Flexible resource multi-level aggregation scheduling method for source network load storage integrated park

Abstract

The application belongs to the technical field of resource scheduling, and specifically discloses a flexible resource multi-level aggregation scheduling method for a source network load storage integrated park, which comprises the following steps: performing multi-source fusion on historical equipment operation data of a target park to obtain time series data information of the target park; performing topology reconstruction on the dependency relationship of the time series data information to obtain a correlation strength matrix of the time series data information; coupling and clustering energy supply characteristics and energy demand characteristics of the target park to obtain a self-balancing relationship group of the target park; performing fitting degree optimization derivation on the self-balancing relationship group to obtain an independent energy storage scheduling plan; performing mutual aid collaborative arrangement on the energy surplus and deficit between the self-balancing relationship groups to obtain a residual energy exchange scheme between the self-balancing relationship groups; and integrating and encoding the independent energy storage scheduling plan and the residual energy exchange scheme to obtain a full-park scheduling instruction of the target park. The application can improve the efficiency of flexible resource multi-level aggregation scheduling for a source network load storage integrated park.

Description

Technical Field

[0001] This invention belongs to the field of resource scheduling technology, and in particular relates to a flexible multi-level aggregation scheduling method for integrated source-grid-load-storage parks. Background Technology

[0002] Integrated energy-grid-load-storage parks are comprehensive energy systems that integrate energy production, grid transmission, load consumption, and energy storage regulation. Their core function is to achieve autonomous energy balance and efficient utilization within the park by integrating distributed power sources, distribution networks, various electrical loads, and energy storage equipment. Flexible and precise resource aggregation and scheduling are crucial for ensuring the system's stable operation and improving energy allocation efficiency. Currently, industry scheduling technologies for such parks primarily rely on traditional decentralized control approaches. Faced with multi-source, heterogeneous equipment operation data within the park, there is a lack of systematic fusion and processing mechanisms. This fails to effectively resolve inherent differences in spatiotemporal benchmarks and semantic descriptions among different equipment data, and it is also difficult to accurately clean and organize abnormal or incomplete raw data. This results in insufficient uniformity and reliability of basic data, hindering effective support for in-depth correlation analysis of energy supply and demand characteristics and making it difficult to accurately uncover the intrinsic dependencies between energy supply and demand.

[0003] Existing scheduling methods lack a scientific aggregation scheduling logic for the multi-level energy structure of integrated energy source-grid-load-storage parks. They also lack quantitative evaluation standards based on data correlation strength for the coupled clustering of energy supply and demand characteristics, making it difficult to form stable and self-balancing relationship groups that are adapted to the actual situation of the parks. At the same time, the formulation of energy storage scheduling plans and the scheduling of inter-group energy surplus and deficit mutual assistance are independent and disconnected, failing to fully combine the temporal variation characteristics and data correlation patterns of various energy resources. This significantly reduces the complementary synergy of energy surplus and deficit within the park, making it impossible to achieve flexible global optimization of resource allocation. Ultimately, this results in low overall energy scheduling efficiency, insufficient control precision and flexibility, and an inability to meet the refined, collaborative, and efficient scheduling development needs of integrated energy source-grid-load-storage parks. Summary of the Invention

[0004] In view of the shortcomings of the prior art, the purpose of this invention is to provide a method for flexible multi-level aggregation scheduling of resources in integrated source-grid-load-storage parks, which can improve the efficiency of flexible multi-level aggregation scheduling of resources in integrated source-grid-load-storage parks.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0006] The flexible multi-level aggregation and scheduling method for integrated source-grid-load-storage parks includes the following steps:

[0007] S01. Perform multi-source fusion on the historical equipment operation data of the target park to obtain the time-series data information of the target park;

[0008] S02. Perform topological reconstruction on the dependencies of time-series data information to obtain the correlation strength matrix of time-series data information;

[0009] S03. Based on the correlation strength matrix, the energy supply characteristics and energy demand characteristics of the target park are coupled and clustered to obtain the self-balancing relationship group of the target park.

[0010] S04. Based on the correlation strength matrix, the fit optimization derivation of the self-balancing relationship group is carried out to obtain the independent energy storage scheduling plan of the self-balancing relationship group.

[0011] S05. Based on the independent energy storage dispatch plan, the energy surplus and deficit among the self-balancing relationship groups are coordinated and arranged in a mutually supportive manner to obtain the remaining energy exchange scheme among the self-balancing relationship groups.

[0012] S06. Integrate and encode the independent energy storage dispatch plan and the remaining energy exchange plan to obtain the whole park dispatch instructions for the target park.

[0013] Preferably, in S01, the process of obtaining the time-series data information of the target park is as follows:

[0014] Spatiotemporal alignment analysis is performed on the historical equipment operation data of the target park to obtain a unified spatiotemporal reference for the historical equipment operation data.

[0015] Semantic conflict resolution is performed on the unified expression of the spatiotemporal reference to obtain a semantically consistent representation of historical equipment operation data;

[0016] Anomaly and defect cleaning are performed on semantically consistent representations to obtain regularized data of historical equipment operation data;

[0017] The regularized data is encapsulated in a unified structure to obtain the time-series data information of the target park.

[0018] Preferably, in S02, the process of obtaining the correlation strength matrix of time-series data information is as follows:

[0019] By performing lag cross-analysis on time series data, potential dependency patterns of the time series data can be obtained.

[0020] Granger causality tests were performed on the potential dependency patterns, and the significance thresholds were used to screen the potential dependency patterns to obtain the significant causal associations of the potential dependency patterns.

[0021] Significant causal relationships are quantified by time lag and intensity weighting to obtain the association weights of significant causal relationships;

[0022] Based on the association weights, significant causal associations are structured and mapped to obtain the association strength matrix of time-series data information.

[0023] Preferably, in S03, the process of obtaining the self-balancing relation set of the target park is as follows:

[0024] Peak intensity detection is performed on the correlation intensity matrix to obtain the peak coordinates of the correlation intensity matrix;

[0025] Based on the peak coordinates, a two-way correlation mapping is performed on the energy supply characteristics and energy demand characteristics of the target park to obtain the basic balance relationship of the target park;

[0026] Based on the correlation strength matrix, the cohesion strength integral of the basic equilibrium relationship is performed to obtain the equilibrium metric value of the basic equilibrium relationship;

[0027] Based on the balance metric, the nearest neighbor feature is explored to obtain candidate features of the basic balance relationship;

[0028] The candidate features are evaluated for balance improvement, and based on the evaluation results, the basic balance relationship is incorporated to obtain an extended balance body of the basic balance relationship;

[0029] By resolving the affiliation conflicts of the extended equilibrium body, a self-balancing relationship group of the target park is obtained.

[0030] Preferably, the equilibrium metric value for obtaining the basic equilibrium relationship is as follows:

[0031] Based on the fundamental balance relationship, the correlation submatrix of the correlation strength matrix is ​​extracted;

[0032] Nonlinear dispersion analysis was performed on the temporal intensity of the correlation submatrix to obtain the temporal volatility characteristics of the correlation submatrix;

[0033] Based on the correlation submatrix and temporal fluctuation characteristics, the cohesion strength of the fundamental equilibrium relationship is calculated. The formula for calculating the cohesion strength is as follows:

[0034] ;

[0035] In the formula, S represents the cohesive strength, N represents the number of equally spaced time points in the time series data, M represents the total number of basic equilibrium relationships, u represents the continuous time variable, and U represents the continuous time series interval. A characteristic pair in the basic equilibrium relationship, where x represents energy supply characteristics, y represents energy demand characteristics, and V represents the basic equilibrium relationship. For feature pairs in the correlation submatrix The strength of the association, Feature pairs in time series volatility characteristics The temporal variance of the association strength, where e is the natural constant;

[0036] By scaling the cohesive strength to a uniform scale, a balance metric value for the basic equilibrium relationship is obtained.

[0037] Preferably, in S04, the process of obtaining the independent energy storage dispatch plan of the self-balancing relationship group is as follows:

[0038] Based on the correlation strength matrix, intra-group temporal intensity mapping is performed on the self-balancing relation group to obtain the temporal supply and demand sequence of the self-balancing relation group;

[0039] Based on the time-series supply and demand sequence, pattern matching is performed on the charging and discharging behavior of energy storage devices in the target park to obtain candidate energy storage actions for the target park.

[0040] The time-coverage fit quantification of candidate energy storage actions yields the intensity maintenance degree of the candidate energy storage actions;

[0041] Based on the intensity maintenance degree, the candidate energy storage actions are prioritized and ranked. Based on the ranking results, the candidate energy storage actions are sequentially connected to obtain the energy storage scheduling sequence of the self-balancing relationship group.

[0042] Based on the correlation strength matrix, the energy storage scheduling sequence is reverse-verified, and the verified sequence is used as an independent energy storage scheduling plan for the self-balancing relation group.

[0043] Preferably, the strength maintenance of the candidate energy storage action is obtained as follows:

[0044] The action time periods of candidate energy storage actions are analyzed to obtain the action time periods of the candidate energy storage actions;

[0045] Based on the correlation strength matrix, the correlation strength values ​​for the action period are extracted;

[0046] The dimensions of the candidate energy storage actions are normalized to obtain the normalized power of the candidate energy storage actions;

[0047] Statistical distribution characteristics analysis of time-period correlation strength values ​​is performed to obtain the local consistency factor of time-period correlation strength values;

[0048] Based on normalized power, time-period correlated intensity values, and local consistency factors, the intensity sustaining degree of candidate energy storage actions is calculated. The formula for calculating the intensity sustaining degree is as follows:

[0049] ;

[0050] In the formula, Z represents the strength retention rate. Let t be a specific time interval of the action period. The normalized power for time period t, This refers to the rated maximum power of the energy storage device. Let be the correlation strength value for time period t, and e be the natural constant. This represents the average of the correlation strength values. This represents the standard deviation of the correlation strength values.

[0051] Preferably, in S05, the process of obtaining the residual energy exchange scheme between the self-balancing relationship groups is as follows:

[0052] By performing an expected state simulation of the independent energy storage dispatch plan, the surplus trajectory and deficit trajectory of the self-balancing relationship group are obtained;

[0053] Based on the surplus trajectory and the gap trajectory, bidirectional collaborative interaction is performed between self-balancing relationship groups to obtain a list of interactive relationships between self-balancing relationship groups.

[0054] Based on the association strength matrix, the effectiveness of the interactive relationship list is evaluated, and the priority weight distribution of the interactive relationship list is obtained.

[0055] Based on priority weight distribution, the interactive relationship list is standardized and integrated to obtain the residual energy exchange scheme between self-balancing relationship groups.

[0056] Preferably, the priority weight distribution of the interactive relationship list is obtained as follows:

[0057] Feature extraction is performed on the interactive relationship list to obtain the relationship pair feature identifiers of the interactive relationship list;

[0058] Based on the feature identification of the relationship pair, the coupling dimension of the association strength matrix is ​​extended to obtain the second-order association matrix of the association strength matrix;

[0059] Based on the correlation strength matrix and the second-order correlation matrix, the influence of relations on feature labels is analyzed to obtain the composite correlation strength between relations and feature labels;

[0060] The overlap between the surplus trajectory and the gap trajectory is inferred by time period coverage, and the overlap between the surplus trajectory and the gap trajectory is obtained.

[0061] The priority weight distribution of the interactive relationship list is obtained by dynamically weighting and fusing the composite association strength and trajectory overlap.

[0062] Preferably, in S06, the process of obtaining the overall scheduling instruction for the target park is as follows:

[0063] The independent energy storage dispatch plan and the surplus energy exchange plan are analyzed in terms of dispatch action items to obtain the dispatch action item set of the target park;

[0064] Align the scheduling action entry set with the reference time to obtain the timing alignment entries of the scheduling action entry set;

[0065] Consistent coordination is performed on the operational conflicts of the timing alignment entries to obtain a conflict-free scheduling flow for the timing alignment entries;

[0066] The conflict-free scheduling flow is encapsulated into instructions to obtain the overall scheduling instructions for the target park.

[0067] The present invention has the following beneficial effects:

[0068] This invention uses multi-source fusion technology to standardize historical equipment operation data, achieving spatiotemporal benchmark unification, semantic conflict resolution, and anomaly removal, ensuring the accuracy and consistency of time-series data information. Then, a correlation strength matrix is ​​constructed through topological reconstruction to accurately capture the dependencies between data. Combined with coupled clustering, a stable self-balancing relationship group is formed, providing a scientific and reliable foundation for scheduling decisions and improving the rationality of scheduling schemes.

[0069] This invention optimizes the fit by relying on the correlation strength matrix, making the independent energy storage scheduling plan highly compatible with the supply and demand characteristics and enhancing the accuracy of single-group scheduling. Through the inter-group energy surplus and deficit mutual assistance arrangement and the integrated coding of scheduling instructions, it realizes the efficient flow and conflict-free scheduling of surplus energy, significantly improves the efficiency of multi-level aggregation scheduling of flexible resources in the park, and optimizes the energy resource allocation effect. Attached Figure Description

[0070] Figure 1 This is a schematic flowchart of the method of the present invention;

[0071] Figure 2 This is a schematic diagram of the scheduling curve of the first group of energy storage devices during the verification process of this invention;

[0072] Figure 3 This is a schematic diagram of the scheduling curve of the second group of energy storage devices during the verification process of this invention;

[0073] Figure 4 This is a schematic diagram of the scheduling curve of the third group of energy storage devices during the verification process of this invention. Detailed Implementation

[0074] Example 1: As Figure 1 As shown, the flexible multi-level aggregation and scheduling method for integrated source-grid-load-storage parks includes the following steps:

[0075] S01. Perform multi-source fusion on the historical equipment operation data of the target park to obtain the time-series data information of the target park;

[0076] S02. Perform topological reconstruction on the dependencies of time-series data information to obtain the correlation strength matrix of time-series data information;

[0077] S03. Based on the correlation strength matrix, the energy supply characteristics and energy demand characteristics of the target park are coupled and clustered to obtain the self-balancing relationship group of the target park.

[0078] S04. Based on the correlation strength matrix, the fit optimization derivation of the self-balancing relationship group is carried out to obtain the independent energy storage scheduling plan of the self-balancing relationship group.

[0079] S05. Based on the independent energy storage dispatch plan, the energy surplus and deficit among the self-balancing relationship groups are coordinated and arranged in a mutually supportive manner to obtain the remaining energy exchange scheme among the self-balancing relationship groups.

[0080] S06. Integrate and encode the independent energy storage dispatch plan and the remaining energy exchange plan to obtain the whole park dispatch instructions for the target park.

[0081] In S01, the time-series data information of the target park is obtained, including:

[0082] Spatiotemporal alignment analysis is performed on the historical equipment operation data of the target park to obtain a unified spatiotemporal reference for the historical equipment operation data.

[0083] Semantic conflict resolution is performed on the unified expression of the spatiotemporal reference to obtain a semantically consistent representation of historical equipment operation data;

[0084] Anomaly and defect cleaning are performed on semantically consistent representations to obtain regularized data of historical equipment operation data;

[0085] The regularized data is encapsulated in a unified structure to obtain the time-series data information of the target park.

[0086] Collect historical operational data from all relevant equipment within the target park, covering various information recorded by different devices during their respective operations. Establish a unified time benchmark within the park, selecting the time records of the park's core control nodes as the standard. Adjust all timestamps of different formats in other equipment data to this standard time format, ensuring complete consistency in the starting point and interval of all data recordings in the time dimension. Define the installation location information of each device, and standardize the spatial location description of all devices according to a pre-set unified location coding rule. This ensures that each piece of data can accurately correspond to a unified spatial identifier, ultimately forming a unified spatiotemporal benchmark expression, providing a consistent temporal and spatial reference for all operational data from different devices.

[0087] We systematically reviewed the descriptive terms and information meanings of all data in the unified spatiotemporal reference representation, comparing how different devices described the same type of operating state or parameter. For terms with different expressions but the same core meaning, we replaced them with standard terms, referring to the data description specifications commonly used within the park. When encountering contradictory data, we consulted the corresponding equipment's operation manual and original data recording rules to verify the true meaning of the data. We replaced the conflicting content with the expression that conforms to the actual operating logic of the equipment, ensuring that all data uses completely consistent terminology and logic when describing the same type of information. Ultimately, this resulted in a semantically consistent representation, eliminating semantic differences and ambiguities between data from different devices.

[0088] Each data point in the semantically consistent representation is checked individually. Based on the normal operating range and common operating states of the equipment, data that does not match the actual operating conditions of the equipment is identified. Abnormal data such as energy consumption or operating parameters recorded even when the equipment is not running are directly removed from the dataset. For missing entries in the data records, the data change patterns are analyzed by combining the continuous operating data of the equipment before and after the missing period. Simultaneously, the corresponding data content of similar equipment under the same operating scenario is referenced to supplement the missing information, ensuring that each data point fully reflects the actual operating state of the equipment in the corresponding time and space. After this processing, a well-organized dataset is obtained, with no abnormal records or missing information, accurately presenting the historical operating status of the equipment.

[0089] The standardized data is categorized and organized according to the attributes it reflects, into different categories such as energy input-related data, energy output-related data, and equipment status-related data. For each category, a unified recording format is established, clearly defining the order of information and field names within each data entry to ensure complete structural consistency within the same data type. All categorized and standardized data is then organized according to this unified format, ensuring that each data entry includes key elements such as time signatures, spatial identifiers, and descriptive content. Subsequently, all data is arranged chronologically to form an ordered and clearly structured dataset, ultimately yielding time-series data information. This information clearly displays the operational status and related data of all equipment within the park at different time points, providing reliable foundational data support for subsequent dependency analysis.

[0090] Acquiring time-series data provides a unified temporal and spatial reference for the historical operational data of different devices within the target park, eliminating differences in the spatiotemporal dimensions of data from different devices. By standardizing data description terminology and logic, semantic conflicts are effectively resolved, ambiguities in data understanding are avoided, and consistency in the expression of similar information is ensured. Simultaneously, outlier data is removed and missing information is supplemented, ensuring that the data accurately and completely reflects the operational status of the devices, thus improving data quality. After classification, organization, and standardized formatting, the resulting time-series data has a clear and orderly structure, showcasing the operational status of various devices at different time points. This provides reliable and high-quality foundational data support for subsequent steps such as time-series data dependency analysis and correlation strength matrix construction, ensuring the smooth progress and accuracy of subsequent stages of the overall scheduling method.

[0091] In S02, the correlation strength matrix of the time-series data information is obtained, including:

[0092] By performing lag cross-analysis on time series data, potential dependency patterns of the time series data can be obtained.

[0093] Granger causality tests were performed on the potential dependency patterns, and the significance thresholds were used to screen the potential dependency patterns to obtain the significant causal associations of the potential dependency patterns.

[0094] Significant causal relationships are quantified by time lag and intensity weighting to obtain the association weights of significant causal relationships;

[0095] Based on the association weights, significant causal associations are structured and mapped to obtain the association strength matrix of time-series data information.

[0096] The potential time delay intervals for the covered time-series data are determined, based on the data recording period and common delay scenarios in actual operation. For all potentially correlated data sequence combinations in the time-series data, two data sequences from each combination are selected one by one. One of the data sequences is then shifted sequentially according to a set lag period, with the shift interval consistent with the data recording interval. At each lag period, the shifted sequence is compared with the non-shifted sequence at the corresponding time point, recording the fit between the two sets of data under that lag state, including the consistency of data change trends and the correlation of numerical fluctuations. A comprehensive analysis of the correlation performance of each data sequence combination under all lag periods is conducted, summarizing those correlation forms with clear time correspondences and signs of mutual influence, ultimately forming the potential dependency patterns of the time-series data.

[0097] For each potential dependency pattern, the roles of the two data sequences involved in the association are clearly defined, distinguishing between possible causal sequences and outcome sequences. The current state of the outcome sequence is described using only its own historical data, versus using both its own historical data and the historical data of the causal sequences. The degree of fit between these two descriptions is observed. If adding the historical data of the causal sequence results in a more accurate description of the current state of the outcome sequence, and this improvement in accuracy meets the preset judgment criteria, then the potential dependency pattern is considered to have a causal relationship. A fixed significance judgment criterion is established, based on the normal fluctuation range of the data and a reference threshold for the validity of the association. The improvement in fit corresponding to each potential dependency pattern that has been tested for causal relationship is compared with this criterion. Causal relationships that fully meet the criterion are retained, thus obtaining the significant causal relationships of the potential dependency patterns.

[0098] For each significant causal relationship, accurately record the time lag between the impact of the causal sequence and the result sequence, i.e., the time interval from the change in the causal sequence to the corresponding change in the result sequence. Assign corresponding time lag weights based on how well the time lag matches the data's operational patterns. The more the time lag matches the typical time characteristics of this type of data relationship, the more the assigned time lag weight reflects its actual impact value.

[0099] Simultaneously, the stability and coverage of the influence of the causal sequence on the outcome sequence in significant causal associations are analyzed to determine the association strength. Corresponding strength weights are assigned based on the actual performance of the association strength; the more stable the influence and the closer the coverage aligns with the data association requirements, the higher the strength weight. The time lag weight and strength weight of each significant causal association are then reasonably integrated, with the integration method determined based on the degree of influence of both on the importance of the association. Finally, an association weight that comprehensively reflects the importance of the significant causal association is formed.

[0100] We analyze all data sequences involved in significant causal relationships, assigning a unique identifier to each sequence to ensure accurate differentiation during subsequent matrix construction. Using the unique identifiers of the data sequences as row and column identifiers, we construct an initial matrix with the same number of rows and columns as the total number of data sequences. For each significant causal relationship, based on the identifiers of the two involved data sequences, we find the intersection of the corresponding row and column in the matrix and accurately fill in the association weight corresponding to that significant causal relationship at that position.

[0101] For data sequence combinations that do not exhibit significant causal relationships, zero values ​​are filled at the corresponding row and column intersections in the matrix to ensure that the matrix can fully represent the relationships between all data sequences. Through this structured arrangement, the association weights of all significant causal relationships are systematically integrated, ultimately forming an association strength matrix that clearly reflects the degree of association between data sequences in time-series data.

[0102] This step accurately captures the potential correlations between data sequences in time-series data. By scientifically defining time delay intervals and performing sequence shift comparisons, it comprehensively analyzes the temporal correspondences and mutual influences between data, forming accurate potential dependency patterns. By clarifying the roles of causal sequences and comparing their fit, combined with fixed significance judgment criteria, it filters out effective significant causal associations and eliminates meaningless correlation interference. By recording time delay durations and assigning appropriate time delay weights, it analyzes the stability and coverage of associations and assigns corresponding strength weights. The association weights formed after reasonable fusion can comprehensively reflect the importance of significant causal associations. Finally, the association strength matrix formed through structured mapping clearly presents the degree of correlation between all data sequences, providing accurate and reliable foundational support for subsequent steps such as energy supply and demand characteristic coupling clustering and energy storage dispatch plan formulation based on the association strength matrix, ensuring the orderly progress of the overall dispatch process.

[0103] In S03, the self-balancing relationship set of the target park is obtained, including:

[0104] Peak intensity detection is performed on the correlation intensity matrix to obtain the peak coordinates of the correlation intensity matrix;

[0105] Based on the peak coordinates, a two-way correlation mapping is performed on the energy supply characteristics and energy demand characteristics of the target park to obtain the basic balance relationship of the target park;

[0106] Based on the correlation strength matrix, the cohesion strength integral of the basic equilibrium relationship is performed to obtain the equilibrium metric value of the basic equilibrium relationship;

[0107] Based on the balance metric, the nearest neighbor feature is explored to obtain candidate features of the basic balance relationship;

[0108] The candidate features are evaluated for balance improvement, and based on the evaluation results, the basic balance relationship is incorporated to obtain an extended balance body of the basic balance relationship;

[0109] By resolving the affiliation conflicts of the extended equilibrium body, a self-balancing relationship group of the target park is obtained.

[0110] Specifically, the equilibrium metric values ​​used to obtain the basic equilibrium relationship include:

[0111] Based on the fundamental balance relationship, the correlation submatrix of the correlation strength matrix is ​​extracted;

[0112] Nonlinear dispersion analysis was performed on the temporal intensity of the correlation submatrix to obtain the temporal volatility characteristics of the correlation submatrix;

[0113] Based on the correlation submatrix and temporal fluctuation characteristics, the cohesion strength of the fundamental equilibrium relationship is calculated. The formula for calculating the cohesion strength is as follows:

[0114] ;

[0115] In the formula, S represents the cohesive strength, N represents the number of equally spaced time points in the time series data, M represents the total number of basic equilibrium relationships, u represents the continuous time variable, and U represents the continuous time series interval. A characteristic pair in the basic equilibrium relationship, where x represents energy supply characteristics, y represents energy demand characteristics, and V represents the basic equilibrium relationship. For feature pairs in the correlation submatrix The strength of the association, Feature pairs in time series volatility characteristics The temporal variance of the association strength, where e is the natural constant;

[0116] By scaling the cohesive strength to a uniform scale, a balance metric value for the basic equilibrium relationship is obtained.

[0117] The system comprehensively scans all correlation strength values ​​in the correlation strength matrix, comparing each value at each position with its adjacent values ​​(up, down, left, and right). Points where all values ​​are higher than their neighbors are identified as intensity peaks. The system accurately records the row and column positions of each intensity peak within the matrix; these positions collectively form the peak coordinates of the correlation strength matrix, ensuring that the location of each peak is precisely captured without omission.

[0118] The correlation strength matrix is ​​defined so that the row labels correspond to the energy supply characteristics of the target park, the column labels correspond to the energy demand characteristics of the target park, the row number of the peak coordinate points to the specific energy supply characteristic, and the column number points to the specific energy demand characteristic. Based on the row and column correspondence of each peak coordinate, the corresponding energy supply characteristics and energy demand characteristics are paired one by one to establish a direct correlation between the two. The set of all these pairings forms the basic balance relationship of the target park.

[0119] Based on the energy supply and energy demand characteristics contained in the basic balance relationship, all rows and columns corresponding to these characteristics are located and extracted in the correlation strength matrix. The matrix fragment formed by the intersection of these rows and columns is the correlation submatrix, ensuring that the correlation submatrix completely contains the correlation strength information of all feature pairs in the basic balance relationship.

[0120] Continuously track the changes in the association strength value of each feature pair in the association submatrix over time, record the specific values ​​of the association strength value at different time points, and observe the fluctuation trend of the values. Analyze the changing patterns of the association strength value over time, and statistically analyze the deviations of the values ​​from the overall average level, including the number of deviations and their specific magnitudes. Systematically organize these changing patterns and deviations to form the temporal volatility characteristics of the association submatrix.

[0121] By combining the specific numerical values ​​of the correlation strength of each feature pair in the correlation submatrix, and the intensity change patterns and deviations recorded in the time-series volatility features, the tightness and stability of the correlation of each feature pair are comprehensively considered. The correlation performance of each feature pair is systematically integrated to form a comprehensive index that can comprehensively reflect the cohesion between the features within the basic equilibrium relationship. This index is the cohesion strength of the basic equilibrium relationship.

[0122] A fixed numerical range is set as the standard range for equalization scaling. This range is determined based on the overall distribution of cohesive strength within all basic equilibrium relationships. The cohesive strength values ​​of each basic equilibrium relationship are adjusted proportionally to the set standard range. During the adjustment process, the relative magnitude of cohesive strength between different basic equilibrium relationships is strictly maintained. The adjusted cohesive strength values ​​are the equilibrium measurement values ​​of the basic equilibrium relationships.

[0123] Based on the core characteristics of the fundamental balance relationship reflected by the balance metric, other energy supply or energy demand characteristics that are associated with these core characteristics and whose association strength meets preset standards are identified in the correlation strength matrix. The rationality of the association between these characteristics and the core characteristics of the fundamental balance relationship is verified one by one to ensure that these characteristics can effectively complement the fundamental balance relationship. These qualified characteristics are the candidate characteristics of the fundamental balance relationship.

[0124] Each candidate feature is temporarily incorporated into its corresponding basic equilibrium relation, simulating the formation of a temporary equilibrium body containing that candidate feature. The equilibrium state of the temporary equilibrium body is then compared with that of the original basic equilibrium relation. The impact of the candidate features on the basic equilibrium relation is evaluated from aspects such as association stability and feature complementarity to determine whether they can improve the overall balance of the basic equilibrium relation. For candidate features that are confirmed to improve the balance, they are formally incorporated into the basic equilibrium relation, expanding the scope of the basic equilibrium relation and forming an extended equilibrium body.

[0125] Examine all extended balancers to identify situations where a particular energy supply or demand characteristic is included in two or more extended balancers simultaneously; such situations constitute attribution conflicts. For each characteristic with an attribution conflict, examine the correlation strength between that characteristic and the core characteristics of each related extended balancer, and explicitly assign the characteristic to the extended balancer with the strongest correlation. Perform the above processing on all attribution conflicts. The resulting set of extended balancers with clear boundaries and no attribution disputes constitutes the self-balancing relationship group for the target industrial park.

[0126] Time series data is obtained through multi-source fusion. The total number of all equally spaced time points is the required quantity. The basic balance relationship is formed through bidirectional correlation mapping of peak coordinates. The total number of basic balance relationships obtained in this process is the corresponding total. The continuous time series interval is the complete time range covered by the time series data. The continuous time variable is any time node within the interval. The set of basic balance relationships is the set of related features. Each pair of energy supply features and energy demand features contained in the set is a feature pair. The correlation sub-matrix is ​​obtained by extracting from the correlation strength matrix. The correlation degree corresponding to each feature pair in this matrix is ​​the correlation strength. The time series volatility characteristics are obtained by performing nonlinear dispersion analysis on the time series strength of the correlation sub-matrix. The time series variance of the correlation strength of each feature pair in the time series volatility characteristics is the corresponding variance value.

[0127] This formula integrates the correlation strength of all feature pairs in the basic equilibrium relationship, while also incorporating the temporal fluctuations of the correlation strength. After normalization, it forms an index that can comprehensively reflect the cohesion between the features within the basic equilibrium relationship, providing a direct basis for subsequent equalization scaling of the basic equilibrium relationship to obtain the equilibrium metric value.

[0128] The higher the correlation strength of the feature pairs and the smaller the temporal variance of the correlation strength, the better the final result of the formula reflects the tight cohesion of the basic equilibrium relationship. The number of equally spaced time points and the total number of basic equilibrium relationships in the time series data are normalized to give the cohesion strength results of basic equilibrium relationships with different time spans and different feature numbers a unified reference standard, accurately presenting the actual situation of internal cohesion characteristics.

[0129] This step accurately captures strong correlation positions in the correlation strength matrix, achieving precise pairing of energy supply and demand characteristics through row and column correspondence, and constructing a basic balance relationship that closely reflects the actual correlation situation. By extracting the correlation submatrix and capturing temporal volatility characteristics, the changing patterns of the internal correlations within the basic balance relationship are fully understood. The resulting cohesion strength accurately reflects the degree of internal cohesion, and the balance metric value after equalization and scaling provides a unified and reasonable standard for subsequent feature selection. The selected candidate features effectively complement the basic balance relationship, expanding the balance body and further improving the overall balance. The resolution of ownership conflicts ensures that the final self-balancing relationship group has clear boundaries and no ownership disputes, possessing overall stability and adaptability. This provides reliable and accurate basic support for the subsequent formulation of independent energy storage dispatch plans and the energy mutual assistance orchestration among self-balancing relationship groups, ensuring the orderly progress of the overall dispatch process.

[0130] This step also ensures that all information relied upon for cohesion strength calculation has clear and reliable acquisition paths, guaranteeing the accuracy and rationality of the calculation basis from the outset. By integrating feature correlation strength and time-series fluctuations, the resulting cohesion strength accurately reflects the degree of cohesion within the basic equilibrium relationship, laying a solid foundation for subsequent equalization scaling to obtain the equilibrium metric value. Normalization processing provides a unified reference standard for the cohesion strength of basic equilibrium relationships with different time spans and different numbers of features, truly presenting the internal cohesion characteristics. This provides accurate and effective data support for subsequent steps such as candidate feature selection and extended equilibrium body construction, ensuring the orderliness and reliability of the overall clustering process.

[0131] In S04, the independent energy storage dispatch plan for the self-balancing relationship group is obtained, including:

[0132] Based on the correlation strength matrix, intra-group temporal intensity mapping is performed on the self-balancing relation group to obtain the temporal supply and demand sequence of the self-balancing relation group;

[0133] Based on the time-series supply and demand sequence, pattern matching is performed on the charging and discharging behavior of energy storage devices in the target park to obtain candidate energy storage actions for the target park.

[0134] The time-coverage fit quantification of candidate energy storage actions yields the intensity maintenance degree of the candidate energy storage actions;

[0135] Based on the intensity maintenance degree, the candidate energy storage actions are prioritized and ranked. Based on the ranking results, the candidate energy storage actions are sequentially connected to obtain the energy storage scheduling sequence of the self-balancing relationship group.

[0136] Based on the correlation strength matrix, the energy storage scheduling sequence is reverse-verified, and the verified sequence is used as an independent energy storage scheduling plan for the self-balancing relation group.

[0137] The strength maintenance of candidate energy storage actions includes:

[0138] The action time periods of candidate energy storage actions are analyzed to obtain the action time periods of the candidate energy storage actions;

[0139] Based on the correlation strength matrix, the correlation strength values ​​for the action period are extracted;

[0140] The dimensions of the candidate energy storage actions are normalized to obtain the normalized power of the candidate energy storage actions;

[0141] Statistical distribution characteristics analysis of time-period correlation strength values ​​is performed to obtain the local consistency factor of time-period correlation strength values;

[0142] Based on normalized power, time-period correlated intensity values, and local consistency factors, the intensity sustaining degree of candidate energy storage actions is calculated. The formula for calculating the intensity sustaining degree is as follows:

[0143] ;

[0144] In the formula, Z represents the strength retention rate. Let t be a specific time interval of the action period. The normalized power for time period t, This refers to the rated maximum power of the energy storage device. This represents the correlation strength value for time period t. This represents the average of the correlation strength values. This represents the standard deviation of the correlation strength values.

[0145] Locate all energy supply and demand feature pairs included in the self-balancing relationship group, find the corresponding row and column positions of these feature pairs in the correlation strength matrix, extract the correlation strength values ​​at each time node in chronological order, correlate these values ​​with the corresponding energy supply and demand change states, and arrange them sequentially along the time axis to form a continuous sequence. This sequence is the time series supply and demand sequence of the self-balancing relationship group.

[0146] By analyzing the changes in the energy supply and demand difference at each time point in the time-series supply and demand sequence, the two basic behavioral modes of energy storage devices—charging and discharging—are identified. The charging mode corresponds to a supply exceeding demand, while the discharging mode corresponds to a demand exceeding supply. The compatibility between the changes in the time-series supply and demand difference and the two basic behavioral modes is compared time-by-time. Charging and discharging behavioral modes that perfectly match the changes in the difference are selected as candidate energy storage actions for the target park.

[0147] Analyze the time-series supply and demand sequence segments corresponding to each candidate energy storage action to determine the start time and end time of the action. The start time is when the difference in the time-series supply and demand sequence reaches the point where the action is needed, and the end time is when the difference recovers to the point where the action is no longer needed. The interval between the start time and the end time is the action period of the candidate energy storage action.

[0148] Based on the start and end times of the action period, all data positions corresponding to that time interval are locked in the association strength matrix. The association strength values ​​of each feature pair in the self-balancing relationship group at these positions are extracted to ensure complete extraction of the association strength data of each time node within the action period. These extracted values ​​are the association strength values ​​of the time period.

[0149] Collect the raw power data of all candidate energy storage actions, find the maximum and minimum values ​​among these raw power data, set a fixed uniform value range, adjust the raw power data of each candidate energy storage action according to the same proportion, so that the adjusted power data all fall into the set uniform value range, and keep the relative relationship between the power of each candidate energy storage action unchanged. The adjusted power data is the normalized power of the candidate energy storage action.

[0150] The distribution of correlation strength values ​​across all time periods within a statistical action period is analyzed. The average correlation strength value within that time period is calculated. Each correlation strength value is compared with the average value, and the deviation of each value from the average value is recorded. All deviations are summarized to determine the degree of concentration of these values ​​around the average value. The more concentrated the values ​​are, the higher the local consistency. An indicator that reflects this degree of concentration is formed, which is the local consistency factor of the correlation strength values ​​of time periods.

[0151] The strength of candidate energy storage actions, as reflected in the normalized power, the characteristics reflected in the time-period correlation strength value, and the intensity concentration reflected in the local consistency factor are comprehensively considered. These three aspects of information are integrated according to a fixed logic, taking into account both the execution strength of the action itself and the tightness and concentration of the correlation strength, to form a comprehensive result that can fully reflect the ability of candidate energy storage actions to play a continuous and stable role during the action period. This result is the intensity maintenance degree of candidate energy storage actions.

[0152] Based on the overall capability corresponding to the intensity maintenance level of each candidate energy storage action, all candidate energy storage actions are ranked sequentially, with stronger capability corresponding to the intensity maintenance level ranking higher. Combining the ranking results with the action time period of each candidate energy storage action, the execution order of the actions is adjusted to ensure that the start time of a later action does not conflict with the end time of a previous action. All candidate energy storage actions are integrated into a continuous, conflict-free execution flow according to time sequence and priority; this flow is the energy storage scheduling sequence of the self-balancing relationship group.

[0153] Substitute the energy storage scheduling sequence into the operation simulation scenario of the self-balancing relationship group. Execute each energy storage action sequentially according to the action order and corresponding action time period in the sequence. Observe the changes in the correlation strength of each feature pair within the self-balancing relationship group in real time during the execution process. Determine whether the changed correlation strength can still maintain the stable state of the self-balancing relationship group. At the same time, verify whether the scheduling sequence can accurately match the changing needs of the time-series supply and demand sequence, and verify its adaptability and effectiveness.

[0154] If, during reverse verification, the correlation strength of the energy storage scheduling sequence remains stable and it can accurately match the changes in the time-series supply and demand sequence, fully meeting the supply and demand balance requirements of the self-balancing relationship group, then the sequence is deemed to have passed verification and is formally designated as an independent energy storage scheduling plan for the self-balancing relationship group, providing clear guidance for the actual charging and discharging operation of energy storage equipment.

[0155] The action period is obtained by analyzing the action periods of candidate energy storage actions, clearly defining the start and end times of the action to form an interval. A specific time period is any time segment divided within the action period. Normalized power is obtained by normalizing the dimensions of candidate energy storage actions. The original power data of all candidate energy storage actions are collected, the maximum and minimum values ​​are found, and a unified numerical range is set. The original power is adjusted to this range by the same proportion while maintaining the relative relationship.

[0156] Rated maximum power is an inherent parameter of the energy storage device itself, extracted directly from the device's specifications. The correlation strength value is extracted based on the correlation strength matrix, locking the matrix position corresponding to the action period and extracting the correlation strength data of the self-balancing relationship feature pairs within that interval.

[0157] The average correlation strength is calculated by analyzing the distribution of all correlation strength values ​​within the statistical action period. It is obtained by summing all values ​​and dividing by the number of values. The standard deviation is calculated by first determining the difference between each correlation strength value and the average, then squaring the absolute value of each difference, summing all squares, dividing the sum by the number of correlation strength values ​​to obtain the mean square, and finally taking the square root of the mean square. The natural constant uses a common, fixed value, and the absolute value operator is applied directly according to standard mathematical operations.

[0158] This formula integrates the normalized power of candidate energy storage actions, the correlation strength value during the action period, and the distribution characteristics of the correlation strength value. By fusing this information through fixed logic, it quantifies the strength maintenance degree of candidate energy storage actions, accurately reflecting the ability of candidate energy storage actions to play a continuous and stable role during their action period. This provides a direct and reliable basis for prioritizing candidate energy storage actions based on the strength maintenance degree.

[0159] The greater the force of the action reflected by the normalized power, the closer the correlation strength value within the time period is to the average correlation strength. The strength maintenance result obtained by the formula better reflects the ability of the candidate energy storage action to maintain stability. The rated maximum power is used as a fixed benchmark to ensure that the power-related data of candidate energy storage actions of different energy storage devices have a unified reference standard. The standard deviation of the correlation strength value reflects the intensity fluctuation and assists in adjusting the result to consider the stability of the correlation strength, so that the final result truly reflects the actual maintenance effect of the candidate energy storage action.

[0160] This step accurately captures the temporal changes in energy supply and demand within the self-balancing relationship group, and the resulting temporal supply and demand sequence provides a clear basis for matching energy storage actions. Through differential changes and matching charging and discharging modes, candidate energy storage actions are ensured to be specifically targeted. Layered analysis of action periods, time-related intensity values, normalized power, and local consistency factors ensures that the intensity maintenance comprehensively reflects the continuous and stable performance capability of candidate energy storage actions. Based on the ranking and temporal sequence of intensity maintenance, a continuous and conflict-free energy storage scheduling sequence is formed. Reverse verification ensures that it maintains the stability of the self-balancing relationship group and accurately matches supply and demand changes. The final determined independent energy storage scheduling plan provides clear and reliable guidance for the actual operation of energy storage equipment, effectively ensuring the supply and demand balance of the self-balancing relationship group and improving the accuracy and stability of energy storage scheduling.

[0161] This step also ensures that all elements required for intensity maintenance calculation have clear and reliable acquisition paths, guaranteeing the accuracy and rationality of the calculation basis from the outset. By comprehensively integrating the power characteristics, correlation intensity values, and distribution patterns of candidate energy storage actions, the quantified intensity maintenance accurately reflects the actual ability of candidate actions to continuously and stably function within the corresponding time period, providing a direct and solid basis for the subsequent prioritization of candidate energy storage actions. By constructing a unified reference benchmark using the rated maximum power and considering the correlation intensity fluctuations in the standard deviation, consistency in the evaluation of candidate actions for different energy storage devices is ensured. Simultaneously, the intensity maintenance results accurately reflect the actual operational effects of candidate actions, laying a crucial foundation for the scientific construction of energy storage scheduling sequences.

[0162] In S05, the remaining energy exchange schemes between the self-balancing relationship groups are obtained, including:

[0163] By performing an expected state simulation of the independent energy storage dispatch plan, the surplus trajectory and deficit trajectory of the self-balancing relationship group are obtained;

[0164] Based on the surplus trajectory and the gap trajectory, bidirectional collaborative interaction is performed between self-balancing relationship groups to obtain a list of interactive relationships between self-balancing relationship groups.

[0165] Based on the association strength matrix, the effectiveness of the interactive relationship list is evaluated, and the priority weight distribution of the interactive relationship list is obtained.

[0166] Based on priority weight distribution, the interactive relationship list is standardized and integrated to obtain the residual energy exchange scheme between self-balancing relationship groups.

[0167] The priority weight distribution of the interactive relationship list is obtained, specifically including:

[0168] Feature extraction is performed on the interactive relationship list to obtain the relationship pair feature identifiers of the interactive relationship list;

[0169] Based on the feature identification of the relationship pair, the coupling dimension of the association strength matrix is ​​extended to obtain the second-order association matrix of the association strength matrix;

[0170] Based on the correlation strength matrix and the second-order correlation matrix, the influence of relations on feature labels is analyzed to obtain the composite correlation strength between relations and feature labels;

[0171] The overlap between the surplus trajectory and the gap trajectory is inferred by time period coverage, and the overlap between the surplus trajectory and the gap trajectory is obtained.

[0172] The priority weight distribution of the interactive relationship list is obtained by dynamically weighting and fusing the composite association strength and trajectory overlap.

[0173] Referring to the charging and discharging sequence and intensity specified in the independent energy storage dispatch plan, and combining the time-series supply and demand sequence of the self-balancing relationship group, the energy supply and demand matching situation after the execution of the dispatch plan in each time period is simulated. The surplus amount of energy supply exceeding demand or the deficit amount of demand exceeding supply in each time period is recorded. The changes of these surplus or deficit amounts are sequentially linked in time to form a surplus trajectory that reflects the trend of energy surplus change in the self-balancing relationship group, and a deficit trajectory that reflects the trend of energy deficit change.

[0174] By comparing the surplus and deficit trajectories of all self-balancing relationship groups one by one, combinations of groups where the time periods of the surplus trajectory and the deficit trajectory overlap are identified. For each group of surplus and deficit groups with overlapping time periods, a two-way relationship is established to clarify the types and quantities of surplus energy that the surplus group can provide, as well as the types and quantities of energy required by the deficit group. This inter-group relationship information is systematically organized to form an interactive relationship list containing group identifiers, energy types, supply and demand quantities, and matching time periods.

[0175] From each interaction in the list of interactive relationships, extract the core energy supply characteristics of the surplus group and the core energy demand characteristics of the deficit group participating in the interaction. Assign a unique identifier to each core characteristic, and combine the surplus group characteristic identifier and the deficit group characteristic identifier in the same interaction relationship to form a relationship pair feature identifier that can uniquely represent the characteristic attributes of the interaction relationship, ensuring that the characteristics of each interaction relationship can be accurately distinguished.

[0176] Based on the relation pair feature identifiers, the dimension of the association strength matrix is ​​expanded. The rows and columns of the newly constructed second-order association matrix are all extracted relation pair feature identifiers. For each row and column intersection position in the second-order association matrix, the association strength between the two sets of core features corresponding to the relation pair feature identifier in the original association strength matrix is ​​queried. Combined with the supply and demand fit of the two sets of features in the interaction relationship, the value of that position in the second-order association matrix is ​​determined, and finally a complete second-order association matrix is ​​formed.

[0177] We extract the association strength between core features within each relation pair feature identifier in the original association strength matrix, and the association strength between the feature identifier of that relation pair and other relation pairs feature identifiers in the second-order association matrix. Taking into account the impact of these two aspects of association strength on the effectiveness of the interaction relationship, we integrate the two parts of strength information according to a fixed logic to form a composite association strength that can comprehensively reflect the tightness of the interaction relationship.

[0178] Align the timelines corresponding to the surplus trajectory and the gap trajectory, compare the coverage of the two trajectories segment by segment, and identify the periods when the two trajectories completely or partially overlap in time. Calculate the total duration of the overlapping periods, determine the proportion of the overlapping periods in the total duration of the surplus trajectory and the gap trajectory, and record the matching of energy surplus and deficit during the overlapping periods to form a trajectory overlap degree that can reflect the degree of temporal fit between the two trajectories.

[0179] A fixed weighting ratio is set based on the influence of composite correlation strength and trajectory overlap on the priority of interaction relationships. The composite correlation strength and trajectory overlap of each interaction relationship are numerically fused according to this ratio, and the fused result is the priority weight of that interaction relationship. All interaction relationship priority weights are organized according to a unified standard to clarify the priority order of each interaction relationship in the overall energy exchange, forming a priority weight distribution for the interaction relationship list.

[0180] All interactive relationships in the list are sorted according to their priority weight distribution, from highest to lowest. Relationships with higher weights are retained first, clearly defining the amount of energy transferred from the surplus group to the deficit group, the specific exchange period, and the responsibilities and cooperation requirements of both parties. The sorted relationships are then formatted, eliminating duplicates and conflicts, and all valid relationships are integrated into a clearly structured and content-defined surplus energy exchange scheme, ensuring that the scheme can directly guide energy allocation between self-balancing relationship groups.

[0181] By accurately grasping the changing patterns of energy surplus and deficit in self-balancing relationship groups, clear data support is provided for energy mutual assistance between groups. An interactive relationship list formed through two-way collaborative interaction between groups comprehensively sorts out the inter-group combinations and core information that meet the conditions for mutual assistance, ensuring the relevance and completeness of the interactive relationships. The extraction of relationship feature identifiers and the construction of a second-order correlation matrix, combined with the calculation of composite correlation strength, significantly improves the accuracy of interactive relationship effectiveness assessment. Meanwhile, the prediction of trajectory overlap ensures a scientific consideration of the adaptability to interaction periods. The priority weight distribution formed by dynamic weighted fusion makes the ranking of interactive relationships more aligned with actual needs. The final standardized and integrated surplus energy exchange scheme has a clear structure and explicit content, possessing practicality to directly guide energy allocation between self-balancing relationship groups, effectively improving overall energy utilization efficiency and the stability of supply and demand balance.

[0182] In S06, the entire target park is dispatched, including:

[0183] The independent energy storage dispatch plan and the surplus energy exchange plan are analyzed in terms of dispatch action items to obtain the dispatch action item set of the target park;

[0184] Align the scheduling action entry set with the reference time to obtain the timing alignment entries of the scheduling action entry set;

[0185] Consistent coordination is performed on the operational conflicts of the timing alignment entries to obtain a conflict-free scheduling flow for the timing alignment entries;

[0186] The conflict-free scheduling flow is encapsulated into instructions to obtain the overall scheduling instructions for the target park.

[0187] Each energy storage device's charging and discharging actions in the independent energy storage dispatch plan are broken down one by one. Key information for each action is identified, including the target energy storage device identifier, action type (charging or discharging), start time, end time, and corresponding power level. Simultaneously, energy transfer actions between all self-balancing groups in the remaining energy exchange schemes are analyzed, determining key details for each transfer action, such as the energy output group identifier, energy input group identifier, energy type, transfer volume, start time, and end time. These individual actions are then organized in a standardized format, with each action as an independent entry. All entries are then aggregated to form the target park's dispatch action entry set.

[0188] The standard time of the core control system of the target park is selected as the unified reference time, which serves as the time reference for the operation of all equipment within the park. The start and end times of each item in the scheduling action item set are checked one by one. The time data of each item, originally based on its own reference standard, is converted to time data referenced to the unified reference time. During the conversion process, the duration of each action is strictly maintained unchanged; only the time coordinate reference is adjusted to ensure that all items are aligned in the same time dimension, ultimately forming the time-aligned items of the scheduling action item set.

[0189] A comprehensive review of all time-series alignment entries was conducted, comparing the actions of the same execution object within the same time period to identify situations where the same equipment or energy channel was assigned multiple different actions at the same time—these situations constitute operational conflicts. For each operational conflict, priority was determined based on the attributes of the actions. In the independent energy storage scheduling plan, actions ensuring supply and demand balance within the self-balancing relationship group had higher priority than cross-group energy transfer actions in the remaining energy exchange plan. High-priority actions were retained, while the start and end times of low-priority conflicting actions were postponed sequentially to avoid the execution period of high-priority actions. This ensured that all adjusted time-series alignment entries had no overlap in time and no execution conflicts, forming a conflict-free scheduling flow for time-series alignment entries.

[0190] Following a pre-defined, unified instruction format, each action item in the conflict-free scheduling flow is standardized and organized. Each instruction item includes a unique instruction identifier, a clear identifier of the execution target, a specific action description, charging / discharging parameters or energy transmission requirements, an execution time interval, and safety operating procedures. All organized instruction items are arranged sequentially according to their execution time, forming a clear and logically coherent instruction sequence. This instruction sequence is then encapsulated, clearly defining the scope of the instruction's effectiveness, its validity period, and the recipients—all relevant equipment control systems within the park. This ultimately forms the park-wide scheduling instruction for the target area, ensuring that the instruction can be accurately identified and executed by all execution units within the park.

[0191] By comprehensively breaking down the various actions in the independent energy storage dispatch plan and the surplus energy exchange scheme, core information is identified and organized in a unified format to ensure the completeness and standardization of the dispatch action item set. Time alignment is performed using the standard time of the park's core control system as a benchmark, ensuring all items remain consistent across the same time dimension and laying a unified time foundation for subsequent dispatch execution. By identifying and coordinating operational conflicts, priorities are clarified and adjusted reasonably based on action attributes, forming a conflict-free dispatch flow and ensuring smooth and interference-free execution of dispatch actions. Instructions are standardized and packaged in a unified format, clarifying core content and scope of application. The resulting park-wide dispatch instruction structure is clear and logically coherent, accurately identifiable and executed by each execution unit, effectively improving the orderliness and accuracy of energy dispatch in the target park and ensuring the efficient implementation of the overall dispatch plan.

[0192] Figures 2-4 The method of this embodiment was applied to three sets of energy storage devices in a source-grid-load-storage integrated park, and the 0-24 hour scheduling power curves were obtained to verify the feasibility and effectiveness of the method of this embodiment in multi-level aggregated optimization scheduling in source-grid-load-storage integrated parks.

[0193] Figure 2 This is the dispatch power curve for the first group of energy storage devices. This group of devices uses a base load (the minimum continuous power required to maintain the park's basic power supply and its own hot standby operation within a 0-24 hour dispatch cycle) of 17.0kW as a benchmark. It achieves precise control under periodic fluctuations in the main frequency (power fluctuation frequency) of 0.083Hz. The noise level (a dimensionless dispatch noise index, referring to disturbances caused by power fluctuations, measurement errors, or control commands) is controlled at 2.8, the operating efficiency reaches 0.84, and the maximum dispatch power is 27.6kW, which does not exceed the rated upper limit. 50.5kW. Figure 2 The curve shows a typical response pattern of charging during off-peak hours and discharging during peak hours, which effectively smooths out load fluctuations in the park and provides stable power support for energy mutual assistance between groups. It is the core execution unit of multi-level aggregation optimization scheduling.

[0194] Figure 3 The second group of energy storage devices has a dispatch power curve. Based on a base load of 17.6kW, this group of devices achieves peak-shifting and coordinated regulation with the first group of energy storage devices under a main frequency fluctuation of 0.093Hz. The noise level is controlled at 2.8, the operating efficiency is 0.82, and the maximum dispatch power is 31.1kW, which does not exceed the rated upper limit of 52.7kW. Figure 3 The curve actively increases the discharge power during peak load periods and supplements charging during off-peak periods, forming a complementary response with the first group of energy storage devices. This significantly improves the overall flexibility and stability of energy allocation in the park and provides a reliable power margin for inter-group mutual support scheduling.

[0195] Figure 4 The power dispatch curve for the third group of energy storage devices is shown. Based on a base load of 7.9kW, this group of devices achieves fine-tuning with a main frequency fluctuation of 0.095Hz, with a noise level controlled at 2.5, an operating efficiency as high as 0.90, and a maximum dispatch power of 22.3kW, which does not exceed the rated upper limit of 48.5kW. Figure 4 The curve responds quickly to dispatch instructions during critical periods, accurately matches the local load demand of the park, and maintains low-power standby during off-peak periods, greatly reducing unnecessary energy loss. It is an important support for realizing the refined allocation and conflict-free dispatch of energy resources in the park.

[0196] Example 2: A flexible multi-level aggregation and scheduling device for integrated source-grid-load-storage parks includes a memory, a processor, and a computer program stored in the memory and capable of running on the processor. The method in Example 1 is implemented by executing the program through the processor.

Claims

1. A flexible multi-level aggregation and scheduling method for integrated source-grid-load-storage parks, characterized in that: Includes the following steps: S01. Perform multi-source fusion on the historical equipment operation data of the target park to obtain the time-series data information of the target park; S02. Perform topological reconstruction on the dependencies of time-series data information to obtain the correlation strength matrix of time-series data information; S03. Based on the correlation strength matrix, coupled clustering is performed on the energy supply characteristics and energy demand characteristics of the target park to obtain the self-balancing relationship group of the target park. The process is as follows: Peak intensity detection is performed on the correlation intensity matrix to obtain the peak coordinates of the correlation intensity matrix; Based on the peak coordinates, a two-way correlation mapping is performed on the energy supply characteristics and energy demand characteristics of the target park to obtain the basic balance relationship of the target park; Based on the correlation strength matrix, the cohesion strength integral of the basic equilibrium relationship is performed to obtain the equilibrium metric value of the basic equilibrium relationship; Based on the balance metric, the nearest neighbor feature is explored to obtain candidate features of the basic balance relationship; The candidate features are evaluated for balance improvement, and based on the evaluation results, the basic balance relationship is incorporated to obtain an extended balance body of the basic balance relationship; The affiliation conflict of the extended equilibrium body is resolved to obtain the self-equilibrium relationship group of the target park; S04. Based on the correlation strength matrix, the fit optimization derivation of the self-balancing relationship group is performed to obtain the independent energy storage scheduling plan for the self-balancing relationship group. The process is as follows: Based on the correlation strength matrix, intra-group temporal intensity mapping is performed on the self-balancing relation group to obtain the temporal supply and demand sequence of the self-balancing relation group; Based on the time-series supply and demand sequence, pattern matching is performed on the charging and discharging behavior of energy storage devices in the target park to obtain candidate energy storage actions for the target park. The time-coverage fit quantification of candidate energy storage actions yields the intensity maintenance degree of the candidate energy storage actions; Based on the intensity maintenance degree, the candidate energy storage actions are prioritized and ranked. Based on the ranking results, the candidate energy storage actions are sequentially connected to obtain the energy storage scheduling sequence of the self-balancing relationship group. Based on the correlation strength matrix, the energy storage scheduling sequence is reverse verified, and the verified sequence is used as an independent energy storage scheduling plan for the self-balancing relation group. S05. Based on the independent energy storage dispatch plan, the energy surplus and deficit among the self-balancing relationship groups are coordinated and arranged in a mutually supportive manner to obtain the remaining energy exchange scheme among the self-balancing relationship groups. S06. Integrate and encode the independent energy storage dispatch plan and the remaining energy exchange plan to obtain the whole park dispatch instructions for the target park.

2. The method for flexible multi-level aggregation and scheduling of resources in an integrated source-grid-load-storage park as described in claim 1, characterized in that, In S01, the process of obtaining the time-series data information of the target park is as follows: Spatiotemporal alignment analysis is performed on the historical equipment operation data of the target park to obtain a unified spatiotemporal reference for the historical equipment operation data. Semantic conflict resolution is performed on the unified expression of the spatiotemporal reference to obtain a semantically consistent representation of historical equipment operation data; Anomaly and defect cleaning are performed on semantically consistent representations to obtain regularized data of historical equipment operation data; The regularized data is encapsulated in a unified structure to obtain the time-series data information of the target park.

3. The method for flexible multi-level aggregation and scheduling of resources in an integrated source-grid-load-storage park as described in claim 1, characterized in that, In S02, the process of obtaining the correlation strength matrix of time-series data information is as follows: By performing lag cross-analysis on time series data, potential dependency patterns of the time series data can be obtained. Granger causality tests were performed on the potential dependency patterns, and the significance thresholds were used to screen the potential dependency patterns to obtain the significant causal associations of the potential dependency patterns. Significant causal relationships are quantified by time lag and intensity weighting to obtain the association weights of significant causal relationships; Based on the association weights, significant causal associations are structured and mapped to obtain the association strength matrix of time-series data information.

4. The method for flexible multi-level aggregation and scheduling of resources in an integrated source-grid-load-storage park as described in claim 1, characterized in that, The equilibrium metric value of the basic equilibrium relationship is obtained as follows: Based on the fundamental balance relationship, the correlation submatrix of the correlation strength matrix is ​​extracted; Nonlinear dispersion analysis was performed on the temporal intensity of the correlation submatrix to obtain the temporal volatility characteristics of the correlation submatrix; Based on the correlation submatrix and temporal fluctuation characteristics, the cohesion strength of the fundamental equilibrium relationship is calculated. The formula for calculating the cohesion strength is as follows: ； In the formula, S represents the cohesive strength, N represents the number of equally spaced time points in the time series data, M represents the total number of basic equilibrium relationships, u represents the continuous time variable, and U represents the continuous time series interval. A characteristic pair in the basic equilibrium relationship, where x represents energy supply characteristics, y represents energy demand characteristics, and V represents the basic equilibrium relationship. For feature pairs in the correlation submatrix The strength of the association, Feature pairs in time series volatility characteristics The temporal variance of the association strength, where e is the natural constant; By scaling the cohesive strength to a uniform scale, a balance metric value for the basic equilibrium relationship is obtained.

5. The method for flexible multi-level aggregation and scheduling of resources in an integrated source-grid-load-storage park as described in claim 1, characterized in that, The strength maintenance of the candidate energy storage action is obtained as follows: The action time periods of candidate energy storage actions are analyzed to obtain the action time periods of the candidate energy storage actions; Based on the correlation strength matrix, the correlation strength values ​​for the action period are extracted; The dimensions of the candidate energy storage actions are normalized to obtain the normalized power of the candidate energy storage actions; Statistical distribution characteristics analysis of time-period correlation strength values ​​is performed to obtain the local consistency factor of time-period correlation strength values; Based on normalized power, time-period correlated intensity values, and local consistency factors, the intensity sustaining degree of candidate energy storage actions is calculated. The formula for calculating the intensity sustaining degree is as follows: ； In the formula, Z represents the strength retention rate. Let t be a specific time interval of the action period. The normalized power for time period t, This refers to the rated maximum power of the energy storage device. Let be the correlation strength value for time period t, and e be the natural constant. This represents the average of the correlation strength values. This represents the standard deviation of the correlation strength values.

6. The method for flexible multi-level aggregation and scheduling of resources in an integrated source-grid-load-storage park as described in claim 1, characterized in that, In S05, the process of obtaining the residual energy exchange scheme between the self-balancing relationship groups is as follows: By performing an expected state simulation of the independent energy storage dispatch plan, the surplus trajectory and deficit trajectory of the self-balancing relationship group are obtained; Based on the surplus trajectory and the gap trajectory, bidirectional collaborative interaction is performed between self-balancing relationship groups to obtain a list of interactive relationships between self-balancing relationship groups. Based on the association strength matrix, the effectiveness of the interactive relationship list is evaluated, and the priority weight distribution of the interactive relationship list is obtained. Based on priority weight distribution, the interactive relationship list is standardized and integrated to obtain the residual energy exchange scheme between self-balancing relationship groups.

7. The method for flexible multi-level aggregation and scheduling of resources in an integrated source-grid-load-storage park as described in claim 6, characterized in that, The priority weight distribution of the interactive relationship list is obtained as follows: Feature extraction is performed on the interactive relationship list to obtain the relationship pair feature identifiers of the interactive relationship list; Based on the feature identification of the relationship pair, the coupling dimension of the association strength matrix is ​​extended to obtain the second-order association matrix of the association strength matrix; Based on the correlation strength matrix and the second-order correlation matrix, the influence of relations on feature labels is analyzed to obtain the composite correlation strength between relations and feature labels; The overlap between the surplus trajectory and the gap trajectory is inferred by time period coverage, and the overlap between the surplus trajectory and the gap trajectory is obtained. The priority weight distribution of the interactive relationship list is obtained by dynamically weighting and fusing the composite association strength and trajectory overlap.

8. The method for flexible multi-level aggregation and scheduling of resources in an integrated source-grid-load-storage park as described in claim 1, characterized in that, In S06, the process of obtaining the overall scheduling instruction for the target park is as follows: The independent energy storage dispatch plan and the surplus energy exchange plan are analyzed in terms of dispatch action items to obtain the dispatch action item set of the target park; Align the scheduling action entry set with the reference time to obtain the timing alignment entries of the scheduling action entry set; Consistent coordination is performed on the operational conflicts of the timing alignment entries to obtain a conflict-free scheduling flow for the timing alignment entries; The conflict-free scheduling flow is encapsulated into instructions to obtain the overall scheduling instructions for the target park.

Images