Wind-solar-hydrogen storage hybrid system double-layer planning method, device and medium considering fuel cell start-stop loss

Abstract

This application discloses a two-layer planning method, equipment, and medium for a wind-solar-hydrogen-storage hybrid system considering fuel cell start-up and shutdown losses, relating to the field of new energy microgrid system modeling technology. The aim of this application is to address the problem of existing energy systems relying on single energy storage devices and having poor coupling and coordination capabilities for different types of units. This application constructs a linear programming model with the objective of minimizing the sum of the total equipment investment cost and operating cost of the wind-solar-hydrogen-storage hybrid system. Based on a fuel cell start-up and shutdown control strategy, the fuel cell is operated, and then the linear programming model is solved to obtain the optimal capacity configuration scheme for the output photovoltaic, wind power, battery, electrolyzer, hydrogen storage tank, and fuel cell. By introducing output power deviation constraints at the operational level, the start-up and shutdown behavior of the fuel cell becomes a response result that meets system-level stability requirements, thereby achieving synergistic optimization of the system's life-cycle economic efficiency and operational reliability.

Description

Technical Field

[0001] This application belongs to the field of new energy microgrid system modeling technology. Background Technology

[0002] With the gradual depletion of fossil fuels and the continued growth in the scale and volatility of electricity load, traditional energy systems face severe challenges in terms of economy, stability, and low carbon emissions. While the increasing proportion of renewable energy generation in the power system has effectively improved the utilization of clean energy, its output volatility and uncertainty also place higher demands on system dispatching and planning.

[0003] Existing energy systems largely rely on fossil fuel backup units or single energy storage technologies for regulation, generally suffering from high carbon emissions, low system coordination efficiency, insufficient operational economics, and poor power supply stability. These systems struggle to meet the comprehensive demands for cleanliness, reliability, and economy in zero-carbon or near-zero-carbon application scenarios. Taking the widely used wind-solar-storage system as an example, this type of system primarily relies on battery energy storage to achieve energy balance on short timescales. When encountering extreme conditions such as continuous rain or no wind, the battery energy storage capacity is rapidly depleted, often requiring the activation of carbon-intensive backup power sources such as diesel generators. This leads to a significant increase in operating costs and carbon emissions, contradicting low-carbon development goals. Furthermore, battery energy storage itself suffers from limited cycle life and capacity decay, further increasing the system's total lifecycle cost and implicit carbon emissions.

[0004] To enhance the system's adaptability across multiple time scales, hybrid energy systems combining wind, solar, hydrogen, and energy storage are gaining increasing attention. Hydrogen energy systems, as long-term, multi-timescale energy carriers, can theoretically compensate for the limitations of battery storage in terms of capacity and lifespan. However, existing research, due to the lack of a systematic collaborative optimization mechanism, often suffers from capacity mismatches and uncoordinated operating strategies among electrolyzers, hydrogen storage tanks, and fuel cells. This results in low utilization rates of hydrogen energy systems, difficulty in guaranteeing return on investment, and failure to fully realize their potential advantages as a zero-carbon energy carrier.

[0005] For the planning and operation of hybrid wind-solar-hydrogen-storage systems, existing research has attempted to introduce two-layer optimization models or uncertainty scenario generation methods, but these methods generally suffer from the following shortcomings: Firstly, existing two-layer models often use static parameters such as equipment capacity as boundary connections between the upper and lower layers, resulting in a loose coupling relationship. Key dynamic economic information exposed during the lower layer's operation is difficult to effectively feed back to the upper layer's planning decisions. Secondly, fuel cells are typically simplified as ideal energy conversion devices with fixed operating costs. Even when considering performance degradation, the focus is often on post-hoc cost adjustments, failing to model fuel cell start-up and shutdown behavior in a unified manner with the overall system operating state. This makes it difficult to reflect the actual impact of start-up and shutdown decisions on system stability and lifecycle economics. Existing research largely focuses on static economic considerations at the single-device level, lacking a comprehensive consideration of the dynamic response of start-up and shutdown behavior and system-level stability constraints. Therefore, start-up and shutdown losses are not adequately reflected in the lifecycle economic optimization, and deep coupling between key components of the system is not achieved.

[0006] Furthermore, existing studies on the handling of uncertainties in wind and solar power output and loads mostly focus on single factors, lacking a unified characterization of source-load composite uncertainties, which limits the applicability and robustness of planning results in complex operating scenarios. Summary of the Invention

[0007] This application aims to address the problem that existing energy systems rely on single energy storage devices and have poor coupling and coordination capabilities for different types of generating units. It provides a two-layer optimization scheduling model driven by system-level operational stability constraints, which characterizes the synergistic coupling relationship between photovoltaic power generation, wind power generation, lithium battery energy storage, and hydrogen energy systems (including electrolyzers, hydrogen storage tanks, and fuel cells) under a unified framework, thereby achieving close linkage between planning decisions and operational behavior.

[0008] The first aspect of this application provides a two-layer planning method for a wind-solar-hydrogen-storage hybrid system that considers fuel cell start-stop losses, including:

[0009] A linear programming model is constructed with the goal of minimizing the sum of the total equipment investment cost and operating cost of the wind-solar-hydrogen-storage hybrid system. The linear programming model includes an upper-level programming model with the goal of minimizing the total equipment investment cost of the wind-solar-hydrogen-storage hybrid system, and a lower-level operating model with the goal of minimizing the operating cost of the wind-solar-hydrogen-storage hybrid system.

[0010] The fuel cell is operated based on the fuel cell start-stop control strategy, and then the linear programming model is solved to obtain the optimal capacity configuration scheme of output photovoltaic, wind power, storage battery, electrolyzer, hydrogen storage tank and fuel cell.

[0011] The fuel cell start-stop control strategy includes:

[0012] When the sum of the power output of the photovoltaic and wind power generation equipment and the output power of the battery is less than the power demand, the fuel cell is started. When the sum of the power output of the photovoltaic and wind power generation equipment and the output power of the battery is greater than or equal to the power demand, it is determined whether the residual value of the fuel cell exceeds its set threshold. If yes, the fuel cell is started; otherwise, the fuel cell is replaced.

[0013] In one possible design, the expression for the upper-level planning model is:

[0014] ,

[0015] in, The total investment cost of equipment for a wind-solar-hydrogen-storage hybrid system. and The investment costs for photovoltaic power generation and wind power generation are respectively. The total deployment cost of the battery. and The total equipment costs for electrolyzers and fuel cells are respectively. The total equipment cost for the hydrogen storage tank.

[0016] In one possible design, the investment cost of the photovoltaic power generation Investment costs of wind power generation Total deployment cost of storage batteries Total equipment cost of electrolytic cells Total equipment cost of fuel cells and the total equipment cost of hydrogen storage tanks. The expressions are as follows:

[0017] ,

[0018] ,

[0019] ,

[0020] ,

[0021] ,

[0022] ,

[0023] in, and The power investment costs for photovoltaic and wind power generation are respectively. and These are the power outputs of photovoltaic and wind power generation equipment, respectively. and These are the fixed investment costs for wind power equipment and photovoltaic equipment, respectively. and These are the capacity cost and power cost of the battery, respectively. and These are the battery capacity and output power, respectively. and The power costs of electrolyzers and fuel cells are respectively. and These are the power outputs of the electrolyzer and the fuel cell, respectively. The cost per unit capacity of the hydrogen storage tank, This refers to the capacity of the hydrogen storage tank.

[0024] In one possible design, the constraints of the upper-level planning model include:

[0025] ,

[0026] ,

[0027] ,

[0028] ,

[0029] in, and They are respectively Battery charging power and battery discharging power during the same period For the capacity of the storage battery, for The state of charge of the battery during a given period. For the capacity of the hydrogen storage tank, for The amount of hydrogen stored during a given period;

[0030] The state-of-charge update equation for the battery is:

[0031] ,

[0032] in, and These are the charging efficiency and discharging efficiency of the battery, respectively.

[0033] The state update equation for a hydrogen energy storage system is:

[0034] ,

[0035] in, for The time period is converted to the hydrogen mass in the hydrogen storage tank. for The mass of hydrogen consumed by the fuel cell during a given period of time.

[0036] In one possible design, the expression for the lower-level operational model is:

[0037] ,

[0038] in, For the operating cost of the wind-solar-hydrogen-storage hybrid system, and Solar and wind power respectively Output power during the time period and The operating costs of solar and wind power generation To sell hydrogen.

[0039] In one possible design, the constraints of the lower-level operating model include:

[0040] ,

[0041] in, for Electricity demand during a given time period for Power of the time-lapse electrolyzer, and for Solar irradiance and wind speed during the time period for The output power of the battery during the period for The power of the fuel cell during a given time period.

[0042] In one possible design, the residual value of the fuel cell is expressed as:

[0043] ,

[0044] in, For the residual value of fuel cells, The total equipment cost of fuel cells, For the number of times the fuel cell is started, The value loss caused by each start-up and shutdown. , This refers to the rated number of start-ups for the fuel cell device.

[0045] The residual value constraint of the fuel cell is expressed as follows: .

[0046] In one possible design, the two-layer planning method for the wind-solar-hydrogen-storage hybrid system is carried out under a scenario that includes load demand, renewable energy characteristics, and techno-economic parameters;

[0047] Methods for generating scenes include:

[0048] Based on historical load data, the K-Shapes clustering algorithm was used to extract daily load curves with typical morphological characteristics.

[0049] Based on the daily load curve, a multivariate statistical method is used to expand the scenarios and generate an initial scenario set;

[0050] An improved K-Means clustering algorithm is used to reduce the initial scene set to obtain representative load scenes and their probability weights;

[0051] Gaussian mixture models are used to perform probability density modeling on historical wind speed and irradiance data to generate typical wind and solar power output scenarios.

[0052] The second aspect of this application provides a two-layer planning device for a wind-solar-hydrogen-storage hybrid system that takes into account fuel cell start-stop losses. The two-layer planning device for a wind-solar-hydrogen-storage hybrid system that takes into account fuel cell start-stop losses includes a processor and a memory. The memory stores at least one instruction, which is loaded and executed by the processor to implement the two-layer planning method for a wind-solar-hydrogen-storage hybrid system that takes into account fuel cell start-stop losses as described above.

[0053] A third aspect of this application provides a computer storage medium storing at least one instruction, which is loaded and executed by a processor to implement the above-described two-layer planning method for a wind-solar-hydrogen-storage hybrid system that takes into account fuel cell start-up and shutdown losses.

[0054] The beneficial effects of this application are:

[0055] This application proposes a tightly coupled two-layer optimization modeling method with system power stability constraints as the core link. By introducing system output power deviation constraints at the operation layer, the start-up and shutdown behavior of the fuel cell becomes an endogenous response result that satisfies system-level stability requirements. The resulting key operational economic information is then fed back to the planning layer, thereby achieving synergistic optimization of system lifecycle economics and operational reliability. Compared with traditional loosely coupled models or models that ignore the dynamic characteristics of key equipment, the advantages of this application are as follows:

[0056] 1. This application proposes a two-layer optimization framework driven by system power stability constraints. Unlike the static capacity configuration of existing technologies, this application introduces a dynamic economic feedback mechanism at the operation layer, making the start-up and shutdown behavior of fuel cells not only determined by operating costs, but also an endogenous decision that coordinates and optimizes economic efficiency and system stability throughout the entire life cycle. Specifically, the start-up and shutdown losses of fuel cells are quantified as residual value over time periods and are endogenously incorporated into the upper-level planning decision as a key cost parameter, thereby achieving a leap from "meeting operational constraints" to "pursuing the optimal throughout the entire life cycle".

[0057] 2. By combining system-level power stability constraints with endogenous economic modeling of fuel cell start-up and shutdown losses, this application not only achieves intelligent optimization decision-making for fuel cell start-up and shutdown behavior, but also feeds back to upper-level planning through this dynamic mechanism, ensuring optimal synergy between economic efficiency and operational reliability throughout the entire lifecycle. This deeply coupled two-layer optimization framework is an innovative contribution not yet covered by existing technologies.

[0058] 3. This application employs a unified method for characterizing the uncertainties of load morphology and wind-solar fluctuations across multiple time scales. Through a three-stage framework of "K-Shapes clustering - multivariate statistical expansion - improved K-Means reduction," a highly representative comprehensive scenario is generated, ensuring the planning scheme possesses excellent robustness in the face of real-world complex fluctuations. Unlike traditional simulation methods based on a single typical scenario, this application can more accurately characterize the multidimensional fluctuation characteristics of source-load composite uncertainties, thereby significantly improving the system's stability and economy. Attached Figure Description

[0059] Figure 1 This is a schematic diagram of the integrated wind, solar, and hydrogen storage system described in the embodiment.

[0060] Figure 2 A three-segment framework diagram is generated for the power load scenario in the example;

[0061] Figure 3 A comparison chart of wind power generation equipment loads with and without a hydrogen system;

[0062] Figure 4 A comparison chart of photovoltaic power generation equipment loads with and without hydrogen systems;

[0063] Figure 5 A line graph showing the operating costs of a hydrogen-free system;

[0064] Figure 6 This is a line graph showing the operating costs of a hydrogen-equipped system.

[0065] Figure 7 Value diagram for a system without fuel cell start-stop control;

[0066] Figure 8 A value diagram showing the cost of fuel cell start-stop control in the system;

[0067] Figure 9 A flowchart of a two-layer planning method for a wind-solar-hydrogen-storage hybrid system that takes into account fuel cell start-up and shutdown losses. Detailed Implementation

[0068] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application. It should be noted that, unless otherwise specified, the embodiments and features in the embodiments of this application can be combined with each other.

[0069] Specific Implementation Method 1: The two-layer planning method for a wind-solar-hydrogen storage hybrid system considering fuel cell start-up and shutdown losses described in this implementation method includes:

[0070] A linear programming model is constructed with the goal of minimizing the sum of the total equipment investment cost and operating cost of the wind-solar-hydrogen-storage hybrid system. The linear programming model includes an upper-level programming model with the goal of minimizing the total equipment investment cost of the wind-solar-hydrogen-storage hybrid system, and a lower-level operating model with the goal of minimizing the operating cost of the wind-solar-hydrogen-storage hybrid system.

[0071] The fuel cell is operated based on the fuel cell start-stop control strategy, and then the linear programming model is solved to obtain the optimal capacity configuration scheme of output photovoltaic, wind power, storage battery, electrolyzer, hydrogen storage tank and fuel cell.

[0072] The fuel cell start-stop control strategy includes:

[0073] When the sum of the power output of the photovoltaic and wind power generation equipment and the output power of the battery is less than the power demand, the fuel cell is started. When the sum of the power output of the photovoltaic and wind power generation equipment and the output power of the battery is greater than or equal to the power demand, it is determined whether the residual value of the fuel cell exceeds its set threshold. If yes, the fuel cell is started; otherwise, the fuel cell is replaced.

[0074] In one implementation, the expression for the upper-level planning model is:

[0075] ,

[0076] in, The total investment cost of equipment for a wind-solar-hydrogen-storage hybrid system. and The investment costs for photovoltaic power generation and wind power generation are respectively. The total deployment cost of the battery. and The total equipment costs for electrolyzers and fuel cells are respectively. The total equipment cost for the hydrogen storage tank.

[0077] In one embodiment, the investment cost of the photovoltaic power generation Investment costs of wind power generation Total deployment cost of storage batteries Total equipment cost of electrolytic cells Total equipment cost of fuel cells and the total equipment cost of hydrogen storage tanks. The expressions are as follows:

[0078] ,

[0079] ,

[0080] ,

[0081] ,

[0082] ,

[0083] ,

[0084] in, and The power investment costs for photovoltaic and wind power generation are respectively. and These are the power outputs of photovoltaic and wind power generation equipment, respectively. and These are the fixed investment costs for wind power equipment and photovoltaic equipment, respectively. and These are the capacity cost and power cost of the battery, respectively. and These are the battery capacity and output power, respectively. and The power costs of electrolyzers and fuel cells are respectively. and These are the power outputs of the electrolyzer and the fuel cell, respectively. The cost per unit capacity of the hydrogen storage tank, This refers to the capacity of the hydrogen storage tank.

[0085] In one implementation, the constraints of the upper-level planning model include:

[0086] ,

[0087] ,

[0088] ,

[0089] ,

[0090] in, and They are respectively Battery charging power and battery discharging power during the same period For the capacity of the storage battery, for The state of charge of the battery during a given period. For the capacity of the hydrogen storage tank, for The amount of hydrogen stored during a given period;

[0091] The state-of-charge update equation for the battery is:

[0092] ,

[0093] in, and These are the charging efficiency and discharging efficiency of the battery, respectively.

[0094] The state update equation for a hydrogen energy storage system is:

[0095] ,

[0096] in, for The time period is converted to the hydrogen mass in the hydrogen storage tank. for The mass of hydrogen consumed by the fuel cell during a given period of time.

[0097] In one implementation, the expression for the lower-level operating model is:

[0098] ,

[0099] in, For the operating cost of the wind-solar-hydrogen-storage hybrid system, and Solar and wind power respectively Output power during the time period and The operating costs of solar and wind power generation To sell hydrogen.

[0100] In one implementation, the constraints of the lower-level operating model include:

[0101] ,

[0102] in, for Electricity demand during a given time period for Power of the time-lapse electrolyzer, and for Solar irradiance and wind speed during the time period for The output power of the battery during the period for The power of the fuel cell during a given time period.

[0103] In one embodiment, the residual value of the fuel cell is expressed as:

[0104] ,

[0105] in, For the residual value of fuel cells, The total equipment cost of fuel cells, For the number of times the fuel cell is started, The value loss caused by each start-up and shutdown. , This refers to the rated number of start-ups for the fuel cell device.

[0106] The residual value constraint of the fuel cell is expressed as follows: .

[0107] In one implementation, the two-layer planning method for the wind-solar-hydrogen-storage hybrid system is carried out under a scenario that includes load demand, renewable energy characteristics, and techno-economic parameters;

[0108] Methods for generating scenes include:

[0109] Based on historical load data, the K-Shapes clustering algorithm was used to extract daily load curves with typical morphological characteristics.

[0110] Based on the daily load curve, a multivariate statistical method is used to expand the scenarios and generate an initial scenario set;

[0111] An improved K-Means clustering algorithm is used to reduce the initial scene set to obtain representative load scenes and their probability weights;

[0112] Gaussian mixture models are used to perform probability density modeling on historical wind speed and irradiance data to generate typical wind and solar power output scenarios.

[0113] To further illustrate the implementation scheme of this application, this embodiment aims to construct a zero-carbon system powered by 100% renewable energy. It proposes a two-layer planning method for a wind-solar-hydrogen-storage hybrid system that considers fuel cell start-up and shutdown losses, as detailed below:

[0114] I. Electrical Load Scenario Handling

[0115] In the planning and operation of wind-solar-hydrogen-storage integrated systems, the randomness of wind and solar power output and the volatility of electricity load are the core uncertainties affecting the system's economics and reliability. The ability to cope with the uncertainties of wind power, solar radiation, and electricity load is crucial. Traditional deterministic planning methods or simulations based on a single typical scenario are insufficient to effectively characterize the inherent volatility characteristics and multiple possible operating states of wind and solar resources, leading to conservative or insufficiently robust planning results. To ensure the system's true reliability, such as... Figure 2 As shown, this embodiment analyzes the time-series data of daily irradiance and wind speed to generate several typical irradiance and wind speed scenarios for power generation simulation. Simultaneously, 300 days of electricity load data, with 24 hours per day, were collected. Cluster analysis was used to classify the 300-day load curves, followed by the generation of multiple scenarios. Representative load scenarios were obtained through scenario reduction and combined with wind and solar scenarios for system simulation experiments.

[0116] (1) Generation of electricity load scenarios

[0117] To efficiently and accurately generate electricity load scenarios suitable for stochastic planning in power systems, this embodiment proposes a three-stage scenario generation and reduction framework. First, based on historical load data, the K-Shapes algorithm is used to extract daily load curves with typical morphological characteristics to accurately depict different electricity consumption patterns (such as weekdays, weekends, and typical seasonal days). Next, multivariate statistical parameters are introduced to randomly expand the typical daily load, generating an initial large-scale scenario set that fully reflects future uncertainties. Finally, an improved K-Means feature clustering method is applied to intelligently reduce this scenario set, obtaining a small-scale, highly representative set of final typical scenarios and their probability weights while retaining their core statistical features. Through the extraction of typical days based on the K-Shapes clustering algorithm, the historical load dataset contains... The daily load data corresponds to electricity load values ​​for 10 months (approximately 30 days per month). The dataset is represented as follows:

[0118] ,

[0119] in, For the first The load vector of the day, Indicates the first Heavenly Load values ​​for the time period ( This embodiment will use the dataset. Clustering Each cluster corresponds to a typical daily load curve, meaning that 10 representative daily load curves are extracted from 300 days, which is equivalent to extracting a typical day for each month.

[0120] The core of K-Shapes clustering is based on shape similarity measurement, using cross-correlation coefficients as the similarity index. For any two load vectors... their cross-relationship numbers Defined as:

[0121] ,

[0122] in, and They are vectors and The mean.

[0123] Based on cross-relation number Define the shape distance function :

[0124] .

[0125] The distance function satisfies The smaller the value, the higher the shape similarity.

[0126] The goal of clustering is to divide the dataset Divided into Cluster And find the centroid of each cluster. (i.e., typical daily load curve), to minimize the sum of shape distances within the cluster:

[0127] .

[0128] This is equivalent to maximizing the sum of intra-cluster cross-correlation coefficients:

[0129] .

[0130] Center of mass The solution is obtained through iterative updates to maximize the sum of cross-correlation coefficients within the cluster. For a cluster... Its centroid update formula is:

[0131] .

[0132] The solution to this optimization problem is given by the following closed-loop expression:

[0133] ,

[0134] ,

[0135] in, This represents the Euclidean norm.

[0136] Center of mass It is a weighted average of all vectors within the cluster, and then normalized to preserve shape features. The extraction of typical days then includes the following steps:

[0137] ① Initialize data: from the dataset Random selection An initial centroid .

[0138] ② Iterative process: For each iteration , each load vector Assign to the nearest cluster:

[0139] .

[0140] Then update the centroid of each cluster:

[0141] .

[0142] ③ Termination condition: The algorithm terminates when the centroid no longer changes significantly.

[0143] centroid vector Each A 24-hour load curve representing a typical day results in 10 typical-day load curves. In this embodiment, the above method was applied to cluster 300 days of historical load data, successfully extracting 10 typical-day load curves. Each typical-day curve captures the temporal shape characteristics of the load. Further, based on scenarios using multivariate statistical parameters, scenarios are extended and generated. Building upon the 10 typical-day load curves, to fully represent the uncertainty and volatility of the load, multivariate statistical methods are used to extend and generate scenarios. Probabilistic perturbations are applied to the typical-day model to generate a large-scale set of random scenarios covering possible situations. For each typical day... Statistical parameters are calculated based on the corresponding historical load data.

[0144] Mean vector: ,

[0145] Standard deviation vector: ,

[0146] Residual distribution characteristics: .

[0147] Subsequently, a multivariate normal distribution was established with the typical daily curve as the mean and historical fluctuations as the variance: .

[0148] Wherein, the covariance matrix The diagonal element is Off-diagonal elements characterize the correlation of loads at different time periods.

[0149] Subsequently, a random sampling process is performed to generate new scenarios from the distribution of each typical day: , .in, The Cholesky decomposition was used to ensure that the new scenarios maintained their spatiotemporal relevance. Three random scenarios were generated from each of 10 typical daily distributions, resulting in a total of 30 new electricity load scenarios. The newly generated scene achieves shape preservation while remaining flat under random perturbations.

[0150] Finally, based on the improved K-Means feature clustering, the scene set is reduced from 30 generated load scenes. To reduce the computational complexity of subsequent optimization models, this embodiment uses an improved K-Means clustering algorithm to reduce the large-scale scene set to a representative scene. Standard K-Means randomly initializes centroids, which is prone to getting trapped in local optima, leading to unstable clustering results each time. This method achieves a significant improvement in computational efficiency while preserving the overall statistical characteristics of the scene set. The 30 daily load scenes generated above are considered as data points in a high-dimensional feature space. Each of the scenes It is a 24-dimensional vector representing the load curve for 24 hours a day. First, the K-Means++ algorithm is used to optimize the selection of the initial centroid to avoid the algorithm getting trapped in local optima. The number of clusters is then set. Solving the cluster partitioning and center of mass To minimize the sum of intra-cluster distances:

[0151] .

[0152] Then, the centroid is solved iteratively:

[0153] ,

[0154] in, Cluster The number of scenes included. Finally, the centroids of each cluster are selected as the final representative typical scenes, and probability weights are assigned according to the cluster size: satisfy Finally, 10 representative scenarios were generated to simulate the power load of this embodiment. For example... Figure 3 and Figure 4 As shown, in a power grid system with hydrogen energy storage, the load on wind power and photovoltaic power generation equipment is significantly reduced during the same period, which can better cope with the fluctuations of wind and solar power and the uncertainty of electricity load.

[0155] (2) Scene generation of irradiance and wind speed

[0156] To represent the uncertainty of wind and solar power output, this embodiment uses a Gaussian mixture model to analyze historical wind speed and irradiance data and extract typical scenarios. For irradiance scenario generation, this embodiment collects 30 consecutive days of historical irradiance data, with each observation day containing 24 consecutive irradiance observations. The preprocessed thick data is then used to construct a sample set:

[0157] ,

[0158] Each sample vector Indicates the first A 24-hour irradiance observation sequence for the day. Representing the Heavenly Irradiance observations over a specific time period. To address the multimodal distribution characteristics of the irradiance data, a Gaussian mixture model was used for probability density modeling. The Gaussian mixture model is established as follows:

[0159] ,

[0160] in, For the first The mixing weights of the Gaussian components satisfy the following conditions: ; For the first The mean vector of Gaussian components has a dimension of 24. Then the first The covariance matrix of Gaussian components is 24×24. Parameter estimation is performed using the expectation-maximization algorithm, the specific process of which includes:

[0161] Calculate the posterior probability of each sample belonging to each component:

[0162] ,

[0163] Update model parameters:

[0164] ,

[0165] ,

[0166] ,

[0167] The above steps are iteratively executed until convergence, yielding the optimal model parameters. Finally, typical scene generation is performed. Based on the trained Gaussian mixture model, three representative irradiance scenes are generated. The generation process is achieved through random sampling from the learned probability distribution:

[0168] .

[0169] Each generated typical scenario All data are 24-dimensional vectors, representing a typical pattern of diurnal irradiance variation. These scenarios maintain the statistical characteristics of historical data while introducing reasonable random fluctuations through probability sampling, thus comprehensively covering possible irradiance variations. Wind speed data is processed using the exact same technical approach and methodology as irradiance data.

[0170] II. Implementation Methods

[0171] First, scenario data including load demand, renewable energy characteristics, and techno-economic parameters are input into a pre-defined system model. This system model is constructed with the linear objective function of minimizing the sum of equipment investment cost and system operating cost, and includes a set of linear constraints such as equipment capacity constraints, power balance constraints, and energy storage state evolution constraints, forming a mixed-integer linear programming (MILP) model. Then, the Gurobi optimization solver is called to solve this MILP model. Leveraging its efficient solution capabilities for both linear and integer programming, the optimal capacity configuration schemes for photovoltaics, wind power, batteries, electrolyzers, hydrogen storage tanks, and fuel cells are directly output. The advantage of this method is that it completely transforms the system optimization problem into a mixed-integer linear programming problem, avoiding the complexity of nonlinear solutions, thus ensuring high efficiency and global optimality. Specifically:

[0172] Step 1: Establish an objective function for the investment cost of the system's upper-level equipment:

[0173] 1) To minimize the investment cost of a wind-solar-hydrogen-storage system (a hybrid system of photovoltaics, wind power, batteries, energy storage, and hydrogen production), establish an objective function for the investment cost of upper-level equipment:

[0174] (1),

[0175] in, The total investment for wind-solar-hydrogen storage system equipment; and These are the investment costs for photovoltaic power generation and wind power generation, respectively. The total deployment cost of the battery; and The total equipment costs for electrolyzers and fuel cells are respectively; The total equipment cost for the hydrogen storage tank.

[0176] , , , , and The expressions are as follows:

[0177] (2),

[0178] (3),

[0179] (4),

[0180] (5),

[0181] (6),

[0182] (7),

[0183] and These are the power investment costs for photovoltaic and wind power generation, respectively; and These refer to the power output of photovoltaic and wind power generation equipment, respectively. and These are the fixed investment costs for wind power equipment and photovoltaic equipment, respectively. and These are the capacity cost and power cost of the battery, respectively. and These are the battery's capacity and output power, respectively. and The power costs are for electrolyzers and fuel cells, respectively. and These are the power outputs of the electrolyzer and the fuel cell, respectively. Cost per unit capacity of hydrogen storage tanks; This refers to the capacity of the hydrogen storage tank.

[0184] In this embodiment, the power constraint of the battery is set to 1000KW~2000KW, and the capacity is 1000KWh~2000KWh; the power constraint of the electrolyzer is 200KW~500KW; the capacity constraint of the hydrogen storage tank is 50KG~200KG; and the power constraint of the fuel cell is 100KW~300KW.

[0185] 2) Construct a state model for the battery system:

[0186] The SOC (State of Charge) update equation for a battery is:

[0187] (8),

[0188] in, for The state of charge of the battery during a given period; and These are the charging efficiency and discharging efficiency of the battery, respectively. and They are respectively The battery charging power and battery discharging power during the time period are both constrained to be non-negative.

[0189] At the same time, the battery is Power during time period Constrained as:

[0190] (9),

[0191] A positive value indicates that a discharge is in progress. A negative value indicates that charging is in progress.

[0192] 3) Modeling of the electric-hydrogen-fuel cell conversion system:

[0193] After the wind and solar power meet the electricity load, any remaining electricity will be supplied to the electrolytic cell. Hydrogen production during the period The expression is:

[0194] (10)

[0195] for The amount of hydrogen produced during a given period is expressed as:

[0196] (11);

[0197] The amount of hydrogen produced per second:

[0198] (12),

[0199] and These represent the operating temperature and power of the electrolytic cell, respectively.

[0200] If the generated hydrogen is stored in a hydrogen storage tank with a certain conversion efficiency. The amount of hydrogen transferred to the hydrogen storage tank during the time period for:

[0201] (13)

[0202] The storage conversion efficiency of the hydrogen storage tank.

[0203] Fuel cells, acting as a backup energy source, begin consuming energy from hydrogen storage tanks to generate electricity when power supply is insufficient. The mass of hydrogen consumed per hour by a fuel cell when it performs its work (generates electricity) is... for:

[0204] (14)

[0205] in, The low calorific value of the electrolytic cell This refers to the utilization efficiency of fuel cells.

[0206] Any remaining hydrogen in the system will be sold.

[0207] (15)

[0208] in, To cover the cost of selling hydrogen, The mass of the remaining hydrogen in all the final hydrogen storage tanks, This refers to the price of hydrogen sold.

[0209] The state update equation for a hydrogen energy storage system is:

[0210] (16)

[0211] in, for Hydrogen storage capacity over time period for The mass of hydrogen stored in the time-limited hydrogen storage tank. for The amount of hydrogen supplied to the fuel cell for power generation in the hydrogen storage tank during a given period.

[0212] The battery charge / discharge constraints, as well as the capacity constraints of the battery and hydrogen storage tank, are as follows:

[0213] (17)

[0214] (18)

[0215] (19)

[0216] (20).

[0217] With equation (1) as the upper objective function and (2)-(16) as the mathematical connections between models, satisfying the constraints of (17)-(20), the upper mathematical model of wind-solar-hydrogen storage system scheduling is formed.

[0218] Step 2: Establish the lower-level operating costs of the system:

[0219] 1) Calculate the operating cost of the lower-level system based on the specific power consumption of the upper-level system during specific time periods.

[0220] To minimize the total operating cost of the lower-level system, a joint objective function for hydrogen sales funds and costs is established:

[0221] (twenty one),

[0222] in, This represents the total operating cost of the lower-level system. and For solar and wind power Output power during the time period; and Operating costs of solar and wind power generation.

[0223] The power balance constraints for each time period are:

[0224] (twenty two),

[0225] in, For the power system in Electricity demand during a given time period; for Power of the time-phase electrolytic cell; for The output power of the battery during a given period; and for Solar irradiance and wind speed during the time period for The power of the fuel cell during a given time period.

[0226] Finally, the total cost of the system is achieved. Minimum:

[0227] (twenty three).

[0228] 2) Fuel cell start-stop control:

[0229] The consumption of fuel cells is controlled by setting up calculations for fuel cell depreciation startup.

[0230] The start-up and shutdown status of a fuel cell is determined as follows:

[0231] when At that time, the fuel cell starts normally;

[0232] when At that time, determine the residual value of the fuel cell. Does it meet the following requirements: If yes, the fuel cell will start normally; otherwise, it indicates the remaining value of the fuel cell. If it is not supported for its next startup, then even with hydrogen storage, it will still be powered by wind and solar power, and the fuel cell is ready to be replaced.

[0233] for Power demand during different time periods

[0234] The formula for calculating the dynamic time-period residual value of fuel cells is as follows:

[0235] (twenty four),

[0236] in, For the number of times the fuel cell is started, The value loss caused by each start-up and shutdown. , This refers to the rated number of start-ups for the fuel cell device.

[0237] With (21) as the lower-level objective function and (23) as the overall objective function, the operation scheduling model is composed of formulas (22) and (24).

[0238] like Figure 3 and Figure 4 The results show that the system effectively mitigates the drastic fluctuations in renewable energy output through the synergistic effect of hydrogen production by electricity and fuel cells, making the net load curves of wind turbines and photovoltaic units smoother and significantly reducing their wind and solar curtailment rates. This enables the system to cope with multiple uncertainties in irradiance, wind speed, and electricity load more efficiently and stably, ensuring the high reliability of the system operation.

[0239] according to Figure 5 and Figure 6 The results analysis shows that, for most of the time periods, the hourly operating cost of the hydrogen-equipped system is generally lower than that of the non-hydrogen-equipped system, demonstrating an immediate cost advantage. However, in terms of cumulative operating costs, the cost of the non-hydrogen-equipped system grows faster, resulting in a significantly higher total cost at the end of all operating periods compared to the hydrogen-equipped system. This indicates that although the hydrogen-equipped system may have higher costs in some periods, its overall operating efficiency is better, and its long-term operation is more economical.

[0240] Meanwhile, by adding start-stop control for hydrogen fuel cells, it is possible to... Figure 7 and Figure 8 The cumulative consumption value of a system with fuel cell start-stop control is lower than that of a system without start-stop control over the same operating cycle. Both cumulative consumption value curves show a linear upward trend, indicating that the fuel cell start-stop control strategy can significantly reduce the resource consumption of the system during long-term operation and has significant economic benefits.

[0241] In summary, the zero-carbon hydrogen-electric coupling system model constructed in this embodiment achieves significant comprehensive advantages through optimized configuration and scheduling. Firstly, under the premise of completely zero-carbon operation, the total cost of the system is comparable to that of a traditional hydrogen-free system, demonstrating excellent economic feasibility. More importantly, the introduction of hydrogen energy greatly enhances the system's flexibility and robustness. As shown in the attached figures, compared to a hydrogen-free baseline system, this system, through the synergistic effect of electro-hydrogen production and fuel cells, effectively mitigates the drastic fluctuations in renewable energy output, resulting in smoother net load curves for wind turbines and photovoltaic units, significantly reducing their curtailment rates. This allows for more efficient and stable handling of multiple uncertainties in irradiance, wind speed, and electricity load, ensuring highly reliable system operation.

[0242] Specific Implementation Method Two: The dual-layer planning device for a wind-solar-hydrogen-storage hybrid system that considers fuel cell start-stop losses described in this implementation method includes a processor and a memory. The memory stores at least one instruction, which is loaded and executed by the processor to implement the dual-layer planning method for a wind-solar-hydrogen-storage hybrid system that considers fuel cell start-stop losses as described in Specific Implementation Method One.

[0243] Specific Implementation Method 3: A computer storage medium described in this implementation method stores at least one instruction, which is loaded and executed by a processor to implement the two-layer planning method for a wind-solar-hydrogen-storage hybrid system considering fuel cell start-up and shutdown losses as described in Specific Implementation Method 1.

[0244] While specific embodiments of this application have been described herein with reference to them, it should be understood that these embodiments are merely examples of the principles and applications of this application. Therefore, it should be understood that many modifications can be made to the exemplary embodiments, and other arrangements can be designed without departing from the spirit and scope of this application as defined by the appended claims. It should be understood that different dependent claims and features described herein can be combined in ways different from those described in the original claims. It is also understood that features described in conjunction with individual embodiments can be used in other described embodiments.

Claims

1. A two-layer planning method for a wind-solar-hydrogen-storage hybrid system considering fuel cell start-stop losses, characterized in that, include: A linear programming model is constructed with the goal of minimizing the sum of the total equipment investment cost and operating cost of the wind-solar-hydrogen-storage hybrid system. The linear programming model includes an upper-level programming model with the goal of minimizing the total equipment investment cost of the wind-solar-hydrogen-storage hybrid system, and a lower-level operating model with the goal of minimizing the operating cost of the wind-solar-hydrogen-storage hybrid system. The fuel cell is operated based on the fuel cell start-stop control strategy, and then the linear programming model is solved to obtain the optimal capacity configuration scheme of output photovoltaic, wind power, storage battery, electrolyzer, hydrogen storage tank and fuel cell. The fuel cell start-stop control strategy includes: When the sum of the power output of the photovoltaic and wind power generation equipment and the output power of the battery is less than the power demand, the fuel cell is started. When the sum of the power output of the photovoltaic and wind power generation equipment and the output power of the battery is greater than or equal to the power demand, it is determined whether the residual value of the fuel cell exceeds its set threshold. If yes, the fuel cell is started; otherwise, the fuel cell is replaced.

2. The two-layer planning method for a wind-solar-hydrogen storage hybrid system considering fuel cell start-up and shutdown losses according to claim 1, characterized in that, The expression for the upper-level planning model is: ， in, The total investment cost of equipment for a wind-solar-hydrogen-storage hybrid system. and The investment costs for photovoltaic power generation and wind power generation are respectively. The total deployment cost of the battery. and The total equipment costs for electrolyzers and fuel cells are respectively. The total equipment cost for the hydrogen storage tank.

3. The two-layer planning method for a wind-solar-hydrogen storage hybrid system considering fuel cell start-up and shutdown losses according to claim 2, characterized in that, The investment cost of photovoltaic power generation Investment costs of wind power generation Total deployment cost of storage batteries Total equipment cost of electrolytic cells Total equipment cost of fuel cells and the total equipment cost of hydrogen storage tanks. The expressions are as follows: ， ， ， ， ， ， in, and The power investment costs for photovoltaic and wind power generation are respectively. and These are the power outputs of photovoltaic and wind power generation equipment, respectively. and These are the fixed investment costs for wind power equipment and photovoltaic equipment, respectively. and These are the capacity cost and power cost of the battery, respectively. and These are the battery capacity and output power, respectively. and The power costs of electrolyzers and fuel cells are respectively. and These are the power outputs of the electrolyzer and the fuel cell, respectively. The cost per unit capacity of the hydrogen storage tank, This refers to the capacity of the hydrogen storage tank.

4. The two-layer planning method for a wind-solar-hydrogen storage hybrid system considering fuel cell start-up and shutdown losses according to claim 2 or 3, characterized in that, The constraints of the upper-level planning model include: ， ， ， ， in, and They are respectively Battery charging power and battery discharging power during the same period For the capacity of the storage battery, for The state of charge of the battery during a given period. For the capacity of the hydrogen storage tank, for The amount of hydrogen stored during a given period; The state-of-charge update equation for the battery is: ， in, and These are the charging efficiency and discharging efficiency of the battery, respectively. The state update equation for a hydrogen energy storage system is: ， in, for The time period is converted to the hydrogen mass in the hydrogen storage tank. for The mass of hydrogen consumed by the fuel cell during a given period of time.

5. The two-layer planning method for a wind-solar-hydrogen storage hybrid system considering fuel cell start-up and shutdown losses according to claim 1, 2, or 3, characterized in that, The expression for the lower-level operating model is: ， in, For the operating cost of the wind-solar-hydrogen-storage hybrid system, and Solar and wind power respectively Output power during the time period and The operating costs of solar and wind power generation To sell hydrogen.

6. The two-layer planning method for a wind-solar-hydrogen storage hybrid system considering fuel cell start-up and shutdown losses according to claim 5, characterized in that, The constraints of the lower-level operating model include: ， in, for Electricity demand during a given time period for Power of the time-lapse electrolyzer, and for Solar irradiance and wind speed during the time period for The output power of the battery during the period for The power of the fuel cell during a given time period.

7. The two-layer planning method for a wind-solar-hydrogen storage hybrid system considering fuel cell start-up and shutdown losses according to claim 1, 2, 3, or 6, characterized in that, The residual value expression of the fuel cell is as follows: ， in, For the residual value of fuel cells, The total equipment cost of fuel cells, For the number of times the fuel cell is started, The value loss caused by each start-up and shutdown. , This refers to the rated number of start-ups for the fuel cell device. The residual value constraint of the fuel cell is expressed as follows: .

8. The two-layer planning method for a wind-solar-hydrogen storage hybrid system considering fuel cell start-up and shutdown losses according to claim 1, 2, 3, or 6, characterized in that, The proposed two-layer planning method for the wind-solar-hydrogen-storage hybrid system is implemented under scenarios that include load demand, renewable energy characteristics, and techno-economic parameters. Methods for generating scenes include: Based on historical load data, the K-Shapes clustering algorithm was used to extract daily load curves with typical morphological characteristics. Based on the daily load curve, a multivariate statistical method is used to expand the scenarios and generate an initial scenario set; An improved K-Means clustering algorithm is used to reduce the initial scene set to obtain representative load scenes and their probability weights; Gaussian mixture models are used to perform probability density modeling on historical wind speed and irradiance data to generate typical wind and solar power output scenarios.

9. A dual-layer planning device for a wind-solar-hydrogen storage hybrid system considering fuel cell start-up and shutdown losses, characterized in that, The dual-layer planning device for a wind-solar-hydrogen-storage hybrid system that considers fuel cell start-stop losses includes a processor and a memory. The memory stores at least one instruction, which is loaded and executed by the processor to implement the dual-layer planning method for a wind-solar-hydrogen-storage hybrid system that considers fuel cell start-stop losses as described in any one of claims 1 to 8.

10. A computer storage medium, characterized in that, The computer storage medium stores at least one instruction, which is loaded and executed by a processor to implement the two-layer planning method for a wind-solar-hydrogen-storage hybrid system considering fuel cell start-up and shutdown losses as described in any one of claims 1 to 8.

Images