A hydrogen energy cogeneration system scheduling method without conventional power supply support

Abstract

The application provides a hydrogen energy cogeneration system scheduling method without conventional power support, comprising the following steps: obtaining a set of prediction errors of decision variables according to a system operation model and a historical data set, adopting a regular vine Copula and Rosenblatt transformation, obtaining a set of prediction errors of limit scenario points according to the system operation model, the historical data set and an uncertain variable set, so as to obtain a fusion set, and adopting an asymmetric directional Wasserstein metric to determine a fuzzy set of prediction error probability distribution of the decision variables, and combining the system operation model to construct a two-stage distribution robust optimization model including day-ahead prediction and intra-day deviation adjustment, and solving to obtain a current decision value set. The method covers extreme working conditions without greatly increasing the conservatism, realizes effective balance of economy and system robustness, avoids generating a large number of pseudo-limit scenarios which are impossible to occur in meteorology, and balances the source and load risk costs.

Description

Technical Field

[0001] This invention relates to the field of hydrogen cogeneration system scheduling technology, specifically to a scheduling method for a hydrogen cogeneration system without conventional power source support. Background Technology

[0002] In hydrogen cogeneration systems that integrate new energy power generation, the inherent volatility and intermittency of new energy sources, coupled with limited transmission capacity, lead to a dual challenge for large-scale new energy bases: unstable power transmission and difficulties in large-scale consumption.

[0003] In the research on optimization scheduling considering source-load uncertainty, existing methods mainly include two categories: stochastic optimization and robust optimization. Stochastic optimization relies on accurate probability distributions, but in the scenario of large-scale new energy bases, due to limited historical data and the existence of distribution drift, it is difficult to accurately obtain the true probability distribution. Robust optimization often reserves excessive adjustment margins when dealing with worst-case scenarios and completely ignores the probability distribution information contained in historical data, leading to a decline in system economy. When dealing with wind and solar load uncertainty, existing scheduling methods, if using traditional robust optimization, result in overly conservative system configurations and poor economy. Summary of the Invention

[0004] This invention addresses the problems of existing technologies by providing a scheduling method for hydrogen cogeneration systems without conventional power source support that enhances robustness to extreme conditions. This method uses regular vine Copula and Rosenblatt transforms to generate extreme scenarios and incorporates tail features of prediction errors in decision variables. It covers extreme conditions without significantly increasing conservatism, achieving an effective balance between economy and system robustness. Furthermore, it can accurately extract tail-dependent limit boundaries that conform to real meteorological and physical coupling characteristics in high-dimensional space, avoiding the generation of numerous pseudo-limit scenarios that are meteorologically impossible. The asymmetric directional Wasserstein metric is used to construct fuzzy sets that can guide convergence through directional constraints, thus balancing the source-load risk costs in hydrogen cogeneration systems without conventional power source support.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows: a scheduling method for a hydrogen cogeneration system without conventional power source support, comprising: Obtain the historical dataset, set of uncertain variables, and current initial state of the hydrogen cogeneration system. The set of uncertain variables represents the collection of uncertain variables. A system operation model is constructed based on a hydrogen-powered combined heat and power system. A first error set is obtained based on the system operation model and historical datasets. This first error set represents the set of first errors, which in turn represent the prediction errors of decision variables. Using the Copula rule and Rosenblatt transformation, a second error set is obtained based on the system operation model, historical datasets, and a set of uncertain variables. This second error set represents the set of prediction errors at extreme scenario points, where extreme scenario points represent scenarios where uncertain variables simultaneously take extreme values. A fusion set is then obtained based on the first and second error sets. The system employs an asymmetric Wasserstein metric and constructs a fuzzy set for the first error probability distribution based on the fusion set. A two-stage sub-Bruker optimization model is then constructed based on the system operation model and the fuzzy set. The two stages include a day-ahead prediction stage and an intraday deviation adjustment stage. The current decision value set is obtained by solving the two-stage sub-Brussels bar optimization model based on the current initial state, the set of uncertain variables, and the system operation model. The current decision value set represents the set of decision variables for the current scheduling period.

[0006] In some embodiments, the Copula and Rosenblatt transformations are used to obtain a second error set based on the system operating model, historical datasets, and a set of uncertain variables, including: The third error set is obtained based on the system operation model, historical dataset, and set of uncertain variables. The third error set represents the set of third errors, and the third error represents the prediction error of uncertain variables. The edge probability distribution set is obtained by fitting the edge distribution based on the third error set; The Copula model is constructed using the Copula regularization method based on the marginal probability distribution set. The second error set is obtained by using the Rosenblatt transform and the regular vine Copula model.

[0007] In some embodiments, a fuzzy set is constructed based on an asymmetric Wasserstein metric and a first error probability distribution, including: The terrain cost function is obtained based on the fusion set. The terrain cost function characterizes the asymmetric directional measure between the actual error and the prediction error of the decision variables. Using the Wasserstein metric, a distance measurement model is constructed based on the terrain cost function. The distance measurement model is used to solve the asymmetric Wasserstein distance between the empirical probability distribution and the true distribution of the first error. Fuzzy sets are obtained based on distance measurement models.

[0008] In some embodiments, a dual transformation method is employed to construct a two-stage sub-Bruker optimization model based on the system operation model and fuzzy sets.

[0009] In some of these embodiments, the uncertainties include renewable energy output and electrical load.

[0010] In some of these embodiments, the renewable energy output includes wind power and photovoltaic power.

[0011] In some embodiments, the two-stage sub-Blu-ray optimization model includes an objective function that represents taking the supremum of the expected deviation adjustment cost in the intraday deviation adjustment stage within the fuzzy set, while minimizing the estimated operating cost in the intraday deviation adjustment stage.

[0012] In some embodiments, the estimated operating cost represents the difference between the total operating cost and the revenue from energy sales, and the total operating cost represents the sum of the equipment operating cost and the cost of energy curtailment and power failure penalties; the deviation adjustment cost represents the difference between the total operating deviation adjustment cost and the revenue from energy sales deviation, and the total operating deviation adjustment cost represents the sum of the equipment operating deviation adjustment cost and the cost of energy curtailment and power failure penalties.

[0013] In some embodiments, the sub-bar optimization model also includes constraints, including energy balance constraints, energy loss constraints, and output and operating state constraints.

[0014] In some embodiments, the hydrogen cogeneration system includes an electrical subsystem, a thermal subsystem, and a hydrogen subsystem; The power subsystem is used for main power generation based on new energy generator sets, as well as for charging and discharging electrical energy; The thermal energy subsystem is used for main heating and auxiliary power generation based on solar thermal power plants, auxiliary heating based on electric heaters, and for charging and discharging thermal energy. The hydrogen energy subsystem is used for hydrogen production based on reversible solid oxide batteries, as well as for the charging and discharging of hydrogen. The reversible solid oxide battery operates in two modes: a solid oxide electrolyzer mode and a solid oxide fuel cell mode. When the reversible solid oxide battery operates in the solid oxide electrolyzer mode, the thermal energy subsystem provides thermal energy to the hydrogen energy subsystem. When the reversible solid oxide battery operates in the solid oxide fuel cell mode, the hydrogen energy subsystem provides thermal energy to the thermal energy subsystem.

[0015] Compared with the prior art, the present invention has the following beneficial effects: 1. This application integrates the prediction error of extreme scenario points based on the set of uncertain variables into the prediction error set of decision variables by using a fusion set, and corrects the tail features of the basic error set. Only when the confidence requirement is extremely high will the probability distribution based on the fusion set tilt towards the extreme scenario points, thus achieving a natural and objective balance between economy and robustness in the underlying mathematical logic.

[0016] 2. This application uses the Copula and Rosenblatt transformation to obtain the second error set, which can accurately extract the tail-dependent limit boundary that conforms to the real meteorological physical coupling characteristics in high-dimensional space. This avoids generating a large number of pseudo-limit scenarios that are meteorologically impossible when dealing with microgrids under complex climatic conditions such as "sand desert".

[0017] 3. This application uses an asymmetric directional Wasserstein metric to construct a fuzzy set, which can guide convergence through directional constraints to balance the source-load risk cost in a hydrogen cogeneration system without conventional power source support. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating the scheduling method for a hydrogen cogeneration system without conventional power supply in the embodiments of this application. Figure 2 This is the power balance dispatch result for scenario 1 in the example of this application; Figure 3 This is the thermal energy balance scheduling result for scenario 1 in this application's example; Figure 4 This is the hydrogen energy balance scheduling result for scenario 1 in this application example. Detailed Implementation

[0019] In addressing the problems in the background technology, some researchers have adopted the data-driven Distributed Robust Optimization (DRO) method. This method constructs fuzzy sets of probability distributions based on historical data, achieving a better trade-off between robustness and economy. Its fuzzy set construction methods are divided into two categories: moment information and probability density. Compared to the moment information method, the Wasserstein distance-based method directly measures the distance between probability distributions, without assuming a probability distribution function. It directly constructs distributions based on historical data samples, limiting the fuzzy set range with a distance threshold, resulting in a more reasonable distribution. It has been applied to scenarios such as unit combination, virtual power plants, and distribution network scheduling. However, when applying this DRO to the scheduling optimization of a hydrogen cogeneration system integrating new energy power generation, the system is prone to load shedding under extreme scenarios. The inventors, through numerous experiments, verified that this load shedding phenomenon occurs because the fuzzy set model did not consider extreme weather conditions.

[0020] Traditional methods, such as using polyhedral vertices to incorporate extreme scenarios into fuzzy set modeling, directly combine the absolute upper and lower bounds of each uncertain variable, resulting in physical extreme values ​​rather than probabilistic extreme values. When dealing with microgrids under complex climatic conditions such as deserts, this generates a large number of pseudo-extreme scenarios that are meteorologically impossible.

[0021] Furthermore, using the standard Wasserstein distance imposes equal penalties on both positive and negative fluctuations in prediction errors. However, in hydrogen cogeneration systems without conventional power support, the risk cost of "sources falling short of load" is far higher than that of "sources exceeding load."

[0022] To clearly illustrate the technical features of this solution, the implementation methods of this application will be described in detail below with reference to the accompanying drawings and embodiments. This will allow for a full understanding and implementation of how this application uses technical means to solve technical problems and achieve corresponding technical effects. The embodiments of this application and the various features within them can be combined with each other without conflict, and the resulting technical solutions are all within the protection scope of this application.

[0023] This application proposes a scheduling method for a hydrogen cogeneration system without conventional power source support, including: Obtain the historical dataset, set of uncertain variables, and current initial state of the hydrogen cogeneration system. The set of uncertain variables represents the collection of uncertain variables. The historical dataset includes multiple sets of first historical sample data. Each set of first historical sample data includes a historical initial state and a set of historical decision values. The historical initial state represents the initial state of the historical scheduling cycle corresponding to the historical sample data, and the set of historical decision values ​​represents the set of decision values ​​at each moment of the historical scheduling cycle corresponding to the historical sample data. A system operation model was constructed based on a hydrogen-powered combined heat and power system. The first error set is obtained based on the system operation model and historical dataset. The first error set represents the set of first errors, and the first errors represent the prediction errors of the decision variables; specifically: The system operation model is simulated based on each historical initial state to obtain the first set of predicted values. The first set of predicted values ​​represents the set of predicted values ​​of decision variables at each moment within the historical scheduling cycle. The first error set is obtained based on the historical decision value set and the historical decision variable prediction value set. The first error set is: ; ; in, Denotes the first error set. Indicates the first The prediction error set of each decision variable. Indicates the index of the decision variable. This represents the total number of decision variables. for Time of the first The prediction error of each decision variable The time sequence number within the scheduling period. The total number of moments within the scheduling period; The historical dataset also includes multiple sets of second historical sample data. Each set of second historical sample data includes a historical initial state and a set of historical uncertain variable values. The historical initial state represents the initial state of the historical scheduling period corresponding to the historical sample data, and the set of historical uncertain variable values ​​represents the set of uncertain variable values ​​at each moment of the historical scheduling period corresponding to the historical sample data. Using the Copula rule and Rosenblatt transform, a second error set is obtained based on the system operation model, historical dataset, and set of uncertain variables. This second error set represents the set of prediction errors for extreme scenario points, where each uncertain variable in the set simultaneously takes its extreme value. The second error set is as follows: ; ; in, This represents the second error set. Indicates the first The prediction error set for each extreme scenario point Indicates the index of the extreme scenario point. This represents the total number of extreme scene points. for Time of the first Prediction error for extreme scenario points.

[0024] The fusion set is obtained based on the first error set and the second error set. The fusion set is as follows: ; in, This is the fusion set, and also the support set using the asymmetric Wasserstein metric. The prediction error set for the first decision variable. For the first The prediction error set of each decision variable. This is the prediction error set for the first extreme scenario point. For the first The prediction error set for each extreme scenario point For the integration of the first One element, The sequence number of the elements in the fusion set; The asymmetric directionality Wasserstein metric is adopted, and a fuzzy set of the first error probability distribution is constructed based on the fusion set. A two-stage sub-Bluhl bar optimization model is constructed based on the system operation model and the fuzzy set. The two stages include the day-ahead prediction stage and the intraday deviation adjustment stage. The construction of the sub-Bluhl bar optimization model follows strict decision-making timing. The predicted values ​​of the decision variables in the day-ahead prediction stage are generated by the day-ahead prediction and determined before the actual error value is revealed. The actual values ​​of the decision variables in the intraday deviation adjustment stage are triggered by the intraday deviation and are a real-time response to the actual error value.

[0025] The objective function of the pluripotent bar optimization model takes the supremum of the expected adjustment cost during the intraday deviation adjustment stage within the fuzzy set, while minimizing the estimated operating cost during the day-ahead forecast stage, strictly satisfying engineering timing and information availability requirements. The objective function is: ; in, To minimize the function, The cost coefficient within the scheduling period. For the set of predicted values ​​of decision variables, To obtain the supremum, we mean finding the probability distribution in the fuzzy set that maximizes the expected adjustment cost. This represents the true distribution of the first error. For fuzzy sets, Represents the mathematical expectation. Cost of adjusting for deviation For the first Each decision variable is Predicted value at time, Indicates the first Each decision variable is Prediction error at time, For the set of actual values ​​of decision variables, superscript Indicates transpose. Indicates constraints. For the first Each decision variable is The actual value at time, As the first coefficient, As the second coefficient, The third coefficient, , and All are greater than 0. It is a constant. ; The current decision value set is obtained by solving the two-stage sub-Brussels bar optimization model based on the current initial state, the set of uncertain variables, and the system operation model. The current decision value set represents the set of decision variables for the current scheduling period.

[0026] Beneficially, by fusing the prediction errors of extreme scenario points based on the set of uncertain variables into the prediction error set of decision variables through a fusion set, and correcting the tail characteristics of the basic error set (i.e., the first error set), the probability distribution based on the fusion set will only tilt towards the extreme scenario points when the confidence requirement is extremely high. This achieves a natural and objective balance between economy and robustness at the underlying mathematical logic level. Simultaneously, by using the Copula rule and Rosenblatt transform to obtain the second error set, the tail-dependent limit boundary that conforms to the coupling characteristics of real meteorological physics can be accurately extracted in high-dimensional space. This avoids generating a large number of pseudo-extreme scenarios that are meteorologically impossible when dealing with microgrids under complex climatic conditions such as desertification. Furthermore, constructing a fuzzy set using the asymmetric directional Wasserstein metric can guide convergence through directional constraints, thus balancing the source-load risk costs in hydrogen cogeneration systems without conventional power source support.

[0027] In some embodiments, the Copula and Rosenblatt transformations are used to obtain a second error set based on the system operating model, historical datasets, and a set of uncertain variables, including: The third error set is obtained based on the system operation model, historical dataset, and set of uncertain variables. The third error set represents the set of third errors, and the third error represents the prediction error of uncertain variables. Marginal probability distribution sets are obtained by fitting the marginal distribution based on the third error set. Optionally, the marginal distribution is fitted using the kernel density estimation (KDE) method. The R-vine Copula model is constructed based on the marginal probability distribution set. A regular vine Copula tree is constructed based on the marginal probability distribution set. The nodes and edges of the regular vine Copula tree reflect the strength of conditional dependence between uncertain variables. The regular vine Copula tree is as follows: ; in, Represents the regular vine Copula tree, Represents the first in the regular vine Copula tree A tree, Indicates the index of the tree in the regular vine Copula tree. This represents the total number of trees in a regular vine (Copula) tree. The total number of uncertain variables; Constructing a Copula model involves building a multidimensional joint probability density function and extracting a multidimensional tail dependence metric.

[0028] The multidimensional joint probability density function constructed based on the rule-based Copula tree is as follows: ; in, Describes the multidimensional joint probability density function. This represents the marginal probability distribution of the prediction error corresponding to the first uncertain variable. This represents the marginal probability distribution of the prediction error corresponding to the second uncertain variable. Indicates the first Marginal probability distribution of prediction errors corresponding to uncertain variables Indicates the first Marginal probability distribution of prediction errors corresponding to uncertain variables For the index of the uncertain variable, Indicates the first n The marginal probability density function of the prediction error corresponding to each uncertain variable Indicates the first The edge set of trees, Indicates the index of the edge in the edge set. Indicates the first Tree No. Pair-copula density function of the strip edge, This represents the cumulative distribution function given the prior conditions. Represents a node The prediction error of the corresponding uncertain variables, Indicates the first m Among the trees e The first node connected, Represents a node The prediction error of the corresponding uncertain variables, Indicates the first m Among the trees e The second node connected, Representing a condition set The prediction error of the included uncertain variables, Represents a condition set; For tail features, Clayton fitting is used for lower tail dependency and Gumbel fitting for upper tail dependency. Extract multidimensional tail dependency metrics, which include upper tail dependency coefficient, lower tail dependency coefficient, and tail confidence threshold. The dependence coefficient of the upper tail is: ; in, The upper tail dependence coefficient, To represent limit operations, The tail cumulative probability threshold is used to define the boundary of the tail region in extreme scenarios. As a probability measure, The value of the marginal cumulative distribution function corresponding to the first uncertain variable. The value of the marginal cumulative distribution function corresponding to the second uncertain variable. The variable takes values ​​that are all thresholds. u The value of the bivariate Copula joint cumulative distribution function at time; The lower tail dependency coefficient is: ; in, The upper tail dependency coefficient; Set the tail confidence threshold This isolates the high-risk probability space located in the tail-dependent region.

[0029] The Rosenblatt transform is employed. Based on the Copula model, a second error set is obtained. The Rosenblatt transform then maps independent, uniformly distributed samples to a tail region with a specific R-vine structure, generating a dynamic extreme scene point set. The conditional cumulative distribution function of the Rosenblatt transform is: ; in, Indicates before Under the condition of observations of uncertain variables, The conditional cumulative distribution function, Indicates the first The joint Copula cumulative distribution function of the mapping variables corresponding to the uncertain variables. Indicates the first Independent standard uniformly distributed variables mapped from uncertain variables Indicates the first Independent standard uniformly distributed variables mapped from uncertain variables Indicates the first Independent standard uniformly distributed variables mapped from uncertain variables Indicates the preceding The joint Copula cumulative distribution function of the mapping variables corresponding to the uncertain variables; In the tail probability space, let Through the inverse Rosenblatt transform The extreme points on the equal probability density profile are the second error.

[0030] In some embodiments, an asymmetric directional Wasserstein metric is employed, and a fuzzy set of the first error probability distribution is constructed based on the fusion set. This includes: obtaining a terrain cost function based on the fusion set, whereby the terrain cost function characterizes an asymmetric directional metric between the actual error and the prediction error of the decision variables; the terrain cost function is: ; in, Let the terrain cost function be... This represents the actual error. For prediction error, For the first The actual error of each decision variable For the first The prediction error of each decision variable For the first The upper limit of the terrain cost penalty weight for each decision variable. For the first The lower bound of the terrain cost penalty weight for each decision variable; optionally, when the error direction points to system energy shortage, a higher weight is assigned, in which case let When the error direction points towards a system with excess energy, a low weight is assigned, and at this time, let .

[0031] A Wasserstein metric is used, and a distance measurement model is constructed based on the terrain cost function. This model is used to solve for the asymmetric Wasserstein distance between the empirical probability distribution and the true distribution of the first error. The distance measurement model is as follows: ; in, express and The asymmetric Wasserstein distance between them This represents the true distribution of the first error. The empirical probability distribution of the first error. This indicates a function that takes a lower bound. This indicates the search area for the first error. Indicates actual error With prediction error All joint distributions, express and Fuzzy sets between; The fuzzy set is: ; in, The radius is measured by Wasserstein. In some embodiments, a dual transformation method is employed to construct a two-stage sub-Bruker optimization model based on the system operation model and fuzzy sets.

[0032] The objective function takes the supremum of the expected adjustment cost for the intraday deviation adjustment stage within the fuzzy set. Based on the duality theorem, the objective function for the intraday deviation adjustment stage is transformed into a solvable semi-infinite programming problem. The transformed objective function for the intraday deviation adjustment stage is as follows: ;

[0033] in, As dual variables, For the first Transformation of the expression of decision variables; The constraints of the objective function in the transformed intraday deviation adjustment phase are: for any extreme scenario point, the following conditions are satisfied: ; By further introducing auxiliary variables and dual norms, the nonlinear max operator in the terrain cost function is transformed into multiple sets of linear inequality constraints: ; ; ; In the formula, For the introduced auxiliary variables, Characterizing the first The decision variable is determined by the asymmetric Wasserstein metric. The upper bound of the weighted absolute deviation between the prediction error and the actual error for each dimension corresponding to an uncertain variable. For the first Asymmetric penalty weights for positive fluctuations in prediction error across dimensions corresponding to each uncertain variable. For the first Asymmetric penalty weights for negative fluctuations in prediction error across dimensions corresponding to each uncertain variable. For the first i The decision variable is the first The prediction error for each dimension corresponding to an uncertain variable. For the first The actual error of an uncertain variable; Through the above transformation, the Min-Max robust problem of a two-stage sub-Bruker optimization model with asymmetric directional metrics and dynamic limit scenario points is perfectly transformed into a large-scale mixed-integer linear programming model that is easy to solve using commercial solvers.

[0034] In some embodiments, the uncertain variables include renewable energy output and electrical load; optionally, renewable energy output includes wind power output and photovoltaic power output.

[0035] The new fuzzy set retains the first error probability distribution information from the historical dataset while also covering the tail risk of extreme scenarios. Although extreme scenarios are included in the historical dataset, they are only used as part of the support set to correct the tail features of the first error probability distribution. The optimization process seeks the expected value of the worst distribution within the fuzzy set, rather than the absolute value for a single extreme scenario. In other words, the method proposed in this application automatically balances the impact of high-probability normal samples and low-probability extreme samples. The distribution only significantly tilts towards extreme scenarios when historical data is extremely scarce or when extremely high confidence is required. Therefore, the method proposed in this application can achieve a better balance between economy and robustness while ensuring resistance to extreme conditions and avoiding excessive economic costs for extremely low-probability events.

[0036] In some embodiments, the estimated operating cost represents the difference between the total operating cost and the revenue from energy sales. The total operating cost represents the sum of equipment operating costs and the cost of energy curtailment / discontinuation penalties. ; The estimated operating cost is: ; In the formula, This is the serial number of the power generation equipment. A collection of power generation equipment. For the first Taiwan power generation equipment The cost of generating electricity at any given time For the first Taiwan power generation equipment Power generation at any given moment This is the serial number of the energy storage device. A collection of energy storage devices, For the first Taiwan energy storage equipment The operating cost at any given time For the first Taiwan energy storage equipment The charging power at any given moment For the first Taiwan energy storage equipment Energy release power at any given time This is the serial number of the separation and purification device. For the assembly of separation and purification devices, For the first Taiwan separation and purification device The cost of hydrogen production at any given time For the first Taiwan separation and purification device Hydrogen production at any given time This is the cost coefficient for power curtailment penalties. for Power curtailment at any given time The penalty cost coefficients for hydrogen loss, heat loss, and power loss are... for Power loss at any given moment for Heat loss over time for The amount of hydrogen lost at each moment. This is the hydrogen sales cost coefficient. This is the heat sales cost coefficient. This is the electricity sales cost coefficient. for Hydrogen sales volume at any given time for Heat sold at any time for Electricity sales power at any given time.

[0037] In some embodiments, deviation adjustment cost represents the difference between total operating deviation adjustment cost and energy sales deviation revenue, and total operating deviation adjustment cost represents the sum of equipment operating deviation adjustment cost and energy curtailment / disabling penalty deviation cost.

[0038] ; In the formula, After taking into account the prediction error, the first Taiwan power generation equipment Power generation at any given moment After taking into account the prediction error, the first Taiwan energy storage equipment The charging power at any given moment After taking into account the prediction error, the first Taiwan energy storage equipment Energy release power at any given time After taking prediction error into account Power curtailment at any given time After taking prediction error into account Power loss at any given moment After taking prediction error into account Electricity sales power at any given time; In some embodiments, the hydrogen cogeneration system includes an electrical subsystem, a thermal subsystem, and a hydrogen subsystem; The power subsystem is used for main power generation based on new energy generator sets, as well as for charging and discharging electrical energy; The thermal energy subsystem is used for main heating and auxiliary power generation based on solar thermal power plants, auxiliary heating based on electric heaters, and for charging and discharging thermal energy. The hydrogen energy subsystem is used for hydrogen production based on reversible solid oxide batteries, as well as for the charging and discharging of hydrogen. The reversible solid oxide battery operates in two modes: a solid oxide electrolyzer mode and a solid oxide fuel cell mode. When the reversible solid oxide battery operates in the solid oxide electrolyzer mode, the thermal energy subsystem provides thermal energy to the hydrogen energy subsystem. When the reversible solid oxide battery operates in the solid oxide fuel cell mode, the hydrogen energy subsystem provides thermal energy to the thermal energy subsystem, thereby achieving the coupling of the solar thermal power plant and the reversible solid oxide battery to realize the cascade utilization of energy.

[0039] The system operation model is constructed based on the hydrogen energy cogeneration system, including the construction of an electrical energy system model based on the electrical energy subsystem, the construction of a thermal energy system model based on the thermal energy subsystem, and the construction of a hydrogen energy system model based on the hydrogen energy subsystem. Taking new energy power generation units, including wind turbines and photovoltaic units, as an example. Wind farms generally adopt a chain feeder structure. The output power of wind turbines is related to wind speed, and there is a non-linear relationship. Large-scale photovoltaic power plants are generally composed of multiple photovoltaic units, and the power generation of photovoltaic units is affected by factors such as installed capacity, sunlight conditions, and power generation efficiency.

[0040] The power system model includes a power supply module and an energy storage module. The power supply module corresponding to the new energy generator set is as follows: ; ; ; In the formula, for The output power of a single wind turbine at any given time. This refers to the rated power of a single wind turbine unit. for The output power of the wind farm at any given time for The actual wind speed at that moment, for The wind speed at the moment of entry. To cut into wind speed, To cut off the wind speed, Rated wind speed, This represents the total number of wind turbine units. for The output power of the photovoltaic power station at any time This represents the total number of photovoltaic (PV) units. For the first The irradiated area of ​​a photovoltaic panel. Local irradiance, for The power generation efficiency of photovoltaic panels at all times.

[0041] The energy storage module is: ; In the formula, for Real-time energy of the energy storage device. for Real-time energy of the energy storage device. The self-loss rate of the energy storage device. The charging and discharging efficiency of electrical energy storage devices. for The charging power of the energy storage device at all times. for The discharge power of the energy storage device at any given time.

[0042] The hydrogen energy system model includes the RSOC model and a hydrogen energy storage module. RSOC stands for Reversible Solid Oxide Cell. RSOC can flexibly switch between SOEC and SOFC modes. SOEC stands for Solid Oxide Electrolysis Cell, and SOFC stands for Solid Oxide Fuel Cell, achieving bidirectional electro-hydrogen conversion. The electrical and thermal power of RSOC is described by a quadratic polynomial model of hydrogen flow rate.

[0043] The RSOC (Return Flow Control) system comprises the RSOC, heat exchanger, air compressor, and return pump. When modeling the RSOC, each component needs to be modeled separately. The RSOC model is as follows: ; ; In the formula, for At any given moment, the output power of the RSOC for At any given time, RSOC consumes electrical power in SOEC mode. for When the RSOC is in SOFC mode, it outputs electrical power. for At any given time, the output thermal power of the RSOC for At any given time, the RSOC consumes thermal power in SOEC mode. for When the RSOC is in SOFC mode, it outputs thermal power. for Hydrogen consumption at time RSOC for Hydrogen production at time point of RSOC in SOEC mode for Hydrogen consumption of RSOC in SOFC mode at any given time. As the first running parameter, This is the second running parameter. This is the third operating parameter. This is the fourth operating parameter. This is the fifth operating parameter; The switching of operating modes in a reversible solid oxide battery is equivalent to the switching of heat exchange states in the heat exchanger. The heat exchanger is modeled as follows: ; In the formula, for The heat exchange power of RSOC in SOEC mode at any given time. The specific heat capacity of salt. for The heat transfer medium mass of RSOC in SOEC mode at any given time. This refers to the heat exchange temperature difference of RSOC in SOEC mode. The heat exchange efficiency of RSOC in SOEC mode, for The heat exchange power of RSOC in SOFC mode at any given time. The specific heat capacity of water, for The mass of the heat transfer medium when RSOC is in SOFC mode at a given time. This refers to the heat exchange temperature difference of RSOC in SOFC mode. The heat exchange efficiency of RSOC in SOFC mode; In addition, the air compressor performs air circulation, and the return pump performs heat exchange circulation. The air compressor and the return pump are modeled separately as follows: ; ; In the formula, for The electrical power of the air compressor at all times. It is a constant. for air velocity at all times for The compression ratio of the air compressor at all times. for The electrical power of the constant flow pump, For the mechanical efficiency of the reflux pump, for Constant heat transfer fluid flow rate for The pressure at the inlet of the reflux pump is constantly being adjusted. for The pressure at the outlet of the reflux pump at all times; The hydrogen energy storage module is: ; In the formula, for Real-time hydrogen storage capacity of the hydrogen energy storage device. for Real-time hydrogen storage capacity of the hydrogen energy storage device. The self-loss rate of hydrogen energy storage equipment. The hydrogen charging and discharging efficiency of hydrogen energy storage devices. for The hydrogen charging quality of hydrogen storage devices at all times for The quality of hydrogen release from hydrogen storage devices at all times.

[0044] The hydrogen energy system model also includes a hydrogen purification module; commonly used hydrogen purification technologies in the chemical industry include pressure swing adsorption (PSA) and membrane separation. Optionally, considering the maturity, high purification efficiency, and relatively low energy consumption of PSA technology, this application adopts a two-stage PSA series purification process, i.e., second-stage PSA. PSA stands for Pressure Swing Adsorption, which meets the requirements of high-purity hydrogen storage and SOFC applications. The hydrogen purification module is as follows: ; ; In the formula, for The hydrogen mass of the PSA is constantly input. For the first Hydrogen production efficiency in stage PSA This is the stage number of the second-order PSA. This represents the percentage of hydrogen gas in the crude hydrogen. for The mass of hydrogen flowing from RSOC to PSA at any given time. for Total PSA power at any given time for The first stage of the second-order PSA electric power, for Second-stage second-order PSA power; The thermal energy system model includes a CSP model and a thermal energy storage module. CSP stands for Concentrated Solar Power, i.e., a solar thermal power plant. CSP (Chemical Solar Power Storage) comprises a receiver and a turbine. The receiver heats the salt to supply the collected solar heat to the turbine and thermal storage device. A CSP model includes: ; In the formula, for The heat absorption device's heat collection power is constantly increasing. for The heat power transferred from the heat absorber to the steam turbine at all times. for The heat power delivered by the heat absorber to the thermal energy storage device at all times; CSP generates steam to drive a steam turbine, which is mechanically connected to a generator to produce electricity. The CSP model also includes: ; In the formula, for CSP power generation at any time For the thermoelectric conversion efficiency of CSP, for The thermal power generated by the steam turbine at any given time; In addition, the thermal subsystem also uses electric heaters for auxiliary heating, which convert electrical energy into heat energy. The CSP model also includes modeling the electric heaters: ; In the formula, for The electric heater outputs heat power at all times. For thermoelectric conversion efficiency, for The electric power input to the electric heater at any given time.

[0045] The thermal energy storage module is: ; In the formula, for Real-time energy of thermal energy storage equipment for Real-time energy of thermal energy storage equipment The self-loss rate of thermal energy storage equipment. The heat charging and discharging efficiency of thermal energy storage equipment, for The charging power of the thermal energy storage device at all times. for The heat release capacity of the thermal energy storage device at all times. for The heat power transferred from the thermal energy storage device to the steam turbine at all times.

[0046] Based on the above system operation model, the estimated operating costs include the operating costs of power generation equipment, energy storage equipment, and hydrogen-electric equipment; the energy curtailment penalty costs include the wind and solar curtailment penalty costs and the load shedding penalty costs.

[0047] The operating cost of the power generation equipment is: ; ; In the formula, for The operating costs of wind and solar power generation at all times This represents the operating cost coefficient for wind and solar power generation. for The operating cost of solar thermal power generation at any given time This is the operating cost coefficient for concentrated solar power (CSP). for The amount of heat used for heating by the CSP at any given time; The operating cost of energy storage equipment is: ; ; ; In the formula, for Operating costs of instantaneous energy storage devices This is the operating cost coefficient for energy storage equipment. for Real-time operating costs of thermal energy storage equipment This is the operating cost coefficient for thermal energy storage equipment. for Operating costs of hydrogen energy storage equipment at any time This represents the operating cost coefficient for hydrogen energy storage equipment.

[0048] The operating cost of the hydrogen electrolysis equipment is: ; In the formula, for Real-time RSOC operating cost This represents the operating cost coefficient of RSOC in SOEC mode. This represents the operating cost coefficient of RSOC in SOFC mode.

[0049] The penalty cost for curtailing wind and solar power is: ; In the formula, for Constantly incurring penalties and costs for abandoning wind and solar power. for Wind power curtailment for Constantly wasting power; The cost of load shedding penalty is: ; In the formula, for Constantly cutting off load incurs penalties and costs. The load shedding penalty cost coefficient, The heat load penalty cost factor, The hydrogen load reduction penalty cost coefficient; Revenue from energy sales is: ; In the formula, for Revenue from electricity sales at all times For electricity sales price, for Revenue from heat sales at all times The price is for selling heat. for Revenue from hydrogen sales at all times This refers to the price of hydrogen sold.

[0050] In some embodiments, the sub-bar optimization model also includes constraints, including energy balance constraints, energy loss constraints, and output and operating state constraints.

[0051] The energy balance constraint is: ; ; ; ; In the formula, for Power purchased from the external power grid at all times for Constant electrical load power, for Constant heat load power, for The hydrogen mass flow rate released from the hydrogen storage tank at all times. for The hydrogen mass flow rate of the hydrogen storage tank at all times. for At any given time, the hydrogen load mass flow rate is [not specified]. The energy loss constraint is: ; ; Output and operating status constraints are as follows: ; In the formula, This is the minimum power generation capacity of the CSP. This represents the maximum power generation capacity of the CSP. for CSP power generation at any time Maximum ramp power for CSP; ; In the formula, for The state variable of RSOC in SOEC mode at any given time is a 0-1 variable, where 0 indicates that RSOC is not in SOEC mode, and 1 indicates that RSOC is in SOEC mode. for The maximum hydrogen production of RSOC in SOEC mode at time 10000. for The state variable for the RSOC's SOFC mode at any given time is a 0-1 variable, where 0 indicates that the RSOC is not in SOFC mode, and 1 indicates that the RSOC is in SOFC mode. for The maximum hydrogen consumption of RSOC in SOFC mode at any given time; ; In the formula, The rated capacity of the energy storage device. for The constant-time charging status variable for the energy storage device is a 0-1 variable, where 0 indicates the device is charging and 1 indicates it is not charging. This represents the maximum charging and discharging power per unit time. for The discharge working state variable of the energy storage device at any time is a 0-1 variable, where 0 indicates that the energy storage device is in the discharge working state and 1 indicates that the energy storage device is not in the discharge working state. ; In the formula, The rated capacity of the thermal energy storage equipment. for The variable representing the charging and operating status of the thermal energy storage device at any given time is a 0-1 variable, where 0 indicates that the thermal energy storage device is in a charging and operating state, and 1 indicates that the thermal energy storage device is not in a charging and operating state. This represents the maximum heat output per unit time. for The variable representing the heat release working state of the thermal energy storage device is a 0-1 variable, where 0 indicates that the thermal energy storage device is in the heat release working state and 1 indicates that the thermal energy storage device is not in the heat release working state. ; In the formula, This refers to the rated capacity of the hydrogen energy storage device. for The variable representing the hydrogen storage device's hydrogen charging status at any given time is a 0-1 variable, where 0 indicates the device is charging and 1 indicates it is not charging. This represents the maximum hydrogen charge / discharge power per unit time. for The variable representing the hydrogen release status of the hydrogen energy storage device at any given time is a 0-1 variable, where 0 indicates that the hydrogen energy storage device is in the hydrogen release state and 1 indicates that the hydrogen energy storage device is not in the hydrogen release state.

[0052] Optionally, decision variables include CSP power generation at any time , Hydrogen consumption at time RSOC , Real-time energy of instantaneous energy storage devices , Real-time energy of thermal energy storage equipment , Real-time hydrogen storage capacity of the hydrogen energy storage device , Electricity sales power at any time , Calories sold at any time and Hydrogen sales volume at any time .

[0053] The split-bulb optimization model includes a first-stage objective function and a second-stage objective function. The first-stage objective function represents the objective function of the day-ahead forecasting stage, and the second-stage objective function represents the objective function of the intraday deviation adjustment stage. The first-stage objective function aims to minimize the estimated operating cost; the second-stage objective function aims to minimize the deviation adjustment cost, and takes the supremum of the expected adjustment cost within the fuzzy set.

[0054] When solving the bilabial bar optimization model: In the day-ahead forecasting phase, the predicted values ​​of the decision variables are generated from the day-ahead forecasts: Based on the current initial state, the set of uncertain variables, and the system operation model, the first-stage objective function is solved to obtain the predicted value set of the decision variables.

[0055] Obtain predicted values ​​for uncertain variables; At the initial moment of the current scheduling cycle, based on the current initial state, the predicted values ​​of uncertain variables, the system operation model, and constraints, the objective function of the first stage is solved to obtain the predicted values ​​of the decision variables, in order to formulate the scheduling plan for the current scheduling cycle. At this time, the actual value of the prediction error has not yet been realized; that is, the predicted values ​​of the decision variables are determined before the uncertainty is realized. After solving, the predicted values ​​of the decision variables are: .

[0056] During the intraday deviation adjustment phase, as the current scheduling cycle progresses, the actual values ​​of the prediction errors of uncertain variables gradually become apparent. It is necessary to adjust the prediction errors of these uncertain variables according to the error adjustment formula to adjust the predicted values ​​of the decision variables from the day-ahead forecasting phase. The error adjustment formula is: ; In the formula, for Timing adjustment error for The actual value of the variable at any given time is uncertain. for Predicted values ​​of variables whose time is uncertain.

[0057] The goal of the intraday deviation adjustment phase is to address deviations with minimal adjustment cost, given the actual error value, to maintain the safe and stable operation of the system. This requires adjusting the system operation based on the predicted values ​​of the decision variables, with the objective of minimizing the deviation adjustment cost. The deviation adjustment cost is: ; In the formula, Represents the actual values ​​of decision variables. The function.

[0058] To adjust the error based on the actual values ​​of the decision variables The real-time response is based on the predicted values ​​of the decision variables, constructing an affine policy function; the affine policy function is: ; In the formula, After taking prediction error into account Time of the first The actual values ​​of the decision variables, for Time of the first The predicted values ​​of each decision variable. for Time of the first The response coefficients of each decision variable to the prediction error; The predicted value of the decision variable For example, the affine policy function is: ; In the formula, After taking prediction error into account Actual CSP power generation at any given time for The response coefficient of CSP power generation to prediction error at time t. After taking prediction error into account The actual hydrogen consumption at time RSOC. for The response coefficient of hydrogen consumption at time RSOC to prediction error. After taking prediction error into account Real-time actual energy value of the energy storage device. for The response coefficient of the instantaneous energy storage device to the prediction error. After taking prediction error into account Real-time actual energy value of thermal energy storage equipment. for The real-time energy response coefficient of a thermal energy storage device to prediction errors. After taking prediction error into account The actual real-time hydrogen storage capacity of the hydrogen energy storage device. for The response coefficient of the real-time hydrogen storage capacity of the hydrogen energy storage device to the prediction error. After taking prediction error into account The actual power output at any given time. for The response coefficient of the electricity sales power at any given time to the prediction error. After taking prediction error into account The actual calorie output at any given time. for The response coefficient of heat sales at any given time to prediction error. After taking prediction error into account The actual amount of hydrogen sold at any given time. for The response coefficient of hydrogen sales at any given time to prediction error.

[0059] In addition to the above-mentioned bias adjustment of decision variables, the intraday bias adjustment phase also adjusts the decision variables according to the error adjustment formula and the affine strategy function. , , , , , , and Adjust the deviation.

[0060] The constraints of the intraday deviation adjustment stage are the same as those of the day-ahead forecast stage. The predicted values ​​of the decision variables are replaced with the actual values ​​of the decision variables after considering the forecast error, and the objective function of the second stage is solved to obtain the decision value set of the current scheduling cycle.

[0061] Calculation example To verify the effectiveness of the proposed scheduling method for hydrogen cogeneration systems without conventional power supply, this example sets up five comparative scenarios, as shown in Table 1. In the table, "√" indicates that relevant components were considered, and "×" indicates that relevant components were not considered.

[0062] Table 1 Scene Comparison

[0063] Scenario 5 employs the traditional DRO scheduling method, where DRO stands for Distributed Robust Optimization. Scenario 3 uses the SO scheduling method, where SO stands for Stochastic Optimization. The SO scheduling method constructs a scenario tree based on 1000 Monte Carlo samples, assuming that the prediction error follows a normal distribution. , It is the standard deviation. Based on historical data estimates, Scenario 4 employs the RO scheduling method, where RO stands for Robust Optimization. The RO scheduling method is based on a box-type uncertainty set to describe the uncertainty of the source load.

[0064] Table 2 compares the economic indicators of the five scenarios. Scenario 1 achieves energy cascade utilization through CSP-RSOC coupling, directly replacing the electric heating demand of the SOEC mode with high-temperature molten salt thermal energy, significantly reducing the system's power consumption in the hydrogen production stage, lowering operating costs by 26.2% compared to Scenario 2, and increasing net profit to 984,700 yuan. Comparing Scenarios 1, 3, 4, and 5, Scenario 1 uses a robust optimization method considering extreme scenarios, which improves profits by 46.4% compared to an overly conservative robust optimization method, achieving a good balance between robustness and economy. Although Scenario 3 has the highest electricity sales revenue of 1,668,900 yuan, the deviation adjustment cost is as high as 297,400 yuan, the total cost climbs to 816,000 yuan, and the net profit drops to 853,200 yuan. This is because stochastic optimization relies on an exact probability distribution, which can easily lead to high adjustment costs when the uncertainty of source load is large. Scenario 4 incurs a high operating cost of 794,500 yuan, with a total cost of 994,700 yuan and a net profit of only 672,400 yuan. This is due to overly conservative robustness optimization, which allocates too much adjustment margin to cope with the worst-case scenario, sacrificing system economy. Although the net profit of Scenario 1 is 82,500 yuan lower than that of Scenario 5, this is a reasonable price to pay for improved robustness. Through further Monte Carlo simulation, the worst-case net profit of Scenario 1 is 820,000 yuan, while that of Scenario 5 may be as low as 650,000 yuan. Therefore, Scenario 1 achieves a significant reduction in risk by sacrificing a smaller expected profit, thus achieving an effective balance between economy and robustness.

[0065] Table 2. Economic Comparison of Five Scenarios

[0066] Table 3 compares the stability indicators of the five scenarios. Scenario 1 achieves a relatively good balance in multidimensional stability indicators. Regarding the fluctuation of transmitted power, Scenario 4 has the smallest standard deviation but an average transmitted power of only 97.01MW, indicating an overly conservative approach that weakens the transmission capacity. Scenario 3 has a high standard deviation of 18.76MW, with the highest transmitted power but also experiencing severe fluctuations and a load shedding rate of 6.63%. Scenario 1 has a transmitted power standard deviation of 12.35MW, which is moderate, while maintaining a relatively high transmitted power of 106.37MW. This indicates that the method controls fluctuations without excessively sacrificing transmission capacity, demonstrating the ability of bipolar optimization to balance robustness and economy. The energy storage device's state of charge (SOC) of 48.7% is close to the median level, preserving sufficient adjustment margin while avoiding deep cycling. With a combined renewable energy utilization rate of 98.17% and a load shedding rate of 2.97%, Scenario 1 effectively controls the risk of load shedding while ensuring a high proportion of renewable energy consumption.

[0067] Table 3. Stability Comparison of Five Scenarios

[0068] The comparison results show that in systems without conventional power supply support, the performance differences between different optimization methods are significantly amplified, mainly due to two reasons: First, limited regulation resources amplify the impact of the reserve strategy. The system's regulation capability relies entirely on the synergy of the solar thermal power plant, RSOC, and energy storage devices. In this context, Scenario 3 relies on a precise probability distribution, but the historical data of the new energy base is limited and exhibits distribution drift, resulting in insufficient reserves. When the actual deviation exceeds expectations, a costly second-stage adjustment or even load shedding is required. Scenario 4, to cope with the worst-case scenario, requires excessive regulation margins throughout the entire time period, forcing the solar thermal power plant to reduce its power generation output and sacrificing electricity sales revenue. Scenario 1 integrates extreme scenarios to cover tail risks and characterize distribution uncertainty. It optimizes the CSP-RSOC coupling strategy in the first stage and flexibly responds to deviations in the second stage through energy storage and RSOC mode switching, achieving the optimal configuration of reserve margin and response capability. Second, the cascade utilization of energy enhances the complexity of the trade-off between economy and robustness. The bidirectional heat exchange of CSP-RSOC enables deep coupling of the three subsystems of electricity, heat, and hydrogen. Under this structure, scheduling decisions not only need to balance the multi-energy flow in the current time period but also need to consider energy transfer paths across time periods. This spatiotemporal coupling characteristic requires optimization methods to consider both source-load time-series correlation and extreme scenarios when characterizing uncertainties. Scenario 3 relies on accurate distributions, and Scenario 4 ignores probabilistic information, both of which are difficult to effectively cope with this complex coupling. Scenario 1 retains the probabilistic characteristics of historical data and expands the distribution support set through extreme scenarios, enabling the first-stage decision-making to fully utilize the economic advantages of energy cascade utilization under robustness constraints. In summary, the results verify the applicability of Scenario 1 in systems without conventional power sources.

[0069] Figures 2 to 4 The scheduling results for Scenario 1 are shown, with the positive and negative directions of the Y-axis representing energy supply and demand, respectively. In the figure, RSOC-E indicates that the RSOC is in SOEC mode, and RSOC-F indicates that the RSOC is in SOFC mode. The results show that the split-bulb optimization method considering extreme scenarios effectively addresses source-load uncertainties. The power balance curve shows that the system achieves stable power output through the coordinated scheduling of the solar thermal power plant, RSOC, and energy storage equipment. During load ramp-up periods, the RSOC flexibly switches its operating mode, and the energy storage equipment responds quickly, effectively smoothing out new energy fluctuations. The heat balance curve shows that the solar thermal power plant achieves cross-period heat allocation through the heat storage equipment. The heat charge during the day is significantly higher than the heat release at night, and the state of charge of the heat storage is basically the same at 0:00 and 24:00, satisfying the intraday energy conservation. Regarding hydrogen balance, the RSOC continuously produces hydrogen during periods of sufficient sunlight, and the hydrogen storage capacity of the hydrogen storage equipment increases in a stepwise manner. At night, the hydrogen storage equipment releases hydrogen to support RSOC power generation, and the hydrogen storage capacity of the hydrogen storage equipment gradually decreases to the initial level, forming a complete closed loop.

[0070] Those skilled in the art will understand that embodiments of this application can be provided as methods, apparatus (systems), or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of this application can be implemented in various computer languages, such as the object-oriented programming language Java and the interpreted scripting language JavaScript.

[0071] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce an instruction that executes via the processor of the computer or other programmable data processing apparatus to create an instruction for implementing the flowchart. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0072] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0073] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0074] Finally, it should be noted that the above content is only used to illustrate the technical solution of the present invention, and is not intended to limit the scope of protection of the present invention. Simple modifications or equivalent substitutions made by those skilled in the art to the technical solution of the present invention do not depart from the essence and scope of the technical solution of the present invention.

Claims

1. A scheduling method for a hydrogen cogeneration system without conventional power source support, characterized in that, include: Obtain the historical dataset, set of uncertain variables, and current initial state of the hydrogen cogeneration system, wherein the set of uncertain variables represents the collection of uncertain variables; A system operation model was constructed based on a hydrogen-powered combined heat and power system. A first error set is obtained based on the system operation model and historical dataset. The first error set represents the set of first errors, which represent the prediction errors of decision variables. A second error set is obtained based on the system operation model, the historical dataset, and the set of uncertain variables using the Copula rule and Rosenblatt transformation. The second error set represents the set of prediction errors for extreme scenario points, where the extreme scenario points represent scenarios where the uncertain variables simultaneously take extreme values. A fusion set is obtained based on the first error set and the second error set; The asymmetric directional Wasserstein metric is used, and a fuzzy set of the first error probability distribution is constructed based on the fusion set. A two-stage sub-Bruker optimization model is constructed based on the system operation model and the fuzzy set. The two stages include a day-ahead prediction stage and an intraday deviation adjustment stage. The current decision value set is obtained by solving the two-stage sub-Blu-ray bar optimization model based on the current initial state, the set of uncertain variables, and the system operation model. The current decision value set represents the set of values ​​of the decision variables in the current scheduling period.

2. The scheduling method for a hydrogen cogeneration system without conventional power source support according to claim 1, characterized in that, Using the Copula rule and Rosenblatt transform, a second error set is obtained based on the system operation model, the historical dataset, and the set of uncertain variables, including: A third error set is obtained based on the system operation model, the historical dataset, and the set of uncertain variables. The third error set represents the set of third errors, and the third error represents the prediction error of the uncertain variables. Based on the third error set, an edge probability distribution set is obtained by fitting the edge distribution; The Copula model is constructed using the Copula model based on the marginal probability distribution set. The second error set is obtained by using the Rosenblatt transform and the Copula model.

3. The scheduling method for a hydrogen cogeneration system without conventional power source support according to claim 2, characterized in that, The process involves constructing a fuzzy set of the first error probability distribution based on the fusion set using an asymmetric Wasserstein metric, including: The terrain cost function is obtained based on the fusion set, and the terrain cost function characterizes the asymmetric directional measure between the actual error and the prediction error of the decision variables; A distance measurement model is constructed based on the terrain cost function using the Wasserstein metric. The distance measurement model is used to solve the asymmetric Wasserstein distance between the empirical probability distribution and the true distribution of the first error. The fuzzy set is obtained based on the distance measurement model.

4. The scheduling method for a hydrogen cogeneration system without conventional power supply support according to claim 3, characterized in that, A two-stage sub-Bruker optimization model is constructed using the dual transformation method based on the system operation model and the fuzzy set.

5. The scheduling method for a hydrogen cogeneration system without conventional power source support according to claim 1, characterized in that, Uncertain variables include new energy output and electricity load.

6. The scheduling method for a hydrogen cogeneration system without conventional power source support according to claim 5, characterized in that, The new energy output includes wind power output and photovoltaic power output.

7. The scheduling method for a hydrogen cogeneration system without conventional power source support according to claim 1, characterized in that, The two-stage sub-Blu-ray optimization model includes an objective function, which represents taking the supremum of the expected deviation adjustment cost in the intraday deviation adjustment stage within the fuzzy set, while minimizing the estimated operating cost in the intraday deviation adjustment stage.

8. The scheduling method for a hydrogen cogeneration system without conventional power source support according to claim 7, characterized in that, The estimated operating cost represents the difference between the total operating cost and the revenue from energy sales, and the total operating cost represents the sum of equipment operating costs and energy curtailment / disruption penalty costs; The deviation adjustment cost represents the difference between the total operating deviation adjustment cost and the energy sales deviation revenue. The total operating deviation adjustment cost represents the sum of the equipment operating deviation adjustment cost and the energy curtailment and power loss penalty deviation cost.

9. The scheduling method for a hydrogen cogeneration system without conventional power source support according to claim 8, characterized in that, The sub-bar optimization model also includes constraints, including energy balance constraints, energy loss constraints, and output and operating state constraints.

10. The scheduling method for a hydrogen cogeneration system without conventional power source support according to any one of claims 1-9, characterized in that, The hydrogen energy combined heat and power system includes an electrical energy subsystem, a thermal energy subsystem, and a hydrogen energy subsystem; The power subsystem is used for main power generation based on new energy generator sets, and for charging and discharging electrical energy. The thermal energy subsystem is used for main heating and auxiliary power generation based on the solar thermal power plant, auxiliary heating based on the electric heater, and charging and discharging of thermal energy. The hydrogen energy subsystem is used for hydrogen production based on a reversible solid oxide battery, and for charging and discharging hydrogen. The reversible solid oxide battery operates in two modes: a solid oxide electrolyzer mode and a solid oxide fuel cell mode. When the reversible solid oxide battery operates in solid oxide electrolysis mode, the thermal energy subsystem provides thermal energy to the hydrogen energy subsystem; when the reversible solid oxide battery operates in solid oxide fuel cell mode, the hydrogen energy subsystem provides thermal energy to the thermal energy subsystem.

Images