Hydrogen energy collaborative scheduling method suitable for distributed renewable energy characteristics

Abstract

The application relates to the technical field of hydrogen energy scheduling, and particularly discloses a hydrogen energy collaborative scheduling method suitable for the characteristics of distributed renewable energy, aiming to solve the problems of high abandoned power rate, poor hydrogen energy industry chain collaboration, high comprehensive cost and insufficient power grid stability caused by the fluctuation of distributed energy output. The application collects renewable energy output, power grid load and hydrogen production, hydrogen storage and transportation equipment state data, constructs a multi-time scale scheduling framework including day-ahead, day-ahead and real-time, establishes a multi-objective optimization model coupled with the full-link cost and abandoned power penalty, adopts a fluctuation grading strategy to adapt to the output change, solves and issues scheduling instructions through a model predictive control algorithm, and dynamically corrects model parameters according to the abandoned power rate and prediction error. The application can efficiently consume surplus power, reduce the abandoned power rate, realize the collaborative optimization and cost optimization of the whole industry chain, suppress the energy output fluctuation, and improve the operation stability of the power grid and the adaptability of the scheduling system.

Description

Technical Field

[0001] This invention relates to the field of hydrogen energy dispatching technology, and in particular to a hydrogen energy collaborative dispatching method adapted to the characteristics of distributed renewable energy. Background Technology

[0002] Distributed renewable energy sources such as wind power and photovoltaics are important sources of power supply for the power grid. However, the output of such energy is highly intermittent, random, and subject to instantaneous fluctuations. Direct grid connection will reduce the power quality and operational stability of the power grid. Moreover, surplus renewable energy cannot be consumed locally, which generally leads to problems of wind curtailment, solar curtailment, and electricity curtailment, resulting in low energy utilization.

[0003] Hydrogen energy, as a clean energy carrier that can efficiently absorb surplus electricity, is a core solution to the grid connection problem of distributed energy. However, the existing hydrogen energy industry chain technology can only realize the static cost calculation of hydrogen production, storage and transportation. It has not formed a dynamic scheduling mechanism adapted to the fluctuation of renewable energy. At the same time, it lacks a multi-time scale scheduling framework and cannot dynamically adjust the operation strategy of the entire industry chain according to the output fluctuation. The hydrogen production, storage and transportation links are independently optimized and have poor coordination. It is neither able to quickly respond to sudden changes in renewable energy output and efficiently absorb abandoned electricity, nor can it achieve the optimal comprehensive cost of the entire industry chain, making it difficult to provide stable support for grid operation.

[0004] Therefore, there is an urgent need for a hydrogen energy collaborative scheduling method that adapts to the characteristics of distributed renewable energy to solve the above problems. Summary of the Invention

[0005] The purpose of this invention is to provide a hydrogen energy collaborative scheduling method adapted to the characteristics of distributed renewable energy, comprising the following steps: Collect real-time output data, predicted output data, grid load data of distributed renewable energy power generation systems, and status data of hydrogen production systems, hydrogen storage systems, and transportation systems in the hydrogen energy industry chain; A multi-timescale collaborative scheduling framework is constructed, which includes a day-ahead scheduling phase, an intraday rolling correction phase, and a real-time adjustment phase. A multi-objective collaborative scheduling optimization model is established with the goals of maximizing the utilization of curtailed electricity and minimizing the overall cost. The overall cost includes hydrogen production cost, hydrogen storage cost, and transportation cost, and a curtailment penalty term is introduced. Based on the fluctuating characteristics of renewable energy output, a fluctuating hierarchical scheduling strategy is designed. The model predictive control algorithm is used to solve the cooperative scheduling optimization model to generate hydrogen production power instructions, hydrogen storage charging and discharging instructions, and transportation scheduling instructions. The instructions are sent to the hydrogen production system, hydrogen storage system, and transportation system, and their execution is controlled. Based on the actual execution feedback and real-time monitoring results of the curtailment rate, the curtailment penalty coefficient in the optimization model is dynamically adjusted, and the prediction error distribution parameters are updated on a rolling basis.

[0006] The beneficial effects of this application are as follows: Firstly, this invention uses a multi-timescale scheduling framework (day-ahead, intraday, and real-time) and a fluctuating hierarchical scheduling strategy to dynamically adapt to the fluctuating characteristics of distributed renewable energy. This allows for the rapid absorption of surplus electricity, filling power gaps, significantly reducing wind and solar curtailment rates, and maximizing the utilization of distributed renewable energy.

[0007] Secondly, this invention constructs a global multi-objective optimization model based on the cost model of the entire hydrogen production, storage and transportation process, which overcomes the shortcomings of independent optimization of each process, realizes collaborative cost control of the entire hydrogen energy industry chain, and effectively reduces the overall operating cost.

[0008] Third, by setting up multiple physical constraints and a closed-loop adaptive correction mechanism, the present invention accurately matches the equipment operating capacity with the grid demand, smooths out the fluctuations in renewable energy output, and significantly improves the stability of grid operation. Attached Figure Description

[0009] Figure 1 This is a schematic diagram of a method flow proposed in an embodiment of this application.

[0010] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0011] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0012] like Figure 1 As shown, this application provides a hydrogen energy collaborative scheduling method adapted to the characteristics of distributed renewable energy, including the following steps: S1: Collect real-time output data, predicted output data, grid load data of distributed renewable energy power generation systems, and status data of hydrogen production systems, hydrogen storage systems, and transportation systems in the hydrogen energy industry chain. S2: Construct a multi-timescale collaborative scheduling framework, which includes a day-ahead scheduling phase, an intraday rolling correction phase, and a real-time adjustment phase; S3: Establish a multi-objective collaborative scheduling optimization model with the goals of maximizing curtailment and minimizing overall costs. The overall costs include hydrogen production costs, hydrogen storage costs, and transportation costs, and a curtailment penalty term is introduced. S4: Based on the fluctuation characteristics of renewable energy output, a fluctuation-level scheduling strategy is designed. This strategy adjusts the hydrogen production power in advance according to the magnitude of output changes. S5: Use the model predictive control algorithm to solve the cooperative scheduling optimization model and generate hydrogen production power command, hydrogen storage charging and discharging command and transportation scheduling command; S6: Issue the instructions to the hydrogen production system, hydrogen storage system, and transportation system, and control their execution; S7: Based on actual execution feedback and real-time monitoring results of curtailment rate, dynamically adjust the curtailment penalty coefficient in the optimization model and continuously update the prediction error distribution parameters.

[0013] As described in steps S1-S7 above, wind and solar power outputs in distributed renewable energy sources are highly intermittent, random, and subject to instantaneous fluctuations. Grid load changes dynamically in real time with user demand. Independent operation of hydrogen production, storage, and transportation can lead to disconnects, resource idleness, and high costs. Furthermore, the inability to effectively absorb surplus renewable energy can result in wind and solar curtailment, directly reducing grid stability. To address these issues, a comprehensive scheduling method is needed that can perceive the status in real time, develop hierarchical plans, balance cost and absorption, respond quickly to fluctuations, and achieve closed-loop self-optimization, thus binding renewable energy absorption to the entire hydrogen energy industry chain.

[0014] Existing hydrogen and renewable energy dispatch technologies only perform static cost calculations or fixed-cycle dispatching for a single link, lacking a multi-timescale collaborative framework, fluctuation adaptation strategies, and closed-loop adaptive correction mechanisms. These technologies are unable to cope with dynamic fluctuations in renewable energy, making it difficult to achieve optimal collaboration across the entire industry chain and efficient curtailment absorption. Consequently, the dispatching schemes exhibit extremely poor practicality and robustness. This invention systematically addresses the core problems of difficult fluctuation adaptation, poor industry chain collaboration, high curtailment rates, and high overall costs through an integrated solution encompassing full-domain data acquisition, multi-scale framework construction, multi-objective model establishment, fluctuation-level dispatching, model prediction and solution, command execution, and adaptive parameter updates.

[0015] This invention constructs a full-link hydrogen energy collaborative scheduling system that integrates data perception, multi-scale scheduling, multi-objective optimization, fluctuation adaptation, algorithm solving, instruction execution, and adaptive correction. This system achieves deep coupling between distributed renewable energy and the entire hydrogen energy production, storage, and transportation industry chain, ultimately maximizing power curtailment, minimizing the overall cost of the industry chain, and stabilizing grid operation.

[0016] This invention utilizes real-time / predictive output of distributed renewable energy, grid load, and the status of equipment across the entire hydrogen energy industry chain as the basis for sensing data. It establishes a multi-timescale scheduling framework encompassing day-ahead long-term planning, intraday rolling correction, and real-time emergency adjustments. A multi-objective optimization model is built, coupling the costs of hydrogen production, storage, and transportation with curtailment penalties. A tiered response strategy is formulated based on the fluctuation range of renewable energy output. A model predictive control algorithm is used to continuously solve for the optimal scheduling command. After the command is issued to the equipment for execution, the curtailment penalty coefficient is dynamically adjusted based on actual execution feedback and real-time curtailment rate. Simultaneously, the prediction error distribution parameters are continuously updated. Ultimately, a closed-loop working mechanism of sensing, planning, optimization, execution, feedback, and learning is formed, dynamically adapting to renewable energy fluctuations throughout the entire process and achieving efficient and coordinated operation of the entire hydrogen energy industry chain and distributed energy.

[0017] In one embodiment, step S1 specifically includes: S11: Collect real-time power output from wind turbines and photovoltaic arrays in distributed renewable energy plants at a sampling frequency of seconds, and obtain predicted power output data for the next 24 hours at a resolution of 15 minutes from the meteorological forecasting system; S12: Obtain the current grid load from the grid dispatch center and calculate the power curtailment. The power curtailment is calculated by subtracting the grid load from the real-time output power and then subtracting the maximum absorption power of the hydrogen production system. If the value is greater than 0, it is recorded as the power curtailment; otherwise, the power curtailment is 0. S13: Collect the lower and upper limits of available power of the hydrogen production system, the current efficiency of the electrolyzer, the current amount of hydrogen in the hydrogen storage system, the upper limit of hydrogen storage capacity, the limit of hydrogen charging rate and the limit of hydrogen discharging rate, and the number of available vehicles, the carrying capacity of a single vehicle, the real-time location of vehicles and their speed in the transportation system.

[0018] As described in steps S11-S13 above, accurate data collection can be completed on the distributed renewable energy power generation system, grid operation status, and equipment status of the entire hydrogen energy industry chain. At the same time, standardized quantitative calculation of curtailed power can be achieved, providing complete and accurate basic data for subsequent multi-timescale collaborative scheduling, optimization model solving, and fluctuation-level scheduling. Ultimately, the core objectives of adapting to the fluctuation characteristics of distributed renewable energy, maximizing the absorption of curtailed power, and minimizing the comprehensive cost of the hydrogen energy industry chain can be achieved.

[0019] The output of distributed renewable energy is intermittent and random, and the grid load is in a state of real-time dynamic change. The operating status of hydrogen production, storage, and transportation systems directly determines the coordinated regulation capability of the hydrogen energy industry chain. If the above data cannot be fully collected and the curtailed power cannot be accurately calculated, the dispatching decision will deviate from the actual operating conditions. This will not only fail to effectively reduce wind and solar curtailment, but also make it difficult to achieve optimal cost control across the entire industry chain. Therefore, it is necessary to provide basic data support that is consistent with the actual operating status through standardized data collection and quantitative calculation methods.

[0020] Traditional hydrogen energy dispatching technologies only collect single-dimensional data from the power generation side or equipment side, failing to achieve synchronous data collection from renewable energy, the power grid, and the hydrogen energy industry chain. Furthermore, they lack a precise calculation method for curtailed power, resulting in insufficient data completeness and metering accuracy, thus hindering the implementation of dynamic collaborative dispatching. This invention addresses the technical problems of missing basic data and inaccurate curtailment metering through a comprehensive solution that includes modular collection of renewable energy output data, quantitative calculation of curtailed power, and comprehensive collection of equipment status data across the hydrogen energy industry chain.

[0021] Step S11 involves collecting real-time power output data from the wind turbines and photovoltaic arrays of the distributed renewable energy plant at a sampling frequency of seconds. This second-level sampling frequency can fully capture the instantaneous fluctuation characteristics of wind turbine and photovoltaic power output, avoiding the omission of details of power output changes due to excessively long sampling intervals. Simultaneously, forecast power output data for the next 24 hours at a resolution of 15 minutes is obtained from the meteorological forecasting system. The 15-minute resolution balances forecast accuracy and data processing efficiency, providing forecast data that matches the time scale requirements for subsequent day-ahead scheduling and intraday rolling corrections. This provides the foundation of original measured and forecasted data for power output fluctuation analysis and scheduling plan formulation.

[0022] Step S12: Obtain the current grid load from the grid dispatch center. Grid load is a core indicator for defining the local consumption capacity of renewable energy. Based on this, calculate the curtailed power. Curtailed power is the value obtained by subtracting the grid load from the real-time power output and then subtracting the maximum absorption power of the hydrogen production system. If this value is greater than 0, it is recorded as curtailed power; otherwise, the curtailed power is 0. The above calculation method can accurately quantify the surplus electricity of renewable energy that cannot be consumed locally and cannot be absorbed by the hydrogen production system, providing a quantitative basis for setting the curtailment penalty term in the optimization model and achieving the curtailment consumption target.

[0023] Step S13 involves collecting the lower and upper limits of available power and the current efficiency of the electrolyzer in the hydrogen production system. These parameters directly determine the power adjustment range and operating efficiency of the hydrogen production system, serving as the core basis for hydrogen production power scheduling. The current hydrogen quantity, upper limit of hydrogen storage capacity, hydrogen charging rate limit, and hydrogen discharging rate limit in the hydrogen storage system are also collected. These settings determine the hydrogen storage and charging / discharging regulation capabilities of the hydrogen storage system, connecting the hydrogen production and transportation links. The number of available vehicles, single-vehicle carrying capacity, real-time vehicle location, and travel speed in the transportation system directly define the hydrogen transfer capacity and efficiency. Comprehensive collection of the above equipment status data ensures that subsequent scheduling instructions are perfectly matched with the actual operating capacity of the equipment, avoiding problems such as excessive hydrogen production power, excessive hydrogen storage capacity, and insufficient transportation capacity, thus guaranteeing the stable execution of coordinated scheduling across the entire hydrogen energy industry chain.

[0024] In one embodiment, step S2 specifically includes: S21: During the day-ahead dispatch phase, with a time resolution of 1 hour, based on the next day's renewable energy output forecast and load forecast, formulate the hydrogen production power plan curve, hydrogen storage system charging and discharging plan and transportation plan for the next 24 hours, and issue them as the baseline plan. S22: During the intraday rolling correction phase, a rolling optimization is triggered every 15 minutes. Starting from the current moment, based on the latest measured data and short-term forecasts for the next 4 hours, the model predictive control algorithm is used to continuously update the scheduling instructions for the next 4 hours. After each optimization, only the control instructions for the first 15 minutes are executed. S23: During the real-time adjustment phase, the deviation trigger threshold is set to 20%. When the measured renewable energy output deviates from the day-ahead forecast by more than this threshold, or the grid frequency deviation exceeds 0.1Hz, a second-level real-time adjustment is initiated: within 1 minute, the hydrogen power is rapidly increased to the maximum available value, or the hydrogen power is decreased to the minimum operating value. If necessary, hydrogen in the hydrogen storage system is released to feed back to the grid through the fuel cell.

[0025] As described in steps S21-S23 above, a multi-timescale collaborative scheduling framework can be constructed, including day-ahead scheduling, intraday rolling correction, and real-time adjustment. This forms a hierarchical scheduling mechanism with long-term planning, medium-term correction, and short-term emergency response, enabling scheduling instructions to continuously adapt to the output fluctuations and prediction errors of distributed renewable energy, ensuring the accuracy and stability of collaborative scheduling in the hydrogen energy industry chain, and helping to achieve the goals of maximizing curtailment and minimizing overall costs.

[0026] Distributed renewable energy output exhibits both short-term volatility and long-term forecasting errors, while grid load also changes in real time. Relying solely on a single timescale for scheduling plans cannot balance the rationality of long-term resource planning with the flexibility of short-term operating condition adjustments. This can lead to a disconnect between scheduling instructions and actual operating conditions, resulting in increased curtailment rates, equipment exceeding operating limits, and decreased grid stability. Therefore, it is necessary to establish a multi-timescale collaborative scheduling framework to complete plan formulation, dynamic correction, and emergency response at different levels.

[0027] Traditional hydrogen energy dispatching technologies mostly employ static dispatching plans with fixed cycles, lacking intraday rolling correction mechanisms and real-time adjustment mechanisms for sudden output fluctuations. The dispatching schemes cannot dynamically optimize to keep pace with actual operating conditions, exhibiting extremely poor adaptability to prediction errors and output fluctuations, making it difficult to meet the collaborative dispatching requirements of distributed renewable energy scenarios. This invention addresses the technical problem of static dispatching's inability to adapt to output fluctuations and prediction errors through a phased approach: developing a baseline dispatching plan, continuously updating dispatching instructions, and triggering real-time emergency adjustments.

[0028] Step S21 involves developing a 24-hour hydrogen production power plan curve, a hydrogen storage system charging / discharging plan, and a transportation plan based on the next day's renewable energy output and load forecasts, using a 1-hour time resolution during the day-ahead scheduling phase. This plan is then issued as the scheduling baseline. Based on long-term forecast data, an overall plan for scheduling across the entire industry chain is completed, rationally allocating resources and tasks across hydrogen production, storage, and transportation. This provides a stable framework for subsequent detailed adjustments, ensuring the long-term rationality of the scheduling work.

[0029] Step S22 involves triggering a rolling optimization every 15 minutes during the intraday rolling correction phase. Starting from the current moment, the latest measured data and short-term forecast data for the next 4 hours are combined to use a model predictive control algorithm to continuously update the dispatch instructions for the next 4 hours. After each optimization, only the control instructions for the first 15 minutes are executed. Through high-frequency rolling correction, the accumulated errors in renewable energy output forecasts are continuously offset, ensuring that dispatch instructions constantly align with the real-time operating status. This avoids excessive deviations between the baseline plan and actual operating conditions, guaranteeing the real-time performance and accuracy of dispatch decisions.

[0030] Step S23: During the real-time adjustment phase, a 20% output deviation trigger threshold is set. When the measured renewable energy output deviates from the day-ahead forecast by more than this threshold, or the grid frequency deviation exceeds 0.1 Hz, a second-level real-time adjustment is initiated. Within one minute, the hydrogen production power is rapidly increased to the maximum available value or decreased to the minimum operating value. If necessary, hydrogen from the hydrogen storage system is released and fed back to the grid through the fuel cell. This system can quickly respond to sudden and severe fluctuations in output and abnormal grid conditions. Through rapid adjustment of hydrogen production power and emergency support from the hydrogen storage system, it ensures stable grid operation while minimizing power curtailment, thus achieving an emergency guarantee function for coordinated dispatch.

[0031] In one embodiment, the multi-objective cooperative scheduling optimization model in step S3 specifically includes an objective function and various cost models: S31: Define the objective function as minimizing the sum of comprehensive cost and curtailment penalty, where comprehensive cost includes hydrogen production cost, hydrogen storage cost and transportation cost, and comprehensive cost is multiplied by a comprehensive cost weighting coefficient, and curtailment penalty term is composed of dynamic curtailment penalty coefficient multiplied by the sum of curtailed power in each time period; S32: The calculation method for the hydrogen production cost model is as follows: First, calculate the available resources, which equals the total resources minus the resources to be consumed locally; then add the available resources to the planned resources, and multiply by the power supply stability factor, the cost coefficient of supporting environment and equipment, the cost of main equipment system, and the system operation cost coefficient. The power supply stability factor is calculated by the product ratio of resource output coefficient, time, total resource power, resource and grid complementarity coefficient, power absorption coefficient of resources, and total power consumption. S33: The calculation method of the hydrogen storage cost model is as follows: multiply the cost of the hydrogen storage equipment system by the maintenance cost coefficient of the hydrogen storage system, then multiply by the maintenance cost coefficient of the hydrogen storage system, and finally divide by the product of the correlation between the application scenario and the correlation between the transportation mode. S34: The transportation cost model is calculated by multiplying the cost of the hydrogen storage equipment system by the maintenance cost coefficient of the hydrogen storage system, adding the operating cost of the hydrogen storage system, and then dividing by the product of the application scenario correlation and the transportation mode correlation.

[0032] As described in steps S31-S34 above, a multi-objective collaborative scheduling optimization model with the core of minimizing comprehensive costs and maximizing curtailment consumption can be built. The objective function composition and cost calculation rules for each link of hydrogen production, storage and transportation are clarified. The economic indicators of the entire industrial chain and curtailment control objectives are quantitatively integrated, providing a standardized calculation basis for subsequent scheduling algorithm solutions.

[0033] The hydrogen energy industry chain involves interconnected costs in hydrogen production, storage, and transportation. At the same time, the dispatch process needs to prioritize the absorption of distributed renewable energy curtailment. Simply pursuing the lowest cost in a single link cannot achieve global optimization, nor can it take into account the core need for curtailment absorption. Therefore, it is necessary to construct an optimization model that couples multi-dimensional costs and curtailment penalties, and balance economic efficiency and curtailment absorption effect through quantitative formulas.

[0034] Traditional hydrogen energy cost calculation techniques only calculate the cost of a single link independently, failing to form a multi-objective optimization model covering the entire industry chain. They also do not incorporate curtailment penalties into the objective function, and the cost calculations do not consider the characteristics of renewable energy consumption scenarios, thus failing to support globally optimal decisions for coordinated dispatch. This invention addresses the technical problems of fragmented cost accounting and the inability to balance economic and consumption objectives by defining multiple objective functions and constructing cost models for hydrogen production, storage, and transportation tailored to different scenarios.

[0035] Step S31 defines the objective function as minimizing the sum of the overall cost and the curtailment penalty. The overall cost includes the costs of hydrogen production, storage, and transportation, and a weighting coefficient is assigned to each cost. The curtailment penalty is obtained by multiplying the dynamic curtailment penalty coefficient by the sum of the curtailed power in each time period. By balancing the priorities of cost control and curtailment consumption through weighting coefficients, and by adjusting the dynamic curtailment penalty coefficient according to real-time operating conditions, the dual optimization objectives are transformed into a solvable mathematical expression, providing a clear optimization direction for coordinated scheduling.

[0036] Step S32 involves constructing a hydrogen production cost model. First, the available resource quantity is obtained by subtracting locally consumed resources from the total resources. Then, the available resource quantity is added to the planned resource quantity and multiplied sequentially by the power supply stability factor, the supporting environment and equipment cost coefficient, the main equipment system cost, and the system operation cost coefficient. The power supply stability factor is calculated as the ratio of the product of the resource output coefficient, time, total resource power, the complementarity coefficient between the resource and the grid, the power consumption coefficient for resources, and the total power consumption. This model fully integrates the resource characteristics, equipment investment, and operational characteristics of renewable energy hydrogen production, accurately quantifying all costs in the hydrogen production process and closely aligning with the actual scenarios of distributed energy hydrogen production.

[0037] Step S33: Construct a hydrogen storage cost model by multiplying the hydrogen storage equipment system cost, the hydrogen storage system maintenance cost coefficient, and the hydrogen storage system maintenance cost coefficient, and then dividing by the product of the application scenario correlation and the transportation mode correlation. This model links hydrogen storage costs to actual application scenarios and subsequent transportation modes, adapting to different hydrogen storage methods such as high-pressure gaseous, low-pressure liquid, and solid adsorption, thereby improving the accuracy and adaptability of hydrogen storage cost calculation.

[0038] Step S34: Construct a transportation cost model by multiplying the cost of the hydrogen storage equipment system by the maintenance cost coefficient of the hydrogen storage system, adding the operating cost of the hydrogen storage system, and then dividing by the product of the application scenario correlation and the transportation mode correlation. The specific calculation formula is as follows: ; Among them, the This indicates the overall cost of hydrogen storage. This indicates the cost of a hydrogen storage equipment system. This represents the maintenance cost coefficient of the hydrogen storage system. This indicates the operating cost of the hydrogen storage system. Indicates the relevance of application scenarios. This model represents the correlation between different modes of transportation. It covers various transportation and transshipment methods, including automobiles, ships, and hydrogen blending via gas pipelines. By combining core elements such as investment, operation, maintenance, applicability, and transmission distance, it accurately quantifies all costs associated with hydrogen transportation and transshipment.

[0039] In one embodiment, the multi-objective cooperative scheduling optimization model in step S3 further includes the following constraints: S35: Power balance constraint: Real-time output power equals the sum of grid load, actual power consumption of hydrogen production system, and power curtailment. S36: Hydrogen production power operation constraints: The actual power consumed by the hydrogen production system must be between its lower and upper limits of available power, and the change in hydrogen production power between two adjacent moments cannot exceed the maximum ramp rate of the hydrogen production system. S37: Dynamic constraints of hydrogen storage system: The amount of hydrogen stored at the next moment is equal to the amount of hydrogen stored at the current moment plus the hydrogen charging efficiency multiplied by the hydrogen production power multiplied by the time step, and then minus the reciprocal of the hydrogen release efficiency multiplied by the hydrogen supply power of the hydrogen storage system multiplied by the time step. At the same time, the amount of hydrogen stored at any moment must be between 0 and the upper limit of hydrogen storage capacity. S38: Transportation coordination constraint: The total transportation volume during the entire scheduling cycle shall not exceed the hydrogen storage volume at the end of the scheduling period minus the minimum safe storage volume, and the total hydrogen load of all transportation vehicles at any time shall not exceed the sum of the maximum hydrogen load of a single vehicle. S39: Upper limit constraint on renewable energy curtailment rate: The ratio of the sum of curtailed power to the sum of renewable energy output during the entire dispatch cycle shall not exceed the preset maximum allowable curtailment rate.

[0040] As described in steps S35-S39 above, complete constraints can be added to the multi-objective collaborative scheduling optimization model. The execution boundary of the scheduling scheme is limited from four dimensions: energy balance, equipment operation, link coordination, and curtailment control. This ensures that the scheduling instructions solved by the optimization model conform to physical laws, equipment safety requirements, and curtailment absorption indicators, making the scheduling scheme practically executable.

[0041] There is a strict energy balance relationship between distributed renewable energy output, grid load, and hydrogen production power consumption. Hydrogen production, storage, and transportation equipment all have fixed operating limits and regulation capabilities. At the same time, the curtailment rate needs to meet the hard indicators of renewable energy consumption. If the optimization model lacks constraints, the solution results will have problems such as energy imbalance, equipment overload, link mismatch, and excessive curtailment, making it impossible to implement. Therefore, complete constraints must be set to ensure that the dispatching scheme is reasonable and feasible.

[0042] Traditional hydrogen energy dispatch optimization only employs single-device operation constraints, failing to construct a comprehensive constraint system covering energy balance, full industry chain collaboration, and curtailment management. The constraints are poorly adapted to distributed renewable energy consumption scenarios, and optimization results often deviate from actual operating conditions, failing to meet the safety and performance requirements of coordinated dispatch. This solution addresses the issues of boundless, unexecutable, and substandard optimization results by setting five types of constraints: power balance, equipment operation, hydrogen storage dynamics, transportation coordination, and curtailment rate ceiling.

[0043] Step S35 is a power balance constraint, which stipulates that the real-time output power is equal to the sum of the grid load, the actual power consumed by the hydrogen production system, and the power curtailed. This constraint can strictly ensure that the output of renewable energy is reasonably allocated among the local consumption, hydrogen production consumption, and power curtailment, maintain the energy balance between the grid and the hydrogen production system, and fundamentally avoid scheduling errors caused by energy calculation deviations.

[0044] Step S36 is the hydrogen production power operation constraint. This constraint requires that the actual power consumption of the hydrogen production system be between the lower limit and the upper limit of the available power, and that the change in hydrogen production power between two adjacent moments does not exceed the maximum ramp rate of the hydrogen production system. This not only limits the safe operating range of the hydrogen production equipment, but also controls the power adjustment speed, preventing damage to the electrolyzer due to overload or frequent large-scale adjustments, and ensuring the continuous and stable operation of the hydrogen production system.

[0045] Step S37 is a dynamic constraint for the hydrogen storage system. This constraint clarifies that the hydrogen storage capacity at the next moment is calculated by the current hydrogen storage capacity, hydrogen charging efficiency, hydrogen production power, time step, hydrogen release efficiency, and the hydrogen supply power of the hydrogen storage system. It also requires that the hydrogen storage capacity at any moment be between 0 and the upper limit of the hydrogen storage capacity. This establishes a dynamic evolution rule for the hydrogen storage capacity, limits the storage capacity boundary, avoids overcharging or venting of the hydrogen storage system, and ensures a stable connection between the hydrogen storage link and the hydrogen production and transportation process.

[0046] Step S38 is a transportation coordination constraint. This constraint stipulates that the total transportation volume during the entire scheduling cycle shall not exceed the hydrogen storage volume at the end of the scheduling period minus the minimum safe storage volume, and the total hydrogen load of all transport vehicles at any time shall not exceed the sum of the maximum hydrogen load of a single vehicle. This binds the hydrogen storage capacity to the transportation capacity, avoids the problem of no hydrogen storage available for transport and overloaded transport vehicles, and achieves efficient coordination between hydrogen storage and transportation.

[0047] Step S39 is the upper limit constraint of renewable energy curtailment rate. This constraint requires that the ratio of the sum of curtailed power to the sum of renewable energy output during the entire dispatch cycle does not exceed the preset maximum allowable curtailment rate. This transforms the core objective of curtailment consumption into a rigid quantitative constraint, ensuring that the dispatch results always meet the renewable energy consumption indicators and achieving the core technical objective of reducing wind and solar curtailment.

[0048] In one embodiment, the fluctuation grading scheduling strategy in step S4 specifically includes: S41: Calculate the rate of change of renewable energy output in real time. The rate of change is the difference between the predicted output at the next moment and the current measured output divided by the current measured output. Set a classification threshold: when the rate of change is greater than 20%, it is judged as a sharp increase in output; when the rate of change is less than -20%, it is judged as a sharp decrease in output; when the absolute value of the rate of change does not exceed 20%, it is judged as a stable state. S42: If the power output is determined to be a sudden increase event, a pre-increase command is triggered 30 minutes in advance: the hydrogen production power setting value for the next 30 minutes is increased, and the increased value is the smaller value between the upper limit of hydrogen production power and the current hydrogen production power plus the rate of change multiplied by the current output, and the hydrogen storage system is notified to reserve receiving capacity at the same time. S43: If the power output is determined to be a sharp drop event, an emergency reduction command is triggered: the hydrogen production power is reduced within 1 minute. The reduced value is the larger of the lower limit of the hydrogen production power and the current hydrogen production power minus the absolute value of the rate of change multiplied by the current output. If the power balance cannot be met after the reduction, the hydrogen storage system is activated to supply hydrogen in reverse and convert hydrogen into electricity through the fuel cell to replenish the grid. S44: If the state is determined to be stable, the current scheduling plan is maintained, but the prediction confidence interval is continuously monitored: Based on the historical prediction error distribution, a 90% confidence interval is set, and adjustable capacity is reserved in the day-ahead scheduling. The adjustable capacity is equal to the output at the upper bound of the confidence interval minus the output at the lower bound, which is used to deal with prediction uncertainty.

[0049] As described in steps S41-S44 above, the fluctuation-level scheduling strategy completes the operating condition classification based on the magnitude of renewable energy output changes, executes differentiated adjustment commands for different fluctuation types, and reserves adjustable capacity based on the predicted confidence interval under stable operating conditions. This enables hydrogen production and storage systems to quickly adapt to scenarios of sharp increases and decreases in output, maximizes the absorption of abandoned power and ensures grid power balance, and improves the ability of coordinated scheduling to adapt to the fluctuation characteristics of distributed renewable energy.

[0050] Distributed renewable energy output can exhibit three typical states: sharp rise, sharp fall, and stable operation. The requirements for hydrogen production power adjustment speed, adjustment range, and hydrogen storage system coordination vary significantly under different operating conditions. If a uniform dispatch rule is adopted, it will be impossible to absorb the surplus electricity generated by the sharp rise in output in a timely manner, nor will it be possible to quickly make up for the power gap caused by the sharp fall in output. Ultimately, this will lead to an increase in curtailment rate and a decrease in grid operation stability. Therefore, it is necessary to formulate a graded dispatch strategy according to the output fluctuation range to achieve precise and real-time fluctuation response.

[0051] Traditional hydrogen energy dispatch strategies fail to address the tiered handling of renewable energy output fluctuations, relying solely on fixed power regulation rules. This lacks both a pre-emptive increase mechanism for sharp output increases and an emergency reduction plan for sharp output decreases. Furthermore, it fails to account for forecast errors and reserve adjustable capacity, resulting in a significant lag in response to sudden fluctuations. Consequently, it cannot meet the core requirements of efficiently absorbing abandoned power and stably supporting the power grid. This invention addresses the technical problems of untimely fluctuation response and a single regulation method through a comprehensive solution that quantifies and tiers output change rates, executes customized regulation commands based on operating conditions, and reserves capacity within the confidence interval of stable operating conditions.

[0052] Step S41 involves calculating the rate of change in renewable energy output in real time. This rate of change is obtained by dividing the difference between the predicted output at the next moment and the current measured output by the current measured output. Using a 20% threshold, the operating conditions are categorized into three types: sharp increase in output, sharp decrease in output, and stable operation. This quantitative indicator enables precise determination of the operating condition, providing a clear basis for the execution of subsequent differentiated dispatch instructions and avoiding resource waste and response deviations caused by indiscriminate adjustments.

[0053] In step S42, when a sudden increase in output is detected, a pre-increase command is triggered 30 minutes in advance. The hydrogen production power setpoint for the next 30 minutes is increased. The adjusted value is the smaller of the upper limit of hydrogen production power, the current hydrogen production power, and the change rate multiplied by the current output. At the same time, the hydrogen storage system is notified to reserve receiving capacity. Increasing hydrogen production power in advance can quickly absorb surplus renewable energy electricity, reducing power curtailment at the source. Reserved hydrogen storage capacity can prevent hydrogen production from exceeding storage capacity, ensuring the coordinated operation of hydrogen production and storage.

[0054] Step S43: When a sharp drop in output is detected, an emergency reduction command is triggered within 1 minute. The hydrogen production power is reduced to the greater of the following two values: the lower limit of hydrogen production power and the current hydrogen production power minus the absolute value of the rate of change multiplied by the current output power. If the power balance cannot be achieved after the reduction, the hydrogen storage system is activated to supply hydrogen in reverse, converting hydrogen into electricity through fuel cells to replenish the grid. Rapidly reducing the power consumption of the hydrogen production system can reduce the pressure on the grid supply, and the reverse hydrogen supply from the hydrogen storage system can promptly fill the power gap, maintaining grid power balance and operational stability.

[0055] Step S44: When the system is determined to be in a stable state, the current dispatch plan remains unchanged. Simultaneously, the forecast confidence interval is continuously monitored, and a 90% confidence interval is set based on the historical forecast error distribution. Adjustable capacity is reserved in the day-ahead dispatch for the difference between the upper and lower bounds of the confidence interval's output. This reserved adjustable capacity can address the uncertainty in renewable energy output forecasting, prevent dispatch plan failures due to forecast deviations, and improve the robustness of the collaborative dispatch system.

[0056] In one embodiment, step S5 employs a model predictive control algorithm to solve the cooperative scheduling optimization model, specifically including: S51: Set the prediction time domain to 16 time steps (corresponding to the next 4 hours, each step is 15 minutes), and the control time domain to 4 time steps (corresponding to the next 1 hour). At each rolling optimization time, construct a quadratic programming problem based on the current state measurement value and the future predicted disturbance sequence. S52: Discretize the objective function and constraints in step S3 into a quadratic programming standard form, where the sequence of control variables includes the hydrogen production power, hydrogen storage and release amount, and transportation scheduling amount at each time point in the future control time domain. S53: Solve the quadratic programming problem using the interior point method or the effective set method to obtain the optimal control sequence, which includes the hydrogen production power command, hydrogen storage charging and discharging command, and transportation scheduling command at each time in the current and future control time domains. S54: Only executes the instructions of the first control cycle in the optimal control sequence. At the next sampling time, it updates the system state measurement value, re-solves the optimization problem, and achieves rolling optimization.

[0057] As described in steps S51-S54 above, the model predictive control algorithm is used to solve the collaborative scheduling optimization model, clarifying the parameter settings, model discretization rules, standard solution methods, and instruction execution logic in the prediction time domain and control time domain. Through the rolling optimization mechanism, scheduling instructions that fit the real-time operating conditions are output, providing implementable control instructions for multi-time-scale collaborative scheduling and fluctuating hierarchical scheduling, ensuring the real-time performance and optimality of scheduling decisions.

[0058] The output of distributed renewable energy, grid load, and the status of hydrogen energy equipment are all in a continuous dynamic state. The collaborative scheduling optimization model contains multiple objective functions and multiple constraints, and needs to be matched with a time scale of 15 minutes of rolling correction within the day to complete the dynamic solution. If a one-time static solution method is used, the optimization results cannot be updated in real time with the operating conditions, and the scheduling instructions will deviate from the actual operating status. Therefore, it is necessary to use a model predictive control algorithm combined with a rolling optimization process to complete the model solution and ensure that the output instructions have dynamic adaptability.

[0059] Traditional hydrogen energy dispatching model solving techniques often employ static, one-time solutions, failing to establish prediction and control time domains for matching short-term dispatch, failing to convert multi-objective, multi-constraint models into a standard quadratic programming format, and lacking a rolling optimization mechanism. Consequently, the solution results lag behind actual operating conditions, and the efficiency and accuracy cannot meet the requirements of real-time collaborative dispatching. This invention addresses the technical problems of untimely model solutions and inapplicable results under dynamic operating conditions through a comprehensive scheme involving fixed time-domain parameter settings, standardized model discretization, efficient algorithm solving, and single-cycle instruction rolling execution.

[0060] Step S51: Set the prediction time domain to 16 time steps, corresponding to the next 4 hours, with each step being 15 minutes. Set the control time domain to 4 time steps, corresponding to the next 1 hour. At each rolling optimization moment, construct a quadratic programming problem based on the current state measurement value and the future predicted perturbation sequence. This parameter setting perfectly matches the time period of intraday rolling correction and the short-term prediction range, clearly defining the time boundary of the optimization calculation and providing a standardized parameter basis for constructing the quadratic programming problem.

[0061] Step S52 involves discretizing the objective function and constraints into a quadratic programming standard form. The sequence of control variables includes the hydrogen production capacity, hydrogen storage and release capacity, and transportation scheduling capacity at each time point in the future control time domain. Discretization transforms the complex multi-objective, multi-constraint optimization model into a general solvable mathematical format, unifies the dimensions of control variables, and effectively improves the algorithm's solution efficiency and stability.

[0062] Step S53: Solve the quadratic programming problem using either the interior-point method or the effective set method to obtain the optimal control sequence encompassing the current time and all future control time points. This sequence includes hydrogen production power commands, hydrogen storage charging and discharging commands, and transportation scheduling commands. Both algorithms are applicable to solving constrained quadratic programming problems and can quickly output scheduling commands that satisfy all operational constraints and optimization objectives, ensuring the accuracy and compliance of the solution results.

[0063] Step S54 executes only the instructions for the first control cycle in the optimal control sequence. At the next sampling time, the system state measurement values ​​are updated and the optimization problem is solved again, achieving rolling optimization. Single-cycle instruction execution combined with real-time state updates can continuously offset the impact of renewable energy output prediction errors and sudden changes in operating conditions, ensuring that scheduling instructions always match the actual operating state and fulfilling the core control requirements of dynamic collaborative scheduling.

[0064] In one embodiment, the step S6 of issuing instructions and controlling execution specifically includes: S61: Convert the hydrogen production power command into the current setting value of the electrolyzer. The conversion method is: divide the hydrogen production power command by the product of the number of electrolyzers in series and the voltage of a single electrolyzer to obtain the current setting value. Then, send it to the electrolyzer controller via industrial Ethernet. The controller uses proportional-integral-derivative regulation to make the actual power follow the setting value. S62: Convert the hydrogen storage charge / discharge command into a compressor or valve opening signal: If the hydrogen storage charge / discharge command is greater than 0, open the hydrogen charging valve and set the compressor speed to be proportional to the charge / discharge command; if the hydrogen storage charge / discharge command is less than 0, open the hydrogen discharging valve and set the pressure reducing valve opening to be proportional to the absolute value of the charge / discharge command; if the hydrogen storage charge / discharge command is equal to 0, close all valves. S63: Convert transportation dispatch instructions into vehicle dispatch orders. The dispatch order includes the specified vehicle number, hydrogen collection location (hydrogen storage station), delivery location (hydrogen refueling station or industrial user), planned departure time, and estimated arrival time. The dispatch order is pushed to the delivery person via mobile terminal and the vehicle status is updated to "dispatchd" in real time.

[0065] As described in steps S61-S63 above, the abstract scheduling instructions output by the model predictive control algorithm are transformed into physical control signals that can be directly identified and executed by hydrogen production, storage, and transportation equipment. This completes the transformation of scheduling decisions from the computational layer to the equipment execution layer, ensuring that the collaborative scheduling instructions accurately drive the operation of equipment across the entire industry chain, and truly realizing the scheduling goals of curtailment, cost optimization, and fluctuation adaptation.

[0066] The hydrogen production power, hydrogen storage and charging / discharging capacity, and transportation scheduling capacity output by the collaborative scheduling optimization model are all numerical abstract instructions. They cannot directly control physical equipment such as industrial-grade hydrogen electrolyzers, hydrogen storage compressors, and transportation vehicles. Hydrogen production, storage, and transportation equipment all have their own dedicated control logic and execution interfaces. If the instructions cannot be accurately converted and reliably issued, it will lead to a disconnect between scheduling decisions and equipment execution. Even if the scheduling algorithm calculates the optimal solution, it will not be able to drive the equipment to complete the corresponding actions. The core effects of fluctuation response, power balance, and power curtailment cannot be achieved. Therefore, it is necessary to convert the abstract scheduling instructions into specific control signals that the equipment can execute and complete the issuance and closed-loop execution control.

[0067] Traditional hydrogen energy systems simply transmit numerical commands without addressing the specific control characteristics of hydrogen production, storage, and transportation equipment. This results in poor command-equipment compatibility, insufficient execution precision, and a lack of a closed-loop adjustment mechanism for command execution. Consequently, accurate implementation of dispatch commands cannot be guaranteed, making it difficult to support the real-time and precision requirements of dynamic collaborative dispatch. This invention addresses the technical problems of abstract dispatch commands failing to directly drive equipment and lacking execution precision through a comprehensive solution that categorizes commands by equipment type, implements dedicated channels for command conversion, and establishes closed-loop adjustment execution.

[0068] Step S61: The hydrogen production power command is converted into the current setpoint of the electrolyzer. The conversion method is to divide the hydrogen production power command by the product of the number of electrolyzers connected in series and the voltage of a single electrolyzer section. After obtaining the current setpoint, it is sent to the electrolyzer controller via industrial Ethernet. The controller uses proportional-integral-derivative (PID) control to ensure that the actual power follows the setpoint. The electrolyzer uses current as the core control parameter. This conversion method perfectly matches the physical control logic of the electrolyzer. PID control can ensure that the hydrogen production power accurately follows the scheduling command, quickly respond to the power adjustment needs caused by output fluctuations, and ensure the execution accuracy of the hydrogen production process.

[0069] Step S62: The hydrogen storage charging / discharging command is converted into a compressor or valve opening signal. When the hydrogen storage charging / discharging command is greater than 0, the charging valve is opened, and the compressor speed is set to be proportional to the charging / discharging command. When the hydrogen storage charging / discharging command is less than 0, the discharging valve is opened, and the pressure reducing valve opening is set to be proportional to the absolute value of the charging / discharging command. When the hydrogen storage charging / discharging command is equal to 0, all valves are closed. This conversion method matches the physical control method of hydrogen charging / discharging in the hydrogen storage system, which can accurately control the hydrogen charging / discharging rate, avoid overcharging or over-discharging of the hydrogen storage system, stably connect the hydrogen production and transportation links, and ensure the safe and efficient operation of the hydrogen storage link.

[0070] Step S63 converts the transportation dispatch instruction into a vehicle dispatch order. The dispatch order includes the specified vehicle number, hydrogen pickup location, delivery location, planned departure time, and estimated arrival time. This order is pushed to the delivery person via a mobile terminal, and the vehicle status is updated in real time to show that it has been dispatched. This process transforms the numerical transportation dispatch instruction into a directly operable delivery task, matching the dispatch management process of the transportation vehicles. This ensures that hydrogen transportation is strictly executed according to the dispatch plan, achieving coordinated implementation of the transportation, hydrogen production, and hydrogen storage processes.

[0071] In one embodiment, step S7, which dynamically adjusts the curtailment penalty coefficient based on actual execution feedback and real-time monitoring results of the curtailment rate, specifically includes: S71: Define the real-time curtailment rate as the ratio of the integral of curtailed power to the integral of renewable energy output in the past hour, calculated every 15 minutes. S72: Set a target threshold for the curtailment rate. If the real-time curtailment rate is greater than the target threshold, the curtailment penalty coefficient will be dynamically increased as follows: multiply the current penalty coefficient by a factor greater than 1. This factor is equal to 1 plus the adjustment sensitivity coefficient multiplied by the difference between the real-time curtailment rate and the target threshold. At the same time, ensure that the increased penalty coefficient does not exceed the preset maximum value. S73: If the real-time power curtailment rate is less than or equal to the target threshold minus 2%, then gradually reduce the power curtailment penalty coefficient: multiply the current penalty coefficient by 0.9, while ensuring that the reduced penalty coefficient is not lower than the preset minimum value, to avoid the penalty coefficient being too small and causing the scheduling to ignore power curtailment; S74: If the curtailment rate continues to exceed the target threshold for more than 2 hours, an alarm will be triggered and the hydrogen production capacity will be forcibly increased to the maximum available value. At the same time, the power grid dispatch center will be notified to request external power transmission support.

[0072] As described in steps S71-S74 above, the curtailment penalty coefficient in the collaborative scheduling optimization model is dynamically adjusted based on actual execution feedback and real-time curtailment rate monitoring results. This constructs a closed-loop feedback adjustment mechanism for curtailment consumption, ensuring that scheduling decisions always prioritize the core objective of curtailment consumption, and stably controlling the actual curtailment rate within the preset target range. This enhances the dynamic adaptability of collaborative scheduling to the demand for distributed renewable energy consumption.

[0073] Curtailment absorption is the core technical objective of this invention to adapt to the fluctuating characteristics of distributed renewable energy. Random fluctuations in renewable energy output during real-time operation will cause the curtailment rate to change continuously. A fixed curtailment penalty coefficient cannot be dynamically adjusted according to the operating conditions, which will lead to problems such as the scheduling strategy being unable to correct the deviation in time after the curtailment rate exceeds the standard and the curtailment absorption effect not meeting the standard. Therefore, it is necessary to adaptively adjust the curtailment penalty coefficient according to the real-time curtailment rate. By optimizing the dynamic change of the target weight, the scheduling scheme is forced to tilt towards curtailment absorption, thereby realizing closed-loop management of the curtailment rate.

[0074] Traditional hydrogen energy dispatch technologies often employ fixed curtailment penalty coefficients, lacking dynamic adjustment mechanisms based on real-time curtailment rates and emergency response logic for persistently exceeding curtailment limits. This prevents the technology from autonomously optimizing dispatch targets based on actual consumption effects, resulting in insufficient flexibility and effectiveness in curtailment rate management, and failing to meet the application requirements for high consumption of distributed renewable energy. A comprehensive solution, employing real-time quantification of curtailment rates, tiered adjustment of penalty coefficients, and mandatory intervention for persistently exceeding limits, specifically addresses the technical problems of fixed curtailment penalty coefficients and the lack of closed-loop management for curtailment consumption.

[0075] Step S71: Define the real-time curtailment rate as the ratio of the integral of curtailed power to the integral of renewable energy output in the past hour, and perform the calculation every 15 minutes. The formula for calculating the real-time curtailment rate is: ; Among them, the This represents the real-time power curtailment rate over the past hour at time t. This represents the integral value of the abandoned power over the past hour. It represents the renewable energy output integral value over the past hour. Using the integral value to calculate the curtailment rate can smooth out the calculation deviation caused by instantaneous output fluctuations and objectively reflect the true curtailment level over a continuous period. The 15-minute calculation cycle is perfectly matched with the daily rolling correction rhythm, providing an accurate and stable quantitative basis for adjusting the curtailment penalty coefficient.

[0076] Step S72: Set a target threshold for the curtailment rate. When the real-time curtailment rate exceeds this target threshold, dynamically increase the curtailment penalty coefficient according to predetermined rules, while ensuring that the adjusted penalty coefficient does not exceed a preset maximum value. Increasing the curtailment penalty coefficient directly increases the weight of curtailment absorption in the optimization objective, forcing the dispatch system to increase hydrogen production capacity to absorb more surplus electricity, quickly bringing the curtailment rate back to the target range, and ensuring the curtailment absorption effect.

[0077] Step S73: When the real-time curtailment rate is less than or equal to the target threshold minus 2%, gradually reduce the curtailment penalty coefficient while ensuring that the adjusted penalty coefficient is not lower than the preset minimum value. Appropriately reducing the penalty coefficient can balance the priority relationship between comprehensive cost control and curtailment consumption, avoid unnecessary cost consumption caused by an excessively high penalty coefficient, and retain basic curtailment control capabilities to maintain the economy of the dispatching system.

[0078] Step S74: When the curtailment rate continues to exceed the target threshold for 2 hours, a system alarm is triggered and the hydrogen production capacity is forcibly increased to the maximum available value. At the same time, the power grid dispatch center is notified to request external power transmission support. This emergency response mechanism can deal with scenarios where the curtailment rate runs out of control due to extreme power output fluctuations. By forcibly intervening, it maximizes the absorption of surplus electricity and prevents the curtailment phenomenon from continuing to worsen.

[0079] In one embodiment, the rolling update of the prediction error distribution parameters in step S7 specifically includes: S75: Construct a prediction error sample set. Each sample is the difference between the predicted output and the measured output at a certain moment divided by the measured output. Use a rolling time window with a length of 168 hours (i.e., 1 week). Update the sample set after the end of each day, remove the earliest 24-hour data, and add the latest 24-hour data. S76: Calculate the mean (i.e., average prediction error) and standard deviation (i.e., the degree of fluctuation of prediction error) of the updated sample set, and update the prediction confidence interval accordingly: the lower bound of the 90% confidence interval is equal to the prediction output multiplied by (1 plus the average prediction error minus 1.645 times the standard deviation), and the upper bound is equal to the prediction output multiplied by (1 plus the average prediction error plus 1.645 times the standard deviation). S77: Recalculate the adjustable reserved capacity in day-ahead scheduling using the updated confidence interval. The adjustable reserved capacity is equal to the upper bound of the confidence interval minus the lower bound. Simultaneously update the resource output coefficient included in the power supply stability factor in the hydrogen production cost model (corrected according to the latest prediction error distribution) to achieve adaptive online learning of model parameters.

[0080] As described in steps S75-S77 above, a rolling time window is used to update the renewable energy output prediction error sample set, calculate the statistical characteristic value of the error distribution, and correct the prediction confidence interval and hydrogen production cost model parameters. This enables the scheduling model to learn adaptively online, continuously improve the prediction accuracy and the scheduling model's ability to adapt to long-term operating condition changes, and ensure the long-term stable operation of the collaborative scheduling scheme.

[0081] Renewable energy output forecasting errors can drift over long periods due to factors such as seasonal changes, weather variations, and equipment aging. Fixed forecasting error distribution parameters can lead to distorted confidence interval calculations, mismatch between adjustable reserved capacity settings and actual operating conditions, and resource output coefficients in hydrogen production cost models can also be miscalculated due to changes in the operating environment. Ultimately, this reduces the robustness of day-ahead scheduling and the calculation accuracy of the optimization model. Therefore, it is necessary to regularly update the forecasting error distribution parameters to ensure that the model parameters always match the latest actual operating conditions.

[0082] Traditional scheduling models often set prediction error parameters initially and only once, without using rolling time windows to dynamically update sample data or linking error distribution characteristics with cost model parameters for correction. This results in long-term deviations between the prediction confidence interval and the cost model from actual operating conditions, leading to a continuous decline in the long-term adaptability and computational accuracy of the scheduling system, failing to meet the requirements for long-term stable scheduling of distributed renewable energy. A comprehensive solution, employing rolling updates to the error sample set, calculation of error statistical characteristics, and linked correction of confidence intervals and cost model parameters, specifically addresses the technical problems of fixed prediction parameters and poor long-term adaptability of the scheduling model.

[0083] Step S75: Construct a prediction error sample set. A single sample is the difference between the predicted and measured output at a given moment, divided by the measured output. A rolling time window of 168 hours is used. At the end of each day, the earliest 24-hour data in the sample set is removed, and the latest 24-hour data is added. The rolling time window retains the most recent and valuable operational data while removing outdated and invalid historical information, ensuring that the prediction error sample set always reflects the latest output prediction deviation characteristics. This provides a clean and current-condition-appropriate data foundation for subsequent error distribution calculations.

[0084] Step S76: Calculate the average prediction error and the degree of prediction error fluctuation of the updated sample set. Based on this, update the 90% prediction confidence interval. The lower bound of the confidence interval is the predicted output multiplied by 1 plus the average prediction error minus 1.645 standard deviations, and the upper bound is the predicted output multiplied by 1 plus the average prediction error plus 1.645 standard deviations. Correcting the confidence interval based on real-time error statistical characteristics ensures that the interval range accurately matches the latest prediction error level, significantly improving the effectiveness and accuracy of the confidence interval and providing a reliable basis for calculating adjustable reserved capacity.

[0085] Step S77 involves recalculating the adjustable reserved capacity in day-ahead scheduling using the updated confidence interval, while simultaneously correcting the resource output coefficients included in the power supply stability factor of the hydrogen production cost model, thus achieving adaptive online learning of the model parameters. Linking the prediction error correction results with the hydrogen production cost model ensures that the cost calculation logic always aligns with the latest renewable energy output characteristics, continuously improving the calculation accuracy of the optimization model and guaranteeing the optimization effect and long-term stability of coordinated scheduling.

[0086] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in this application and in the embodiments can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0087] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, apparatus, article, or method that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, apparatus, article, or method. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, apparatus, article, or method that includes that element.

[0088] The above description is merely a preferred embodiment of the present invention and does not limit the scope of this application. Any equivalent results or equivalent process transformations made based on the content of the present invention's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the scope of protection of this application.

Claims

1. A hydrogen energy collaborative scheduling method adapted to the characteristics of distributed renewable energy, characterized in that, Includes the following steps: Collect real-time output data, predicted output data, grid load data of distributed renewable energy power generation systems, and status data of hydrogen production systems, hydrogen storage systems, and transportation systems in the hydrogen energy industry chain; A multi-timescale collaborative scheduling framework is constructed, which includes a day-ahead scheduling phase, an intraday rolling correction phase, and a real-time adjustment phase. A multi-objective collaborative scheduling optimization model is established with the goals of maximizing the utilization of curtailed electricity and minimizing the overall cost. The overall cost includes hydrogen production cost, hydrogen storage cost, and transportation cost, and a curtailment penalty term is introduced. Based on the fluctuating characteristics of renewable energy output, a fluctuating hierarchical scheduling strategy is designed. The model predictive control algorithm is used to solve the cooperative scheduling optimization model to generate hydrogen production power instructions, hydrogen storage charging and discharging instructions, and transportation scheduling instructions. The instructions are sent to the hydrogen production system, hydrogen storage system, and transportation system, and their execution is controlled. Based on the actual execution feedback and real-time monitoring results of the curtailment rate, the curtailment penalty coefficient in the optimization model is dynamically adjusted, and the prediction error distribution parameters are updated on a rolling basis.

2. The hydrogen energy collaborative scheduling method adapted to the characteristics of distributed renewable energy as described in claim 1, characterized in that, The steps for collecting real-time output data, predicted output data, grid load data of distributed renewable energy power generation systems, and status data of hydrogen production systems, hydrogen storage systems, and transportation systems in the hydrogen energy industry chain specifically include: Real-time power output is collected from distributed renewable energy power plants, and forecast power output data for future preset time periods is obtained from meteorological forecasting systems. Obtain the current grid load from the grid dispatch center, and calculate the power curtailment based on real-time power output, grid load, and the maximum absorption power of the hydrogen production system; Collect data on the upper and lower limits of the operating power and current efficiency of the hydrogen production system, the current hydrogen storage capacity, the upper limit of the hydrogen storage capacity, and the hydrogen charging and discharging rate limits of the hydrogen storage system, and the number of available vehicles, the carrying capacity of a single vehicle, the real-time location of the vehicles, and their driving speed of the transportation system.

3. The hydrogen energy collaborative scheduling method adapted to the characteristics of distributed renewable energy as described in claim 1, characterized in that, The specific steps for constructing the multi-timescale collaborative scheduling framework include: During the day-ahead scheduling phase, based on the next day's renewable energy output forecast and load forecast, a hydrogen production power plan curve, hydrogen storage system charging and discharging plan, and transportation plan for the next preset period are formulated with a preset first time resolution and issued as a baseline plan. During the intraday rolling correction phase, rolling optimization is triggered at a preset second time interval. Starting from the current moment, based on the latest measured data and short-term forecast data, the model predictive control algorithm is used to continuously update the scheduling instructions for the future preset time period. After each optimization, only the control instructions for the current time interval are executed. During the real-time adjustment phase, a deviation trigger threshold is set. When the deviation between the measured renewable energy output and the day-ahead forecast exceeds this threshold, or when the grid frequency deviation exceeds the preset frequency deviation threshold, real-time adjustment is initiated.

4. The hydrogen energy collaborative scheduling method adapted to the characteristics of distributed renewable energy as described in claim 1, characterized in that, The steps for establishing a multi-objective collaborative scheduling optimization model with the objectives of maximizing power curtailment and minimizing overall costs specifically include the objective function and various cost models: The objective function is defined as minimizing the sum of the overall cost and the curtailment penalty, where the overall cost includes hydrogen production cost, hydrogen storage cost, and transportation cost; A hydrogen production cost model is constructed, which calculates the hydrogen production cost based on available resources, planned resources, power supply stability factor, supporting environmental and equipment cost coefficients, main equipment system cost, and system operation cost coefficient. A hydrogen storage cost model is constructed, which calculates the hydrogen storage cost based on the cost of hydrogen storage equipment system, the maintenance cost coefficient of hydrogen storage system, the operating cost of hydrogen storage system, as well as the correlation with application scenarios and transportation methods. A transportation cost model is constructed, which calculates transportation costs based on input costs, operating costs, maintenance cost coefficients, applicability coefficients, and transmission distance factors.

5. The hydrogen energy collaborative scheduling method adapted to the characteristics of distributed renewable energy as described in claim 1, characterized in that, The multi-objective cooperative scheduling optimization model also includes the following constraints: Power balance constraint: Real-time output power equals the sum of grid load, actual power consumption of hydrogen production system, and power curtailment. Hydrogen production power operation constraints: The actual power consumption of the hydrogen production system must be between its lower and upper limits of available power, and the change in hydrogen production power between adjacent moments cannot exceed the maximum ramp rate of the hydrogen production system; Dynamic constraints of hydrogen storage system: The amount of hydrogen stored at the next moment is calculated based on the amount of hydrogen stored at the current moment, hydrogen charging efficiency, hydrogen production power, time step, hydrogen release efficiency and the hydrogen supply power of the hydrogen storage system to the outside world, and the amount of hydrogen stored at any moment must be between zero and the upper limit of hydrogen storage capacity. Transportation coordination constraints: The total transportation volume during the entire scheduling cycle shall not exceed the hydrogen storage volume at the end of the scheduling period minus the minimum safe storage volume, and the total hydrogen load of all transportation vehicles at any time shall not exceed the sum of the maximum hydrogen load of a single vehicle. Renewable energy curtailment rate cap constraint: The ratio of the sum of curtailed power to the sum of renewable energy output during the entire dispatch cycle shall not exceed the preset maximum allowable curtailment rate.

6. The hydrogen energy collaborative scheduling method adapted to the characteristics of distributed renewable energy as described in claim 1, characterized in that, The steps for designing a tiered scheduling strategy based on the fluctuation characteristics of renewable energy output specifically include: The rate of change of renewable energy output is calculated in real time, a classification threshold is set, and the operating condition is determined as a sharp increase in output, a sharp decrease in output, or a stable state based on the comparison between the rate of change and the classification threshold. If the event is determined to be a sudden increase in output, a pre-increase command is triggered: the hydrogen production power setting value for the future preset time period is increased, and the increased value is the smaller of the upper limit of hydrogen production power and the current hydrogen production power plus the adjustment amount related to the rate of change, and the hydrogen storage system is notified to reserve receiving capacity. If the power output is determined to be a sharp drop event, an emergency reduction command will be triggered: the hydrogen production power will be reduced within the preset emergency response time. The reduced value will be the larger of the lower limit of hydrogen production power and the current hydrogen production power minus the adjustment amount related to the rate of change. If the power balance still cannot be met after the reduction, the hydrogen storage system will be activated to supply hydrogen in reverse and convert hydrogen into electricity through fuel cells to replenish the grid. If the state is determined to be stable, the current scheduling plan will be maintained and the prediction confidence interval will be continuously monitored: based on the historical prediction error distribution, a confidence interval with a preset confidence level will be set, and adjustable capacity will be reserved in the day-ahead scheduling. The adjustable capacity is equal to the output at the upper limit of the confidence interval minus the output at the lower limit.

7. The hydrogen energy collaborative scheduling method adapted to the characteristics of distributed renewable energy as described in claim 1, characterized in that, The steps of solving the cooperative scheduling optimization model using the model predictive control algorithm specifically include: Set a prediction time domain and a control time domain of preset lengths. At each rolling optimization time, construct a quadratic programming problem based on the current state measurement value and the future predicted disturbance sequence. The objective function and constraints are discretized into a standard quadratic programming form, where the sequence of control variables includes the hydrogen production capacity, hydrogen storage and release capacity, and transportation scheduling capacity at each time point in the future control time domain. The quadratic programming problem is solved by an optimization algorithm to obtain the optimal control sequence, which includes the hydrogen production power command, hydrogen storage charging and discharging command, and transportation scheduling command at each time in the current and future control time domains. Only the instructions of the first control cycle in the optimal control sequence are executed. The system state measurement is updated at the next sampling time, and the optimization problem is solved again to achieve rolling optimization.

8. The hydrogen energy collaborative scheduling method adapted to the characteristics of distributed renewable energy as described in claim 1, characterized in that, The steps of issuing the instructions to the hydrogen production system, hydrogen storage system, and transportation system, and controlling their execution, specifically include: The hydrogen production power command is converted into the current set value of the electrolyzer and sent to the electrolyzer controller via industrial Ethernet. The controller uses closed-loop regulation to make the actual power follow the set value. Convert hydrogen storage charging / discharging commands into compressor or valve opening signals: If the hydrogen storage charge / discharge command indicates hydrogen charging, then open the hydrogen charging valve and set the compressor speed according to the command ratio; If hydrogen release is indicated, open the hydrogen release valve and set the pressure reducing valve opening according to the command ratio; If the command is zero, then close all valves; Transportation dispatch instructions are converted into vehicle dispatch orders, which include the designated vehicle number, hydrogen collection location, delivery location, planned departure time, and estimated arrival time. These orders are then pushed to delivery personnel via mobile devices, and vehicle status is updated in real time.

9. The hydrogen energy collaborative scheduling method adapted to the characteristics of distributed renewable energy as described in claim 1, characterized in that, The step of dynamically adjusting the curtailment penalty coefficient in the optimization model based on actual execution feedback and real-time monitoring results of the curtailment rate specifically includes: The real-time curtailment rate is defined as the ratio of the curtailed power integral to the renewable energy output integral within a preset historical time window, and is calculated according to a preset period. Set a target threshold for the curtailment rate. If the real-time curtailment rate exceeds the target threshold, the curtailment penalty coefficient will be dynamically increased. If the real-time curtailment rate is less than or equal to the target threshold minus the preset backlash threshold, the curtailment penalty coefficient will be gradually reduced. If the curtailment rate continues to exceed the target threshold for an extended period of time, an alarm will be triggered and the hydrogen production capacity will be forcibly increased to the maximum available value. At the same time, the power grid dispatch center will be notified to request external power transmission support.

10. The hydrogen energy collaborative scheduling method adapted to the characteristics of distributed renewable energy according to claim 1, characterized in that, The specific steps for rolling up and updating the prediction error distribution parameters include: Construct a prediction error sample set, where each sample represents the relative deviation between the predicted output and the measured output. Use a rolling time window of a preset length to periodically update the sample set, removing data from the earliest time window and adding the latest data. Calculate the mean prediction error and standard deviation of the prediction error for the updated sample set, and update the prediction confidence interval accordingly. The adjustable reserved capacity in day-ahead scheduling is recalculated using the updated confidence interval. The adjustable reserved capacity is equal to the upper bound of the confidence interval minus the lower bound. The resource output coefficients included in the power supply stability factor in the hydrogen production cost model are updated simultaneously to achieve adaptive online learning of model parameters.

Images