Optimization method and device of electrothermal hydrogen hybrid energy storage system for coping with extreme scene

Abstract

The invention relates to the technical field of electricity, heat and hydrogen co-production and hybrid energy storage systems, and particularly discloses an optimization method and device of an electricity, heat and hydrogen hybrid energy storage system for coping with extreme scenes. Comprising the following steps: identifying a core uncertainty source of the electrothermal hydrogen hybrid energy storage system; quantifying the random characteristics of the uncertainty source; performing uncertainty influence analysis based on Monte Carlo simulation, including sample generation, model introduction and index calculation and risk scene identification; constructing a two-stage random robust capacity optimization model; solving the model and verifying robustness; according to the method, the influence of three uncertainties of wind and light output prediction error, load fluctuation and equipment parameter drift on the economy and reliability of the system is emphatically solved, so that the ability of the system to cope with extreme scenes is improved; uncertain sources are fully covered, and the system identifies and quantifies three types of core uncertainties of wind and light output, load and equipment parameters for the first time; the method can achieve the balance of robustness and economy, and is higher in engineering practicability.

Description

Technical Field

[0001] This invention relates to the field of electrothermal-hydrogen cogeneration and hybrid energy storage systems, specifically to an optimization method and apparatus for an electrothermal-hydrogen hybrid energy storage system to cope with extreme scenarios. Background Technology

[0002] Hybrid electrothermal-hydrogen energy storage systems, capable of combining power regulation and energy storage, have become a key technology for mitigating the fluctuations in new energy sources. Among existing technologies, some scholars aim to maximize daily net profit by combining battery life, molten salt thermal storage, and electrolytic hydrogen production constraints to achieve capacity optimization. However, this approach is based on deterministic scenario assumptions and fails to consider three core uncertainties in actual operation: ① Uncertainty in wind and solar power output, i.e., random variations in wind speed and solar irradiance lead to deviations between actual and predicted output, easily causing power supply and demand imbalances in the system; ② Uncertainty in load, i.e., changes in residential electricity consumption habits and industrial production plans cause random fluctuations in base load and time-shiftable load, further exacerbating supply and demand mismatches; ③ Uncertainty in equipment parameters, i.e., battery life index drifts ± equipment parameters due to temperature effects, and electrolytic hydrogen production energy consumption drifts ± response drift due to operation and maintenance levels, resulting in shortened planned battery life and increased hydrogen production costs, leading to the problem of "optimistic planning."

[0003] The aforementioned uncertainties can cause actual system performance indicators (such as daily net revenue, battery life, and renewable energy utilization rate) to deviate significantly from planned values. In extreme scenarios, this could even lead to safety incidents such as exceeding power deviation limits or oxygen depletion in aquatic organisms (if coupled with aquaculture scenarios). Therefore, there is an urgent need for a method to quantify the impact of uncertainties and achieve robust capacity optimization, so as to ensure that the system can still maintain both economic efficiency and reliability under extreme scenarios. Summary of the Invention

[0004] The purpose of this invention is to provide an optimized method and apparatus for an electrothermal hydrogen hybrid energy storage system to cope with extreme scenarios, and to solve the following technical problems: Existing capacity optimization technologies for electrothermal-hydrogen hybrid energy storage systems do not consider the shortcomings of uncertainty and insufficient robustness. They are particularly suitable for capacity optimization and operation of electrothermal-hydrogen hybrid energy storage systems in the context of high renewable energy penetration, and can improve the economy and reliability of the system in extreme scenarios.

[0005] The objective of this invention can be achieved through the following technical solutions: An optimization method for an electrothermal-hydrogen hybrid energy storage system to cope with extreme scenarios includes the following steps: Step 1: Identify the core uncertainty sources of the electrothermal-hydrogen hybrid energy storage system. The core uncertainty sources include wind and solar power output uncertainty, load uncertainty, and equipment parameter uncertainty. Step 2: Quantify and statistically analyze the core uncertainty sources through Monte Carlo simulation. The quantification of the core uncertainty sources includes quantifying the uncertainty of wind and solar power output, quantifying the uncertainty of load, and quantifying the uncertainty of equipment parameters. Step 3: Randomly sample according to the probability distribution parameters of the quantified core uncertainty sources, generate samples, and input the samples into the basic optimization model for simulation analysis and identification of risk scenarios; Step 4: Use a sliding window to run historical risk scenario data in real time to train the probability distribution correction model, and use an online learning algorithm to correct and obtain the probability distribution parameters of the uncertainty source; Step 5: Construct a two-stage stochastic robust capacity optimization model with uncertain budget based on the simulation analysis results of the basic optimization model; Step 6: Solve the two-stage stochastic robust capacity optimization model and verify its robustness.

[0006] Preferably, the uncertainty of wind and solar power output is due to the random changes in wind speed and solar irradiance, which manifests as the deviation between the actual output of wind and solar power and the predicted value; the uncertainty of load is due to the random fluctuations in electricity demand, which is divided into non-transferable base load and transferable flexible load; the uncertainty of equipment parameters is due to the performance drift of energy storage batteries and electrolysis hydrogen production equipment, which manifests as the deviation of battery life index and electrolysis hydrogen production energy consumption from the rated value.

[0007] Preferably, to quantify the uncertainty of wind and solar power output, the expression for the actual wind and solar power output is:

[0008]

[0009] in, , They are respectively t The actual contribution of wind and light at all times; , They are respectively t The predictive power of wind and sunlight at all times; , These are the relative errors of wind and solar power output, respectively. =0.2 and =0.15 represents the standard deviation of wind and solar power errors, respectively; The expression for quantifying the base load uncertainty is as follows:

[0010] The expression for time-shiftable load quantization is:

[0011] in, , They are respectively tActual values ​​of base load and time-shiftable load at any given time; , They are respectively t Forecast values ​​of base load and time-shiftable load at any given time; =0.05、 =0.1 represents the amplitude of fluctuations in base load and time-shiftable load, respectively; U It is uniformly distributed.

[0012] Preferably, generating samples refers to generating samples based on the given sources of uncertainty. The method for calculating the probability distribution parameters of the uncertainty source is as follows: To quantify the uncertainty of equipment parameters, the expression for quantifying the battery life index is:

[0013] in, , These are the battery's actual life index and rated life index, respectively. =2; This is the battery life index, with a value of 0.1. The distribution follows a triangular pattern, where a is the minimum value, b is the most likely value, and c is the maximum value. The expression for quantifying the energy consumption of hydrogen production by electrolysis is:

[0014] in, , These are the actual energy consumption and rated energy consumption for hydrogen production via electrolysis, respectively. =5 kWh / Nm 3 ; The value is 0.05, representing the energy consumption drift of hydrogen production via electrolysis.

[0015] Preferably, the standard deviation correction for wind and solar errors is calculated by dividing the wind speed range into intervals, and the statistical method is as follows: Historical data is divided into three intervals based on wind speed: low wind speed interval 1, medium wind speed interval 2, and high wind speed interval 3. The formula for calculating the relative error of the intervals in low wind speed interval 1, medium wind speed interval 2, and high wind speed interval 3 is as follows:

[0016] in, , Indicates an interval; Correction for standard deviation of wind and light error:

[0017] in, For interval The number of samples, ,and ∈ ; Correction for the triangular distribution parameters of the battery life index: First, calculate the moving average of the prediction error:

[0018] in, The mean absolute error of the battery life index; This indicates the total number of days in a quarter. It represents one day; The prediction error for a single day represents the difference between the actual life index and the predicted life index. Secondly, update the triangular distribution parameters:

[0019]

[0020]

[0021] in, This is a new value from a triangular distribution. The old value is a triangular distribution; These are possible values ​​from a triangular distribution. Rated life index; It is the maximum value of the triangular distribution; and ≥0.85 , <1.15 ; Correction of the triangular distribution parameters for energy consumption in hydrogen production by electrolysis:

[0022]

[0023]

[0024] in, The actual energy consumption for hydrogen production by electrolysis; Rated energy consumption for hydrogen production by electrolysis; This represents the minimum energy consumption. This represents the median energy consumption. This represents the maximum energy consumption. It is the minimum value of the triangular distribution. These are possible values ​​from a triangular distribution. It represents the maximum value of the triangular distribution.

[0025] Preferably, the variables in the two-stage stochastic robust capacity optimization model for uncertain budgeting include deterministic variables, deterministic random variables, and deterministic robust parameters; (1) The known variables are: battery capacity E B MWh; battery power P B MW; molten salt thermal storage capacity E TS MWh; Electrolyzer power P H2 MW; (2) The random variables to be determined are: actual output of wind and solar power. , Actual load Battery life index Energy consumption for hydrogen production by electrolysis ; (3) The robust parameters to be determined include: uncertainty budget , of which 0 ≤ where represents the maximum number of responses. Extreme uncertainty at any given moment; The objective function is defined as follows:

[0026] in, The daily depreciation cost of the first phase of energy storage investment is calculated using the following formula:

[0027] in, For battery capacity, the value is taken as 1500 yuan / (kWh); The power cost is set at 500 yuan / kW; The cost of molten salt thermal storage capacity is set at 200 yuan / (kWh); The power cost of the electrolytic cell is set at 800 yuan / kW; Y The system lifespan is set to 20 years. Let E be a vector of random variables; E[the expected operating cost of the second phase of the mechanism, including maintenance costs]. Battery replacement cost Losses from wind and solar power curtailment ; λ This is the robustness penalty coefficient, used to balance economy and robustness, and its value is 0.3; The conditional value of risk at the ββ3 penalty confidence level is used to measure extreme costs, and its calculation formula is as follows:

[0028] in, v For CVaR auxiliary variables; Core constraints are given for the robustness of the integration, including robust constraints for wind-solar-load matching to cope with extreme supply-demand imbalances, the expression of which is: st in, This refers to the battery discharge power. Battery charging power; For the output of hydrogen fuel cells; The heating power for molten salt thermal storage; This is an indicator function that takes the value 1 when the condition is met. The threshold for extreme load fluctuations is set to 0.1.

[0029] Preferably, it also includes a robust constraint on battery life, requiring a lifespan of not less than 10 years at a 95% confidence level, expressed as:

[0030] in, This refers to the actual cycle life of the battery, in years. This represents the number of cycles at 100% battery discharge depth, with a value of 3000. This represents the actual depth of battery discharge. It is a probability function; Including robust constraints on hydrogen production via electrolysis, requiring that hydrogen production capacity can meet demand under extreme energy consumption, expressed as:

[0031] in, The maximum energy consumption for hydrogen production by electrolysis is taken as 5.25 kWh / Nm³. 3 ; The minimum daily hydrogen production requirement needs to be determined based on the hydrogen load.

[0032] Preferably, a sample average approximation + GAMS / BARON approach is used to solve the two-stage stochastic robust model. The specific steps are as follows: While considering both accuracy and computational speed, generate... M =500 random samples; the expected value and CVaR are transformed into sample mean and quantiles to establish a deterministic equivalent model; a hierarchical solution method is adopted: in the first stage, the energy storage capacity is optimized; in the second stage, the scheduling strategy is optimized for each sample, iterating until the objective function converges, where the convergence threshold is... ε =10 -4 ; The steps to verify robustness are as follows: compare the core indicators of the robust optimization scheme with those of the deterministic scheme, analyze the impact of the uncertain budget Γ on robustness and economy, and thus determine the optimal Γ.

[0033] An optimization device for an electrothermal-hydrogen hybrid energy storage system designed for extreme scenarios, comprising: The core uncertainty source identification module is used to identify uncertainties in wind and solar power output, load uncertainty, and equipment parameter uncertainty; The uncertainty source random characteristic quantification module is used to quantify the uncertainty of wind and solar power output, load uncertainty, and equipment parameter uncertainty; The uncertainty impact analysis module is used to generate samples according to the probability distribution parameters of uncertainty sources, input them into the model, calculate indicators, and identify risk scenarios. The scenario adaptation module is used to train a probability distribution correction model by running historical risk scenario data in real time using a sliding window, and corrects and obtains the probability distribution parameters of uncertainty sources through online learning algorithms. A two-stage stochastic robust capacity optimization module is used to define model variables, determine the objective function, and provide the core constraints for fusion robustness. The Solving and Robustness Verification module is used to solve a two-stage stochastic robust model and verify its robustness.

[0034] The beneficial effects of this invention are: Compared with the prior art, the beneficial effects of the present invention are as follows: (1) The capacity optimization configuration of the electrothermal hydrogen hybrid energy storage system is integrated into uncertainty analysis and robust optimization. The focus is on solving the impact of three types of uncertainties on the system's economy and reliability: wind and solar power output prediction error, load fluctuation, and equipment parameter drift, thereby improving the system's ability to cope with extreme scenarios. At the same time, the model probability distribution parameters are adjusted quarterly through dynamic error analysis, expanding the model's applicability.

[0035] (2) It comprehensively covers the sources of uncertainty, and for the first time systematically identifies and quantifies the three core uncertainties of wind and solar power output, load and equipment parameters, avoiding optimization deviations caused by single uncertainty analysis; (3) Achieving a balance between robustness and economy: By using uncertainty budgeting and CVaR (conditional value at risk of operating costs) penalty terms, a balance is achieved between "reliability assurance in extreme scenarios" and "economic optimization in normal scenarios". The daily net income loss at a 95% confidence level is reduced from 18.2% to 1.3%. (4) It has strong engineering practicality. Based on the existing electrothermal hydrogen hybrid energy storage system framework, the model solving tools are mature (GAMS / BARON is commonly used) and can be directly embedded into the existing scheduling platform. At the same time, the robust parameter Γ can be flexibly adjusted depending on the degree of regional uncertainty. It is particularly suitable for Northwest China, North China and other places where the penetration of new energy is very high.

[0036] Of course, any product implementing this invention does not necessarily need to achieve all the advantages described above at the same time. Attached Figure Description

[0037] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0038] Figure 1 This is a schematic diagram of the process of an electrothermal hydrogen hybrid energy storage system in an embodiment of the present invention, which uses uncertainty analysis and robust capacity optimization to cope with extreme scenarios.

[0039] Figure 2 This is a system block diagram illustrating how uncertainty analysis and robust capacity optimization are used to address extreme scenarios in an embodiment of the invention. Detailed Implementation

[0040] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0041] Example See Figure 1 This embodiment proposes an electrothermal hydrogen hybrid energy storage system device and method to cope with extreme scenarios through uncertainty analysis and robust capacity optimization. The technical solution adopted includes the following steps: S1 identifies the core uncertainty sources of the electrothermal-hydrogen hybrid energy storage system, including uncertainties in wind and solar power output, load uncertainty, and equipment parameter uncertainty.

[0042] Among them, the uncertainty of wind and solar power output is due to the random changes in wind speed and solar irradiance, which manifests as the deviation between the actual output of wind and solar power and the predicted value; the uncertainty of load is due to the random fluctuation of electricity demand, which is divided into non-time-shiftable base load (such as residential lighting) and time-shiftable flexible load (such as industrial electricity); the uncertainty of equipment parameters is due to the performance drift of energy storage batteries and electrolysis hydrogen production equipment, which manifests as the deviation of battery life index and electrolysis hydrogen production energy consumption from the rated value.

[0043] S2, quantifies the stochastic characteristics of uncertainty sources; For the uncertainty sources identified in S1, their random characteristics are quantified by combining historical operating data and probability distribution theory.

[0044] S2.1, Quantifying the uncertainty of wind and solar power output; The actual power output of wind and solar power is defined as the predicted value × (1 + relative error), where the relative error follows a normal distribution, and the expression is as follows:

[0045] In the formula, , They are respectively t The actual contribution of wind and light at all times; , They are respectively t The predictive power of wind and sunlight at all times; , These are the relative errors of wind and solar power output, respectively. =0.2 and =0.15 represents the standard deviation of wind and solar power errors, respectively.

[0046] S2.2, Quantifying load uncertainty; The load is divided into basic load and time-shiftable load, which are quantized using interval distribution and uniform distribution, respectively.

[0047] (1) Quantification of basic load Using interval distribution, the expression is as follows: (3) (2) Time-shiftable load quantization Using a uniform distribution, the expression is as follows: (4) in, , They are respectively t Actual values ​​of base load and time-shiftable load at any given time, in MW; , They are respectively t Forecast values ​​of base load and time-shiftable load at any given time, in MW; =0.05、 =0.1 represents the amplitude of fluctuations in base load and time-shiftable load, respectively; U It is uniformly distributed.

[0048] S2.3, Quantifying the uncertainty of equipment parameters; The drift of the battery life index and the energy consumption of hydrogen production by electrolysis were quantified using a triangular distribution.

[0049] (1) Quantification of battery life index (5) in, , These are the battery's actual life index and rated life index, respectively. =2; This is the battery life index, with a value of 0.1. It follows a triangular distribution, where a is the minimum value, b is the most likely value, and c is the maximum value.

[0050] (2) Energy consumption quantification of hydrogen production by electrolysis (6) in, , These are the actual energy consumption and rated energy consumption for hydrogen production via electrolysis, respectively. =5 kWh / Nm 3 ; The value is 0.05, representing the energy consumption drift of hydrogen production via electrolysis.

[0051] S3, Analysis of the impact of uncertainties based on Monte Carlo simulation; The impact of uncertainty on the system's core indicators is quantified using Monte Carlo simulation, following these steps: S3.1, Generate samples; The sources of uncertainty are given in S2. Random sampling is performed according to its probability distribution, and the number of sampling times is... N =1000 to ensure statistical significance.

[0052] S3.2, Substitute into the model and calculate the indicators; Each time, random parameters are assigned to the basic optimization model. The goal of this model is to maximize the daily comprehensive benefit, which includes constraints on new energy output, battery life, molten salt thermal storage, and electrolytic hydrogen production. The model is used to calculate the system's daily comprehensive net benefit, battery life, new energy utilization rate, and power deviation rate.

[0053] S3.3, Identify risk scenarios; Statistical analysis of 1000 simulation results identifies extreme risk scenarios, such as daily net profit loss and battery life reduction at the 95th percentile, and clarifies the types of uncertainties that need to be addressed for robust optimization.

[0054] S4. A sliding window is used to train a probability distribution correction model using historical risk scenario data in real time. The probability distribution parameters of uncertainty sources are corrected and obtained through an online learning algorithm. After identifying the risk scenarios in step S3, the probability distribution parameters are corrected through an online learning algorithm based on real-time data and a sliding window for one quarter (3 months) using adaptive scenario settings. This is to adjust the quantitative values ​​of uncertainty sources, ensure that the real-time updated probability distribution parameters meet the actual operating index requirements, and improve the seasonal adaptability of the model.

[0055] S4.1, the standard deviation correction for wind and solar errors is calculated by dividing the data into intervals according to wind speed levels. The statistical method is as follows: Historical data is divided into three intervals based on wind speed: low wind speed interval 1, medium wind speed interval 2, and high wind speed interval 3. The formula for calculating the relative error of the intervals in low wind speed interval 1, medium wind speed interval 2, and high wind speed interval 3 is as follows: (7) in, , Indicates an interval; Correction for standard deviation of wind and light error: (8) in, For interval The number of samples, ,and ∈ ; Correction for the triangular distribution parameters of the battery life index: First, calculate the moving average of the prediction error: (9) in, The mean absolute error of the battery life index; This indicates the total number of days in a quarter. It represents one day; The prediction error for a single day represents the difference between the actual life index and the predicted life index. Secondly, update the triangular distribution parameters:

[0056]

[0057] (10) in, This is a new value from a triangular distribution. The old value is a triangular distribution; These are possible values ​​from a triangular distribution. Rated life index; It is the maximum value of the triangular distribution; and ≥0.85 , <1.15 ; Correction of the triangular distribution parameters for energy consumption in hydrogen production by electrolysis:

[0059]

[0060] (11) in, The actual energy consumption for hydrogen production by electrolysis; Rated energy consumption for hydrogen production by electrolysis; This represents the minimum energy consumption. This represents the median energy consumption. This represents the maximum energy consumption. It is the minimum value of the triangular distribution. These are possible values ​​from a triangular distribution. It represents the maximum value of the triangular distribution.

[0062] S5. Establish a two-stage stochastic robust capacity optimization model; A two-stage stochastic robust capacity optimization model with uncertain budget is established, aiming to balance system economy and reliability under extreme scenarios. It is divided into the first stage (capacity configuration decision) and the second stage (real-time scheduling response).

[0063] S5.1, Define model variables; (1) Determined variables (first-stage decision-making) include: battery capacity E B MWh; battery power P B MW; molten salt thermal storage capacity E TS MWh; Electrolyzer power P H2 ,MW.

[0064] (2) Determine the random variables (second-stage parameters), including: actual wind and solar power output. , Actual load Battery life index Energy consumption for hydrogen production by electrolysis .

[0065] (3) Determine the robust parameters, including: uncertainty budget , of which 0 ≤ where represents the maximum number of responses. Extreme uncertainty at any given moment.

[0066] S5.2, Determine the objective function; The objective function aims to maximize the overall net return on the robust day. Considering the adjustment costs in the second stage due to uncertainty, the objective function is expanded to minimize the sum of the investment cost in the first stage, the expected operating cost in the second stage, and the robustness penalty term, as shown in the following expression: (12) in, The daily depreciation cost of the first phase of energy storage investment is calculated using the following formula: (13) In the formula, For battery capacity, the value is taken as 1500 yuan / (kWh); The power cost is set at 500 yuan / kW; The cost of molten salt thermal storage capacity is set at 200 yuan / (kWh); The power cost of the electrolytic cell is set at 800 yuan / kW; Y The system lifespan is set to 20 years. Let E be a vector of random variables; E[the expected operating cost of the second phase of the mechanism, including maintenance costs]. Battery replacement cost Losses from wind and solar power curtailment ; λ This is the robustness penalty coefficient, used to balance economy and robustness, and its value is 0.3; for β The conditional value of risk at a confidence level of 0.95 is used to measure extreme costs, and its specific calculation formula is as follows: (14) v For CVaR auxiliary variables; S5.3 provides the core constraints and requires robust integration. (1) To cope with extreme supply and demand imbalances, a robust constraint for wind-solar-load matching is given, the expression of which is as follows: (15) st (16) in, This refers to the battery discharge power. Battery charging power; For the output of hydrogen fuel cells; The heating power for molten salt thermal storage; This is an indicator function that takes the value 1 when the condition is met. The threshold for extreme load fluctuations is set to 0.1.

[0067] (2) Robust constraint on battery life, requiring a lifespan of no less than 10 years at a 95% confidence level, is expressed as follows: (17) in, This refers to the actual cycle life of the battery, in years. This represents the number of cycles at 100% battery discharge depth, with a value of 3000. This represents the actual depth of battery discharge. Let be a probability function.

[0068] (3) Robust constraints for electrolytic hydrogen production require that the hydrogen production capacity can meet the demand under extreme energy consumption, as expressed in the following expression: (18) in, The maximum energy consumption for hydrogen production by electrolysis is taken as 5.25 kWh / Nm³. 3 ; The minimum daily hydrogen production requirement needs to be determined based on the hydrogen load.

[0069] S6, Solve the model and verify robustness; S6.1, Solution algorithm; This patent employs Sample Average Approximation (SAA) + GAMS / BARON to solve a two-stage stochastic robust model. The specific steps are as follows: ① Taking into account both accuracy and computational speed, generate M =500 random samples; ② Transform the expected value and CVaR into sample mean and quantiles to establish a deterministic equivalent model; ③ A hierarchical solution method is adopted: In the first stage, the energy storage capacity is optimized; in the second stage, the scheduling strategy is optimized for each sample, iterating until the objective function converges, where the convergence threshold is... ε =10 -4 .

[0070] S6.2, verify robustness; By comparing the core metrics (such as daily net revenue, power deviation rate, and battery life) of the robust optimization scheme with those of the deterministic scheme, we analyze the impact of the uncertain budget Γ on robustness and economy, thereby determining the optimal Γ. Here, we recommend Γ here, where robustness is significantly improved and economic loss is controlled within 6%.

[0071] Example 2 See Figure 2 An optimization device for an electrothermal-hydrogen hybrid energy storage system designed for extreme scenarios, comprising: The core uncertainty source identification module is used to identify uncertainties in wind and solar power output, load uncertainty, and equipment parameter uncertainty; The uncertainty source random characteristic quantification module is used to quantify the uncertainty of wind and solar power output, load uncertainty, and equipment parameter uncertainty; The uncertainty impact analysis module is used to generate samples according to the probability distribution parameters of uncertainty sources, input them into the model, calculate indicators, and identify risk scenarios. The scenario adaptation module is used to train a probability distribution correction model by running historical risk scenario data in real time using a sliding window, and corrects and obtains the probability distribution parameters of uncertainty sources through online learning algorithms. A two-stage stochastic robust capacity optimization module is used to define model variables, determine the objective function, and provide the core constraints for fusion robustness. The Solving and Robustness Verification module is used to solve a two-stage stochastic robust model and verify its robustness.

[0072] Example 3 Based on the same inventive concept, embodiments of the present invention also provide a computer-readable storage medium storing at least one instruction, at least one program, code set, or instruction set, wherein at least one instruction, at least one program, code set, or instruction set is loaded and executed by a processor to implement the optimization method and apparatus for an electrothermal hydrogen hybrid energy storage system for extreme scenarios as described in Embodiment 1.

[0073] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing related hardware. This program can be stored in a computer-readable storage medium, including read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electronically erasable rewritable read-only memory (EEPROM), compact disc readable read-on (CDmpac) or other optical disc storage, disk storage, magnetic tape storage, or any other computer-readable medium capable of carrying or storing data.

[0074] Since the storage medium is the storage medium of the optimized method and device for an electrothermal hydrogen hybrid energy storage system in extreme scenarios according to an embodiment of the present invention, and the principle of the storage medium in solving the problem is similar to that of the method, the implementation of the storage medium can refer to the implementation process of the above-described method embodiment one, and the repeated parts will not be described again.

[0075] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the embodiments of apparatus, devices, and non-volatile computer storage media are basically similar to the method embodiments, and therefore described more simply; relevant parts can be referred to the descriptions of the method embodiments.

[0076] The above content is merely an example and illustration of the concept of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described or use similar methods to replace them, as long as they do not deviate from the concept of the invention or exceed the scope defined in this application, they should all fall within the protection scope of the present invention.

Claims

1. A method for optimizing an electro-thermal hydrogen hybrid energy storage system for extreme scenarios, characterized in that, The method comprises the following steps: Step one, identifying the core uncertainty source of the electric heat hydrogen hybrid energy storage system, wherein the core uncertainty source comprises wind and light output uncertainty, load uncertainty and equipment parameter uncertainty; Step two, quantitatively analyzing and counting the core uncertainty source by Monte Carlo simulation, wherein the quantification of the core uncertainty source comprises quantification of wind and light output uncertainty, quantification of load uncertainty and quantification of equipment parameter uncertainty; Step three, randomly sampling according to the probability distribution parameters of the quantified core uncertainty source, generating samples and bringing the samples into a basic optimization model for simulation analysis and risk scenario identification; Step four, training a probability distribution correction model by using a sliding window real-time operation historical risk scenario data, and correcting and obtaining the probability distribution parameters of the uncertainty source by an online learning algorithm; Step five, constructing a two-stage random robust capacity optimization model with uncertainty budget based on the simulation analysis results of the basic optimization model; Step six, solving the two-stage random robust capacity optimization model and verifying the robustness.

2. The method of claim 1, wherein, The wind and light output uncertainty is caused by the random change of wind speed and solar irradiance, and is manifested as the deviation of actual wind power and photovoltaic output from the predicted value; the load uncertainty is caused by the random fluctuation of electricity demand, and is divided into basic load which cannot be time-shifted and flexible load which can be time-shifted; the equipment parameter uncertainty is caused by the performance drift of energy storage batteries and electrolytic hydrogen production equipment, and is manifested as the deviation of battery life index and electrolytic hydrogen production energy consumption from the rated value.

3. The method of claim 1, wherein the optimization method is used for an electro-thermal hydrogen hybrid energy storage system for extreme scenarios, characterized in that, The quantification of wind and light output uncertainty is expressed as: where , are the actual wind and solar power at time t , are the predicted wind and solar power at time t , are the relative errors of wind and solar power; = 0.2 and = 0.15 are the standard deviations of wind and solar power errors.​​ The quantification of basic load is expressed as: The quantification of time-shifted load is expressed as: Wherein, , are respectively t actual values of the base load and the time-shiftable load at the time point t; , are respectively t predicted values of the base load and the time-shiftable load at the time point t; =0.05, =0.1 are respectively amplitudes of the base load and the time-shiftable load fluctuation; U is a uniform distribution.

4. The method for optimizing an electro-thermal hydrogen hybrid energy storage system for extreme scenarios of claim 3, wherein, The generating sample refers to obtaining the probability distribution parameters of the given uncertainty source The calculation manner of the probability distribution parameters of the uncertainty source is that The quantification of battery life index is expressed as: wherein, , are the actual and rated life indices of the battery, respectively, = 2; is the battery life index, taking the value 0.1; is a triangular distribution, a is the minimum value, b is the most likely value, and c is the maximum value. The quantification of electrolytic hydrogen production energy consumption is expressed as: wherein, , are the actual and rated energy consumption of electrolytic hydrogen production, respectively, = 5 kWh / Nm 3 ; is the energy consumption drift amplitude of electrolytic hydrogen production, taking the value 0.

05.

5. The optimization method of the electro-thermal hydrogen hybrid energy storage system for extreme scenarios according to claim 4, characterized in that, The wind and light error standard deviation correction is statistically calculated according to the wind speed grade interval, and the statistical method is: The historical data is divided into three intervals according to the wind speed: low wind speed interval 1, medium wind speed interval 2 and high wind speed interval 3; the calculation formula of the relative error of the low wind speed interval 1, the medium wind speed interval 2 and the high wind speed interval 3 is: wherein , represents an interval; The wind and light error standard deviation correction is statistically calculated according to the wind speed grade interval, and the statistical method is: wherein is the number of samples in the interval , , and ∈ ; The battery life index triangular distribution parameter correction is: Firstly, the prediction error sliding average is calculated: wherein, is the battery life index mean absolute error; represents the total number of quarter days, represents one day; is the single day prediction error, which represents the difference between the actual life index and the predicted life index; Secondly, the triangular distribution parameters are updated: wherein, is a triangular distribution new value, is a triangular distribution old value; is a triangular distribution possible value; is a rated life index; is a triangular distribution maximum value; and ≥ 0.85 , < 1.15 ; The electrolytic hydrogen production energy consumption triangular distribution parameter correction is: wherein, is the actual energy consumption for electrolytic hydrogen production; is the rated energy consumption for electrolytic hydrogen production; is the minimum energy consumption; is the median energy consumption; is the maximum energy consumption; is the minimum value of the triangular distribution, is the possible value of the triangular distribution; is the maximum value of the triangular distribution.

6. The method of claim 1, wherein, The variables of the two-stage random robust capacity optimization model with uncertainty budget include deterministic variables, deterministic random variables and deterministic robust parameters; (1) Determined variables are: battery capacity E B , MWh; battery power P B , MW; molten salt thermal storage capacity E TS , MWh; electrolyzer power P H2 , MW; (2) Determine the random variables: wind and solar actual output , , actual load , battery life index , electrolytic hydrogen energy consumption ; (3) The robust parameters are determined as: uncertainty budget where 0 < 1 ≤ where denotes the extreme uncertainty at most time instants. The expression of the deterministic objective function is: wherein, is the daily depreciation cost of the first stage energy storage investment, calculated as: wherein, is the battery capacity; is the power cost; is the molten salt thermal storage capacity cost; is the electrolyzer power cost; Y is the system lifetime; is the random variable vector; E[O] is the expected second-stage operating cost, which contains the operating cost , battery replacement cost ; wind and solar curtailment loss ; λ is the robustness penalty coefficient, which is used to balance the economy and robustness, and is set to 0.

3. The conditional value at risk at a confidence level for ββ3penalty is a measure of extreme cost and is calculated as: wherein, v is the CVaR auxiliary variable; The core constraint condition for fusing robustness is given, including the wind and light-load matching robust constraint for coping with extreme supply and demand imbalance, the expression of which is: s.t. where, Pbat is the battery discharge power, Pbat is the battery charge power; Pfuel is the hydrogen fuel cell output; Pheat is the molten salt thermal storage electric heating power; is an indicator function that takes value 1 when the condition is satisfied; is the load extreme fluctuation threshold, taken as 0.

1.

7. The method of optimizing an electro-thermal hydrogen hybrid energy storage system for extreme scenarios of claim 6, wherein, The robust constraint of battery life is also included, which requires that the life is not less than 10 years at a confidence level of 95%, the expression of which is: wherein, is the actual cycle life of the battery, years; is the number of cycles at 100% depth of discharge of the battery, taking a value of 3000; is the actual depth of discharge of the battery; is a probability function; The robust constraint of electrolytic hydrogen production is included, which requires that the hydrogen production amount can meet the demand under extreme energy consumption, the expression of which is: wherein, is the maximum energy consumption for electrolytic hydrogen production, taking the value 5.25 kWh / Nm 3 ; is the daily minimum hydrogen production requirement, which needs to be determined based on the hydrogen load.

8. The method for optimizing an electro-thermal hydrogen hybrid energy storage system for extreme scenarios of claim 1, wherein, The method of solving the two-stage stochastic robust model using sample average approximation + GAMS / BARON involves the following steps: While considering both accuracy and computational speed, generate... M =500 random samples; convert the expected value and CVaR into sample mean and quantiles, and establish a deterministic equivalent model; adopt a hierarchical solution method: in the first stage, optimize the energy storage capacity; In the second stage, the scheduling policy is optimized for each sample, iterating until the objective function converges, where the convergence threshold is ε = 10 -4 ; The step of verifying the robustness is: comparing the robust optimization scheme with the core indexes of the deterministic scheme, analyzing the influence of the uncertainty budget Γ on the robustness-economy, and determining the optimal Γ.

9. An optimization device for an electro-thermal hydrogen hybrid energy storage system for coping with extreme scenarios, characterized by The application discloses an optimization method for realizing an electrothermal hydrogen hybrid energy storage system for coping with extreme scenes. A core uncertainty source identification module is configured to identify wind and light output uncertainty, load uncertainty and equipment parameter uncertainty. An uncertainty source random characteristic quantification module is configured to quantize the wind and light output uncertainty, the load uncertainty and the equipment parameter uncertainty. An uncertainty influence analysis module is configured to generate samples according to the probability distribution parameters of the uncertainty sources and substitute the samples into a model to calculate indexes and identify risk scenes. A scene self-adaption module is configured to train a probability distribution correction model by using a sliding window to run historical risk scene data in real time, and correct and obtain the probability distribution parameters of the uncertainty sources by using an online learning algorithm. A two-stage random robust capacity optimization module is configured to define model variables, determine a target function and give core constraint conditions of the fusion robustness. A solution and robustness verification module is configured to solve the two-stage random robust model and verify the robustness.

Images