Flexible balanced resource allocation method and device considering uncertainty of new energy

Abstract

The invention provides a flexible balanced resource allocation method and device considering uncertainty of new energy, and the method comprises the steps: obtaining natural day time scale data and typical extreme scene label data of a whole year, and decomposing an energy storage main state into a daytime energy component and a scene-driven intra-day energy component according to the data; establishing a first coupling relationship between the daytime energy component and a seasonal energy storage model, and a second coupling relationship between the scene-driven intraday energy component and a short-time energy storage model; based on the second coupling relationship, using a short-time energy storage model to generate charging and discharging power and constraining the charging and discharging power not to exceed the rated capacity; and based on the first coupling relationship, updating the daytime energy component through the seasonal energy storage model and constraining the daytime energy component not to exceed the investment capacity. According to the method, the energy storage charging and discharging power and capacity are effectively restrained, and flexible balance resource optimization configuration under the uncertainty of new energy is achieved.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of new energy resource regulation, and particularly relates to a flexible balancing resource configuration method and device considering new energy uncertainty. BACKGROUND

[0002] With the transformation of global energy structure and the promotion of the "double carbon" target, the proportion of new energy represented by wind power and photovoltaic power in the power system is increasing.

[0003] However, new energy output has significant intermittent, volatile and seasonal imbalance characteristics. Its output is greatly affected by natural conditions such as wind speed, light intensity and weather changes, resulting in possible sharp fluctuations in new energy output in a short period of time, increasing the difficulty of real-time balancing and stability control of the power system. Especially in extreme weather conditions such as "extreme heat without wind" or "extreme cold with little light", new energy output may drop sharply at a moment, posing a serious threat to the power supply reliability of the power system. In addition, there are significant differences in new energy output between different seasons, such as higher photovoltaic output in summer and higher wind power output in winter, which further increases the risk of supply and demand balance of the power system in different seasons.

[0004] Traditional power system planning methods are mostly based on historical data and deterministic models, which are difficult to effectively cope with these uncertainty characteristics of new energy output, especially in dealing with long time scale planning problems with high computational complexity, making it difficult to ensure the accuracy and reliability of the planning scheme. SUMMARY

[0005] The purpose of the present application is to overcome the defects in the prior art, and to provide a flexible balancing resource configuration method and device considering new energy uncertainty.

[0006] The present application provides a flexible balancing resource configuration method considering new energy uncertainty, comprising:

[0007] Obtaining annual natural day time scale data and typical extreme scenario label data;

[0008] According to the natural day time scale data and the typical extreme scenario label data, the energy storage main state is decomposed into a daily energy component and a scenario-driven daily energy component;

[0009] Establishing a first coupling relationship between the daily energy component and the seasonal energy storage model;

[0010] Establishing a second coupling relationship between the scenario-driven daily energy component and the short-time energy storage model;

[0011] Based on the second coupling relationship, the scene-driven intraday energy component is converted through the short-time energy storage model to generate charge and discharge power, so as to constrain the charge and discharge power from exceeding the short-time energy storage rated capacity;

[0012] Based on the first coupling relationship, the daily energy component is updated through the seasonal energy storage model, so as to constrain the daily energy component from exceeding the seasonal energy storage investment capacity.

[0013] Optionally, the annual natural day time scale data and the typical extreme scene label data are obtained, including:

[0014] The annual wind power output data, the photovoltaic output data and the load data of the power system are collected to construct an original data set; and the original data set is processed by principal component analysis dimension reduction;

[0015] The typical extreme scene label data is generated according to the data set after dimension reduction.

[0016] Optionally, the typical extreme scene label data is generated by the following steps:

[0017] The data after dimension reduction is clustered by using a K-medoids algorithm, including determining the optimal clustering number based on an elbow rule;

[0018] Each cluster center point is selected as a typical scene;

[0019] The extreme scene is identified from the typical scene by using a quantile threshold.

[0020] Optionally, the method further includes:

[0021] A power system planning model is constructed, and the power system planning model takes minimizing investment cost and operation cost as an objective;

[0022] The intraday energy balance and the inter-annual energy balance result are input into the power system planning model, and a generator unit installed capacity, a power transmission line construction scale and an energy storage configuration scheme are obtained by solving.

[0023] Optionally, the energy storage main state is decomposed into a daily energy component and a scene-driven intraday energy component, including:

[0024] The scene-driven intraday energy component at an initial time of each day is set to zero.

[0025] The application also provides a power system energy storage state reconstruction device, including:

[0026] An obtaining module is configured to obtain annual natural day time scale data and typical extreme scene label data;

[0027] a decomposition module configured to decompose the main state of the energy storage into a day-ahead energy component and a scenario-driven intra-day energy component according to the natural day-time scale data and the typical extreme scenario label data;

[0028] a coupling module configured to establish a first coupling relationship between the day-ahead energy component and a seasonal energy storage model, and establish a second coupling relationship between the scenario-driven intra-day energy component and a short-time energy storage model;

[0029] a constraint module configured to, based on the second coupling relationship, convert the scenario-driven intra-day energy component through the short-time energy storage model to generate a charging and discharging power, so as to constrain the charging and discharging power from exceeding a short-time energy storage rated capacity; and based on the first coupling relationship, update the day-ahead energy component through the seasonal energy storage model, so as to constrain the day-ahead energy component from exceeding a seasonal energy storage investment capacity.

[0030] Optionally, the acquisition module acquires the annual natural day-time scale data and the typical extreme scenario label data by:

[0031] collecting annual wind power output data, photovoltaic output data and load data of a power system to construct an original data set, and performing principal component analysis dimension reduction processing on the original data set;

[0032] generating the typical extreme scenario label data according to the data set after dimension reduction.

[0033] Optionally, the typical extreme scenario label data is generated by the following steps:

[0034] adopting a K-medoids algorithm to cluster the data after dimension reduction, including determining an optimal clustering number based on an elbow rule;

[0035] selecting each clustering center point as a typical scenario;

[0036] identifying an extreme scenario from the typical scenarios through a quantile threshold.

[0037] Optionally, the method further comprises:

[0038] a model module configured to construct a power system planning model, the power system planning model taking minimization of investment cost and operation cost as a target; input the intra-day energy balance and the inter-annual energy balance result into the power system planning model, and solve to obtain a generator unit installed capacity, a power transmission line construction scale and an energy storage configuration scheme.

[0039] Optionally, the decomposition module configured to decompose the main state of the energy storage into a day-ahead energy component and a scenario-driven intra-day energy component comprises:

[0040] setting the scenario-driven intra-day energy component at an initial time of each day as zero.

[0041] The beneficial effects of the present application are:

[0042] The present application provides a flexible balancing resource configuration method considering new energy uncertainty, comprising: obtaining annual natural day time scale data and typical extreme scenario label data; decomposing the energy storage main state into an inter-day energy component and a scenario-driven intra-day energy component according to the natural day time scale data and the typical extreme scenario label data; establishing a first coupling relationship between the inter-day energy component and a seasonal energy storage model; establishing a second coupling relationship between the scenario-driven intra-day energy component and a short-time energy storage model; based on the second coupling relationship, converting the scenario-driven intra-day energy component through the short-time energy storage model to generate charging and discharging power, so as to constrain the charging and discharging power not to exceed the short-time energy storage rated capacity; based on the first coupling relationship, updating the inter-day energy component through the seasonal energy storage model, so as to constrain the inter-day energy component not to exceed the seasonal energy storage investment capacity. The present application effectively constrains the energy storage charging and discharging power and capacity by decomposing the energy storage state and coupling the seasonal and short-time energy storage models, and realizes the optimal configuration of flexible balancing resources under new energy uncertainty. BRIEF DESCRIPTION OF DRAWINGS

[0043] Figure 1 is a flexible balancing resource configuration process schematic diagram considering new energy uncertainty in the present application. DETAILED DESCRIPTION

[0044] Exemplary embodiments of the present disclosure will be described in greater detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the drawings, it is understood that various forms implement the present disclosure and should not be limited by the embodiments set forth herein. Rather, the embodiments are provided to enable a more thorough understanding of the present disclosure and to convey the scope of the present disclosure to those skilled in the art.

[0045] The present application mainly includes four steps of original data set preprocessing, typical scenario selection, extreme scenario selection, and seasonal energy storage operation model establishment. Based on the combination of annual 8760 hours of operation simulation and extreme scenario optimization, not only the time sequence coupling of the power system can be comprehensively described, but also the seasonal imbalance and extreme events can be effectively responded to, thereby providing support for the safe, stable, and economic operation of the high-proportion renewable energy power system.

[0046] The present application determines the optimal number of clusters based on the elbow rule, and performs cluster dimension reduction on 8760 time sequences; and then, based on the typical scenario and extreme scenario data obtained by clustering, reconstructs the typical day time sequence according to the mapping relationship between the typical scenario, the extreme scenario, and the natural day established by the clustering function modeling.

[0047] The dimension reduction and reconstruction method of the time sequence mainly includes two steps:

[0048] 1) Clustering dimension reduction is performed on the natural time series, and the target scene day set is obtained by clustering dimension reduction through a clustering function;

[0049] 2) The sequence reconstruction of the target scene day forms a new reconstructed time series.

[0050] Through clustering dimension reduction and reconstruction of the time series, feature modeling of the time series under different time scales is realized, and the scale of the optimization problem to be solved is effectively reduced. The panoramic time sequence operation simulation of the power system can be performed based on the reconstructed time series, while considering the system operation constraints within and between days. By using a small number of decision variables, the seasonal fluctuation characteristics of the power system throughout the year can be accurately described in the optimization model.

[0051] Referring to Figure 1 , the application provides a flexible balancing resource configuration method considering new energy uncertainty, comprising:

[0052] S101: Obtain annual natural day time scale data and typical extreme scene label data;

[0053] The data collection unit is responsible for collecting electrical parameters of each branch of the power system, voltage phase angles of each node, connection relationships of generator units and new energy devices, active power vectors, active load vectors, and admittance matrices, and selecting wind power, photovoltaic output, and load data of each node of the system planning year 8760 hours to construct a basic data set.

[0054] The scene x in the data set is defined as follows:

[0055]

[0056] where q, w, and pv represent load demand, wind power, and photovoltaic output, respectively, and N n and N T are the number of nodes and the time length of the scene, respectively.

[0057] The data set X contains N s scenes, and the dimension number of the characteristic variables of each scene is represented by N sl .

[0058] The data dimension reduction unit is used for systems with dozens of nodes, and the characteristic variable dimension of the data set X can reach thousands of dimensions, which contains a large amount of redundant features. In order to improve data processing efficiency, principal component analysis is used to reduce the dimension of the collected data, remove redundant information, and reduce the data dimension.

[0059] The characteristics are z-score standardized:

[0060]

[0061] where σ f is the variance of the feature F f , and X sc is the normalized dataset.

[0062] The covariance matrix ∑cor of the dataset is calculated:

[0063]

[0064] The covariance matrix is decomposed to obtain the eigenvalues λ1, λ2, …, λ Nsl in descending order, and the corresponding eigenvectors μ λ1 , μ λ2 , …, μ λNsl form a matrix U λ :

[0065]

[0066] The purpose of dimensionality reduction is to compress the number of feature variables of the dataset while trying to maintain the differences between different scenarios. When the first K feature variables are selected to represent the original data, the proportion of the retained data variance ΛK is:

[0067]

[0068] Further, according to the expected proportion of the retained data variance, the eigenmatrix Uλ used to represent the original data is uniquely determined, and the dimensionally reduced and compressed dataset X r is:

[0069]

[0070] The scenario selection and weight determination module is used to select typical scenarios and extreme scenarios and determine the weights of the scenarios, including using the K-medoids algorithm to select typical scenarios and using the quantile-based dynamic weight extreme scenario selection method to select extreme scenarios:

[0071] a. A typical scenario extraction unit uses the K-medoids algorithm to perform cluster analysis on the preprocessed data, determines the optimal number of clusters according to the elbow rule, and selects the cluster centers as typical scenarios: The K-medoids algorithm is used to select typical scenarios, which has the advantage that the cluster centers obtained are actual scenarios rather than the average of all scenarios, avoiding the interference of outliers in the selection of cluster centers. Each scenario in the dataset X is divided into different clusters, denoted as Ω1, Ω2, …, Ω Np , and the number of each type of scenario is denoted as S1, S2, …, S Np . The classification goal of the K-medoids algorithm is to minimize the Euclidean distance of all scenarios in each cluster to its cluster center, i.e.

[0072]

[0073] where o i is the center point of the cluster Ω i , and x s is a scene in the reduced dimension dataset X r .

[0074] The elbow method is used to determine the number of typical scenes. The Within-Cluster Sum of Squares (WCSS) is an indicator that measures the sum of squared distances of each sample point to its cluster center, reflecting the tightness within the cluster. As the number of clusters k increases, WCSS will generally decrease gradually, because more clusters mean that the points within each cluster are more concentrated. However, when k increases to a certain extent, the speed of WCSS decrease will slow down significantly, forming an "elbow" or "kink". This elbow point is usually considered as the optimal number of clusters, because it balances the complexity of clustering and the tightness within the cluster. The method specifically includes two steps:

[0075] a. Calculate WCSS for different numbers of clusters; b. Plot the WCSS curve and select the elbow point.

[0076] Based on the above clustering algorithm and cluster number determination principle, the cluster center o i of each cluster is selected as a typical scene, and the typical scene dataset is constructed as follows:

[0077]

[0078] where n is the final number of typical scenes.

[0079] b. The extreme scene selection unit calculates the Euclidean distance of each data point in the cluster to the cluster center, sets a quantile threshold, distinguishes between the tight point set and the discrete point set, judges the data points in the discrete point set as extreme scenes and assigns weights, and simultaneously dynamically adjusts the weight of the typical scene: a dynamic weight extreme scene selection method based on quantile is used to optimize the selection of extreme scenes. The core of this method is to use the quantile threshold to distinguish between extreme scenes and typical scenes, and after selecting the extreme scenes, the weight of the typical scenes is dynamically adjusted. This dynamic adjustment mechanism is one of the core innovations of this method.

[0080] The definition of tight points and discrete points is for all points in a cluster Ω i obtained by clustering. The Euclidean distance of any scene x s in a cluster to the cluster center o i is calculated, and a quantile threshold ε is set to form a tight point set Ω ic and a discrete point set Ω i d , a set of scenes Ω s is established. For the close point set Ω i c , the clustering center point o i is taken as a typical scene, and the typical scene weight is determined based on the close point set Ω i c The number of data points determines the typical scene weight. The data points in the discrete point set Ω i d are all taken as extreme scenes and are respectively assigned weights, and the weights of the extreme scenes are dynamically adjusted according to the quantile threshold. After identifying the extreme scenes, the weights of the typical scenes are dynamically adjusted to ensure the accuracy and practicability of the scene analysis.

[0081] The method aims to efficiently identify representative extreme scenes by combining clustering analysis and quantile threshold, and dynamically adjust the weights to ensure the reasonableness of the total weight.

[0082] The specific steps include:

[0083] a. Quantile threshold setting;

[0084] b. Weight calculation;

[0085] c. Dynamic weight adjustment.

[0086] S102: According to the natural day time scale data and the typical extreme scene label data, the energy storage main state is decomposed into a day energy component and a scene-driven intra-day energy component;

[0087] The time series reconstruction module is used to time series reconstruct the typical scenes and the extreme scenes according to the scene and natural day mapping relationship, forming a new reconstructed time series.

[0088] The application constructs an intra-day balanced short-time energy storage and an annual balanced seasonal energy storage operating state, wherein the seasonal energy storage has the same charging and discharging behavior in the same typical day, the same typical day corresponds to the same net change amount of the seasonal energy storage SOC, and the seasonal energy storage has an accumulation relationship in the day. For the short-time energy storage, the energy storage SOC fluctuation curve in the same typical day is the same, and the intra-day energy balance is ensured.

[0089] The state of charge (SOC) of the energy storage b is decomposed into a day component and an intra-day component

[0090]

[0091] Wherein, b is the energy storage number, is the energy storage SOC, is the daily component of the energy storage SOC, reflecting the daily energy storage change of the energy storage b in the dth natural day, unit: MWh; is the daily component of the energy storage SOC, reflecting the daily energy storage change of the energy storage b in the kth moment of the st typical day or extreme day, unit: MWh.

[0092] Set the scenario-driven daily energy component at the initial moment of each day to zero:

[0093]

[0094] S103: Establish the first coupling relationship of the daily energy component and the seasonal energy storage model;

[0095] The capacity conversion process of the daily energy storage component is:

[0096]

[0097] wherein, represents the net increment of the daily component of the energy storage after the daily operation simulation.

[0098] S104, establish the second coupling relationship of the scenario-driven daily energy component and the short-time energy storage model:

[0099] The energy conversion process of the daily component of the energy storage at the hourly level is:

[0100]

[0101] wherein, represents the charging / discharging efficiency of the energy storage b, represents the charging / discharging power of the energy storage b at the kth moment of the st typical day or extreme day, unit: MW.

[0102] S105: Based on the second coupling relationship, convert the scenario-driven daily energy component through the short-time energy storage model to generate charging / discharging power, so as to constrain the charging / discharging power not to exceed the rated capacity of the short-time energy storage;

[0103] The upper and lower limit constraints of the energy storage charging / discharging probability are:

[0104]

[0105] wherein, is the upper limit of the energy storage charging / discharging probability.

[0106] Based on the first coupling relationship, update the daily energy component through the seasonal energy storage model, so as to constrain the daily energy component not to exceed the investment capacity of the seasonal energy storage:

[0107] The upper and lower limit constraints of the energy storage SOC are:

[0108]

[0109] wherein B B,Sto is the energy storage investment capacity.

[0110] S106, based on the first coupling relationship, updating the day-time energy component through the seasonal energy storage model to constrain the day-time energy component from exceeding a seasonal energy storage investment capacity.

[0111] The energy balance constraints of short-time energy storage and seasonal energy storage are:

[0112]

[0113] The short-time energy storage is set as the intra-day energy balance, and the seasonal energy storage is set as the intra-year energy balance.

[0114] wherein the superscript B respectively denotes the short-time energy storage SSB and the seasonal energy storage LSB.

[0115] Solving the planning scheme:

[0116] The specific steps of the planning model establishment and solution module include:

[0117] The model construction unit is configured to construct a power system planning model with the sum of investment cost and operation cost minimized as an objective function, comprehensively considering the constraints of power supply and transmission line investment decision, power system production simulation, and related constraints of flexible balancing resources:

[0118]

[0119] wherein, is the generation and line investment cost, is the operation cost.

[0120] The model solution unit is configured to input the planning model with the time-series reconstructed data, call an optimization algorithm to solve the model, and obtain the planning scheme of the power system, including the installed capacity of various types of power generation resources, the construction scale of transmission lines, and the configuration capacity and operation strategy of flexible balancing resources.

[0121] Furthermore, the constraints of the planning model cover multiple aspects, including power generation, energy storage, grid operation, and load balancing, aiming to ensure the safe, economical, and stable operation of the power system. Power generation-related constraints include upper and lower limits on the power output of conventional thermal power and ramp rate limits, as well as maximum output constraints. Energy storage-related constraints involve the charging and discharging power of short-term energy storage, intraday energy balance, SOC range limits, and requirements for the charging and discharging power, intra-year energy balance, and SOC returning to its initial value at the end of the year for seasonal energy storage. Grid operation constraints include branch power flow limits and node power balance constraints. Load balancing constraints emphasize the matching of total power generation with total load power and consider load shedding limits. These constraints work together to ensure that the power system maintains its stability and reliability while meeting various demands.

[0122] Regarding power generation constraints, conventional thermal power generation is limited by its maximum and minimum capacity, and the rate of change in output cannot exceed the ramp rate; the output power of new energy power generation (wind power, photovoltaic) cannot exceed its installed capacity and resource conditions.

[0123] Among the constraints related to energy storage, the charging and discharging power of short-term energy storage cannot exceed the rated maximum value, and must meet the requirements of intraday energy balance and state of charge (SOC) range; seasonal energy storage is also limited by charging and discharging power and energy balance SOC range, and must return to the initial SOC at the end of the year.

[0124] Power grid operation constraints require that branch power flow be kept within limits to prevent transmission line overload; power at each node must be kept in balance to ensure grid frequency and voltage stability.

[0125] Load balance constraints emphasize that total power generation must meet total load demand, while also considering load reduction limits to ensure the safe and stable operation of the power grid.

[0126] Furthermore, the flexibility balancing resources include short-term energy storage and seasonal energy storage. The device can optimize the configuration of the flexibility balancing resources to cope with the uncertainty of new energy output and the volatility of load demand, and ensure the stable operation of the power system at different time scales.

[0127] The above description of the embodiments is provided to enable those skilled in the art to understand and apply this application. Those skilled in the art will readily make various modifications to the above embodiments and apply the general principles described herein to other embodiments without inventive effort. Therefore, this application is not limited to the above embodiments, and any improvements and modifications made to this application based on the disclosure thereof should be within the scope of protection of this application.

Claims

1. A flexible resource allocation method considering the uncertainty of new energy sources, characterized in that, include: Acquire year-round natural day time scale data and typical extreme scenario label data; Based on the natural day time scale data and the typical extreme scenario label data, the main state of energy storage is decomposed into daytime energy components and scenario-driven intraday energy components. Establish the first coupling relationship between the daytime energy component and the seasonal energy storage model; Establish a second coupling relationship between the scenario-driven intraday energy component and the short-term energy storage model; Based on the second coupling relationship, the scenario-driven intraday energy component is converted through the short-time energy storage model to generate charging and discharging power, so as to constrain the charging and discharging power to not exceed the rated capacity of short-time energy storage. Based on the first coupling relationship, the daytime energy component is updated through the seasonal energy storage model to constrain the daytime energy component from exceeding the seasonal energy storage investment capacity. Based on the constraints that the charging and discharging power should not exceed the short-term energy storage rated capacity and the constraints that the daytime energy component should not exceed the seasonal energy storage investment capacity, the installed capacity, transmission lines and energy storage configuration scheme are obtained by solving the problem.

2. The flexible resource allocation method considering the uncertainty of new energy sources according to claim 1, characterized in that, Obtain year-round natural day time scale data and typical extreme scenario label data, including: Collect annual wind power output data, photovoltaic power output data, and load data from the power system to construct the original dataset; Principal component analysis was performed to reduce the dimensionality of the original dataset. The typical extreme scenario label data is generated based on the dimensionality-reduced dataset.

3. The flexible resource allocation method considering the uncertainty of new energy sources according to claim 2, characterized in that, The typical extreme scenario label data is generated through the following steps: The K-medoids algorithm is used to cluster the dimensionality-reduced data, including determining the optimal number of clusters based on the elbow rule; Each cluster center point was selected as a typical scenario; Extreme scenarios are identified from the typical scenarios using quantile thresholds.

4. The flexible resource allocation method considering the uncertainty of new energy sources according to claim 1, characterized in that, Also includes: A power system planning model is constructed, with the objective of minimizing investment and operating costs. The intraday and interannual energy balance results are input into the power system planning model to solve for the generator installed capacity, transmission line construction scale, and energy storage configuration scheme.

5. The flexible resource allocation method considering the uncertainty of new energy sources according to claim 1, characterized in that, The primary state of energy storage is decomposed into a daytime energy component and a scenario-driven intraday energy component, including: The scenario-driven daily energy component is set to zero at the initial moment of each day.

6. A power system energy storage state reconfiguration device, characterized in that, include: The acquisition module retrieves year-round natural day time scale data and typical extreme scenario label data; The decomposition module decomposes the main state of energy storage into daytime energy components and scene-driven intraday energy components based on the natural day time scale data and the typical extreme scenario label data. The coupling module establishes the first coupling relationship between the daytime energy component and the seasonal energy storage model. Establish a second coupling relationship between the scenario-driven intraday energy component and the short-term energy storage model; The constraint module, based on the second coupling relationship, converts the intraday energy components driven by the scenario into charging and discharging power through the short-term energy storage model, so as to constrain the charging and discharging power to not exceed the rated capacity of the short-term energy storage. Based on the first coupling relationship, the daytime energy component is updated through the seasonal energy storage model to constrain the daytime energy component from exceeding the seasonal energy storage investment capacity.

7. A power system energy storage state reconfiguration device according to claim 6, characterized in that, The acquisition module acquires annual natural day time scale data and typical extreme scenario label data, including: Collect annual wind power output data, photovoltaic power output data, and load data from the power system to construct the original dataset; Principal component analysis was performed to reduce the dimensionality of the original dataset. The typical extreme scenario label data is generated based on the dimensionality-reduced dataset.

8. A power system energy storage state reconfiguration device according to claim 7, characterized in that, The typical extreme scenario label data is generated through the following steps: The K-medoids algorithm is used to cluster the dimensionality-reduced data, including determining the optimal number of clusters based on the elbow rule; Each cluster center point was selected as a typical scenario; Extreme scenarios are identified from the typical scenarios using quantile thresholds.

9. A power system energy storage state reconfiguration device according to claim 6, characterized in that, Also includes: The model module constructs a power system planning model, which aims to minimize investment and operating costs. The intraday and interannual energy balance results are input into the power system planning model to solve for the generator installed capacity, transmission line construction scale, and energy storage configuration scheme.

10. A power system energy storage state reconfiguration device according to claim 6, characterized in that, The decomposition module decomposes the energy storage master state into daytime energy components and scene-driven intraday energy components, including: The scenario-driven daily energy component is set to zero at the initial moment of each day.

Images