An electric-hydrogen hybrid energy storage cross-seasonal hierarchical optimization scheduling method, device and medium

Abstract

The present application relates to a kind of electric hydrogen hybrid energy storage cross-seasonal hierarchical optimization scheduling method, equipment and medium, comprising: collecting photovoltaic, load, electricity price, battery energy storage, hydrogen storage, equipment constraint and grid interaction data;Annual planning layer model is constructed, monthly hydrogen storage target, monthly battery energy reference and life loss control requirement are obtained;Monthly scheduling layer model is constructed, monthly hydrogen storage target is converted into weekly hydrogen storage reference trajectory, equipment benchmark output plan and grid interaction plan, and output hourly reference trajectory;According to photovoltaic and load data generation short-term prediction data, combine current hydrogen storage amount, battery state of charge and weekly hydrogen storage reference trajectory, construct intra-day execution layer model to carry out rolling optimization, obtain the power control instruction of each equipment in current scheduling period;The control instruction is executed, and subsequent scheduling is updated according to the state after execution.It has the advantages of high executability, strong robustness and long running life compared with prior art.

Description

Technical Field

[0001] This invention relates to the field of optimized scheduling and control technology, and in particular to a cross-seasonal hierarchical optimized scheduling method, equipment and medium for hybrid electric-hydrogen energy storage. Background Technology

[0002] With a high proportion of photovoltaic (PV) power being integrated into low-carbon communities, industrial parks, and microgrids, the source-load matching problem is becoming increasingly complex, exhibiting not only short-term intraday fluctuations but also significant cross-monthly and cross-seasonal energy mismatches. Specifically, there is a surplus of PV power in spring and summer, while a persistent energy shortage is likely to occur in autumn and winter or during periods of continuous low irradiance. While single electrochemical energy storage offers advantages such as fast response and high efficiency, making it suitable for hourly power smoothing and short-term regulation, its capacity, cost, and cycle life limitations make it unsuitable for long-term cross-seasonal energy transfer. In contrast, hydrogen energy storage features large capacity, long storage period, and low self-discharge rate, making it suitable for cross-monthly and cross-seasonal energy transfer. Therefore, hybrid electro-hydrogen energy storage is gradually becoming an effective solution to these problems.

[0003] Existing scheduling methods for hybrid electric-hydrogen energy storage systems primarily focus on capacity planning, day-ahead / intra-day optimization, typical daily operation, or handling uncertainties at a single time scale. However, for hybrid electric-hydrogen energy storage systems with significant cross-seasonal coupling characteristics, existing technologies still have shortcomings in terms of scheduling coordination, target executability, and equipment lifespan management. First, long-term seasonal hydrogen storage targets are usually limited to annual or monthly levels, lacking an effective connection mechanism with hourly scheduling execution. This leads to structural deviations when long-term targets are transmitted to intra-day scheduling, making it difficult to guarantee executability in actual operation. Significant uncertainties exist in photovoltaic output and load demand at monthly and weekly scales. If sufficient safety margins are not reserved in the plan, intra-day scheduling is highly susceptible to disturbances, deviating from the established hydrogen storage trajectory. This reduces the system's ability to cope with source-load fluctuations, increasing the risk of low hydrogen storage and insufficient power supply. Furthermore, existing scheduling models often fail to adequately consider the relationship between battery life degradation characteristics and net load frequency distribution. This makes batteries susceptible to over-utilization by optimization algorithms for long-cycle energy transfer, leading to unnecessary deep discharges and frequent cycling, accelerating battery aging, increasing system operation and maintenance costs, and blurring the functional positioning of batteries and hydrogen energy storage. Therefore, how to construct an electric-hydrogen hybrid energy storage scheduling scheme that can effectively transfer hydrogen storage targets across multiple time scales (annual, monthly, and daily), reserve risk buffers, and constrain battery life degradation is a technical problem that needs to be solved. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of the existing technology by providing a cross-seasonal hierarchical optimization scheduling method, device, and medium for hybrid electric-hydrogen energy storage. This is achieved by constructing a three-layer hierarchical feedback scheduling framework consisting of an annual planning layer, a monthly scheduling layer, and a daily execution layer. At the monthly layer, two-stage stochastic planning is used to reserve a safety margin for hydrogen storage. At the daily layer, the decomposition of net load fast and slow components decouples the functions of short-term battery adjustment and long-term hydrogen storage relocation. Simultaneously, nonlinear battery degradation costs are introduced to suppress deep battery discharge and excessive cycling, effectively improving the feasibility of cross-seasonal hydrogen storage targets in hourly scheduling.

[0005] The objective of this invention can be achieved through the following technical solutions: According to one aspect of the present invention, a cross-seasonal hierarchical optimization scheduling method for hybrid electric-hydrogen energy storage is provided, the specific steps of which include: S1. Collect photovoltaic output data, load data, electricity price data, battery energy storage status data, hydrogen storage status data, equipment operation constraint data, and external power grid interaction constraint data within the scheduling cycle; S2. Based on the collected data, taking months as the stage, and using the beginning hydrogen storage amount in the hydrogen storage status data, the battery state of charge in the battery energy storage status data, and the cumulative life loss in the battery energy storage status data as state variables, construct an annual planning layer model and solve it to obtain the monthly hydrogen storage target, monthly battery energy reference, and life loss control requirements. S3. Based on the monthly hydrogen storage target, generate multiple disturbance scenarios based on the uncertainty distribution of source and load, and construct and solve a monthly scheduling layer model based on the disturbance scenarios. Transform the monthly hydrogen storage target into a weekly hydrogen storage reference trajectory, equipment baseline output plan and grid interaction plan, and output an hourly reference trajectory. S4. Generate short-term photovoltaic forecast data based on the photovoltaic output data, generate short-term load forecast data based on the load data, and construct an intraday execution layer model based on the short-term photovoltaic forecast data, short-term load forecast data, current hydrogen storage capacity, current battery state of charge, and weekly hydrogen storage reference trajectory, and perform rolling optimization solution to obtain control commands for the electrolyzer power, fuel cell power, battery charging and discharging power, and grid interaction power during the current scheduling period. S5. Execute the control command for the current scheduling period, and update the subsequent scheduling status based on the hydrogen storage capacity, battery state of charge, and cumulative lifetime loss after execution.

[0006] Furthermore, the annual planning layer model in S2 is solved using a finite-stage Markov decision process or an equivalent dynamic programming model. The decision actions of the annual planning layer model include the monthly net change in hydrogen storage, the electrolyzer reference power, the fuel cell reference power, the battery reference adjustment, and the grid interaction. The immediate cost of the annual planning layer model is determined based on the grid interaction cost, equipment operation and maintenance cost, battery degradation cost, hydrogen storage target deviation penalty, and low hydrogen storage risk penalty. The low hydrogen storage risk penalty is calculated through a low hydrogen storage relaxation variable, which is activated when the hydrogen storage is lower than the monthly minimum hydrogen storage safety margin. The annual planning layer outputs the monthly hydrogen storage target, the monthly battery energy reference, and the life loss control requirements.

[0007] Furthermore, the monthly scheduling layer model in S3 is solved using a two-stage stochastic programming model. Uncertainty scenarios are divided into first-stage variables and second-stage variables. The first-stage variables are determined before photovoltaic and load disturbances occur, including weekly hydrogen storage reference, equipment baseline output, battery baseline net energy, and grid baseline interaction. The second-stage variables are determined after the disturbance scenarios occur, including equipment correction, supply shortage relaxation variables, power curtailment relaxation variables, and low hydrogen storage relaxation variables. The objective function of the monthly scheduling layer model is determined based on the first-stage baseline plan cost, the second-stage scenario expected correction cost, hydrogen storage target tracking cost, battery energy tracking cost, and low hydrogen storage risk cost, and satisfies the unexpectedness constraint, ensuring that the weekly baseline plan does not depend on future disturbance scenarios. The monthly scheduling layer model outputs a weekly hydrogen storage reference trajectory, equipment baseline output plan, and grid interaction plan, and also outputs an hourly reference trajectory. The monthly scheduling layer model satisfies power balance constraints, equipment output boundary constraints, hydrogen storage state transition constraints, battery energy state transition constraints, hydrogen storage safety margin constraints, and second-stage correction constraints under various disturbance scenarios. The hydrogen storage safety margin constraint measures the amount of hydrogen storage below the monthly minimum hydrogen storage safety margin through a low hydrogen storage relaxation variable, and will be included in the low hydrogen storage risk penalty term.

[0008] Furthermore, in step S4, before constructing the intraday execution layer model and performing rolling optimization, a net load fast / slow component decomposition operation is performed, which includes: Net load is determined based on the difference between the short-term load forecast data and the short-term photovoltaic forecast data; a slow net load variable is extracted using a moving average window, and a fast net load component is obtained based on the difference between the net load and the slow net load variable; a battery reference net power is generated based on the fast net load component, enabling the battery energy storage unit to track the fast net load component to compensate for hourly rapid fluctuations, while the electrolyzer, fuel cell, and external power grid bear the energy balance of the slow net load variable; the battery reference net power is limited to the range of the battery's maximum charging power and maximum discharging power by a saturation function.

[0009] Furthermore, the intraday execution layer model in S4 is solved using a model predictive control model, with several hours as the prediction time domain and one or more scheduling periods as the rolling step size. At each rolling moment, only the control quantity of the current scheduling period is executed, and it is re-initialized according to the actual state at the next rolling moment.

[0010] Furthermore, the objective function of the intraday execution layer model includes a nonlinear battery degradation cost, which is determined based on the single-period discharge depth, degradation parameters, single-period maximum discharge penalty threshold, and battery replacement cost. When the single-period discharge depth exceeds the single-period maximum discharge penalty threshold, the marginal degradation cost increases. The hourly battery degradation increments are accumulated on a monthly scale to obtain the cumulative lifetime loss at the beginning of the next month, and the cumulative lifetime loss is fed back to the annual planning layer as a state variable.

[0011] Furthermore, the control commands include the hydrogen production power of the electrolyzer, the power generation power of the fuel cell, the charging power of the battery, the discharging power of the battery, the power purchased from the external power grid, and the power sold from the external power grid; the control commands satisfy the upper and lower limits of hydrogen storage capacity, the upper and lower limits of battery energy, and the upper and lower limits of equipment power.

[0012] Furthermore, the equipment operation constraint data includes upper and lower limits for electrolyzer power, upper and lower limits for fuel cell power, upper and lower limits for battery charge and discharge power, upper and lower limits for hydrogen storage capacity, and upper and lower limits for battery energy; the external power grid interaction constraint data includes upper and lower limits for purchased power and upper and lower limits for sold power.

[0013] According to a second aspect of the present invention, an electronic device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the program to implement the method described thereon.

[0014] According to a third aspect of the present invention, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the method described thereon.

[0015] Compared with the prior art, the present invention has the following beneficial effects: (1) This invention constructs a three-layer hierarchical feedback scheduling framework consisting of an annual planning layer, a monthly scheduling layer, and an intraday execution layer. The annual planning layer generates monthly hydrogen storage targets in monthly phases. The monthly scheduling layer transforms the monthly hydrogen storage targets into weekly hydrogen storage reference trajectories with hydrogen storage safety margins and outputs hourly reference trajectories. The intraday execution layer continuously optimizes and solves the control instructions for the current scheduling period based on the hourly reference trajectories and short-term forecast data. After execution, the actual status is fed back up level by level, so that the cross-seasonal hydrogen storage targets can be passed from the annual layer to the monthly layer and then decomposed into executable intraday instructions. At the same time, the actual execution deviation of the intraday layer affects the correction of the upper-level plan through status feedback, thereby effectively coordinating the connection between long-term energy transfer plans and short-term scheduling execution, and improving the executability of cross-seasonal hydrogen storage targets in hourly scheduling.

[0016] (2) In the monthly scheduling layer, the present invention generates multiple disturbance scenarios based on the uncertainty distribution of source and load, and solves them using a two-stage stochastic programming model. In the first stage, the variables determine the weekly hydrogen storage reference and grid benchmark interaction before the disturbance is realized. In the second stage, the variables are adjusted by using low hydrogen storage relaxation variables to ensure that the hydrogen storage is not lower than the monthly minimum hydrogen storage safety margin. Thus, a safety margin is reserved in the monthly plan for the random fluctuations of photovoltaic output and load demand, which can enhance the robustness of the weekly plan under external source and load disturbances, thereby reducing the risk of low hydrogen storage and insufficient power supply in low-carbon communities during periods of continuous low irradiance or high load.

[0017] (3) In the intraday execution layer, the difference between short-term photovoltaic forecast data and short-term load forecast data is used as the net load. The net load is decomposed into slow variables and fast components through a moving average window, so that the battery energy storage unit tracks the fast component of the net load to bear the compensation of hourly rapid fluctuations. At the same time, the electrolyzer, fuel cell and external power grid bear the energy balance of the slow variable of the net load, realizing the decoupling of the functions of short-term battery regulation and long-term energy transfer of hydrogen energy storage. In addition, a nonlinear battery degradation cost determined based on the single-period discharge depth and the single-period large discharge penalty threshold is introduced into the objective function of the intraday execution layer. When the single-period discharge depth exceeds the penalty threshold, the marginal degradation cost increases. The hourly battery degradation increment is accumulated on the monthly scale and fed back to the annual planning layer as the state variable of the annual planning layer. In this way, the deep discharge and excessive cycling of the battery on the single-period scale are suppressed in the scheduling decision, the actual operating life of the battery is extended, and the operation and maintenance cost of the energy storage system throughout its entire life cycle is reduced. Attached Figure Description

[0018] Figure 1 A flowchart for a cross-seasonal hierarchical optimization scheduling method for hybrid electric-hydrogen energy storage; Figure 2 A hierarchical scheduling data flow diagram for the annual planning layer, monthly scheduling layer, and daily execution layer; Figure 3 This is a schematic diagram illustrating the decoupling of the fast and slow components of net load and the feedback of battery degradation. Figure 4 This is a graph showing the monthly net distribution of hydrogen storage and changes in hydrogen inventory. Figure 5 A comparative chart showing the hydrogen storage trajectory tracking at the annual, monthly, and intraday levels; Figure 6 A comparison chart of hydrogen storage trajectory tracking under baseline and disturbance scenarios; Figure 7 A comparison chart of hydrogen storage tracking errors between the M1 and M3 models; Figure 8 This is a probability distribution map of the depth of discharge of a battery in a single time period; Figure 9 A comparison chart of normalized operating indicators for batteries; Figure 10 A comparison chart of planned minimum hydrogen storage and actual minimum hydrogen storage; Figure 11 This is a comparison chart of supply shortages and low hydrogen storage hours. Figure 12 A comparison chart of hydrogen storage margin and tracking indicators; Figure 13 A chart comparing economic costs and risk indicators; Figure 14 This is a comparative chart showing the supply shortage statistics under disturbance scenarios. Detailed Implementation

[0019] 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. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0020] Hybrid energy storage systems, combining battery and hydrogen storage, offer advantages in addressing short-term intraday fluctuations in energy supply and load and seasonal energy mismatches, and have gained widespread attention in low-carbon communities, industrial parks, and microgrids in recent years. However, given that hybrid energy storage systems involve scheduling decisions across multiple timescales (annual, monthly, and intraday), establishing an effective mechanism to bridge long-term energy management goals with short-term operational control remains a major challenge in current engineering applications. Existing scheduling methods typically separate annual and intraday plans, lacking a coordinated framework for hierarchical transmission and feedback correction, leading to deviations in seasonal hydrogen storage targets when decomposed to short-term scheduling. Furthermore, monthly photovoltaic output and load demand exhibit significant uncertainty. If sufficient safety margins are not reserved during the planning phase, intraday execution is highly susceptible to disturbances, deviating from the expected trajectory and increasing system operational risks. Simultaneously, without reasonable lifetime constraints and functional positioning, battery storage units are easily exploited by optimization algorithms for long-cycle energy transfer, resulting in unnecessary deep cycles and shortening their service life. To address the aforementioned issues, this invention constructs a hierarchical optimization scheduling scheme that includes annual target generation, monthly risk buffering, and intraday rolling execution. The scheme establishes monthly hydrogen storage targets at the annual planning layer, reserves safety margins based on two-stage stochastic programming at the monthly scheduling layer, and employs model predictive control combined with net load fast / slow component decomposition and nonlinear battery degradation costs at the intraday execution layer. This allows batteries to centrally handle hourly rapid fluctuation compensation, while hydrogen storage handles cross-seasonal energy transfer, thereby achieving multi-timescale coordinated optimization, improving the feasibility of hydrogen storage targets, and suppressing excessive battery wear.

[0021] like Figure 1 As shown in this embodiment, a cross-seasonal hierarchical optimization scheduling method for hybrid electric-hydrogen energy storage is provided. The specific steps include: S1. Collect photovoltaic output data, load data, electricity price data, battery energy storage status data, hydrogen storage status data, equipment operation constraint data, and external power grid interaction constraint data within the scheduling cycle; S2. Based on the collected data, taking months as the stage, and using the beginning hydrogen storage amount in the hydrogen storage status data, the battery state of charge in the battery energy storage status data, and the cumulative life loss in the battery energy storage status data as state variables, construct an annual planning layer model and solve it to obtain the monthly hydrogen storage target, monthly battery energy reference, and life loss control requirements. S3. Based on the monthly hydrogen storage target, multiple disturbance scenarios are generated based on the uncertainty distribution of source and load. Based on the disturbance scenarios, a monthly scheduling layer model is constructed and solved. The monthly hydrogen storage target is transformed into a weekly hydrogen storage reference trajectory, equipment baseline output plan and grid interaction plan, and an hourly reference trajectory is output. S4. Generate short-term photovoltaic forecast data based on photovoltaic output data, generate short-term load forecast data based on load data, and construct an intraday execution layer model based on short-term photovoltaic forecast data, short-term load forecast data, current hydrogen storage capacity, current battery state of charge, and weekly hydrogen storage reference trajectory, and perform rolling optimization solution to obtain control commands for electrolyzer power, fuel cell power, battery charging and discharging power, and grid interaction power during the current scheduling period. S5. Execute the control instructions for the current scheduling period, and update the subsequent scheduling status based on the hydrogen storage capacity, battery state of charge, and cumulative lifespan loss after execution.

[0022] The method in this embodiment is used in an electro-hydrogen hybrid energy storage system, which includes a photovoltaic power generation unit, an electrical load unit, a battery energy storage unit, an electrolyzer, a hydrogen storage device, a fuel cell, an external power grid interface, and a hierarchical optimization scheduling controller. The hierarchical optimization scheduling controller carries a program that implements the method of this embodiment, used to collect source load and energy storage status, and to issue power control commands to the battery energy storage unit, the electrolyzer, the fuel cell, and the external power grid interface.

[0023] like Figure 2 As shown, this embodiment adopts a hierarchical scheduling sequence of annual target generation, monthly risk buffering, and intraday rolling execution: the annual planning layer generates monthly hydrogen storage targets at the beginning of the year or during rolling updates; the monthly scheduling layer forms weekly plans with a monthly planning cycle and a weekly rolling update step size; the intraday execution layer uses hourly scheduling resolution and performs model predictive control based on short-term forecasts and real-time status.

[0024] The annual planning layer model in S2 is solved using a finite-stage Markov decision process or an equivalent dynamic programming model. The decision actions of the annual planning layer model include the monthly net change in hydrogen storage, the reference power of the electrolyzer, the reference power of the fuel cell, the reference adjustment of the battery, and the grid interaction. The immediate cost of the annual planning layer model is determined based on the grid interaction cost, equipment operation and maintenance cost, battery degradation cost, hydrogen storage target deviation penalty, and low hydrogen storage risk penalty. The low hydrogen storage risk penalty is calculated through a low hydrogen storage relaxation variable, which is activated when the hydrogen storage is lower than the monthly minimum hydrogen storage safety margin. The annual planning layer outputs the monthly hydrogen storage target, the monthly battery energy reference, and the lifetime loss control requirements.

[0025] Specifically, the annual planning level is based on months. This is a phase used to generate cross-seasonal hydrogen storage targets. It represents the system state at the annual planning level. Decision-making actions The state transitions are represented as follows: , , , in, Indexed by month; For the first System status at the beginning of the month, including hydrogen storage at the beginning of the month. Battery energy at the beginning of the month and cumulative lifespan loss at the beginning of the month ; For the first Monthly decision-making actions, including monthly net change in hydrogen storage. Electrolytic cell reference power Fuel cell reference power Battery reference adjustment Interaction with the power grid ; This refers to monthly net load disturbance. To execute actions and take into account monthly net load disturbances The next System status at the beginning of the month; For the first The system state transition function for the month.

[0026] Immediate costs at the annual planning level Power grid interaction cost Equipment operation and maintenance costs Battery degradation costs The components are: hydrogen storage target deviation penalty and low hydrogen storage risk penalty. , , , , , in, To execute actions and take into account monthly net load disturbances The next Hydrogen storage at the beginning of the month; The first given for the annual tier Monthly hydrogen storage reference target; For the first Minimum monthly hydrogen storage safety margin; For the first Hydrogen storage at the beginning of the month; For the first Low monthly hydrogen storage relaxation variables; This is the penalty coefficient for deviation from the hydrogen storage target; The penalty coefficient for low hydrogen storage risk; Indexed by time period; For the first Monthly hourly collection; For the first Time-of-use electricity pricing; For the first Power purchased during specific time periods; For the first Electricity price during specific time periods; For the first Electricity sales volume during specific time periods; The duration of the scheduling period; For device indexing; A collection of devices; For equipment The operation and maintenance cost coefficient; For the first Time-of-use equipment The power; For the first Battery degradation costs over time.

[0027] The goal of the annual planning layer is to minimize the expected total annual cost while satisfying constraints on equipment power, hydrogen storage boundary, battery energy boundary, and lifespan degradation. The Bellman recursive model for the annual planning layer is used to solve this problem. , , in, For the first The moon is in a state of flux. The value function of time; For state The set of feasible actions that satisfy the constraints of equipment power, hydrogen storage boundary, battery energy boundary and lifetime loss; For the first The moon is in a state of flux. The value function of time; Regarding monthly net load disturbance The expectation operator; To perform the action And take into account monthly net load disturbances. Later by the beginning of the month status Transfer to the beginning of the month status State transition function; To minimize the expected total cost for the year The monthly optimal decision-making action annual planning layer outputs monthly hydrogen storage targets, monthly battery energy references, and lifespan loss control requirements; the monthly scheduling layer generates weekly hydrogen storage reference trajectories based on the monthly hydrogen storage targets.

[0028] The monthly scheduling layer model in S3 employs a two-stage stochastic programming model for solution. Uncertainty scenarios are divided into first-stage variables and second-stage variables. First-stage variables are determined before photovoltaic and load disturbances occur, including weekly hydrogen storage reference, equipment baseline output, battery baseline net energy, and grid baseline interaction. Second-stage variables are determined after the disturbance scenarios occur, including equipment correction, supply shortage relaxation variables, power curtailment relaxation variables, and low hydrogen storage relaxation variables. The objective function of the monthly scheduling layer model is determined based on the first-stage baseline plan cost, the second-stage scenario expected correction cost, hydrogen storage target tracking cost, battery energy tracking cost, and low hydrogen storage risk cost, while satisfying the unexpectedness constraint to ensure that the weekly baseline plan does not depend on future disturbance scenarios. The monthly scheduling layer model outputs a weekly hydrogen storage reference trajectory, equipment baseline output plan, and grid interaction plan, and also outputs an hourly reference trajectory. The monthly scheduling layer model satisfies power balance constraints, equipment output boundary constraints, hydrogen storage state transition constraints, battery energy state transition constraints, hydrogen storage safety margin constraints, and second-stage correction constraints under various disturbance scenarios. The hydrogen storage safety margin constraint measures the amount of hydrogen storage below the monthly minimum hydrogen storage safety margin through a low hydrogen storage relaxation variable and includes a low hydrogen storage risk penalty term.

[0029] Specifically, the monthly scheduling layer translates the annual hydrogen storage target into a weekly executable plan with risk buffering capabilities. Let Ω be the set of uncertain scenarios. Number the scene. For the scene The probability of [the event / event]. The monthly scheduling layer model is solved using a two-stage stochastic programming model. The variables in the first stage are determined before the photovoltaic and load disturbances occur, and the variables in the second stage are compensated for after the disturbance scenarios occur. The first stage baseline planned cost. And the second-stage correction cost under scenario ω They are respectively: , , in, The first phase of the plan includes a variable vector comprising weekly hydrogen storage reference, equipment baseline output, battery baseline net energy, and grid baseline interaction. For the scene The second-stage correction variable vector includes equipment correction amounts. Lack of slack variables Slack variables of abandoned power and low hydrogen storage relaxation variables ; Indexed by time period; For the hour of week w; For the first Time-of-use grid interaction costs; For the first Equipment operation and maintenance costs during specific time periods; For the first Battery degradation costs over time; This is the penalty coefficient for supply shortage; For the scene Next The time period lacks slack variables; This is the penalty coefficient for power curtailment; For the scene Next Slack variables for time-period power curtailment; For equipment Correction action penalty coefficient; For the scene Next Time-of-use equipment The power correction amount; This refers to the duration of the scheduling period.

[0030] The objective function of the monthly scheduling layer model is: , , , , in, Cost of tracking the target; For the risks and costs of hydrogen storage; It is a weekly index; For the first Weekly hydrogen storage tracking penalty coefficient; For the first Weekly hydrogen storage reference value; The first batch of documents issued at the annual level Weekly hydrogen storage reference target; For the first Weekly battery energy tracking penalty coefficient; For the first Weekly battery baseline net energy; The first batch of documents issued at the annual level Weekly battery energy reference; The penalty coefficient for low hydrogen storage risk; For the scene Next Weekly low hydrogen storage relaxation variables; For the scene Next The first phase of the week's plan variables; For the first The first-stage planning variables for the week; unexpected constraints, i.e., the baseline plan for the same week cannot depend on future disturbance scenarios.

[0031] The main constraints under each disturbance scenario include power balance constraints, hydrogen storage state transition constraints, battery energy state transition constraints, hydrogen storage safety margin constraints, and the relationship between the second-stage correction amount: , , , , , in, For the scene Next Photovoltaic power output data for different time periods; For the scene Fuel cell power in the next time period t; For the scene Next Battery discharge power over a given period; For the scene Next Power purchased during specific time periods; For the scene Next The time period lacks slack variables; For the scene Next Time-period load data; For the scene Next Electrolytic cell power during specific time periods; For the scene Next Battery charging power during a given period; For the scene Next Electricity sales volume during specific time periods; For the scene Next Slack variables for time-period power curtailment; For the scene Next Hydrogen storage capacity over time period; For the scene Next Hydrogen storage capacity over time period; For electrolytic cell efficiency; For fuel cell efficiency; For the scene Next Battery energy status over a period of time; For the scene Next Battery energy status over a period of time; Improve battery charging efficiency; For battery discharge efficiency; For the first Minimum hydrogen storage safety margin for a given period; For the scene Next Low hydrogen storage relaxation variables during the period; For the first The baseline output of device k during the time period; For the scene Next Time-of-use equipment The power correction amount.

[0032] As a result, the monthly scheduling layer outputs weekly hydrogen storage plans, weekly battery state of charge plans, equipment baseline output, grid interaction plans, and hourly reference trajectories.

[0033] In S4, before constructing the intraday execution layer model and performing rolling optimization, a net load fast / slow component decomposition operation is also performed, such as... Figure 3 As shown, the net load fast and slow component decomposition operation includes: Net load is determined based on the difference between short-term load forecast data and short-term photovoltaic forecast data; slow net load variable is extracted based on a moving average window, and fast net load component is obtained based on the difference between net load and slow net load variable; battery reference net power is generated based on fast net load component, enabling battery energy storage units to track fast net load component to compensate for hourly rapid fluctuations, while electrolyzer, fuel cell and external power grid bear the energy balance of slow net load variable; battery reference net power is limited to the range of maximum battery charging power and maximum battery discharging power by saturation function.

[0034] Specifically, the intraday execution layer model, with an hourly resolution, solves the predictive control problem based on short-term photovoltaic forecast data, short-term load forecast data, current hydrogen storage, battery state of charge, and cumulative lifetime loss. To highlight the battery's short-term regulation function, the net load is decomposed into a slow net load variable and a fast net load component: , , , , in, For the first Net load during the period; For the first Time-period load data; For the first Photovoltaic power output data for different time periods; For the first Slow-moving variables in net load during time periods; The moving average window length; For the index of the summation variable; For the first Net load during the period; For the first Rapid component of net load during time period; For the first Battery reference net power for the time period; It is a saturation function; Maximum charging power for the battery; Maximum discharge power of the battery; reference net power of the battery. Through saturation function Limiting the battery's maximum charging power and battery maximum discharge power Within this range, the battery energy storage unit primarily handles hourly rapid fluctuation compensation, while hydrogen energy storage and the external power grid handle energy balance for slow net load variations.

[0035] The intraday execution layer model in S4 is solved using a model predictive control model. It uses several hours as the prediction time domain and one or more scheduling periods as rolling steps. At each rolling moment, only the control quantity for the current scheduling period is executed, and it is reinitialized based on the actual state at the next rolling moment. The objective function of the intraday execution layer model includes nonlinear battery degradation cost, which is determined based on the single-period discharge depth, degradation parameters, the single-period maximum discharge penalty threshold, and battery replacement cost. When the single-period discharge depth exceeds the single-period maximum discharge penalty threshold, the marginal degradation cost increases. The hourly battery degradation increments are accumulated on a monthly scale to obtain the cumulative lifetime loss at the beginning of the next month, and this cumulative lifetime loss is fed back to the annual planning layer as a state variable.

[0036] The model prediction control objective function of the intraday execution layer model is: , , in, For predicting the time period index within the time domain; This is the current scheduling period; For the prediction time domain of model predictive control; For the first Time-of-use grid interaction costs; For the first Equipment operation and maintenance costs during specific time periods; For the first Battery degradation costs over time; This is the penalty coefficient for hydrogen storage tracking; For the first Hydrogen storage capacity over time period; The first one issued by the monthly scheduling layer Reference trajectory for hydrogen storage over time; This is the penalty coefficient for battery power tracking; For the first Battery net power during the period; For the first Battery reference net power for the time period; The control quantity is smoothed by a penalty coefficient; For the first Time-period control variable vector, including electrolyzer power Fuel cell power Battery charging power Battery discharge power Power purchased and electricity sales capacity ; For the first Time period control variable vector; It is the squared L2 norm of the control variable vectors in adjacent time periods, used to suppress frequent fluctuations in control commands.

[0037] To avoid the scheduling model over-utilizing battery storage units for short-term economic gains, a nonlinear battery degradation cost related to the depth of discharge is introduced. (Depth of discharge in a single time period) Hourly battery degradation increment , No. Battery degradation costs over time and cumulative lifespan loss They are respectively: , , , , , in, For the first Discharge depth in a single time period; For the first Battery discharge power during the time period; Δt is the duration of the scheduling period. This represents the battery's maximum energy capacity. For the first Battery degradation increment per hour over time period; , and Battery degradation parameters; This is the penalty threshold for a large discharge in a single time period; Single-period discharge depth Exceeding the single-period large discharge penalty threshold The partial square term, when the single-time discharge depth Exceeding the single-period large discharge penalty threshold At that time, the cost of marginal degradation increases; For battery replacement costs; For the first Battery degradation costs over time; For the first Cumulative degradation on an hourly scale over a given period; For the first Cumulative degradation on an hourly scale over a given period; For the first Cumulative lifespan loss at the beginning of the month; For the first Cumulative lifespan loss at the beginning of the month; For the first Monthly hourly collection; For the first Battery degradation increment per hour over time period This represents a monthly-scale accumulation over the 12 months of the year. A hierarchical feedback chain is formed between battery discharge actions, single-period discharge depth, hourly degradation increment, cumulative lifetime status, and annual decisions: the battery discharge power of the intraday execution layer model. Determine the depth of discharge in a single time period Single-period discharge depth By degenerate parameters , , and penalty threshold Calculation of hourly battery degradation increment Hourly battery degradation increment The cumulative lifetime loss at the beginning of the following month is obtained by summing the data on a monthly scale. and the cumulative lifespan loss As one of the system state variables of the annual planning layer model, it is fed back to the annual planning layer model.

[0038] The control commands issued include the hydrogen production power of the electrolyzer, the power generation power of the fuel cell, the charging power of the battery, the discharging power of the battery, the power purchased from the external power grid, and the power sold from the external power grid; the control commands meet the upper and lower limits of hydrogen storage capacity, the upper and lower limits of battery energy, and the upper and lower limits of equipment power.

[0039] The equipment operation constraint data in this embodiment includes upper and lower limits for electrolyzer power, fuel cell power, battery charge and discharge power, hydrogen storage capacity, and battery energy; the external power grid interaction constraint data includes upper and lower limits for purchased power and sold power.

[0040] The annual planning layer, monthly scheduling layer, and intraday execution layer all satisfy unified energy balance constraints, state transition constraints, and equipment boundary constraints. The system's hourly power balance constraints, hydrogen storage state transition constraints, battery energy state transition constraints, and major equipment boundary constraints are as follows: , , , , , , , , , in, Indexed by time period; For the first Photovoltaic power output data for different time periods; For the first Fuel cell power during specific time periods; For the first Battery discharge power over a given period; For the first Power purchased during specific time periods; For the first The time period lacks slack variables; For the first Time-period load data; For the first Electrolytic cell power during specific time periods; For the first Battery charging power during a given period; For the first Electricity sales volume during specific time periods; For the first Slack variables for time-period power curtailment; For the first Hydrogen storage capacity over time period; For the first Hydrogen storage capacity over time period; For electrolytic cell efficiency; The duration of the scheduling period; For fuel cell efficiency; For the first Battery energy status over a period of time; For the first Battery energy status over a period of time; Improve battery charging efficiency; For battery discharge efficiency; This is the maximum power of the electrolytic cell; For the first Electrolytic cell power during specific time periods; This represents the maximum power of the fuel cell; For the first Fuel cell power during specific time periods; Maximum charging power for the battery; For the first Battery charging power during a given period; This represents the battery's maximum discharge power. For the first Battery discharge power over a given period; This represents the lower limit of hydrogen storage capacity. For the first Hydrogen storage capacity over time period; This represents the upper limit of hydrogen storage capacity. This represents the lower limit of battery energy. For the first Battery energy status over a period of time; This represents the upper limit of battery energy.

[0041] To verify the method provided in this embodiment, a verification experiment was conducted based on 8760 hours of photovoltaic power output and load data from a low-carbon community throughout the year. Compared with hydrogen-free energy storage scheduling methods, the hierarchical optimization scheduling method of the present invention using hybrid electric-hydrogen energy storage can significantly reduce overall operating costs and supply shortages; compared with a two-layer hierarchical structure lacking a monthly scheduling layer model, it can reduce hydrogen storage target tracking errors; and compared with scheduling methods that do not consider battery degradation costs, it can reduce the large discharge ratio of batteries in a single period, annual throughput, equivalent cycle count, and cumulative lifespan loss. Therefore, the present invention can coordinate long-term energy transfer of hydrogen storage, short-term adjustment of battery energy storage units, and buffering of source-load uncertainty risks.

[0042] like Figure 4 As shown in the monthly net hydrogen storage distribution map, the distribution characteristics of net hydrogen storage are characterized by surplus months for hydrogen production and shortage months for hydrogen consumption. Hydrogen storage inventory exhibits cross-seasonal replenishment and release characteristics, indicating that the monthly target generation of the annual planning layer model and the rolling execution of the monthly scheduling layer model can establish cross-seasonal energy transfer channels. Figure 5 As shown in the comparison chart of annual, monthly, and intraday hydrogen storage trajectory tracking, the results indicate that a relatively continuous hydrogen storage target transfer relationship can be formed among the annual, monthly, and intraday layers. The monthly risk buffer layer can transform the annual seasonal hydrogen storage target into a smoother, more executable intraday operating trajectory. Figure 6 As shown in the comparison chart of hydrogen storage trajectory tracking under the baseline and perturbation scenarios, the results indicate that monthly two-stage stochastic programming can improve hydrogen storage margin under perturbation and reduce the risks of supply shortages and low hydrogen storage. Figure 7As shown in the comparison chart of hydrogen storage tracking errors between the M1 and M3 models, the monthly risk buffer layer can effectively reduce the tracking error of the hydrogen storage target. Figure 8 As shown in the figure, the probability distribution of the battery's single-period discharge depth indicates that the introduction of nonlinear degradation costs suppresses the large discharge ratio in a single period. Figure 9 As shown in the comparison chart of battery normalized operating indicators, the results indicate that after introducing nonlinear degradation costs, both the annual battery throughput and equivalent cycle count are suppressed, allowing the battery to more effectively handle hourly rapid fluctuation compensation. Figure 10 As shown in the diagram, the comparison between the planned minimum hydrogen storage and the actual minimum hydrogen storage indicates that monthly two-stage stochastic programming can improve hydrogen storage margin under perturbations. Figure 11 As shown in the diagram, the comparison between supply shortage and low hydrogen storage hours indicates that monthly two-stage stochastic programming can reduce the risks of supply shortage and low hydrogen storage. Figure 12 As shown in the figure, the comparison between hydrogen storage margin and tracking indicators indicates that monthly two-stage stochastic programming can enhance operational robustness under external source load disturbances. Figure 13 As shown in the comparison chart of economic costs and risk indicators, the hierarchical optimization scheduling method for hybrid electric-hydrogen energy storage of this invention can significantly reduce overall operating costs. Figure 14 As shown in the comparison chart of supply shortage statistics under disturbance scenarios, the results indicate that monthly two-stage stochastic programming can effectively reduce the supply shortage under external source load disturbances.

[0043] Therefore, the cross-seasonal hierarchical optimization scheduling method for hybrid electric-hydrogen energy storage of the present invention constructs a three-layer hierarchical feedback scheduling framework consisting of an annual planning layer, a monthly scheduling layer, and an intraday execution layer. The annual planning layer generates monthly hydrogen storage targets in monthly phases. The monthly scheduling layer generates multiple disturbance scenarios based on the uncertainty distribution of source and load and uses two-stage stochastic programming to solve the problem, transforming the monthly hydrogen storage targets into weekly hydrogen storage reference trajectories with safety margins and outputting hourly reference trajectories. The intraday execution layer performs rolling optimization based on short-term photovoltaic forecast data, short-term load forecast data, current hydrogen storage, current battery state of charge, and weekly hydrogen storage reference trajectories to obtain power control instructions for each device during the current scheduling period. After execution, feedback is provided up the hierarchy based on actual hydrogen storage, battery state of charge, and cumulative lifetime loss, enabling the cross-seasonal hydrogen storage targets to be transmitted from the annual layer through the monthly layer to the intraday layer and decomposed into executable scheduling instructions. At the same time, actual execution deviations affect the correction of the upper-level plan through state feedback. Thus, in engineering applications, this effectively solves the problem of connecting long-term cross-seasonal hydrogen storage targets with short-term scheduling execution and improves the executability of hydrogen storage targets in hourly scheduling. The monthly scheduling layer, through the division of first-stage and second-stage variables in a two-stage stochastic programming model and the introduction of unexpected constraints, ensures that the hydrogen storage level is not lower than the monthly minimum hydrogen storage safety margin under scenarios of photovoltaic output and load demand disturbances. This reserves risk buffer space for weekly planning, enabling the system to cope with adverse conditions such as continuous low irradiance or high load in actual operation, reducing the risks of low hydrogen storage and insufficient power supply, and enhancing the operational robustness of the low-carbon community electric-hydrogen hybrid energy storage system under external source and load disturbances. The intraday execution layer decomposes the net load into slow variables and fast components, enabling battery energy storage units to track the fast component of the net load to compensate for hourly rapid fluctuations. Electrolyzers, fuel cells, and the external power grid handle the energy balance of the slow variable of the net load. This decouples the functions of short-term battery regulation and long-term energy transfer of hydrogen energy storage, avoiding overlap in the functional positioning of the two types of energy storage. Simultaneously, by introducing a nonlinear battery degradation cost determined based on the single-period discharge depth and the single-period maximum discharge penalty threshold, the marginal degradation cost increases when the single-period discharge depth exceeds the penalty threshold. The hourly battery degradation increment is accumulated on a monthly scale and used as one of the state variables for feedback in the annual planning layer. This creates a complete hierarchical feedback chain between battery discharge actions, single-period discharge depth, hourly degradation increment, cumulative lifespan status, and annual decisions. In scheduling decisions, this effectively suppresses deep discharge and excessive cycling of batteries on a single-period scale, extending the actual operating life of the batteries. This reduces the operation and maintenance costs and equipment replacement frequency of the energy storage system throughout its entire lifespan, demonstrating good economic benefits and engineering practical value.

[0044] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working process of the described module can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0045] The electronic device of this invention includes a central processing unit (CPU), which can perform various appropriate actions and processes according to computer program instructions stored in read-only memory (ROM) or loaded from a storage unit into random access memory (RAM). The RAM may also store various programs and data required for device operation. The CPU, ROM, and RAM are interconnected via a bus. Input / output (I / O) interfaces are also connected to the bus.

[0046] Multiple components in the device are connected to an I / O interface, including: input units such as a keyboard, mouse, etc.; output units such as various types of displays, speakers, etc.; storage units such as disks, optical disks, etc.; and communication units such as network interface cards, modems, wireless transceivers, etc. The communication unit allows the device to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks. The processing unit performs the various methods and processes described above, such as the method of the present invention. For example, in some embodiments, the method of the present invention may be implemented as a computer software program tangibly contained in a machine-readable medium, such as a storage unit. In some embodiments, part or all of the computer program may be loaded and / or installed on the device via ROM and / or the communication unit. When the computer program is loaded into RAM and executed by the CPU, one or more steps of the method of the present invention described above may be performed. Alternatively, in other embodiments, the CPU may be configured to execute the method of the present invention by any other suitable means (e.g., by means of firmware).

[0047] The functions described above in this document can be performed, at least in part, by one or more hardware logic components. For example, exemplary types of hardware logic components that can be used, without limitation, include: Field Programmable Gate Arrays (FPGAs), Application-Specific Integrated Circuits (ASICs), Application Standard Products (ASSPs), System-on-Chip (SoCs), Complex Programmable Logic Devices (CPLDs), and so on.

[0048] The program code used to implement the methods of the present invention can be written in any combination of one or more programming languages. This program code can be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing device, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code can be executed entirely on the machine, partially on the machine, as a standalone software package partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0049] In the context of this invention, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. Machine-readable media can include, but are not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0050] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A cross-seasonal hierarchical optimization scheduling method for hybrid electric-hydrogen energy storage, characterized in that, The specific steps include: S1. Collect photovoltaic output data, load data, electricity price data, battery energy storage status data, hydrogen storage status data, equipment operation constraint data, and external power grid interaction constraint data within the scheduling cycle; S2. Based on the collected data, taking months as the stage, and using the beginning hydrogen storage amount in the hydrogen storage status data, the battery state of charge in the battery energy storage status data, and the cumulative life loss in the battery energy storage status data as state variables, construct an annual planning layer model and solve it to obtain the monthly hydrogen storage target, monthly battery energy reference, and life loss control requirements. S3. Based on the monthly hydrogen storage target, generate multiple disturbance scenarios based on the uncertainty distribution of source and load, and construct and solve a monthly scheduling layer model based on the disturbance scenarios. Transform the monthly hydrogen storage target into a weekly hydrogen storage reference trajectory, equipment baseline output plan and grid interaction plan, and output an hourly reference trajectory. S4. Generate short-term photovoltaic forecast data based on the photovoltaic output data, generate short-term load forecast data based on the load data, and construct an intraday execution layer model based on the short-term photovoltaic forecast data, short-term load forecast data, current hydrogen storage capacity, current battery state of charge, and weekly hydrogen storage reference trajectory, and perform rolling optimization solution to obtain control commands for the electrolyzer power, fuel cell power, battery charging and discharging power, and grid interaction power during the current scheduling period. S5. Execute the control command for the current scheduling period, and update the subsequent scheduling status based on the hydrogen storage capacity, battery state of charge, and cumulative lifetime loss after execution.

2. The cross-seasonal hierarchical optimization scheduling method for hybrid electric-hydrogen energy storage according to claim 1, characterized in that, The annual planning layer model in S2 is solved using a finite-stage Markov decision process or an equivalent dynamic programming model. The decision actions of the annual planning layer model include the monthly net change in hydrogen storage, the reference power of the electrolyzer, the reference power of the fuel cell, the reference adjustment of the battery, and the grid interaction. The immediate cost of the annual planning layer model is determined based on the grid interaction cost, equipment operation and maintenance cost, battery degradation cost, hydrogen storage target deviation penalty, and low hydrogen storage risk penalty. The low hydrogen storage risk penalty is calculated through a low hydrogen storage relaxation variable, which is activated when the hydrogen storage is lower than the monthly minimum hydrogen storage safety margin. The annual planning layer outputs the monthly hydrogen storage target, the monthly battery energy reference, and the life loss control requirements.

3. The cross-seasonal hierarchical optimization scheduling method for hybrid electric-hydrogen energy storage according to claim 1, characterized in that, The monthly scheduling layer model in S3 is solved using a two-stage stochastic programming model. Uncertainty scenarios are divided into first-stage variables and second-stage variables. The first-stage variables are determined before photovoltaic and load disturbances occur, including weekly hydrogen storage reference, equipment baseline output, battery baseline net energy, and grid baseline interaction. The second-stage variables are determined after the disturbance scenarios occur, including equipment correction, supply shortage relaxation variables, power curtailment relaxation variables, and low hydrogen storage relaxation variables. The objective function of the monthly scheduling layer model is determined based on the first-stage baseline plan cost, the second-stage scenario expected correction cost, hydrogen storage target tracking cost, battery energy tracking cost, and low hydrogen storage risk cost, and satisfies the unexpectedness constraint so that the weekly baseline plan does not depend on future disturbance scenarios. The monthly scheduling layer model outputs weekly hydrogen storage reference trajectory, equipment baseline output plan and grid interaction plan, and outputs hourly reference trajectory; The monthly scheduling layer model satisfies power balance constraints, equipment output boundary constraints, hydrogen storage state transition constraints, battery energy state transition constraints, hydrogen storage safety margin constraints, and second-stage correction constraints under various disturbance scenarios. The hydrogen storage safety margin constraint measures the amount of hydrogen storage below the monthly minimum hydrogen storage safety margin through a low hydrogen storage relaxation variable and includes a low hydrogen storage risk penalty term.

4. The cross-seasonal hierarchical optimization scheduling method for hybrid electric-hydrogen energy storage according to claim 1, characterized in that, In step S4, before constructing the intraday execution layer model and performing rolling optimization, a net load fast / slow component decomposition operation is performed, which includes: Net load is determined based on the difference between the short-term load forecast data and the short-term photovoltaic forecast data; a slow net load variable is extracted using a moving average window, and a fast net load component is obtained based on the difference between the net load and the slow net load variable; a battery reference net power is generated based on the fast net load component, enabling the battery energy storage unit to track the fast net load component to compensate for hourly rapid fluctuations, while the electrolyzer, fuel cell, and external power grid bear the energy balance of the slow net load variable; the battery reference net power is limited to the range of the battery's maximum charging power and maximum discharging power by a saturation function.

5. The cross-seasonal hierarchical optimization scheduling method for hybrid electric-hydrogen energy storage according to claim 1, characterized in that, The intraday execution layer model in S4 is solved using a model predictive control model. It uses several hours as the prediction time domain and one or more scheduling periods as the rolling step size. At each rolling moment, it only executes the control quantity of the current scheduling period and re-initializes it according to the actual state at the next rolling moment.

6. The cross-seasonal hierarchical optimization scheduling method for hybrid electric-hydrogen energy storage according to claim 5, characterized in that, The objective function of the intraday execution layer model includes nonlinear battery degradation cost, which is determined based on single-period discharge depth, degradation parameters, single-period maximum discharge penalty threshold, and battery replacement cost. When the single-period discharge depth exceeds the single-period maximum discharge penalty threshold, the marginal degradation cost increases. The hourly battery degradation increments are accumulated on a monthly scale to obtain the cumulative lifetime loss at the beginning of the next month, and the cumulative lifetime loss is fed back to the annual planning layer as a state variable.

7. The cross-seasonal hierarchical optimization scheduling method for hybrid electric-hydrogen energy storage according to claim 1, characterized in that, The control commands include the hydrogen production power of the electrolyzer, the power generation power of the fuel cell, the charging power of the battery, the discharging power of the battery, the power purchased from the external power grid, and the power sold from the external power grid; the control commands satisfy the upper and lower limits of hydrogen storage capacity, the upper and lower limits of battery energy, and the upper and lower limits of equipment power.

8. The cross-seasonal hierarchical optimization scheduling method for hybrid electric-hydrogen energy storage according to claim 1, characterized in that, The equipment operation constraint data includes upper and lower limits for electrolyzer power, fuel cell power, battery charge and discharge power, hydrogen storage capacity, and battery energy; the external power grid interaction constraint data includes upper and lower limits for purchased power and sold power.

9. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the program, it implements the method as described in any one of claims 1 to 8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the method as described in any one of claims 1 to 8.

Images