampling period. The max function is used to ensure that the minimum feasible current does not become negative, avoiding abnormal electrolysis current commands due to calculation deviations. When the initial power gap is small and the amount of current reduction is less than the planned current, the minimum feasible current is the difference between the planned current and the amount of current reduction. When the initial power gap is extremely large and the amount of current reduction is greater than the planned current, the minimum feasible current is zero. In this case, the operation of the electrolytic cell needs to be suspended to ensure system stability.
[0042] It should be noted that the initial power gap refers to the difference between the total power supply required by the off-grid power supply network to meet the total system load power (excluding electrolysis load) and the electrolysis cell power demand when the electrolyzer is driven by the planned current during the current sampling period, and the available wind power. This difference reflects the degree of instantaneous supply and demand imbalance in the system before considering energy storage compensation and determines the instability trend of the bus voltage frequency or DC bus voltage. The reduced electrolysis power refers to the amount of power that must be reduced from the electrolysis load side, corresponding to the remaining power gap that cannot be covered by wind power and energy storage even after considering the compensation capacity allowed by the upper limit of energy storage charging and discharging power. It reflects the need to ensure power supply. Under the premise of network stability, the power share that the electrolytic load needs to give up; the reduced current is the amount of reduction in electrolytic current calculated by converting the reduced electrolytic power according to the current electrolytic cell voltage. It is used to convert the power reduction demand into the amount of change in current command that the electrolytic power supply can execute and directly determines the reduction in the electrolytic reaction rate; the minimum feasible current refers to the minimum electrolytic current that the electrolytic cell is allowed to maintain under the constraints of the available wind power and energy storage compensation capacity in the current sampling period, in order to avoid power shortage leading to power grid instability. It is obtained by subtracting the reduced current from the planned current and serves as the basis current boundary for subsequent reductions to balance electrical feasibility and gas-side risk control.
[0043] The risk normalization index calculated for the current sampling period is retrieved from the storage unit. This risk normalization index is used to quantify the risk level of the transient peak in gas purity induced by the negative ramp under the current operating condition. The larger the value, the higher the risk of the purity peak. Then, the constant 1 is added to the risk normalization index to obtain the risk weighting coefficient. The constant 1 serves as the benchmark weight, ensuring that the risk weighting coefficient is 1 when the risk normalization index is zero. At this time, the downward adjustment of the electrolytic current is not slowed down. When the risk normalization index increases, the risk weighting coefficient increases synchronously to enhance the weight of the electrolytic current in the previous sampling period, thereby slowing down the current adjustment. This calculation logic can achieve an adaptive correlation between risk and current adjustment speed. The higher the risk, the smoother the adjustment, effectively suppressing the purity peak.
[0044] After calculating the risk weighting coefficient, the minimum feasible current of the current sampling period and the electrolytic current of the previous sampling period are retrieved. The minimum feasible current is the minimum electrolytic current required to compensate for the power gap under the current operating conditions, serving as the baseline for reducing the electrolytic current. The electrolytic current of the previous sampling period is the actual operating current of the electrolytic cell at the previous moment, stored in the historical data unit of the main controller, and used as a benchmark for current reduction to avoid instantaneous current jumps. Subsequently, the risk weighting coefficient is used as a weighting factor to calculate a weighted average of the minimum feasible current and the electrolytic current of the previous sampling period to obtain the target electrolytic current command. The core logic of this weighted average calculation is that the larger the risk weighting coefficient, the greater the weight of the electrolytic current of the previous sampling period, the closer the target electrolytic current command is to the current of the previous period, and the slower the current reduction speed, thereby weakening the purity spike during negative ramp-up; the smaller the risk weighting coefficient, the greater the weight of the minimum feasible current, and the closer the current reduction speed is to the normal reduction speed under no-risk conditions, balancing power gap compensation and gas safety. The formula for calculating the target electrolytic current command is as follows:
[0045] in the formula The target electrolysis current command for the current sampling period is used to directly send it to the electrolysis power supply to control the operating current of the electrolytic cell; The lowest feasible current for the current sampling period; The value is the electrolytic current of the previous sampling period; the denominator 1 is added to the risk weighting coefficient to normalize the weighted average result, ensuring that the value of the target electrolytic current command is between the minimum feasible current and the electrolytic current of the previous sampling period, thus avoiding abnormal current jumps.
[0046] It should be noted that the target electrolysis current command refers to the current setting value that is finally issued to the electrolysis power supply to drive the electrolytic cell within the current sampling period. This setting value uses the lowest feasible current as the electrical feasible boundary and combines the risk normalization index to shape the steepness of the current reduction edge, so that it can still meet the reduction requirements required for the stability of the power supply network when the available wind power is insufficient or a power gap occurs. At the same time, it mitigates the inconsistency between the rapid decline in oxygen generation rate and the delayed release of dissolved hydrogen caused by the sudden drop in electrolysis current, thereby reducing the probability and amplitude of peak fluctuations in the purity of the anode side in a short period of time and providing direct input for subsequent energy storage power sharing calculations.
[0047] In an embodiment of the present invention, the real-time power gap of the power supply network is assessed based on the target electrolysis current command, and an energy storage power command is generated, including: Retrieve the operating data of the current sampling period and the calculated target electrolysis current command to calculate the real-time total power demand of the system. The real-time total power demand of the system is the sum of the total system load power and the current actual electrolysis power. The current actual electrolysis power is obtained by multiplying the target electrolysis current command and the electrolytic cell voltage. This product can accurately reflect the actual power consumed by the electrolytic cell under the current target operating conditions. Then, subtract the currently available wind power from the real-time total power demand of the system to obtain the real-time power gap of the power supply network. The real-time power gap of the power supply network is used to characterize the power difference between the current operating conditions where the available wind power cannot meet the real-time total power demand of the system (gap is positive) or the current operating conditions where the available wind power is excessive (gap is negative). The formula used to calculate the real-time power gap of the power supply network is as follows:
[0048] in This represents the real-time power gap of the power supply network during the current sampling period. When the value is positive, it indicates that the available wind power is insufficient and the energy storage system needs to discharge to make up for the gap. When the value is negative, it indicates that the available wind power is excessive and the excess energy can be stored by charging the energy storage system. This represents the total system load power during the current sampling period. This represents the electrolytic cell voltage during the current sampling period. The target electrolysis current command for the current sampling period; This represents the available wind power during the current sampling period.
[0049] After completing the real-time power gap calculation of the power supply network, the main controller retrieves the upper limit of energy storage charging and discharging power for the current sampling period. This parameter is the maximum charging and discharging power that the energy storage unit can safely output and input, calculated in real time by the energy storage converter control unit based on the current SOC state of the energy storage unit, the voltage and temperature parameters of the individual battery cells. It is used to limit the operating power of the energy storage system and prevent damage to the energy storage units due to overcharging, over-discharging, or power overload. Subsequently, the real-time power gap of the power supply network is limited between the upper limit of energy storage charging and discharging power and the negative upper limit of energy storage charging and discharging power, resulting in an energy storage power command. The core logic of this limiting operation is to ensure that the energy storage system always operates within a safe power range, which can fully utilize energy storage capacity to make up for the power gap or store excess wind power, while protecting the energy storage units. The power supply extends its service life and is used to directly control the charging and discharging behavior of the energy storage unit by sending a command to the energy storage converter. When the value is positive, the command instructs the energy storage system to discharge to make up for the power gap; when the value is negative, the command instructs the energy storage system to charge and store excess wind power. When the initial real-time power gap is greater than the upper limit of the energy storage charging and discharging power, the energy storage power command takes the upper limit of the energy storage charging and discharging power to ensure that the energy storage system does not discharge under load. When the initial real-time power gap is less than the negative upper limit of the energy storage charging and discharging power, the energy storage power command takes the negative upper limit of the energy storage charging and discharging power to ensure that the energy storage system does not charge under load. When the initial real-time power gap is between the positive and negative upper limits of the energy storage charging and discharging power, the energy storage power command is equal to the real-time power gap of the power supply network, realizing accurate compensation of the power gap or full storage of excess wind power.
[0050] It should be noted that the real-time power gap of the power supply network refers to the difference between the total power supply required by the off-grid power supply network to meet the total system load power (excluding electrolysis load) and the power demand of the electrolytic cell when the electrolytic cell is driven according to the target electrolysis current command within the current sampling period, and the available wind power. This difference reflects the magnitude of the instantaneous supply and demand imbalance of the power supply network under the condition that the target electrolysis load is determined, and corresponds to the deviation trend of the bus voltage frequency or DC bus voltage. The energy storage power command refers to the charging and discharging power setting value generated and issued to the energy storage converter according to the real-time power gap of the power supply network. It is used to compensate for the power gap or absorb the surplus wind power within its available charging and discharging power range, thereby realizing the power sharing of parallel power supply of wind power and energy storage and providing instantaneous support to the power supply network to maintain stable operation.
[0051] like Figure 2 The diagram shown is a functional block diagram of a multi-timescale coordinated control system for an off-grid wind power hydrogen production system provided in an embodiment of the present invention.
[0052] In this embodiment, the functions of each module / unit are as follows: The negative ramp calculation module is used to acquire the system's operating data and calculate the negative ramp intensity based on the rate of decrease of the electrolytic current in the operating data. The peak prediction module is used to construct the dissolved hydrogen inventory proxy state based on the electrolysis current and the negative ramp intensity, and generate the purity peak prediction. The risk normalization module is used to robustly normalize the predicted purity peak within a rolling window to obtain the risk normalization index. The current shaping module is used to determine the minimum feasible current based on the planned current, operating data and energy storage capacity of the current cycle, and to shape the minimum feasible current by a downward adjustment according to the risk normalization index to generate the target electrolysis current command. The energy storage scheduling module is used to assess the real-time power gap of the power supply network based on the target electrolysis current command and generate energy storage power commands.
[0053] like Figure 3 The diagram shown is a schematic topology of an off-grid wind power hydrogen production system according to an embodiment of the present invention. It includes a wind turbine generator, a wind power converter, an energy storage unit, an energy storage converter, and an off-grid power supply bus. The wind power converter and the energy storage converter are connected in parallel to the off-grid power supply bus to supply power to the system's conventional loads and user power consumption. Simultaneously, the off-grid power supply bus supplies power to an electrolysis power source, which provides controlled DC power to the electrolyzer to drive the electrolysis process. The electrolyzer is equipped with a cooling system to maintain a stable operating temperature. Electrolysis current measurement points and electrolyzer voltage measurement points are set on the electrolyzer side, and an anode gas purity monitoring point is set on the anode gas pipeline. After the gas produced by the electrolyzer is separated into gas and liquid by a gas-liquid separator, the hydrogen on the anode side passes through a hydrogen purification device to obtain a hydrogen outlet and enters the hydrogen storage unit, while the gas on the cathode side is separated into gas and liquid to form an oxygen outlet. The multi-timescale coordination controller is connected to the above measurement points to obtain operating data such as electrolysis current, electrolyzer voltage, and anode gas purity. At the same time, it obtains operating information from the wind power side and the energy storage side, and issues target electrolysis current commands to the electrolysis power source and energy storage power commands to the energy storage converter. It can also coordinate the operation of the wind power converter, thereby realizing power sharing of wind power and energy storage parallel power supply and dynamic adjustment of hydrogen production load under off-grid power supply conditions.
[0054] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.
Claims
1. A multi-timescale coordinated control method for an off-grid wind power hydrogen production system, characterized in that, include: S1. Obtain the system's operating data and calculate the negative ramp intensity based on the rate of decrease of the electrolytic current in the operating data; S2. Construct the dissolved hydrogen inventory proxy state based on electrolysis current and negative ramp intensity, and generate purity peak prediction. S3. The predicted purity peak is robustly normalized within the rolling window to obtain the risk normalization index. S4. Determine the minimum feasible current based on the planned current, operating data and energy storage capacity of the current cycle, and shape the minimum feasible current by adjusting the lower edge according to the risk normalization index to generate the target electrolysis current command. S5. Evaluate the real-time power gap of the power supply network based on the target electrolysis current command, and generate energy storage power command.
2. The multi-timescale coordinated control method for an off-grid wind power hydrogen production system according to claim 1, characterized in that, The operational data includes available wind power, upper limit of energy storage charging and discharging power, electrolysis current, electrolytic cell voltage, and total system load power excluding electrolysis load.
3. The multi-timescale coordinated control method for an off-grid wind power hydrogen production system according to claim 2, characterized in that, The negative ramp intensity is calculated based on the rate of decrease in electrolytic current from the operating data, including: Calculate the difference between the electrolytic current of the previous sampling period and the electrolytic current of the current sampling period. If the difference is less than zero, the negative ramp intensity is set to zero; otherwise, the ratio of the difference to the sampling period is taken as the negative ramp intensity.
4. The multi-timescale coordinated control method for an off-grid wind power hydrogen production system according to claim 1, characterized in that, Based on the electrolysis current and negative ramp intensity, a dissolved hydrogen inventory proxy state is constructed, and a purity peak prediction is generated according to the dissolved hydrogen inventory proxy state, including: The dissolved hydrogen inventory proxy status of the previous sampling period is recursively updated by using the online updated equivalent time constant and the electrolysis current of the current sampling period to obtain the dissolved hydrogen inventory proxy status of the current sampling period. The upside term is obtained based on the ratio of the dissolved hydrogen inventory proxy status to the electrolysis current during the current sampling period; The purity peak prediction is generated by linearly combining the uplift term and negative ramp intensity by identifying and updating parameters online.
5. The multi-timescale coordinated control method for an off-grid wind power hydrogen production system according to claim 4, characterized in that, The equivalent time constant and the parameters updated online are updated online using the recursive least squares method. The online update includes: fitting the online measured anode-side gas purity with the purity peak prediction to generate a prediction error; and updating the parameter vector using the gain calculated from the regression vector and the prediction error.
6. The multi-timescale coordinated control method for an off-grid wind power hydrogen production system according to claim 1, characterized in that, The purity peak prediction is robustly normalized within the rolling window to obtain a risk normalization index, including: The median of the predicted purity spike within the set rolling window is used as the baseline value; Calculate the absolute deviation between the predicted purity peak and the baseline value, and use the median of this absolute deviation within the rolling window as the deviation baseline value; Calculate the difference between the purity peak prediction and the baseline value for the current sampling period. If the difference is less than zero, the risk normalization index is set to zero; otherwise, the ratio of the difference to the deviation benchmark value is used as the risk normalization index.
7. The multi-timescale coordinated control method for an off-grid wind power hydrogen production system according to claim 2, characterized in that, The minimum feasible current is determined based on the planned current, operating data, and energy storage capacity for the current cycle, including: The initial power gap is obtained by subtracting the available wind power from the sum of the products of the total system load power and planned current in the operating data and the electrolyzer voltage in the operating data. Subtract the upper limit of energy storage charging and discharging power in the operating data from the initial power gap, and take the maximum value of the difference and zero as the reduced electrolytic power; The amount of current to be reduced is obtained by the ratio of the reduced electrolytic power to the electrolytic cell voltage. The difference between the planned current and the reduced current is taken as the minimum feasible current.
8. The multi-timescale coordinated control method for an off-grid wind power hydrogen production system according to claim 7, characterized in that, Based on the risk normalization index, the minimum feasible current is shaped by a downward adjustment to generate the target electrolysis current command, including: Add 1 to the risk normalization index to obtain the risk weighting coefficient; Using the risk weighting coefficient as a weighting factor, the target electrolysis current command is obtained by weighted averaging the minimum feasible current and the electrolysis current of the previous sampling period.
9. The multi-timescale coordinated control method for an off-grid wind power hydrogen production system according to claim 2, characterized in that, Based on the assessment of the real-time power gap in the power supply network using the target electrolysis current command, an energy storage power command is generated, including: The real-time power gap of the power supply network is obtained by subtracting the available wind power from the sum of the products of the total system load power, the target electrolysis current command and the electrolyzer voltage in the operating data. The energy storage power command is obtained by limiting the real-time power gap of the power supply network between the upper limit of energy storage charging and discharging power and the negative upper limit of energy storage charging and discharging power.
10. A multi-timescale coordinated control system for an off-grid wind power hydrogen production system, applied in the multi-timescale coordinated control method for an off-grid wind power hydrogen production system as described in any one of claims 1-9, characterized in that, The off-grid wind power hydrogen production system's multi-timescale coordinated control system includes: The negative ramp calculation module is used to acquire the system's operating data and calculate the negative ramp intensity based on the rate of decrease of the electrolytic current in the operating data. The peak prediction module is used to construct the dissolved hydrogen inventory proxy state based on the electrolysis current and the negative ramp intensity, and generate the purity peak prediction. The risk normalization module is used to robustly normalize the predicted purity peak within a rolling window to obtain the risk normalization index. The current shaping module is used to determine the minimum feasible current based on the planned current, operating data and energy storage capacity of the current cycle, and to shape the minimum feasible current by a downward adjustment according to the risk normalization index to generate the target electrolysis current command. The energy storage scheduling module is used to assess the real-time power gap of the power supply network based on the target electrolysis current command and generate energy storage power commands.
