Pumped storage global optimization control method and device and electronic equipment

Abstract

The invention provides a pumped storage global optimization control method and device and electronic equipment, and the method comprises the steps: building an operation mode perception and reconstruction mechanism through information-physical fusion modeling, and generating a structured scene set which can comprehensively represent the uncertainty of a power system; constructing a global optimization control model in combination with the structured scene set, and obtaining an optimal pumping and storage planning scheme of the power system through the global optimization control model; the global optimization control model adopts a main-subspace collaborative optimization architecture and comprises a main space model and a subspace model. According to the invention, the method can effectively cope with the operation constraint of a complex power system, guarantees the optimality and solving efficiency of a planning result, and solves a problem that the planning control effect of the power system is not good in the prior art.

Description

Technical Field

[0001] This invention relates to the field of power system resource planning technology, and in particular to a method, device and electronic equipment for global optimization control of pumped storage. Background Technology

[0002] The output of renewable energy sources such as wind and solar power is highly unpredictable and intermittent, posing a continuous pressure on the real-time balance and safe, stable operation of the system. Under these circumstances, pumped storage, due to its significant advantages in technological maturity, economics, and scalability, is considered a key supporting technology for enhancing system flexibility and regulation capabilities. However, its planning and control still face challenges such as accurately describing uncertainties, coordinating multi-level decision-making, and efficiently solving complex models. This necessitates establishing a comprehensive optimization control method for pumped storage power stations, starting from cyber-physical fusion modeling and strategic collaborative optimization.

[0003] In the field of pumped storage planning research, existing methods mainly follow the operations research optimization paradigm and can be divided into two technical routes. The first type of method is based on mathematical programming frameworks such as stochastic optimization or robust optimization, which uses precise mathematical models to coordinate the optimization of pumped storage capacity configuration and operation strategies. Although this type of method has a rigorous mathematical foundation, its modeling process fails to achieve deep integration of information and physical systems. The characterization of uncertainties still relies on traditional probabilistic frameworks, making it difficult to effectively characterize extreme operating conditions and abnormal events, resulting in insufficient adaptability of the planning scheme to complex operating environments.

[0004] The second type of method employs scenario generation and reduction techniques, generating typical operating scenarios through traditional clustering algorithms. While this type of method can reflect certain random characteristics, its scenario construction process has significant limitations: on the one hand, it relies on manually pre-setting the number of clusters, making it highly subjective; on the other hand, it completely ignores low-probability, high-risk events caused by factors such as abnormal weather, leading to insufficient resilience of the planning scheme when dealing with extreme operating conditions. Furthermore, this type of method generally lacks sufficient consideration of the special operating constraints of pumped storage systems, and its effectiveness in practical applications has not yet been fully verified.

[0005] In terms of model solving, pumped storage planning models are characterized by high dimensionality, multiple constraints, and mixed integers, making them complex large-scale optimization problems. Existing research often employs a decoupled-coordinated computational architecture, decomposing the original problem into a main problem and sub-problems for iterative solutions. While this approach alleviates computational complexity to some extent, its decomposition mechanism lacks intelligent collaborative strategies, and effective strategic interaction between the main problem and sub-problems fails to be established, making it difficult to meet the needs of comprehensive optimization planning for pumped storage under high-proportion renewable energy environments.

[0006] There is currently no effective solution to the problem of poor power system planning and control performance in existing related technologies. Summary of the Invention

[0007] This invention provides a method, device, and electronic equipment for comprehensive optimization control of pumped storage power generation, which addresses the shortcomings of existing related technologies in terms of poor power system planning and control performance.

[0008] In a first aspect, the present invention provides a method for global optimization control of pumped storage, comprising: By using cyber-physical fusion modeling, a mechanism for operational modality perception and reconstruction is constructed to generate a set of structured scenarios that can comprehensively characterize the uncertainties of the power system; A global optimization control model is constructed by combining the structured scenario set, and the optimal pumped storage planning scheme of the power system is obtained through the global optimization control model. The global optimization control model adopts a master-subspace collaborative optimization architecture, which includes a master space model and a subspace model.

[0009] According to the pumped storage power generation global optimization control method provided by the present invention, an operational modality perception and reconstruction mechanism is constructed through cyber-physical fusion modeling to generate a structured scenario set that can comprehensively characterize the uncertainties of the power system, including: The multi-source time-series data of the power system are standardized and feature-fused. Density cluster generation and singular value screening are performed on the feature-fused data to identify global and local anomalies; Based on the anomaly identification results, multimodal scene reconstruction and probability weight assignment are performed to generate the structured scene set.

[0010] According to the pumped storage power system global optimization control method provided by the present invention, the multi-source time-series data of the power system are standardized and feature-fused, including: A collaborative processing mechanism for time-series data is established to perform time-scale alignment and data quality control on the input multi-source time-series data, eliminating missing values ​​and outliers; the multi-source time-series data includes net load power sequence, centralized photovoltaic power sequence, and wind power curve; The net load power sequence, centralized photovoltaic power sequence, and wind power curve are integrated into the same feature representation to obtain the feature vector of the daily comprehensive operating status of the power system's equipment. Dynamic range normalization is performed on the multi-source time-series data of each feature channel.

[0011] According to the present invention, a method for global optimization control of pumped storage includes density cluster generation and singular value screening of the feature-fused data to identify global and local anomalies, including: By using a density-sensitive message passing clustering model, the optimal number of clusters is adaptively determined, and clusters are divided to generate a preliminary set of typical daily scenarios. For each typical day scene cluster in the preliminary typical day scene set, the neighborhood relative cluster degree evaluation algorithm is used to identify abnormal samples that are significantly different from the main operation mode.

[0012] According to the present invention, a pumped storage global optimization control method adaptively determines the optimal number of clusters through a density-sensitive message passing clustering model, performs cluster division, and generates a preliminary typical daily scene set, including: Set the initial value of the attribution message among all daily samples to 0; The representativeness and affiliation messages of the daily samples are iteratively updated and clustered until the convergence condition is met.

[0013] According to the pumped storage global optimization control method provided by the present invention, the representativeness message and belongingness message of the daily sample are iteratively updated, and cluster classification is performed, including: Update the representativeness message based on the current affiliation degree and similarity metric matrix; Based on the updated representativeness message, recalculate the attribution message for each daily sample; The daily sample that maximizes the sum of the attribution and representativeness messages is selected as the representative paradigm, and all daily samples sharing the same paradigm are grouped into the same cluster.

[0014] According to the pumped storage power generation system optimization control method provided by the present invention, for each typical day scene cluster in the preliminary typical day scene set, a neighborhood relative clustering degree evaluation algorithm is used to identify anomalous samples that differ significantly from the main operating mode, including: For each object in the typical day scene set, calculate the K-distance of the object and determine its K-neighborhood distance; Calculate the reachability distance of the object relative to each object in its neighborhood; Calculate the local reachability density within the neighborhood of the object; By comparing the local reachability density of the object with the local reachability density of its K-nearest neighbors, a relative outlier evaluation value is obtained, and the object is determined to be an outlier anomalous sample based on the relative outlier evaluation value.

[0015] According to the present invention, a pumped storage power generation system with global optimization control is constructed by combining the structured scenario set to build a global optimization control model, and the optimal pumped storage planning scheme of the power system is obtained through the global optimization control model, including: Power grid modeling based on the existing power transmission distribution factor model; A strategic subspace game-solving mechanism is adopted, with the main space model as the decision-making subject. The coordination and constraint satisfaction between the main space and subspace are achieved through an adaptive feasible cut mapping strategy, so as to obtain the optimal pumped storage planning scheme of the power system.

[0016] Secondly, the present invention also provides a pumped storage power generation system with optimized control across the entire area, comprising: The module is used to build and operate a modal perception and reconstruction mechanism through cyber-physical fusion modeling, and generate a set of structured scenarios that can comprehensively characterize the uncertainties of the power system; The optimization module is used to construct a global optimization control model by combining the structured scenario set, and to obtain the optimal pumped storage planning scheme of the power system through the global optimization control model. The global optimization control model adopts a master-subspace collaborative optimization architecture, which includes a master space model and a subspace model.

[0017] Thirdly, the present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the pumped storage global optimization control method as described in the first aspect above.

[0018] Fourthly, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the pumped storage global optimization control method as described in the first aspect above.

[0019] Fifthly, the present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the pumped storage global optimization control method as described in the first aspect above.

[0020] Compared with the prior art, the present invention has the following beneficial effects: 1. The pumped storage global optimization control method provided by this invention introduces an adaptive feasible cut mapping strategy to construct a game-theoretic interaction framework between the main space and subspaces, thereby enhancing decision-making collaboration and convergence efficiency. Numerical examples demonstrate that this method can effectively address complex power system operational constraints, ensuring the optimality of planning results and solution efficiency, and solving the problem of poor power system planning and control performance in existing related technologies.

[0021] 2. This invention constructs a scenario combination optimization framework of "density cluster generation → singular value screening → secondary cluster optimization," and innovatively introduces the AP algorithm, which does not require a preset number of clusters, enabling the identification of normal and abnormal operating conditions of wind, solar, and load from historical data. This method, while adhering to the principle of maximizing scenario diversity, achieves more objective quantification of complex scenarios in pumped storage planning. Attached Figure Description

[0022] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0023] Figure 1 This is a flowchart of the pumped storage full-domain optimization control method provided by the present invention; Figure 2 This is a schematic diagram of the information-physical fusion-driven multimodal scene reconstruction process in an embodiment of the present invention; Figure 3 This is a schematic diagram of the global optimization control model in an embodiment of the present invention; Figure 4 This is the flow of the strategy-based subspace collaborative optimization algorithm in this embodiment of the invention; Figure 5 This is a schematic diagram of the modified power system structure in an embodiment of the present invention; Figure 6 This is a schematic diagram of the original cluster center samples after removing outliers in an embodiment of the present invention; Figure 7 This is a schematic diagram of outlier cluster center samples in an embodiment of the present invention; Figure 8 This is a schematic diagram showing the operation of each device in the outlier clustering 3-center sample of this invention; Figure 9 This is a schematic diagram of the effective energy storage curve of the reservoir on the outlier cluster 3-center sample in this embodiment of the invention; Figure 10 This is a schematic diagram of the wind and solar power curtailment curves for outlier cluster 3-center samples in an embodiment of the present invention; Figure 11 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0024] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0025] This invention provides a method for global optimization control of pumped storage. Figure 1This is a flowchart of the pumped storage global optimization control method provided by the present invention, such as... Figure 1 As shown, it includes the following steps: Step S101: Through cyber-physical fusion modeling, construct an operational modality perception and reconstruction mechanism to generate a structured scenario set that can comprehensively characterize the uncertainties of the power system; Step S102: Construct a global optimization control model by combining the structured scenario set, and obtain the optimal pumped storage planning scheme of the power system through the global optimization control model; The global optimization control model adopts a master-subspace collaborative optimization architecture, which includes a master space model and a subspace model.

[0026] In this method, firstly, an operational modality perception and reconstruction mechanism is constructed through cyber-physical fusion modeling to generate a structured scenario set covering multiple operational modalities. Then, a global optimization control model is built based on the constructed structured scenario set, and the optimal pumped storage planning scheme for the power system is obtained through this model. The global optimization control model adopts a master-subspace collaborative optimization architecture, including a master space model and a subspace model. By introducing an adaptive feasible cut mapping strategy, a game-theoretic interaction framework is constructed between the master space model and the subspace model, enhancing decision-making collaboration and convergence efficiency. This improves the reliability of the optimal pumped storage planning scheme and effectively addresses complex power system operational constraints, ensuring the optimality of the planning results and solution efficiency. This solves the problem of poor power system planning and control performance in existing related technologies.

[0027] In some embodiments, step S101 involves constructing a modal perception and reconstruction mechanism through cyber-physical fusion modeling to generate a structured scenario set that can comprehensively characterize the uncertainties of the power system. This includes: standardizing and fusing the multi-source time-series data of the power system; generating density clusters and screening outliers on the fused data to identify global and local anomalies; and reconstructing multimodal scenarios and assigning probability weights based on the anomaly identification results to generate the structured scenario set.

[0028] For example, to achieve accurate identification of complex operating modes composed of multiple sources such as load, photovoltaic, and wind power in a power system, this embodiment proposes a multimodal scenario reconstruction method that integrates normal and abnormal operating days. Abnormal operating modes are mainly caused by extreme weather conditions (such as typhoons, blizzards, continuous rain or no wind, etc.), and their renewable energy output characteristics differ significantly from those of normal operating days. Although such events are low-probability events, they are important factors in enhancing the resilience and reliability of the system under extreme operating conditions.

[0029] This embodiment mainly includes three core stages: multi-source time-series data standardization and feature fusion, density cluster generation and singular value screening, and multimodal scene reconstruction and probability weight assignment. Ultimately, it generates a structured scene set that can comprehensively characterize system uncertainties, providing reliable input for the full-domain optimization planning of pumped storage power stations in high-proportion new energy power systems. The optimization framework is as follows: Figure 2 As shown, Figure 2 This is a schematic diagram of the information-physical fusion-driven multimodal scene reconstruction process in an embodiment of the present invention.

[0030] Specifically, the multi-source time-series data of the power system undergoes standardization processing and feature fusion, including: establishing a collaborative processing mechanism for time-series data; performing time-scale alignment and data quality control on the input multi-source time-series data to eliminate missing values ​​and outliers; the multi-source time-series data includes net load power sequences, centralized photovoltaic power sequences, and wind power curves; fusing the net load power sequences, centralized photovoltaic power sequences, and wind power curves into a single feature representation to obtain a feature vector of the comprehensive daily operating status of the power system's equipment; and performing dynamic range normalization processing on the multi-source time-series data of each feature channel.

[0031] For example, in order to achieve deep feature extraction and structured characterization of the dynamic sequence of multi-source operation of the power system, this embodiment first establishes a time-series data collaborative processing mechanism to perform time-scale alignment and data quality control on the input net load, centralized photovoltaic and wind power curves, eliminate missing values ​​and outliers, and ensure accurate synchronization of multi-source power sequences in the time dimension.

[0032] Subsequently, the three types of power sequences each day are merged into a unified feature representation. Assume each type of power sequence contains [number of sequences] daily. For each sampling time point, the daily net load, centralized photovoltaic power, and wind power can be respectively constructed as feature vectors, as shown in the following formulas:

[0033]

[0034]

[0035] in, Represents the net load power sequence. This represents a centralized photovoltaic power sequence. This represents the wind power sequence. The three types of feature vectors mentioned above are fused using multimodal feature fusion to construct the feature vector y representing the comprehensive daily operating status of the equipment. The specific formula is as follows:

[0036] Through this embodiment, the original The data was converted to a dimension of Feature tensor Each sample point corresponds to the combined operating conditions of wind, light, and load during a day.

[0037] To address the significant differences in numerical range and fluctuation characteristics among different energy sources, and to eliminate the influence of dimensions and ensure the objectivity of cluster analysis, this embodiment employs dynamic range normalization, as detailed in the following formula:

[0038] in, Represents the normalized eigenvalues. Indicates the first m The sample at the th n The values ​​of the feature channels, Indicates the first n The mean of the feature channels, Indicates the first n The standard deviation of each feature channel. This process standardizes the data distribution across all feature channels.

[0039] The data after feature fusion is subjected to density cluster generation and singular value screening to identify global and local anomalies. This includes: using a density-sensitive message passing (DSMP) clustering model to adaptively determine the optimal number of clusters and perform cluster division to generate a preliminary set of typical daily scene patterns; for each typical daily scene cluster in the preliminary set of typical daily scene patterns, a neighborhood relative clustering degree evaluation algorithm is used to identify anomalous samples that are significantly different from the main operating mode.

[0040] The optimal number of clusters is adaptively determined using a density-sensitive message passing clustering model. Clustering is performed to generate a preliminary set of typical daily scenarios. This includes: setting the initial value of the attribution message among all daily samples to 0; iteratively updating the representativeness message and attribution message of daily samples and performing clustering until the convergence condition is met.

[0041] The representativeness and attribution messages of daily samples are iteratively updated, and clustering is performed, including: updating the representativeness message based on the current attribution and similarity metric matrix; recalculating the attribution message of each daily sample based on the updated representativeness message; selecting the daily sample that maximizes the sum of the attribution and representativeness messages as the representative example, and assigning all daily samples sharing the same example to the same cluster.

[0042] For each typical day scene cluster in the preliminary typical day scene set, the Neighborhood Relative Outlmerness Evaluatmon (NROE) algorithm is used to identify anomalous samples that differ significantly from the main operating mode in each cluster. This includes: for each object in the typical day scene set, calculating the object's K-distance and determining its K-neighborhood distance; calculating the reachability distance of the object relative to each object in its neighborhood; calculating the local reachability density (LRD) in the object's neighborhood; obtaining the neighborhood relative outliveness evaluation value by comparing the object's local reachability density with the local reachability density of its K-nearest neighbors; and determining whether the object is an outlier anomalous sample based on the relative outliveness evaluation value.

[0043] For example, the preprocessed daily samples are input into a density-sensitive message passing clustering model. This model adaptively determines the optimal number of clusters through a message autonomous passing mechanism between data points, completes the clustering of each daily sample, and forms a preliminary set of typical daily scenarios, i.e., the "density cluster generation" stage.

[0044] DSMP clustering is an autonomous clustering algorithm based on iterative message propagation. It adaptively selects the most representative example samples as cluster centers through bidirectional message passing and competition mechanisms between samples. The specific steps are as follows: 1. Initialization: Set the attribution messages among all samples. Initially 0; 2. Update representativeness information Based on current affiliation and similarity measure matrix Update representativeness message ; 3. Update affiliation messages Based on the updated representativeness information Recalculate the belonging degree of each sample. ; 4. For each sample Choose to make The sample with the largest value As a representative example, all samples sharing the same paradigm belong to the same cluster. If the convergence condition is not met, return to step 2 to continue iterating.

[0045] Among them, similarity measurement Representativeness news and affiliation information The calculation method is shown in the following formula:

[0046]

[0047]

[0048] in, m , g , m ', g 'Indicates the daily sample set index, This indicates the degree of representativeness of the message. This indicates affiliation information. Similarity metric. Reflecting the sample As The fit of the examples is measured by the squared negative Euclidean distance; a larger value indicates that the operating conditions of the two samples are more similar. (Representativeness message) Indicates sample right The cumulative support level for its paradigm is considered in conjunction with the competitive situation of other candidate paradigms; attribution message express The applicability of an example is reflected in the degree to which it is recognized as representative by other samples.

[0049] The process then proceeds to the "outlier screening" phase: For each typical daily scene cluster obtained from the initial clustering, a neighborhood relative outlier evaluation algorithm is further employed to identify anomalous samples that significantly differ from the main operating mode in each cluster. This algorithm quantifies the degree of anomalousness of each object by comparing the local density of the object with its nearest neighbors. Its core calculation process can be summarized in the following steps: 1. Calculation Object K-distance ( ), and determine its K-distance neighborhood. The specific formula is as follows:

[0050] in, express a To a certain object b The distance between the objects. b The following conditions must be met: Condition 1: At least one exists objects , making ; Condition 2: At most exist objects , making ; The specific formula is as follows:

[0051] Here, c represents a data object in dataset D, and D represents the original dataset.

[0052] 2. Calculation Relative to objects in the neighborhood Reachability Distance: Introducing reachability distance to smooth statistical fluctuations, it is defined as an object With object Distance between The maximum value of the K-distance:

[0053] in, express Relative to objects in the neighborhood The reachable distance.

[0054] 3. Calculate the local reachability density, i.e., the density of objects. The reciprocal of the average reachable distance within the neighborhood, as shown in the following formula:

[0055] in, This indicates locally accessible density.

[0056] 4. Calculate the relative outlier evaluation value in the neighborhood: Finally, compare the objects... Locally achievable density and its The local reachability density of nearest neighbors is used to obtain the relative outlier assessment value of the neighborhood. The larger the value, the more abnormal the object is. If the value is higher than a set threshold, it indicates that the sample's operating mode deviates significantly from the main characteristics of its cluster, and is therefore judged as an outlier sample. The specific formula is as follows:

[0057] in, This represents the relative outlier assessment value within the neighborhood.

[0058] This algorithm depends only on parameters. It is applicable to datasets with different density distributions, can effectively identify global and local anomalies, and has a solid theoretical foundation and good interpretability.

[0059] Based on this, all identified abnormal daily sample vectors are integrated into an independent abnormal sample set. In order to further explore the inherent laws of abnormal operation modes, the density-sensitive message passing clustering method is used again to reconstruct the mode of the set, that is, to perform "secondary cluster optimization", which summarizes the discrete abnormal patterns into several abnormal scenarios with clear physical meaning.

[0060] After removing outliers from the initial clustering, typical scenarios are extracted from each cluster, and these scenarios, together with the centers of the outlier scenarios, form a complete scenario set. The probability of occurrence of each scenario is assigned a mixed weight. This indicates that its value is the proportion of the number of cluster samples in this scenario to the total number of samples, satisfying the following condition: Finally, a structured scene set is formed by the center vectors of each scene and their corresponding weights, represented as:

[0061] in, Y c Represents a structured scene set. k c Indicates the total number of scenes. Indicates the first c The center vector of each scene.

[0062] In some embodiments, step S102 involves constructing a global optimization control model based on a structured scenario set, and obtaining the optimal pumped storage planning scheme for the power system through the global optimization control model. This includes: modeling the power grid based on the existing power transmission distribution factor model; adopting a strategic subspace game-solving mechanism, using the main space model as the decision-making subject, and achieving coordination and constraint satisfaction between the main and subspaces through an adaptive feasible cut mapping strategy to obtain the optimal pumped storage planning scheme for the power system.

[0063] For example, Figure 3 This is a schematic diagram of the global optimization control model in an embodiment of the present invention, as shown below. Figure 3 As shown, the global optimization control model constructed by this method adopts a master-subspace collaborative optimization architecture. The master space model aims at optimal investment economy, considering the capacity configuration and construction constraints of pumped storage power stations. The subspace model is further divided into a scenario generation layer and an operation optimization layer: the scenario generation layer aims to maximize the difference in system operation scenarios, generating a comprehensive scenario set covering typical and abnormal operating conditions through a multimodal scenario reconstruction method; the operation optimization layer aims to minimize operating costs, renewable energy curtailment, and load reduction, considering conditions such as transmission section safety, unit operation limits, renewable energy consumption constraints, and load guarantee requirements. The global optimization control model belongs to a class of multi-scenario stochastic optimization problems. It is based on the linear power transfer distribution factor (PTDF) model for power grid modeling and adopts a strategic subspace game solution mechanism. The master space model is used as the decision-making body, and the operation optimization models under each scenario are used as subordinate subspaces. The collaboration and constraint satisfaction between the master and subspaces are achieved through an adaptive feasible cut mapping strategy, ultimately obtaining a pumped storage planning scheme that combines economy, safety, and adaptability to all scenarios.

[0064] The principal space model aims to minimize the total system cost, and its mathematical expression is as follows:

[0065] in, G JS This represents the total system cost. Represents a set of multimodal scenarios. Represents the set of system nodes. Indicates at node a The capacity cost of constructing a pumped storage unit. Represents continuous decision variables, i.e., at nodes a The pumping capacity to be constructed at the site, Indicates the total number of days in the planning period. This represents the mixed weights of scenario c. This represents an auxiliary variable, characterizing the scenario under the current investment decision. c Daily operating costs.

[0066] Furthermore, the master space model conforms to the following constraints: Investment decision constraints include capacity limitations and site selection logic constraints, as shown in the following formula:

[0067] in, and These represent the upper and lower limits of the pumped storage capacity, respectively. It is a binary variable, representing whether it is in the node. a The pumping station was constructed. It is a binary variable, representing a node. a Whether it belongs to the candidate construction node.

[0068] In the strategic subspace collaborative optimization mechanism, two types of cut constraints are introduced: optimality cut and collaborative cut. The former is added when the subspace problem has a solution to approximate the global optimum; the latter is introduced when the subspace is infeasible to eliminate ineffective investment schemes. Traditional collaborative cut relies on extracting Farkas multipliers of all constraints, but in power system models with complex operational constraints, this method is difficult to implement due to multiplier combinatorial explosion and numerical instability. Therefore, this embodiment proposes a dynamic collaborative compensation strategy, which approximates the feasible region by introducing an investment amplification mechanism, the mathematical expression of which is as follows:

[0069]

[0070] in, Indicates the first k Investment decisions in round iterations Indicates the firstk -1 Investment decision-making scenario in round-one iteration c The optimal operating cost, Denotes the dual variable corresponding to the capacity constraint. This represents the synergistic amplification factor, used to guide increased investment when constraints conflict. This represents a reference value for total investment. k Indicates the number of iterations.

[0071] In this embodiment, convergence is determined based on the premise that solutions exist for all scenarios in the subspace. In the... In this iteration, the algorithm calculates the upper bound. and the lower world The deviation between the two factors determines whether convergence has occurred, including both absolute and relative errors, and its formal expression is as follows:

[0072]

[0073] in, Indicates absolute error. and Indicates the preset tolerance. This represents the relative error. When any error is less than the preset tolerance, the convergence condition is considered met.

[0074] The upper and lower bounds are updated as follows:

[0075]

[0076] Among them, the upper boundary This indicates that, under the planning schemes provided by the current master space, the weighted total cost across the entire scenario corresponds to the actual cost of a feasible solution, and therefore is always no less than the globally optimal solution; lower bound This represents the theoretical minimum total cost that the system can achieve under the current cooperative cut constraint; Z PSH This represents the capacity cost of constructing a pumped storage unit. Represents continuous decision variables, i.e., at nodes a The pumping capacity to be constructed at the site, Indicates the total number of days in the planning period. This represents the mixed weights of scenario c. This represents an auxiliary variable, characterizing the scenario under the current investment decision. c Daily operating costs Indicates the first k Investment decision-making scenarios in round iteration cThe optimal operating cost is obtained. As iterations proceed, the collaborative cutting mechanism accumulates, leading to increasingly accurate estimates of the operating cost. The lower bound exhibits a monotonically non-decreasing trend, gradually approaching the true optimal value, such as... Figure 4 As shown, Figure 4 This is the flow of the strategy-based subspace collaborative optimization algorithm in this embodiment of the invention.

[0077] For the subspace model, the objective is to minimize the overall system operating cost under a specified scenario c. This cost comprehensively considers the operating cost of traditional thermal power, the penalty for curtailment of renewable energy, and the penalty for load shedding. Its expression is as follows:

[0078] in, C mp,c This represents the overall operating cost of the system. Indicates the time period within the day. This represents a collection of thermal power units. Indicates thermal power unit j During the period t Those who have made contributions Indicates thermal power unit j The power generation cost coefficient, and These represent the unit penalty costs for renewable energy curtailment and load reduction, respectively. and These represent the curtailment of renewable energy and the reduction of load power, respectively.

[0079] The subspace model must satisfy the following simulation constraints to ensure the safe, stable, and economical operation of the system: 1. Node power balance constraints:

[0080] in, This represents a collection of hydroelectric generating units. This refers to a collection of centralized photovoltaic power plants. This represents a collection of wind turbines. Indicates the time period t Net load power, Represents a node a During the period t Reduce load power, and Representing nodes respectively a The power generation and pumping power of the pumped storage station. Indicates the outflow node of the line. a power, Indicates the time period of centralized photovoltaic power plants t Those who have made contributions Indicates thermal power unitj During the period t Those who have made contributions Indicates the time period of the hydropower unit t Those who have made contributions Indicates the time period of the wind turbine. t Those who have made contributions.

[0081] 2. Conventional unit operating constraints: The formulas for the upper and lower limits of thermal power unit output and the ramping constraint are as follows:

[0082]

[0083] in, This indicates the lower limit of the output of thermal power units. This indicates the upper limit of the output of the thermal power unit. Indicates thermal power unit j During the period t Those who have made contributions This indicates the change in power output of a thermal power unit over a given period of time. This indicates the maximum climbing height.

[0084] The formula for the daily power generation constraint of thermal power units is as follows:

[0085] in, Indicates the time period of the hydropower unit t Those who have made contributions Indicates a unit of time. This indicates the daily power generation limit of a thermal power unit.

[0086] 3. Operating constraints of pumped-storage units: The formula for dynamic balance of storage capacity is as follows:

[0087] in, This indicates that the pumped storage power station reservoir is at the node. Time period The energy storage state quantity under the following conditions and These represent the charging efficiency and discharging efficiency of the pumped storage system, respectively. and Representing nodes respectively a The power generation and pumping power of the pumped storage station.

[0088] The formulas for storage capacity and power limits are as follows:

[0089]

[0090] in, Indicates the first k Investment decisions in round 1 iteration Indicates the maximum discharge power. and These are binary operating status flag variables, representing whether the pumped storage is in a charging or discharging state. This indicates the maximum charging power.

[0091] The formulas for mutual exclusion of operating modes and constraints on the number of daily start-stop cycles are as follows:

[0092]

[0093] in, Indicates the k-th The binary investment decision variable determined in the first iteration, concerning whether to construct a pumped storage power station at node a. and These represent the maximum number of charging starts and discharging starts allowed for a pumped storage power station per day.

[0094] The formula for the consistency constraint of storage capacity at the beginning and end of the cycle (to achieve daily cycle sustainability) is as follows:

[0095] in, As a variable for investment decisions regarding pumped storage capacity from the main space, it is an important coupling variable connecting the optimization of the main space and the subspace. Let represent the energy storage state quantity of the pumped storage power station reservoir at node a, time period T, where T is the last moment of the day. This represents the energy storage status of the pumped storage power station reservoir at node a, time period 1, where 1 represents the initial time of the day. w This represents the ratio of the initial / final reservoir capacity of a pumped storage power station to its rated capacity.

[0096] 4. Constraints on renewable energy consumption and power supply reliability:

[0097]

[0098] in, This indicates the minimum absorption rate of new energy sources. Indicates the maximum allowable load reduction percentage. This represents the unutilized photovoltaic power generation during time period t. Indicates the time period of the wind turbine. t Those who have made contributions This represents the unutilized wind power generation capacity during time period t.

[0099] To demonstrate the effectiveness of the above method, numerical experiments were conducted using an improved Nas-6 node test system. This system includes various conventional generator sets, energy storage units, and wind turbines, making it suitable for verifying pumped storage power station planning problems. Centralized photovoltaic and conventional hydropower units were added to the original data. The modified system structure is as follows: Figure 5 As shown, Figure 5 This is a modified power system structure diagram in an embodiment of the present invention, and it is assumed that nodes A1–A6 are all candidate construction locations for pumped storage power stations.

[0100] Centralized photovoltaic (PV) and wind power output and load data were obtained from the full-year actual operating data of 2024 published by Belgian grid operator Elia, with a raw resolution of 15 minutes. After sampling and normalization, time-series data with a time resolution of 1 hour and PV and wind power penetration rates of 60% and 40% respectively, defined by peak load ratio, were generated.

[0101] 1. Multimodal scene analysis The multimodal scene reconstruction method proposed in this invention ultimately generates 10 typical operating scenarios, including 4 original cluster center sample scenarios after removing outliers and 6 outlier cluster center sample scenarios. Their power time-series curves are shown below. Figure 6 and Figure 7 As shown, Figure 6 This is a schematic diagram of the original cluster center samples after removing outliers in an embodiment of the present invention. Figure 7 This is a schematic diagram of outlier cluster center samples in an embodiment of the present invention. Figure 6 A schematic diagram containing the original cluster center samples 0-3. Figure 7 The diagram illustrates the curves representing outlier clusters with centers 0-5. The red curve represents load power, the yellow curve represents photovoltaic power, and the blue curve represents wind power. Based on different operating characteristics, they can be categorized into four representative modes to facilitate understanding the actual physical conditions corresponding to each cluster: The first category is "PV-dominated days" with wind power as a supplement, taking the original cluster 3 as an example. These days are mostly during periods of clear weather and abundant sunshine. During the day, PV resources are plentiful, and the output curve is typically "bell-shaped," reaching its peak at midday, making it the core force for power supply. Wind power output is relatively stable and at a medium or low level, playing a supporting role. Attention should be paid to the excess power generation at midday and the power gap filled after PV power is phased out in the evening.

[0102] The second category is "wind-dominated days" where wind power is dominant and solar power is weak, with outlier cluster 2 being a representative example. This is common in situations where wind resources are abundant but sunlight is average. Wind power output remains at a high level at night, making it the main source of electricity supply; solar power output contributes only to a limited extent, and "wind curtailment" is likely to occur when the load is low. Efficiently absorbing wind power is a key challenge.

[0103] The third category is "traditional power supply guarantee days" where both wind and solar power are weak and the load relies on traditional power sources, taking outlier cluster 0 as an example. These days often occur in cloudy, rainy weather with little sunshine and low wind speeds. The output of photovoltaic and wind power is extremely low, almost unable to support the power system. The load demand is entirely met by traditional power sources such as thermal power and hydropower, resulting in high peak-shaving pressure on traditional power sources.

[0104] The fourth category is the "complex changing day" where wind and solar power and load interact in a complex way. Outlier cluster 4 is the representative of this category. The output of photovoltaic and wind power and load changes are intertwined. In the evening, there may be situations such as a rapid drop in photovoltaic output, a sharp rise in load, and fluctuations in wind power output. This requires high grid coordination and dispatch capabilities and needs to ensure a stable power supply.

[0105] 2. Solution Verification Using the cyber-physical fusion-driven strategic subspace game method proposed in this invention for global optimization calculations, the final planning scheme determines the deployment of pumped storage power stations with rated capacities of 1420MW and 1500MW at nodes A2 and A4, respectively. Multi-scenario operation simulation analysis shows that the system's annual renewable energy absorption rate reaches 99.19%, with solar and wind curtailment rates controlled at 1.05% and 0.60%, respectively. No active load shedding occurred throughout the entire operating cycle, fully demonstrating the comprehensive advantages of the planning scheme in terms of economy, power supply reliability, and renewable energy acceptance capacity.

[0106] We specifically selected the central sample of outlier cluster 3 for operational characteristic analysis. This sample corresponds to the combined operating conditions of high irradiance and strong winds in summer. Its typical characteristics are a sharp drop in photovoltaic power, violent fluctuations in wind power output, and a rapid increase in load demand in the evening. Figure 8 This diagram illustrates the operational status of each device in the outlier clustering 3-center sample of this invention, showcasing the output response of each unit under the optimal pumped storage configuration. During the midday peak photovoltaic power generation period from 09:00 to 12:00, thermal power output is suppressed to the technical lower limit. During the evening peak, thanks to the continuous injection of wind power resources, the system successfully meets peak load demand through the coordinated discharge of hydropower units and pumped storage, effectively reducing dependence on high-priced thermal power. During the midday off-peak period, the pumped storage units operate at full capacity, significantly improving the absorption level of excess photovoltaic power.

[0107] Figure 9This is a schematic diagram of the effective energy storage curve of the reservoir on the outlier cluster 3-center sample in this embodiment of the invention, further revealing the coupling relationship between the operation strategy of the pumped storage power station and the net load of the system. Both pumped storage power stations pump water at maximum capacity during the peak photovoltaic power generation period, effectively smoothing out the net load trough. Specifically, the small-capacity units at node A2 mainly serve the regulation of the evening peak load, while the large-capacity units at node A4 focus on power support during the morning peak period.

[0108] Figure 10 This is a schematic diagram of the wind and solar power curtailment curves for outlier cluster 3-center samples in an embodiment of the present invention. Figure 10 The data shows that a small amount of solar and wind power curtailment still occurred between 07:00 and 15:00. During this period, pumped-storage units continued to operate at full capacity, indicating that the power generation from renewable energy sources had exceeded the upper limit of the system's real-time regulation capacity. This phenomenon highlights the need to introduce diversified flexible resources, in addition to pumped-storage power stations, in high-proportion renewable energy systems to comprehensively improve the system's regulation capacity and operational resilience.

[0109] In summary, this method first constructs a multimodal operation feature extraction and reconstruction framework through cyber-physical system coupling modeling. It then employs a density-sensitive message passing clustering algorithm to autonomously identify typical system operating states, and combines this with a neighborhood relative outlier evaluation algorithm to detect abnormal operating modes, forming a structured set of scenarios representing the system's multidimensional uncertainties. Based on this, a global optimization model considering investment decisions and multi-scenario operation collaboration is established to achieve cross-level collaborative optimization between planning and operation. Addressing the model's high dimensionality, nonlinearity, and complex constraints, a strategic subspace game-theoretic solution architecture is designed. Through an adaptive feasible cut mapping mechanism, a game interaction and collaborative optimization relationship is established between the principal and subspaces. Analysis of simulation results obtained under the optimal pumped storage investment scheme demonstrates that the proposed method exhibits superior performance in terms of economy, power supply reliability, renewable energy absorption capacity, and global adaptability.

[0110] This invention provides a pumped-storage energy storage global optimization control device. The following description of the pumped-storage energy storage global optimization control device is provided. The pumped-storage energy storage global optimization control device described below can be referred to in correspondence with the pumped-storage energy storage global optimization control method described above. The device includes: The module is used to build and operate a modal perception and reconstruction mechanism through cyber-physical fusion modeling, and generate a set of structured scenarios that can comprehensively characterize the uncertainties of the power system; The optimization module is used to construct a global optimization control model by combining a set of structured scenarios, and to obtain the optimal pumped storage planning scheme for the power system through the global optimization control model. The global optimization control model adopts a master-subspace collaborative optimization architecture, which includes a master space model and a subspace model.

[0111] In operation, this device first constructs a cyber-physical model to build an operational modality perception and reconstruction mechanism, generating a structured scenario set covering multiple operational modalities. Then, the optimization module combines this structured scenario set to construct a global optimization control model, which then obtains the optimal pumped storage planning scheme for the power system. The global optimization control model adopts a master-subspace collaborative optimization architecture, including a master space model and a subspace model. By introducing an adaptive feasible cut mapping strategy, a game-theoretic interaction framework is built between the master space model and the subspace model, enhancing decision-making collaboration and convergence efficiency. This improves the reliability of the optimal pumped storage planning scheme and effectively addresses complex power system operational constraints, ensuring the optimality of the planning results and solution efficiency. This solves the problem of poor power system planning and control performance in existing related technologies.

[0112] Figure 11 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 11 As shown, the electronic device may include: a processor 1101, a communication interface 1102, a memory 1103, and a communication bus 1104, wherein the processor 1101, the communication interface 1102, and the memory 1103 communicate with each other through the communication bus 1104. The processor 1101 can call logic instructions in the memory 1103 to execute a pumped storage full-domain optimization control method, which includes: By using cyber-physical fusion modeling, a mechanism for operational modality perception and reconstruction is constructed to generate a set of structured scenarios that can comprehensively characterize the uncertainties of the power system; A global optimization control model is constructed by combining a set of structured scenarios, and the optimal pumped storage planning scheme for the power system is obtained through the global optimization control model. The global optimization control model adopts a master-subspace collaborative optimization architecture, which includes a master space model and a subspace model.

[0113] Furthermore, the logical instructions in the aforementioned memory 1103 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0114] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer is able to execute the pumped storage full-domain optimization control method provided by the above methods, the method including: By using cyber-physical fusion modeling, a mechanism for operational modality perception and reconstruction is constructed to generate a set of structured scenarios that can comprehensively characterize the uncertainties of the power system; A global optimization control model is constructed by combining a set of structured scenarios, and the optimal pumped storage planning scheme for the power system is obtained through the global optimization control model. The global optimization control model adopts a master-subspace collaborative optimization architecture, which includes a master space model and a subspace model.

[0115] In another aspect, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the pumped-storage full-domain optimization control method provided by the above methods, the method comprising: By using cyber-physical fusion modeling, a mechanism for operational modality perception and reconstruction is constructed to generate a set of structured scenarios that can comprehensively characterize the uncertainties of the power system; A global optimization control model is constructed by combining a set of structured scenarios, and the optimal pumped storage planning scheme for the power system is obtained through the global optimization control model. The global optimization control model adopts a master-subspace collaborative optimization architecture, which includes a master space model and a subspace model.

[0116] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0117] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0118] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A pumped storage global optimization control method, characterized by, The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling.

2. The pumped storage global optimization control method according to claim 1, characterized by, The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling.

3. The pumped storage global optimization control method according to claim 2, characterized by, The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling.

4. The pumped storage global optimization control method according to claim 2, characterized by, The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling.

5. The pumped storage global optimization control method according to claim 4, characterized by, The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling.

6. The pumped storage global optimization control method according to claim 5, characterized by, The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling.

7. The pumped storage global optimization control method according to claim 4, characterized by, The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal pumped storage planning method based on information-physical fusion modeling. The application relates to a power system optimal calculating reachable distances of the object relative to each object in the neighborhood; calculating local reachable density in the neighborhood of the object; obtaining a neighborhood relative outlying degree evaluation value by comparing the local reachable density of the object with the local reachable density of its K-neighbors, and determining whether the object is an outlying abnormal sample according to the relative outlying degree evaluation value.

8. The pumped hydro global optimization control method of claim 1, wherein, constructing a global optimization control model in combination with the structured scenario set, and obtaining an optimal pumped storage planning scheme of the power system through the global optimization control model, including: modeling the power grid based on an existing power transmission distribution factor model; adopting a strategic subspace game solving mechanism, taking the main space model as a decision subject, and realizing coordination and constraint satisfaction between the main space and the subspace through an adaptive feasible cut mapping strategy to obtain the optimal pumped storage planning scheme of the power system.

9. A pumped storage all-domain optimization control device characterized by, including: a construction module configured to construct a structured scenario set capable of comprehensively representing uncertainty of the power system by information-physical fusion modeling and constructing a running mode perception and reconstruction mechanism; an optimization module configured to construct a global optimization control model in combination with the structured scenario set, and obtain an optimal pumped storage planning scheme of the power system through the global optimization control model; the global optimization control model adopts a main-subspace coordinated optimization architecture, including a main space model and a subspace model.

10. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, the processor implements the pumped storage global optimization control method according to any one of claims 1 to 8 when executing the program.

Images