A method for controlling the rectifier power supply of an electrolyzer for hydrogen production from renewable energy sources.

Abstract

This invention discloses a rectifier power supply control method for electrolyzers used in renewable energy hydrogen production, relating to the field of water electrolysis hydrogen production control technology. The method calculates output power by collecting voltage and current data and combining them with an adaptive weighted moving average algorithm. It collects the electrolyzer's start-up and shutdown status, current, voltage, temperature, and bus electrical parameters to construct an input parameter set. This set is then input into a scheduling model, which, combined with historical operating data and environmental prediction data, predicts the available power for the next cycle, generating start-up and shutdown control commands and target operating current settings. Dynamic current regulation and access control are executed to achieve real-time matching between the load and power supply. The system has a limit correction and operational feedback mechanism, supporting adaptive model updates. This invention possesses forward-looking prediction, adaptive scheduling, multi-objective optimization, and fault-tolerant management capabilities, improving the system's adaptability to fluctuating power supplies and the intelligence level of rectifier control. It is applicable to hydrogen production systems from renewable energy sources such as wind power and photovoltaics.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of hydrogen production by electrolysis of water control, and particularly relates to a rectifier power supply control method suitable for hydrogen production by electrolysis of water based on renewable energy. BACKGROUND

[0002] With the widespread deployment of renewable energy such as wind power and photovoltaic power, systems for directly driving hydrogen production by electrolysis of water based on renewable energy have become the focus of current green hydrogen development. However, unlike conventional power grid power supply, renewable energy has significant volatility, intermittency and unpredictability in output power, which puts higher requirements on the steady-state operation and power supply matching of electrolytic cells.

[0003] In the prior art, the control system of the electrolytic cell rectifier power supply usually adopts a fixed parameter setting or a periodic static scheduling strategy. Its control is based on the simple collection of the current power value, and lacks the ability to judge the power fluctuation trend, link analysis of the device state and dynamic correction. The specific problems are as follows: power response lag: the existing system does not perform dynamic smoothing processing on fluctuating power, and directly uses the instantaneous power value for scheduling, which leads to lag or drastic changes in current control, affecting hydrogen production efficiency and equipment life; single control model: the traditional strategy only makes start-stop judgments based on the current power level, without considering historical operation state, environmental factors and available power prediction and other multi-dimensional inputs, and the scheduling model lacks foresight and precision; fragmented execution mechanism: the current control and unit start-stop control paths are separated, and there is no closed-loop logic of parameter collection-scheduling calculation-execution control-feedback correction, which leads to deviation of actual output from the set target and accumulation of errors in long-term operation; lack of fault tolerance and self-adaptive ability: when facing atypical working conditions such as voltage abnormalities and load mutations, the system cannot quickly switch control strategies or perform self-recovery scheduling, and also lacks the ability to adjust and learn model parameters online; low system intelligence: no multi-objective optimization mechanism is established, which cannot balance energy consumption, hydrogen production, electrolytic cell service life and other indicators, the overall control strategy is extensive, and the scheduling precision is limited.

[0004] Therefore, there is an urgent need for a rectifier power supply control method that can accurately identify fluctuating power, link multiple dimensions of electrolytic cell operation state, adaptively update control models and optimize multiple targets, to realize efficient dynamic matching control of electrolytic load and ensure safe, stable and economic operation of the system. SUMMARY

[0005] The present application aims to solve the technical problems of output power response lag, control model rigidity, scheduling execution decoupling, lack of fault tolerance and lack of adaptive and optimization mechanism in the prior art, and proposes an electrolytic cell rectifier power control method suitable for renewable energy hydrogen production, which realizes accurate identification of dynamic power input, efficient control matching of electrolytic load and multi-objective optimization of system scheduling performance.

[0006] In order to achieve the above-mentioned purpose, the present application is realized by the following technical scheme:

[0007] An electrolytic cell rectifier power control method suitable for renewable energy hydrogen production, comprising the following steps:

[0008] Obtain the output power data of the current period from the renewable energy output end, which is calculated by adaptive weighted moving average algorithm from the collected voltage and current; obtain the start-stop state, current, voltage and operating temperature of the electrolytic cell unit from a plurality of electrolytic cell units, and collect the voltage and frequency of the system bus, to form an input parameter set for control scheduling calculation;

[0009] Input the input parameter set into the control scheduling model, which combines historical operation data and environmental prediction data to predict the available power of renewable energy in the next period, and based on the prediction result, generate the start-stop control instruction and target working current set value of each electrolytic cell unit as the basis for control execution;

[0010] According to the start-stop control instruction, control the access and exit state of each electrolytic cell unit, and control the output current of the electrolytic cell unit to the target working current set value, to realize the dynamic matching between the load side output current and the renewable energy output power;

[0011] Collect the actual operating parameters of each electrolytic cell unit, and compare the deviation with the output result of the control scheduling model; if the deviation exceeds the set tolerance threshold, the output current is limited and corrected by the local controller, and the collected operating parameters are fed back to the control scheduling model for next control period update.

[0012] Further improvement of the present application is that the output power data is obtained by collecting voltage and current and calculating by adaptive weighted moving average algorithm, and the expression of adaptive weighted moving average algorithm is:

[0013] ;

[0014] Wherein: is the length of the moving average window in the current control period; is the maximum moving window length; Adjust the sensitivity coefficient for the window; , This refers to the output power of the current cycle and the previous cycle. The sampling time interval;

[0015] The adaptive weighted moving average algorithm dynamically shortens the sliding window when the output power fluctuates drastically and automatically extends the window when the power is stable.

[0016] A further improvement of the present invention is that the generation of start-stop control commands and target operating current settings for each electrolytic cell unit includes the following steps:

[0017] Based on the current cycle's input parameter set, historical operating data, and environmental forecast data, predict the available renewable energy power for the next control cycle. The calculation formula is:

[0018] ;

[0019] in: The current period's moving average output power; This is a correction term based on the load response error of the previous cycle; This is the predicted power term calculated based on the current cycle's environmental data; , , For the weighting coefficients, satisfying ;

[0020] The control and scheduling model is based on the predicted power. Based on the operating status of each electrolytic cell unit in the current cycle, perform load optimization allocation and load optimization scheduling, and generate start / stop control commands and target operating current settings for each electrolytic cell unit.

[0021] A further improvement of the present invention is that the execution of the start / stop control command and the target operating current setpoint is subject to the following control conditions:

[0022] When the system bus voltage is lower than the set safety threshold, the electrolytic cell unit with a large load response delay is shut down first, and all target operating current adjustment commands are temporarily suspended.

[0023] If the change in the target operating current setting relative to the actual output current of the previous control cycle exceeds the set rate threshold, a ramp function limiting strategy is adopted to perform current regulation in stages.

[0024] The control commands are validated by the local electrolytic cell controller. The validation includes hardware constraints such as operating temperature, inter-electrode voltage drop, and current response rate. When the validation passes, the corresponding start / stop or current adjustment command is executed.

[0025] A further improvement of the present invention is that the local controller performs amplitude limiting correction on the output current, collects the actual output current, voltage and temperature data of the current cycle, and compares the deviation of the collected data with the target value output by the control scheduling model. If the deviation exceeds the set tolerance threshold, the target output current is amplitude limited and corrected at the local controller level. The operating parameters collected in the current cycle are fed back to the control scheduling model to update the input parameter set for the next cycle. The feedback data structure is consistent with the original input parameter set structure.

[0026] A further improvement of the present invention is that when the system bus voltage or frequency deviates from the set normal operating range, the control scheduling model executes a fault-tolerant control compensation strategy, specifically including:

[0027] Initiate a rapid load reduction mechanism to prioritize the shutdown of some electrolytic cell units with higher power but slower response rates;

[0028] Linear backoff limiting is applied to the target operating current setting value of the remaining operating units;

[0029] The compensation and adjustment process maintains a dynamic balance without affecting the overall hydrogen production target of the system.

[0030] A further improvement of the present invention is that, when generating the target operating current setpoint for each electrolytic cell unit, the control scheduling model dynamically allocates load weights based on the priority and historical operating performance of each electrolytic cell unit. The formula is:

[0031] ;

[0032] in: Indicates the first The load allocation weight of each electrolytic cell unit in the current cycle; This indicates the number of electrolytic cell units currently in operation; , These are the minimum and maximum values ​​for the load allocation weight, respectively.

[0033] The load allocation weight is used to proportionally allocate the total target output current corresponding to the predicted available power to each electrolytic cell unit.

[0034] A further improvement of the present invention is that, when performing load optimization scheduling, the control scheduling model employs a multi-objective constraint optimization mechanism to jointly optimize the target operating current setpoint, and the optimization objectives include:

[0035] Improve the matching degree between the total load current in the current cycle and the predicted power of renewable energy;

[0036] Balance the cumulative operating time of each electrolytic cell unit to slow down the aging rate of the unit;

[0037] Reduce the energy consumption per unit of hydrogen production;

[0038] The objective function is optimized using a multi-objective weighted objective function method. Represented as:

[0039] ;

[0040] in: To calculate the difference between the predicted power and the target load power; The variance of the cumulative operating time of each electrolytic cell; Electricity consumption per unit of hydrogen production; , , To optimize the weighting coefficients and satisfy .

[0041] A further improvement of the present invention is that the control scheduling model supports periodic adaptive updates, and the running parameter update methods include:

[0042] Based on historical control errors, system response delays, and environmental change trends, a performance evaluation function is constructed to obtain evaluation indicators.

[0043] When the evaluation metric falls below a set threshold, the model structure or parameters are automatically updated, including:

[0044] Adjust the length of the moving average window;

[0045] Weighting coefficients , , Automatic reallocation;

[0046] A local fitting algorithm is introduced to optimize the environment prediction function;

[0047] The update process is deployed online without interrupting control execution.

[0048] A further improvement of the present invention is that the control scheduling model supports dynamic switching and redundancy management at the electrolyzer unit level, specifically including:

[0049] When an electrolytic cell unit is detected to have an operational malfunction, over-temperature, response mismatch, or communication abnormality, the electrolytic cell unit will be automatically removed from the scheduling pool and put into standby mode.

[0050] Select the electrolytic cell unit with the best status and shortest running time from the standby unit pool to take over the operation task, and automatically inherit the original unit's target operating current setting value;

[0051] After the malfunctioning electrolytic cell unit is restored, it enters the observation area and is reinstated into the scheduling system once it meets the requirements for a stable operating cycle.

[0052] The beneficial effects of this invention are as follows: Addressing the technical challenges of large fluctuations in renewable energy output power and high dispatch response requirements, this invention achieves refined dynamic management of the electrolytic cell group by constructing an adaptive and predictive driven control mechanism. Firstly, by collecting voltage and current data and combining them with an adaptive weighted moving average algorithm, the output power of the current cycle is dynamically smoothed, making the power input data more stable and reliable, and improving the control system's adaptability to external disturbances. Based on this, the system synchronously collects the start-up and shutdown status, current, voltage, temperature, and bus electrical parameters of each electrolytic cell unit, forming a complete set of input parameters, providing comprehensive data support for subsequent dispatch. A significant feature of this invention is the introduction of a control dispatch model that combines historical operating data and environmental prediction data, enabling forward-looking prediction of available power for the next cycle. After obtaining the prediction results, the dispatch model generates start-up and shutdown control commands and target operating current setpoints for each electrolytic cell unit based on multi-objective optimization logic, thereby achieving on-demand allocation and refined control of electrolytic load resources, effectively avoiding resource waste and system losses caused by frequent start-ups and shutdowns. Furthermore, this invention dynamically adjusts the access and exit states of each electrolyzer unit through scheduling commands, and guides the output current to change according to the target value, establishing an efficient matching relationship between power output and load demand. By continuously collecting actual operating parameters of the electrolyzer and comparing them with the model output, the system can activate a local limiting correction strategy when the deviation exceeds the limit, promptly limiting the risk of current surges and ensuring equipment operation safety. It also supports periodic feedback of operating parameters and model updates, possessing a certain degree of self-learning and adaptive capabilities, enabling the control system to continuously optimize control effects as operating conditions change. This invention constructs a highly stable, highly adaptive, and highly energy-efficient hydrogen production power supply regulation method through a multi-level linkage mechanism including power smoothing prediction, state perception, scheduling optimization, deviation correction, and feedback updates. It is particularly suitable for renewable energy scenarios with significant fluctuations, such as wind and photovoltaic power, and has significant engineering application value and promotion prospects. Attached Figure Description

[0053] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Wherein:

[0054] Figure 1 This is a flowchart of the method in an embodiment of the present invention. Detailed Implementation

[0055] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the described embodiments of the present invention are within the scope of protection of the present invention.

[0056] To facilitate understanding of the technical solution of this invention, the following terms are explained in this specification and claims:

[0057] Current cycle: refers to a complete operating cycle in the control system of this invention used for power sampling, parameter updating and control command generation. Its duration can be set according to the system configuration, and the typical value is from several seconds to tens of seconds.

[0058] Target operating current setpoint: refers to the target current parameter generated by the control scheduling model based on predicted power results, system operating status and optimization strategy to guide the operation of each electrolytic cell unit. It is different from the real-time output current and can be used as an adjustment reference or limit the upper limit.

[0059] Input parameter set: refers to all input variables involved in control and scheduling calculations, including output voltage and current from the renewable energy side, voltage and frequency of the system bus, and information such as start-up and shutdown status, electrolysis current, voltage and operating temperature of each electrolyzer unit.

[0060] Adaptive weighted moving average algorithm: This refers to an algorithm mechanism that smooths power signals by adjusting the length of the sliding window and the weighting factor. It can adjust the window length in real time according to the rate of power change, thereby enhancing output stability while ensuring response speed.

[0061] Fault-tolerant control compensation strategy: refers to the rapid response mechanism triggered by the system when abnormalities are detected in the system bus voltage, frequency or electrolytic cell unit operating status. Specifically, it includes operations such as load degradation, unit shutdown, and standby switching to maintain the safe and stable operation of the system.

[0062] Limiting correction: This refers to the process of limiting and adjusting the target current through the local controller when the difference between the target current output by the control scheduling model and the current actual current exceeds a set threshold, in order to avoid overshoot or hardware impact in the tank.

[0063] Multi-objective optimization scheduling: refers to the control scheduling model taking into account multiple optimization objectives when generating the target operating current setpoint, including minimizing power prediction error, balancing electrolytic cell aging and minimizing energy consumption, and achieving dynamic balance through weighting coefficients.

[0064] System bus: refers to the main voltage output channel in an electrolytic rectifier power supply system used to connect multiple electrolytic cell units. It is used to carry and distribute the total output power from the rectifier side. Its voltage and frequency parameters can reflect the system load status and operating fluctuations.

[0065] like Figure 1 As shown, this is an embodiment of the present invention, which provides a method for controlling the rectifier power supply of an electrolyzer suitable for renewable energy hydrogen production, including:

[0066] Step S1: Obtain the output power data for the current cycle from the renewable energy output terminal. The output power data is calculated by the collected voltage and current through an adaptive weighted moving average algorithm.

[0067] In one feasible implementation, the output power data is obtained by collecting voltage and current and calculating using an adaptive weighted moving average algorithm, the expression of which is:

[0068] ;

[0069] in: The moving average window length within the current control period; This is the maximum sliding window length; Adjust the sensitivity coefficient for the window; , This refers to the output power of the current cycle and the previous cycle. The sampling time interval;

[0070] The adaptive weighted moving average algorithm dynamically shortens the sliding window when the output power fluctuates drastically and automatically extends the window when the power is stable.

[0071] Compared to the power smoothing method using a fixed sliding window length commonly used in existing technologies, this embodiment introduces an adaptive adjustment mechanism to dynamically scale the window length according to the current power change rate, which significantly improves the system's response flexibility and smoothness robustness under different power fluctuation conditions.

[0072] Specifically, when a large change in output power is detected between the current cycle and the previous cycle (i.e., a large difference), the system will automatically shorten the sliding window length to improve the ability to track sudden changes in output power; while when the power change is slow and stable, the window length will be automatically widened to enhance the smoothness of the average value and the ability to resist interference.

[0073] This adaptive sliding window mechanism maintains the real-time response capability of output power calculation under high-frequency disturbance scenarios, and ensures the stability and continuity of the power signal within the stable range, effectively avoiding the dual risks of "slow response" or "excessive jitter" in traditional methods.

[0074] Furthermore, the algorithm only requires the introduction of a weight factor based on the rate of change adjustment, which can be flexibly embedded into existing control systems. It has low computational complexity, high real-time performance, and good industrial promotion value and application prospects.

[0075] Step S2: Obtain the start-up and shutdown status, current, voltage and operating temperature of the electrolytic cell units from multiple electrolytic cell units, and collect the voltage and frequency of the system bus to form a set of input parameters for control and scheduling calculations;

[0076] In one feasible implementation, generating the start-stop control command and target operating current setting value for each electrolytic cell unit includes the following steps:

[0077] Based on the current cycle's input parameter set, historical operating data, and environmental forecast data, predict the available renewable energy power for the next control cycle. The calculation formula is:

[0078] ;

[0079] in: The current period's moving average output power; This is a correction term based on the load response error of the previous cycle; This is the predicted power term calculated based on the current cycle's environmental data; , , For the weighting coefficients, satisfying ;

[0080] The control and scheduling model is based on the predicted power. Based on the operating status of each electrolytic cell unit in the current cycle, perform load optimization allocation and load optimization scheduling, and generate start / stop control commands and target operating current settings for each electrolytic cell unit.

[0081] Compared to traditional hydrogen production power systems that rely on the instantaneous power of the current cycle to directly determine start-stop or set the target current based on experience, this embodiment proposes a method for predicting available power for the next cycle, realizing a substantial shift from "response control" to "predictive control".

[0082] This prediction model not only considers the current period's moving average power, but also comprehensively incorporates:

[0083] Based on the historical load error correction term, it is used to offset the scheduling deviation of the previous cycle;

[0084] Environmental power forecasts based on external weather forecasts or load forecasts allow for early perception of future available power trends.

[0085] By introducing the above three components, a prediction formula can be constructed to achieve robust prediction of the renewable energy power supply capacity in the next control cycle.

[0086] Based on the predicted power results and the current operating status of each electrolytic cell (whether it is in operation, aging, cold start, etc.), the control system can use a preset optimization strategy to allocate start-stop control rights and target operating current values ​​to ensure balanced operation of each cell unit load and reasonably distribute the risk of high-frequency fluctuations.

[0087] This scheduling logic not only improves the forward-looking nature of power response, but also ensures the operational stability and life cycle consistency of each electrolyzer unit. While improving the utilization rate of hydrogen production, it effectively avoids electrolyzer fatigue and aging caused by frequent start-ups and shutdowns.

[0088] Step S3: Input the set of input parameters into the control scheduling model. The control scheduling model combines historical operating data and environmental prediction data to predict the available renewable energy power in the next cycle. Based on the prediction results, it generates start-stop control commands and target operating current settings for each electrolyzer unit as the basis for control execution.

[0089] In one feasible implementation, the execution of the start / stop control command and the target operating current setpoint is subject to the following control conditions:

[0090] When the system bus voltage is lower than the set safety threshold, the electrolytic cell unit with a large load response delay is shut down first, and all target operating current adjustment commands are temporarily suspended.

[0091] If the change in the target operating current setting relative to the actual output current of the previous control cycle exceeds the set rate threshold, a ramp function limiting strategy is adopted to perform current regulation in stages.

[0092] All control commands output by the control scheduling model are validated for legality by the local electrolytic cell controller. The validity validation includes hardware constraints such as operating temperature, inter-electrode voltage drop, and current response rate. When the validity validation passes, the corresponding start / stop or current adjustment command is executed.

[0093] When generating the target operating current setpoint for each electrolytic cell unit, the control scheduling model dynamically allocates load weights based on the priority and historical operating performance of each electrolytic cell unit. The formula is:

[0094] ;

[0095] in: Indicates the first The load allocation weight of each electrolytic cell unit in the current cycle; This indicates the number of electrolytic cell units currently in operation; , These are the minimum and maximum values ​​of the weights, respectively.

[0096] The weights are used to proportionally allocate the total target output current corresponding to the predicted available power to each electrolytic cell unit.

[0097] Unlike existing electrolytic cell control methods that "only set the target current value and execute it directly without verification", this embodiment significantly enhances the precision of system scheduling and security protection capabilities through multi-level logical judgment and dynamic restriction mechanisms.

[0098] First, after the target current setpoint is generated, it is not directly applied to the electrolytic cell. Instead, a "maximum allowable rate of change" is introduced as a current ramp-up limiting condition to ensure that the target current does not change too quickly, thus avoiding problems such as electrode impact, abnormal local temperature rise, or uncontrolled gas evolution caused by sudden current surges. This limiting process adopts a segmented slope adjustment strategy, which can flexibly adapt to the cell's response capability under different operating conditions.

[0099] Secondly, before each control command is issued, it must undergo a validity check by the local electrolytic cell controller. The system will check whether the current temperature, voltage, and current rate of the corresponding unit are all within the set safety threshold range. Only when all parameters meet the operating requirements will the target current command be actually executed. This "safety pre-verification mechanism" is generally lacking in traditional methods, which greatly improves the safety and fault tolerance of the operation process.

[0100] Furthermore, to prevent some electrolytic cells from aging prematurely due to prolonged overload, this embodiment sets a reasonable target current allocation weight for each electrolytic cell based on historical operating data such as the load level and usage time of each unit. All weights sum to 1, but the proportion of each cell can dynamically fluctuate within a set minimum and maximum range. This optimization mechanism not only ensures the fairness and stability of power allocation but also achieves "differentiated control" based on the unit's health status, effectively extending the overall lifespan of the equipment and reducing maintenance frequency.

[0101] This invention introduces multiple control levels, including prediction, amplitude limiting, verification, and optimization, into the scheduling strategy. This changes the traditional strategy's crude logic of "one-size-fits-all, no feedback, and no fault tolerance," and achieves fine-grained control at the system level. It balances scheduling accuracy, safety, and equipment economy, and significantly improves the stable operation and intelligent response level of renewable energy hydrogen production systems under fluctuating conditions.

[0102] Step S4: Control the access and exit status of each electrolytic cell unit according to the start-stop control command, and control the output current of the electrolytic cell unit to adjust to the target operating current setting value, so as to realize the dynamic matching between the load-side output current and the renewable energy output power;

[0103] When the system bus voltage or frequency deviates from the set normal operating range, the control scheduling model executes a fault-tolerant control compensation strategy, specifically including:

[0104] Initiate a rapid load reduction mechanism to prioritize the shutdown of some electrolytic cell units with higher power but slower response rates;

[0105] Linear backoff limiting is applied to the target operating current setting value of the remaining operating units;

[0106] The compensation and adjustment process maintains a dynamic balance without affecting the overall hydrogen production target of the system.

[0107] The control and scheduling model supports dynamic switching and redundancy management at the electrolyzer unit level, specifically including:

[0108] When an electrolytic cell unit is detected to have an operational malfunction, over-temperature, response mismatch, or communication abnormality, the electrolytic cell unit will be automatically removed from the scheduling pool and put into standby mode.

[0109] Select the electrolytic cell unit with the best status and shortest running time from the standby unit pool to take over the operation task, and automatically inherit the original unit's target operating current setting value;

[0110] After the faulty unit is restored, it enters the observation area and can only be reinstated into the scheduling system after a stable operating cycle is met.

[0111] In existing hydrogen production systems, the electrolyzer unit often operates as a rigid load, lacking the ability to respond in real time to abnormal fluctuations or equipment failures, which can easily lead to system outages or loss of control. In this embodiment, fault-tolerant control compensation and dynamic backup management mechanisms enable the system to have high adaptability and operational robustness.

[0112] When the bus voltage or frequency deviates, the system will not shut down entirely or reduce power uniformly. Instead, it will prioritize identifying electrolytic cells with slower response times but higher power outputs and selectively reduce their load or temporarily shut them down. This "tiered rapid load reduction mechanism" ensures that the total electrolytic load of the system decreases smoothly without affecting the continued operation of more responsive cells, thus better responding to short-term grid fluctuations.

[0113] Furthermore, for electrolytic cell units still in operation, this embodiment also incorporates a linear backoff strategy for current output to prevent sudden equipment overload caused by excessively high target current settings. By limiting the change in the current cycle target value compared to the previous cycle to no more than a set threshold, a "soft scheduling" response is achieved, which is particularly suitable for high-power cells.

[0114] If the system detects an anomaly in a certain electrolytic cell unit (such as excessive operating temperature, abnormal voltage fluctuations, power outage, or communication failure), it automatically triggers the redundancy switching mechanism to quickly take that unit out of operation and activate the standby unit to perform the current task. The selection of the standby unit is based on its "status score" and "remaining lifespan," prioritizing the module with the shortest operating time and lowest operating temperature to be connected to the system, reducing the additional stress caused by the switchover.

[0115] At the same time, the backup unit automatically inherits the target current setting value of the previous faulty unit, avoiding a completely new rescheduling and improving operational continuity under emergency fault conditions.

[0116] Once the original faulty unit recovers to normal operation after isolation and stabilizes after a short period of observation, the system can reinstate it as a candidate for the next round of scheduling, ensuring full utilization and cyclical succession of equipment resources.

[0117] This embodiment breaks through the technical bottleneck of existing "single point failure affecting the system" or "manual switching" mechanisms by adding fast response, dynamic redundancy and self-healing mechanisms to the control layer, and realizes a high availability control architecture at the system level, which is particularly suitable for electrolyzer systems with high requirements for continuous hydrogen supply.

[0118] Step S5: Collect the actual operating parameters of each electrolytic cell unit and compare the deviation with the output results of the control scheduling model; if the deviation exceeds the set tolerance threshold, the local controller will limit and correct the output current, and feed back the collected operating parameters to the control scheduling model for use in the next control cycle update.

[0119] When performing load optimization scheduling, the control scheduling model employs a multi-objective constraint optimization mechanism to jointly optimize the target operating current setpoint. The optimization objectives include:

[0120] Improve the matching degree between the total load current in the current cycle and the predicted power of renewable energy;

[0121] Balance the cumulative operating time of each electrolytic cell unit to slow down the aging rate of the unit;

[0122] Reduce the energy consumption per unit of hydrogen production;

[0123] The objective function is optimized using a multi-objective weighted objective function method. Represented as:

[0124] ;

[0125] in: To calculate the difference between the predicted power and the target load power; The variance of the cumulative operating time of each electrolytic cell; Electricity consumption per unit of hydrogen production; , , To optimize the weighting coefficients and satisfy .

[0126] In one feasible implementation, the control scheduling model supports periodic adaptive updates, and the running parameter update methods include:

[0127] A performance evaluation function is constructed based on historical control errors, system response delays, and environmental change trends.

[0128] When the evaluation metric falls below a set threshold, the model structure or parameters are automatically updated, including:

[0129] Adjust the length of the moving average window;

[0130] Weighting coefficients , , Automatic reallocation;

[0131] A local fitting algorithm is introduced to optimize the environment prediction function;

[0132] The local fitting algorithm includes, but is not limited to:

[0133] Polynomial interpolation based on nearest neighbor sampling points;

[0134] Linear least squares fitting within a sliding time window;

[0135] Radial basis function (RBF) neural network local approximation.

[0136] Its input is real-time data from environmental sensors, and its output is a correction term for the environmental prediction function.

[0137] The update process is deployed online without interrupting control execution.

[0138] Unlike traditional control methods that rely solely on fixed parameters, this embodiment constructs a closed-loop control mechanism with real-time learning and adaptive updating capabilities, enabling continuous optimization control of the electrolytic cell rectifier system.

[0139] Specifically, the control system continuously collects actual operating data for each electrolytic cell unit, especially the deviation between its output current and the predicted value of the control model. If the deviation exceeds the preset tolerance range, the system immediately activates the limiting correction mechanism to gradually adjust the output current, avoiding equipment risks caused by sudden jumps. At the same time, all operating parameters for that cycle are fed back to the scheduling model data buffer for input updates in subsequent cycles.

[0140] During each control cycle, the system evaluates the prediction accuracy and operating trend over a period of time to determine whether the model structure needs to be updated or the operating parameters adjusted. For example, when the prediction error increases over several consecutive cycles, or when the load response trend changes significantly, the system will automatically adjust key variables such as the moving average window length, weighting coefficient allocation method, or error correction parameters.

[0141] Furthermore, this embodiment introduces a multi-objective coordinated optimization mechanism to determine the target current value of the electrolyzer unit. During the setup process, the control model not only pursues a high degree of matching with the predicted renewable energy power, but also considers:

[0142] Balance the operating time among electrolytic cell units to avoid individual equipment from continuously operating under overload;

[0143] Minimize the energy consumption required for each unit of hydrogen production to improve overall energy efficiency.

[0144] By weighing and allocating the above multiple scheduling objectives, the system can dynamically determine the importance of each objective based on actual operation and optimization priorities, and generate scheduling instructions for the current period accordingly.

[0145] This embodiment significantly improves the adaptability of the hydrogen production system to environmental fluctuations and its overall operating efficiency by integrating a closed-loop logic of "prediction-execution-feedback-update" and combining it with amplitude limiting protection, adaptive learning and multi-objective control strategies. It has good industrial usability and prospects for promotion.

[0146] In summary, this invention, by introducing an adaptive weighted moving average algorithm, a power prediction model based on historical and environmental information, a multi-objective current optimization allocation strategy, and a fault-tolerant switching mechanism, constructs a complete dynamic scheduling and feedback correction control closed loop, significantly improving the stability, responsiveness, and energy efficiency of the electrolyzer system under fluctuating power conditions. This method not only possesses excellent intelligent control capabilities but also exhibits self-learning, adaptive, and fault-tolerant characteristics, making it particularly suitable for large-scale hydrogen production systems powered by renewable energy sources such as wind and solar power, and demonstrating broad engineering application value and promising prospects for widespread adoption.

[0147] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of those different embodiments or examples.

[0148] Any process or method description in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing a particular logical function or process. Furthermore, the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functionality involved.

[0149] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various variations or substitutions within the technical scope disclosed in this application, and these should all be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for controlling the rectifier power supply of an electrolyzer suitable for hydrogen production from renewable energy sources, characterized in that, The method comprises the following steps: obtaining output power data of the current control period from the renewable energy output end, the output power data being calculated by adaptive weighted moving average algorithm based on collected voltage and current; obtaining the start-stop state, current, voltage and operating temperature of the electrolytic cell unit from a plurality of electrolytic cell units, collecting the voltage and frequency of the system bus to form an input parameter set for control scheduling calculation; inputting the input parameter set into a control scheduling model, the control scheduling model combining historical operation data and environmental prediction data to predict the available power of the renewable energy in the next control period, and based on the prediction result, generating start-stop control instructions and target working current set values of each electrolytic cell unit as the basis for control execution; predicting the renewable energy available power of a next control period based on a set of input parameters of a current control period, historical operation data and environment prediction data , the formula is: ； wherein: is the current control cycle sliding average output power; is a correction term based on the load response error of the previous control cycle; is a predicted power term calculated based on the current control cycle environmental data; , , is a weighting factor, satisfying ; The control scheduling model is based on predicted renewable energy available power of a next control period and operating states of each electrolytic cell unit in the current control period, performs load optimized distribution; controlling the access and exit state of each electrolytic cell unit according to the start-stop control instructions, and controlling the output current of the electrolytic cell unit to adjust to the target working current set value to dynamically match the load side output current and the renewable energy output power; collecting the actual operating parameters of each electrolytic cell unit, and comparing the deviation with the output result of the control scheduling model; if the deviation is greater than the set tolerance threshold, the local controller limits and corrects the output current, and feeds back the collected operating parameters to the control scheduling model.

2. The electrolytic cell rectifier power supply control method for renewable energy hydrogen production according to claim 1, characterized in that, The output power data is calculated by adaptive weighted moving average algorithm based on collected voltage and current, and the expression of the adaptive weighted moving average algorithm is: ； wherein: is a length of a sliding average window in the current control period; is a maximum sliding window length; is a window adjustment sensitivity coefficient; , is a current and previous control period output power; is a sampling time interval; Through the adaptive weighted moving average algorithm, the sliding window is dynamically shortened when the output power fluctuates sharply, and the window is automatically lengthened when the power is stable.

3. The electrolytic cell rectifier power supply control method for renewable energy hydrogen production of claim 1, wherein, The execution of the start-stop control instructions and the target working current set values is limited by the following control conditions: When the voltage of the system bus is lower than the set safety threshold, the electrolytic cell units with large response delay are preferentially closed, and all target working current adjustment instructions are temporarily suspended; If the change amplitude of the target working current set value relative to the actual output current of the previous control period exceeds the set rate threshold, a slope function limiting strategy is adopted to execute the current adjustment in stages; All control instructions output by the control scheduling model are subjected to legality verification by the local electrolytic cell controller, which includes hardware constraint conditions of operating temperature, inter-electrode voltage drop and current response rate; When the legality verification is passed, the corresponding start-stop or current adjustment instruction is executed.

4. The electrolytic cell rectifier power supply control method for renewable energy hydrogen production of claim 1, wherein, The local controller limits and corrects the output current, collects the actual output current, voltage and temperature data of the current control period, and compares the deviation of the collected data with the target value output by the control scheduling model; if the deviation exceeds the set tolerance threshold, the target output current is limited and corrected at the local controller level, and the operating parameters collected in the current control period are fed back to the control scheduling model for updating the input parameter set of the next control period.

5. The electrolyzer rectifier power supply control method for renewable energy hydrogen production of claim 4, wherein, When the voltage or frequency of the system bus deviates from the set normal operating range, the control scheduling model executes a fault-tolerant control compensation strategy, which specifically includes: starting a rapid load shedding mechanism to preferentially disable some electrolytic cell units with high power but slow response rate; linearly retreating and limiting the target working current set value of the remaining operating units; The compensation adjustment process maintains dynamic balance without affecting the overall hydrogen production target of the system.

6. The electrolyzer rectifier power supply control method for renewable energy hydrogen production of claim 1, wherein, The control scheduling model dynamically allocates a load weight based on the priority and historical operation performance of each electrolytic cell unit when generating a target working current setting value of each electrolytic cell unit , the formula is: ； wherein: represents the load distribution weight of the i-th electrolyzer cell unit in the current control period; represents the number of electrolyzer cell units currently in operation; , are the minimum and maximum values of the load distribution weight, respectively.​ The load distribution weight is used to proportionally distribute the total target output current corresponding to the predicted available power to each electrolytic cell unit.

7. The electrolyzer rectifier power supply control method for renewable energy hydrogen production of claim 1, wherein, The control scheduling model adopts a multi-objective weighted objective function method to optimize an objective function when performing load optimization scheduling is represented as: ； wherein: is the difference between the predicted power and the target load power; is the variance of the cumulative running time of each electrolytic cell; is the unit hydrogen production power consumption; , , is the optimization weight coefficient, satisfying .

8. The electrolytic cell rectifier power supply control method for renewable energy hydrogen production of claim 1, wherein, The control scheduling model supports periodic adaptive updates, and the operation parameter update mode includes: A performance evaluation function is constructed based on historical control errors, system response delays, and environmental change trends to obtain evaluation indexes. When the evaluation index is less than a set threshold, triggering model structure or parameter self-updating, including: sliding average window length adjustment; automatic redistribution of weighted coefficients , , ; introducing local fitting algorithm optimization environment prediction function.

9. The electrolytic cell rectifier power supply control method for renewable energy hydrogen production of claim 1, wherein, The control scheduling model supports dynamic switching and redundancy management at the electrolytic cell unit level, specifically including: When detecting that a certain electrolytic cell unit has operational faults, over-temperature, response mismatch, or communication abnormalities, the electrolytic cell unit is automatically excluded from the scheduling pool and becomes a backup state; The electrolytic cell unit with the optimal state and the shortest running time is selected from the backup unit pool to replace the running task, and automatically inherits the original electrolytic cell unit target working current setting value; The failed electrolytic cell unit enters the observation area after recovery, and is re-included in the scheduling after meeting the stable operation period.

Images