A multi-level power grid collaborative planning method and system
By dividing the multi-level power grid into grids and performing cluster analysis and optimization model construction, the power supply deviation problem caused by the unidirectional information transmission between the main grid and the distribution network is solved, and efficient, economical and stable power supply of the power system is achieved.
Patent Information
- Application Number
- CN202611584696.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-10-14
- Publication Date
- 2026-11-10
AI Technical Summary
In existing technologies, under the multi-level power grid planning model, information transmission between the main grid, distribution network, and microgrid is unidirectional and lacks real-time data, resulting in a systematic deviation between power supply capacity configuration and actual demand, which affects the power supply stability and reliability of the power system.
By dividing the multi-level power grid into grids based on power grid topology information, obtaining historical operating data for cluster analysis, determining the adjustable range of power resources, constructing a planning model to minimize power grid planning costs, and optimizing power supply facility planning schemes through balance index, two-way information interaction and closed-loop feedback at the main and distribution network levels are achieved.
It enables precise configuration of multi-level power grid facilities, reduces computational complexity and planning costs, improves the adaptability and economy of the power supply system, avoids power supply capacity configuration deviations, and enhances power supply stability and reliability.
Smart Images

Figure CN123070646A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of power system planning technology, and in particular to a multi-level power grid collaborative planning method and system. Background Technology
[0002] With the large-scale integration of diverse new elements such as distributed renewable energy, energy storage systems, and electric vehicle charging facilities into the power grid, the power supply characteristics of distribution networks and microgrids have shifted from rigid loads to flexible nodes with adjustable resources. To ensure a balanced supply and demand match between the main grid's power supply capacity configuration and the actual net load demand of the distribution network, coordinated planning of the main grid, distribution network, and microgrids is necessary.
[0003] In existing technologies, a top-down, sequential approach is used to plan multi-level power grids. This means the main grid layer first determines the capacity and distribution of substations; the distribution network layer performs local optimization under the boundary constraints of the main grid planning results; and the microgrid layer plans independently within the distribution network framework. Information is transmitted unidirectionally between levels only through planning boundary conditions. However, the increasing penetration of distributed renewable energy in the distribution network, the bidirectional regulation capabilities of energy storage systems, and the spatiotemporal uncertainty of electric vehicle charging loads cause significant fluctuations and distributional differences in the actual net load demand of the distribution network. Determining the layout of power supply facilities without this information leads to a systematic deviation between power supply capacity configuration and actual demand. Some areas have redundant and idle substation capacity, while others experience power shortages. This approach severely impacts the stability and reliability of the power system. Summary of the Invention
[0004] This invention provides a multi-level power grid collaborative planning method and system to solve the technical problem of deviation between the power grid power supply capacity configuration and actual demand caused by the serial planning mode, so as to achieve the effect of generating a safe and reliable power system power supply planning scheme.
[0005] To address the aforementioned technical problems, this invention provides a multi-level power grid collaborative planning method, comprising: Based on power grid topology information, the multi-level power grid is divided into various power grid grids; Obtain historical operating data of power equipment within each of the aforementioned power grid grids; Cluster analysis is performed on each of the historical operating data to obtain the historical load curve corresponding to each of the power grid grids; Based on the operational status data of all energy storage devices that communicate with each grid grid, the adjustable range of power resources within each grid grid is determined. Based on the historical load curves and the adjustable range of power resources, the net load curves of each power grid grid are obtained, and based on the changing patterns of the net load curves, an equilibrium index for characterizing the changes in power supply and demand within the power grid grid is determined. A planning model is constructed with the goal of minimizing the power grid planning cost. The adjustable range of power resources is used as a constraint condition. The planning model is solved to obtain the power supply facility planning scheme of the multi-level power grid. The power supply facility planning scheme is then optimized based on the balance index to obtain the optimized power supply facility planning scheme.
[0006] As one preferred embodiment, the step of performing cluster analysis on each of the historical operating data to obtain the historical load curve corresponding to each power grid grid includes: Extract the operating load data from each of the historical operating data sets, and perform feature analysis on the operating load data to obtain time-series load characteristics; Based on the K-means clustering algorithm, the time-series load characteristics are processed to obtain historical load scenarios; The historical load scenarios are analyzed to obtain the historical load curves corresponding to each power grid grid.
[0007] As one preferred embodiment, obtaining the net load curve for each of the power grid grids based on the historical load curve and the adjustable range of power resources includes: The adjustable range of the power resources is analyzed to obtain the optimal power resource scheduling quantity; Based on the electricity demand of each of the power grid grids and the power resource optimization scheduling, the historical load curves are processed to obtain the net load curves of each of the power grid grids.
[0008] As one preferred embodiment, determining the equilibrium index for characterizing the power supply and demand changes within the power grid based on the variation pattern of the net load curve includes: The variation pattern of the net load curve is analyzed, and the net power supply demand proposed by each grid grid to the main grid layer is determined based on the analysis results. Determine the power supply margin allocated by the main grid layer to each of the power grid grids; By analyzing the net power supply demand and the power supply margin, the balance index is obtained to characterize the changes in power supply and demand within the power grid.
[0009] As one preferred embodiment, optimizing the power supply facility planning scheme based on the balance index to obtain the optimized power supply facility planning scheme includes: Determine the power supply demand information for each of the power grid grids in the power supply facility planning scheme; Based on the balance index, the power supply demand information is optimized to obtain the optimized power supply facility planning scheme.
[0010] To address the aforementioned technical problems, this invention also provides a multi-level power grid collaborative planning system, comprising: The grid partitioning module is used to divide a multi-level power grid into various grid grids based on power grid topology information; The data acquisition module is used to acquire historical operating data of power equipment within each of the power grid grids; The clustering analysis module is used to perform clustering analysis on each of the historical operating data to obtain the historical load curve corresponding to each of the power grid grids; The data analysis module is used to determine the adjustable range of power resources within each grid grid based on the operating status data of all energy storage devices that communicate and interact with each grid grid. The curve analysis module is used to obtain the net load curve of each grid grid based on the historical load curve and the adjustable range of power resources, and to determine the balance index used to characterize the changes in power supply and demand within the grid grid based on the changing pattern of the net load curve. The model solving module is used to construct a planning model with the goal of minimizing the power grid planning cost, and to solve the planning model with the adjustable range of power resources as a constraint to obtain the power supply facility planning scheme of the multi-level power grid. Based on the balance index, the power supply facility planning scheme is optimized to obtain the optimized power supply facility planning scheme.
[0011] As one preferred embodiment, the clustering analysis module includes: The feature analysis unit is used to extract the operating load data from each of the historical operating data and perform feature analysis on the operating load data to obtain time-series load features; The clustering analysis unit is used to process the time-series load characteristics based on the K-means clustering algorithm to obtain historical load scenarios; The scenario analysis unit is used to analyze the historical load scenarios and obtain the historical load curve corresponding to each power grid grid.
[0012] As one preferred embodiment, the curve analysis module includes: The scheduling unit is used to analyze the adjustable range of the power resources to obtain the optimized scheduling quantity of the power resources; The net load curve unit is used to process the historical load curve based on the electricity demand of each of the power grid grids and the power resource optimization scheduling amount, so as to obtain the net load curve of each of the power grid grids.
[0013] As one preferred embodiment, the curve analysis module includes: The pattern analysis unit is used to analyze the variation pattern of the net load curve and determine the net power supply demand proposed by each grid grid to the main grid layer based on the analysis results. A power supply margin unit is used to determine the power supply margin allocated by the main grid layer to each of the power grid grids; The balance index unit is used to analyze the net power supply demand and the power supply margin to obtain the balance index used to characterize the changes in power supply and demand within the power grid.
[0014] As one preferred embodiment, the model solving module includes: A power supply demand information unit is used to determine the power supply demand information of each of the power grid grids in the power supply facility planning scheme; The scheme optimization unit is used to optimize the power supply demand information based on the balance index to obtain the optimized power supply facility planning scheme.
[0015] Compared with existing technologies, the beneficial effects of the embodiments of the present invention are at least one of the following: The present invention divides a multi-level power grid into various grid grids based on power grid topology information, realizing spatial decoupling of large-scale power supply areas, decomposing the global planning problem into local planning problems of multiple grid units, reducing the variable scale and computational complexity of a single optimization solution, and providing a computable infrastructure for subsequent hierarchical collaborative planning; The present invention obtains the historical load curves corresponding to each grid by clustering analysis of historical operating data, significantly reducing the scale of time-period variables in the optimization model while retaining the load time-series characteristics, and improving the planning solution efficiency; The present invention obtains the adjustable power resource range corresponding to each grid grid by aggregating and analyzing the operating status data of each energy storage device, realizing a unified quantitative representation of multiple types of heterogeneous resources such as microgrid energy storage devices, distributed power sources, and controllable loads, abstracting the complex source-storage-load coupling characteristics within the microgrid into equivalent elastic parameters that can be directly called by the distribution network layer, enabling the planning model to fully utilize the adjustment capabilities of distributed resources, and improving the adaptability and economy of the planning scheme to the power system; The present invention, based on historical load curves and power resources The adjustable range yields the net load curves of each grid, enabling the actual power supply demand of the distribution network to accurately reflect the spatiotemporal distribution characteristics and adjustability of distributed resources. This provides a precise demand basis for the capacity configuration of the main grid substations, avoiding power supply capacity configuration deviations caused by information asymmetry. This invention constructs a planning model with the minimum grid planning cost as the objective function and solves it using the adjustable range of power resources as a constraint. This achieves joint optimization of feeder investment and operating costs at the distribution network level, ensuring that the planning scheme achieves optimal life-cycle cost while meeting safety constraints and flexible resource adjustment capabilities. This avoids the economic losses caused by the separation of investment and operating costs in traditional planning. Furthermore, this invention optimizes the power supply facility planning scheme based on the equilibrium index, realizing bidirectional information interaction and closed-loop feedback between the main grid and the distribution network. This allows the planning results at the distribution network level to transmit demand information upwards to influence the layout of main grid power supply facilities, and the power supply margin information at the main grid level to feed back downwards to guide the distribution network in adjusting its planning boundaries. This overcomes the technical defects of information separation at each level in the existing unidirectional serial planning mode, making the multi-level grid planning results approach a globally optimal supply-demand matching state. Attached Figure Description
[0016] Figure 1 This is a flowchart illustrating a multi-level power grid collaborative planning method in one embodiment of the present invention; Figure 2 This is a schematic diagram of the structure of a multi-level power grid collaborative planning system in one embodiment of the present invention; Figure label: The modules are: 11. Grid generation module; 12. Data acquisition module; 13. Cluster analysis module; 14. Data analysis module; 15. Curve analysis module; and 16. Model solving module. Detailed Implementation
[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The purpose of providing these embodiments is to make the disclosure of the present invention more thorough and comprehensive. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.
[0018] In the description of this application, the terms "first," "second," "third," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Therefore, a feature defined with "first," "second," "third," etc., may explicitly or implicitly include one or more of that feature. In the description of this application, unless otherwise stated, "a plurality of" means two or more.
[0019] In the description of this application, it should be noted that, unless otherwise expressly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to fixed connections, detachable connections, or integral connections; they can refer to mechanical connections or electrical connections; they can refer to direct connections or indirect connections through an intermediate medium; and they can refer to the internal communication between two components. The terms "vertical," "horizontal," "left," "right," "upper," "lower," and similar expressions used herein are for illustrative purposes only and do not indicate or imply that the device or component referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as limiting the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances.
[0020] In the description of this application, it should be noted that, unless otherwise defined, all technical and scientific terms used in this invention have the same meaning as commonly understood by one of ordinary skill in the art. The terminology used in this specification is for the purpose of describing specific embodiments only and is not intended to limit the invention. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances.
[0021] Power grid planning is a core prerequisite for power system construction and development. The scientific nature of its planning directly determines the power grid's reliability, operational economy, and long-term adaptability. Under the traditional power grid architecture, the distribution side is dominated by rigid industrial and residential loads, with distributed power sources and adjustable resources accounting for a very low proportion. The distribution network and microgrid layers exhibit a "passive receiving end" characteristic. At this stage, the industry generally adopts a top-down serial planning model: the main grid layer determines the substation location and capacity configuration based on macro-load forecasts, forming power supply boundary conditions that are then unidirectionally transmitted to the distribution network layer; the distribution network layer, under the power supply capacity constraints given by the main grid, carries out local network optimization such as feeder construction and expansion; the microgrid layer further independently completes the internal source-load storage planning and configuration within the distribution network framework. Under this model, each layer only needs to complete its own independent solution, completing the entire planning process without bidirectional interaction, demonstrating strong engineering practicality in the traditional power grid stage dominated by rigid loads.
[0022] However, with the continuous advancement of new power system construction, diverse new elements such as distributed renewable energy, electrochemical energy storage systems, and electric vehicle charging facilities are being integrated into the distribution network on a large scale, profoundly changing the power supply characteristics at the distribution network and microgrid levels. The distribution network is transforming from a traditional pure receiving-end load node into an active network with "source-load symbiosis," where the stochastic output of distributed photovoltaics can significantly alter the temporal distribution and peak level of the node's net load. The microgrid is transforming from a passive load terminal into a flexible unit with bidirectional regulation capabilities; the charging and discharging control of energy storage systems and the flexible response of controllable loads can mitigate load fluctuations and reduce peak demand within a certain range. Against this backdrop, the traditional top-down sequential planning model is gradually revealing multi-dimensional technical deficiencies and is no longer suitable for the planning requirements of new power systems.
[0023] To address the aforementioned technical problems, one embodiment of the present invention provides a multi-level power grid collaborative planning method. For details, please refer to [link to specific documentation]. Figure 1 , Figure 1 The diagram shown is a flowchart of a multi-level power grid collaborative planning method according to one embodiment of the present invention. The method includes steps S1 to S6: S1. Based on the power grid topology information, the multi-level power grid is divided into various power grid grids; S2. Obtain historical operating data of power equipment within each of the power grid grids; S3. Perform cluster analysis on each of the historical operating data to obtain the historical load curve corresponding to each of the power grid grids; S4. Based on the operating status data of all energy storage devices that communicate and interact with each grid grid, determine the adjustable range of power resources within each grid grid. S5. Based on the historical load curve and the adjustable range of power resources, obtain the net load curve of each power grid grid, and based on the changing pattern of the net load curve, determine the balance index used to characterize the changes in power supply and demand within the power grid grid. S6. With the minimum power grid planning cost as the objective function, construct a planning model and use the adjustable range of power resources as a constraint condition to solve the planning model, thereby obtaining the power supply facility planning scheme of the multi-level power grid. Based on the balance index, optimize the power supply facility planning scheme to obtain the optimized power supply facility planning scheme.
[0024] Under the traditional unified planning model, the decision variables for the layout of main grid substations, distribution network feeder architecture, and microgrid resource and storage configuration are coupled across the entire domain. The decision space expands nonlinearly with the area of the power supply region and the number of devices, ultimately forming a large-scale mixed integer programming problem. This problem has high solution complexity and long computation time, making it difficult to support engineering planning applications for power supply areas at the prefecture-level city level and above. At the same time, coarse-grained planning across the entire domain cannot accurately match the differences in load characteristics and distributed resource distribution features of different areas, easily leading to structural supply-demand mismatches where local power supply redundancy and power supply shortages coexist. Therefore, step S1 needs to decompose the large-scale power supply area into multiple relatively independent grid computing units before conducting multi-level grid collaborative planning. This decomposes the globally coupled planning problem into independent sub-problems within each grid, significantly reducing the scale of decision variables in a single optimization and providing a feasible computing architecture for multi-level hierarchical parallel solutions.
[0025] Specifically, firstly, the main network-distribution network-branch topology matrix is read from the GIS power grid topology database to obtain the set of 10kV feeders within the power supply range of each substation. The geographical coordinates of the load nodes at the end of the feeders are extracted as spatial data points. Using each substation site as the cluster center, Voronoi polygon partitioning based on Euclidean distance is performed to form the initial grid boundary. The computational complexity of Voronoi partitioning is O(NlogN), which has efficient spatial segmentation capabilities and can ensure that each load node belongs to the power supply range of the nearest substation, naturally fitting the radial power supply characteristics of the distribution network. After the initial partitioning is completed, the peak load of each grid is calculated, and a load balancing threshold is set. When the ratio of the peak loads of any two grids exceeds this threshold, a grid partitioning operation is triggered, dividing the grid with the larger load into two sub-grids along the load density gradient direction. This process is iteratively verified until the peak loads of all grids are at the same order of magnitude. In engineering implementation, the typical value for the load balancing threshold is 3.0. This value strikes an optimal balance between planning accuracy and computational efficiency: a threshold that is too small will lead to excessive grid segmentation, significantly increasing the computational load of subsequent optimization; a threshold that is too large will result in excessive differences in load characteristics within the grid, weakening the representativeness of typical load scenarios. After completing the load balancing verification, the grid boundary is aligned with the administrative boundary of the jurisdiction, and the grid boundary is shifted to the nearest district / county boundary. The shift distance is the shortest Euclidean distance between the Voronoi boundary and the district / county boundary, ensuring that no single grid crosses the district / county administrative boundary. This operation will cause the grid boundary to deviate slightly from the strict Voronoi polygon, but the load balancing deviation can be controlled within 5%, which will not have a substantial impact on the subsequent planning accuracy.
[0026] As a preferred embodiment, consider the power distribution area of a prefecture-level city, which includes five 110kV substations and has a planned area of approximately 200 km². 2 Using five substations as anchor points, Voronoi mapping was performed to obtain five initial grids. Load balance was then verified. The peak load is , for , for , for , for Maximum ratio Exceeding the equilibrium threshold ,trigger Segment. Divided along the load density gradient (peak value) )and (peak value) ), and finally obtained One grid. Verification: Balance is satisfied.
[0027] Grid partitioning can significantly reduce computational complexity. By spatially decoupling, the scale of decision variables in global planning is broken down from a globally coupled magnitude to an independent magnitude for each grid. Subsequent distribution network layer planning and optimization can be solved in parallel on a grid-by-grid basis, greatly improving computational efficiency and making multi-level collaborative planning for large-scale regions engineering feasible. Furthermore, using the grid as a refined computational unit allows for precise matching of load timing characteristics, renewable energy distribution patterns, and elastic resource endowments in different areas, avoiding local supply-demand mismatches caused by global average planning, and significantly improving the alignment between power supply capacity configuration and actual regional needs.
[0028] Furthermore, to support the independent load clustering, flexible resource aggregation, and distribution network optimization of each grid grid unit, step S2 involves the targeted collection of operation and ledger data for corresponding power equipment from various power business systems. Specifically, hourly active and reactive power data for the past three years for the corresponding feeders and load nodes in each grid are extracted from the electricity consumption information collection system and the distribution network automation system, covering different load types such as industrial, commercial, and residential, serving as the original feature input for subsequent clustering of typical load scenarios. Hourly output data and corresponding meteorological data for distributed photovoltaic, small wind turbines, and other equipment within each grid are collected from the distributed power source operation and maintenance management system for subsequent output probability distribution fitting and confidence interval estimation. Hourly charging and discharging power, state of charge, and charging and discharging efficiency data of energy storage devices within each grid are collected from the energy storage management system for calibrating the parameter boundaries of the time-series adjustable power model. Ledger information such as the model, rated capacity, impedance parameters, and commissioning years of feeders and transformers within each grid are extracted from the power grid equipment asset management system to provide basic parameters for subsequent power flow calculation and investment cost accounting. Then, outlier removal, interpolation completion, timestamp alignment, and unit standardization are performed on the collected multi-source data. Finally, a unified historical operation dataset of the power grid grid is formed.
[0029] Furthermore, in step S3, a small number of typical load scenarios are used to replace hourly data throughout the year, significantly reducing the number of time-series constraints in the planning model. This significantly improves solution efficiency within an acceptable range of accuracy loss, providing a feasible basis for iterative calculations in multi-level collaborative planning. Specifically, feature analysis is performed on the hourly operating load data for the past three years within each grid to obtain time-series load features. These features refer to a set of characteristics reflecting the pattern of load power changes over time, primarily reflecting the inherent attributes of intraday load fluctuations, peak and valley distributions, and time-period change rates. They serve as the basic feature vectors for load clustering analysis. Subsequently, the time-series load features are clustered using the K-means clustering algorithm to obtain historical load scenarios. Historical load scenarios refer to representative typical load pattern categories summarized from the annual daily load samples using clustering algorithms. Each scenario corresponds to a set of daily load curves with similar shapes, covering a specific type of operating condition, such as high-load air conditioning conditions in summer, heating load conditions in winter, normal weekday conditions, and low-load conditions on holidays. Finally, historical load curves for the corresponding scenarios are generated through cluster center restoration and dimensional mapping.
[0030] Preferably, the K-means clustering algorithm is used to perform sample partitioning in order to avoid the local optimum problem caused by random initialization. The K-means++ strategy is used to select the initial cluster centers: the first cluster center is randomly selected from the samples, and the probability of selecting each subsequent cluster center is proportional to the square of the shortest distance from the sample to the selected center, ensuring that the initial centers are as dispersed as possible in the sample space and improving the global optimum of the clustering results.
[0031] The time-series load characteristics of the 365 days of the year are clustered into K. scn The clustering objective function can be expressed as: In the formula, For the first The 24-hour load vector for a given day, in MW; For the first Cluster centers of a class; For the first The cluster contains a set of days. The cluster center of each cluster is a typical daily load curve, along with the percentage of days included in that cluster. As a weight, it satisfies Normalization was performed on the load data of each grid before clustering. This is to eliminate the impact of differences in load magnitude between different grids.
[0032] Preferably, the number of typical scenarios The value 4 corresponds to high-load days in summer, high-load days in winter, days during the spring and autumn transition season, and low-load days on holidays. The selection of the number of scenes is validated using silhouette coefficients. The silhouette coefficient is calculated for each sample; the closer the global average silhouette coefficient is to 1, the clearer the clustering structure. At that time, it was impossible to distinguish the differences in load characteristics between seasons and holidays. While clustering improves quality to a limited extent, it significantly increases the computational load of subsequent planning. Therefore, four clusters are chosen as the optimal value for the project.
[0033] After clustering to obtain various historical load scenarios, historical load curves for the corresponding scenarios are generated through cluster center reconstruction and dimensional mapping. Specifically, the cluster center vector of each scenario is taken as the normalized typical load curve for that scenario. Its 24-dimensional value corresponds to the normalized load level at each time of day, representing the standard variation pattern of the load under that scenario. Based on the historical peak load benchmark of the corresponding grid, dimensional reconstruction is performed on the normalized cluster centers. The normalized value at each time is multiplied by the historical peak load of the grid to obtain a typical daily load curve with physical dimensions, in MW. This curve is the grid historical load curve corresponding to the k-th historical load scenario.
[0034] Furthermore, in step S4, based on the operational status data of all energy storage devices communicating with each grid grid, the adjustable power resource range within each grid is determined. This is a core quantitative step in realizing the coordination of microgrid-level elastic resources with higher levels and breaking down the barriers between the main and distribution microgrid levels. Based on the real-time operational status and inherent parameters of the energy storage devices within the grid, and through physical constraint modeling and multi-device aggregation, the dispersed power regulation capabilities within the microgrid are transformed into standardized range parameters that can be directly called upon by the planning model. This provides a quantitative benchmark for elastic regulation capabilities for subsequent net load calculation, distribution network optimization solutions, and supply-demand coordination iterations.
[0035] Specifically, based on the relationship between the defined grid cell boundaries and the power grid topology, a list of all energy storage devices with communication capabilities within each grid is compiled. Two types of operational status data are collected from the corresponding energy management systems: first, the inherent static parameters of the devices, including rated charging and discharging power, rated capacity, charging efficiency, discharging efficiency, and upper and lower limits of safe state of charge; second, time-series dynamic operational data. Power rating constraints are considered for each individual energy storage device. Safety constraints on the state of charge With energy sustainability constraints in mind, the adjustable power limits for each time period are calculated by taking the intersection of the three constraints. The upper discharge limit is obtained by minimizing the rated discharge power and the maximum discharge power that can be released under the current state of charge. In the formula, The time step is typically 1 hour.
[0036] The lower limit of rechargeability is obtained by taking the negative of the minimum value between the rated charging power and the maximum charging power that can be released under the current state of charge: The recursive relationship for the state of charge during discharge is as follows: The recursive relationship for the state of charge during charging is as follows: The actual state of charge consumed during discharge is the output power divided by the discharge efficiency, while the actual state of charge injected into the battery during charging is the input power multiplied by the charging efficiency.
[0037] After calculating the time-adjustable range for a single energy storage unit, the ranges of all energy storage devices within the same grid are linearly superimposed to obtain the basic adjustment range for energy storage contribution. Based on this, the output confidence range of distributed power sources within the grid and the adjustable capacity of controllable loads can be further incorporated to collectively form a complete grid power resource adjustment range.
[0038] Generating net load curves for each grid based on historical load curves and the adjustable range of power resources, and further extracting supply-demand balance indicators, is a core link connecting distribution network demand calculation and main grid supply allocation, supporting multi-level bidirectional collaborative iteration. Step S5, on the one hand, incorporates the adjustment capacity of elastic resources into the load calculation system to obtain net power supply demand that conforms to the actual operating characteristics of the power grid; on the other hand, by constructing quantitative balance evaluation indicators, it provides a basis for judgment and optimization direction for the iterative convergence of multi-level planning, which is a key support for realizing the leap from "hierarchical independent solution" to "global collaborative optimization".
[0039] Specifically, firstly, using the adjustable power resource range of each grid as a rigid constraint boundary, and minimizing distribution network losses and load peak-to-valley differences as optimization objectives, combined with safety constraints such as node voltage deviation and line current carrying capacity, the optimal power resource scheduling quantity for each time period under various typical scenarios is obtained through the optimal power flow algorithm. The optimal power resource scheduling quantity refers to the actual elastic resource power value actually called upon in each time period within the constraint boundary of the adjustable power resource range, with the goal of optimizing distribution network operation safety and economy. It is the core decision variable for lower-level distribution network operation optimization. This variable is a time-series variable, and its value must strictly fall within the adjustable power resource range of the corresponding time period: positive values represent the injection of power from elastic resources into the distribution network (such as energy storage discharge and controllable load reduction), which can reduce the grid's external power supply demand; negative values represent the absorption of power from the distribution network by elastic resources (such as energy storage charging and controllable load growth), which will increase the grid's external power supply demand. By adjusting this scheduling quantity in a time series, peak shaving and valley filling and fluctuation smoothing effects can be achieved, making it the core carrier for transforming elastic resources into actual supply and demand regulation capabilities.
[0040] Secondly, for each typical load scenario, the net load of the grid is calculated by overlaying it time-by-time: In the formula, The historical load power within time period t, This represents the time-series power of the new load added during time period t. Let be the expected output power of the distributed power source in time period t. This refers to the optimized scheduling of microgrid elastic resources.
[0041] Next, the variation pattern of the net load curve is analyzed, and the net power supply demand proposed by each grid grid to the main grid layer is determined based on the analysis results. Net power supply demand refers to the peak power of the grid's net load curve under each typical load scenario. It represents the maximum power supply that the grid needs to obtain from the upper-level main grid under the most severe operating conditions, and is a core requirement parameter for the main grid substation capacity configuration and power supply margin allocation. This value is taken as the maximum value of the net load across all typical scenarios and all time periods, ensuring that the power supply capacity can cover extreme peak conditions. It is the core requirement variable transmitted from the distribution network layer to the main grid layer, and also a key bridge connecting distribution network planning and main grid planning.
[0042] After the main grid layer completes substation site selection, capacity determination, and N-1 safety verification based on the net power supply demand uploaded by each grid, it calculates the remaining power supply capacity of each substation and allocates it according to the proportion of net power supply demand of the grid it supplies, thus obtaining the power supply margin of each grid. The power supply margin refers to the power quota allocated by the main grid substation to the corresponding grid after meeting the net power supply demand of all grids, representing the available power supply redundancy space of that grid. It is a core supply variable fed back from the main grid layer to the distribution network layer. The power supply margin is derived from the difference between the rated power supply capacity of the substation and the total power supply demand, allocated according to the grid demand proportion. The larger the margin, the more sufficient the power supply redundancy of the grid, and the stronger its ability to cope with load growth, abnormal fluctuations, and fault transfer. This parameter is the core basis for assessing the supply-demand matching status and adjusting demand constraints.
[0043] Finally, the margin ratio for each grid is calculated by dividing the power supply margin by the net power supply demand. The maximum and minimum margin ratios for all grids are then calculated, and the difference between the two is the balance index. The balance index is a quantitative indicator used to characterize the degree of balance between power supply and demand across all grids. A smaller balance index value indicates that the power supply redundancy levels of each grid are more similar, and the global supply and demand match is more balanced; a larger value indicates a more severe structural imbalance, with some areas experiencing excessive power redundancy and others experiencing power shortages.
[0044] Step S6 is the core solution and optimization closed-loop link of multi-level power grid collaborative planning. It follows the load characteristic analysis and elastic resource quantification results from the front end, constructs a planning model with the goal of optimizing the whole life cycle cost, incorporates the elastic resource adjustment capacity as a rigid constraint into the solution system, and dynamically corrects the planning boundary through bidirectional iterative feedback between the main and distribution levels, guided by the supply and demand matching balance index, and finally obtains a multi-level power grid power supply facility planning scheme with global supply and demand balance and optimal economic performance.
[0045] Specifically, the planning model aims to minimize the average annual total cost over the entire lifecycle within the planning period. The objective function encompasses two parts: the annualized equivalent of the feeder investment cost and the annual network operation loss cost. In the formula, For the candidate feeder set, For decision variables ( Indicates the selection of candidate feeders ), For feeder One-time investment cost (unit: RMB 10,000) The coefficient for equal annuities is denoted by r, where r is the discount rate and T is the planning period in years, which transforms a one-time investment into an average annual cost. For the scene Time period Network loss cost (ten thousand yuan / h). For scene weights.
[0046] The planning model adopts a two-layer optimization architecture: an upper investment decision layer and a lower operation verification layer. The constraint system includes three core constraints: investment decision constraints, operational safety constraints, and flexible resource adjustment constraints. Investment decision constraints include the selection logic of candidate feeders and power supply radiation range constraints, ensuring that the distribution network topology meets engineering requirements such as radial power supply and full power supply coverage. Operational safety constraints include node voltage deviation constraints (the deviation of each node's voltage amplitude from the nominal value does not exceed 7%), line current carrying capacity constraints (the transmission power of each feeder does not exceed the rated current carrying capacity), and power flow balance constraints, verifying the operational feasibility of the scheme through AC optimal power flow calculation. Flexible resource adjustment constraints incorporate the adjustable range of power resources as a rigid constraint into the operation optimization; that is, the optimized scheduling amount of power resources in each time period must fall within the boundary of the adjustable range for the corresponding time period, ensuring that the flexible resource adjustment capacity invoked in the planning scheme has actual operational feasibility and avoiding the problem of planning and operation being disconnected.
[0047] Secondly, a column generation algorithm with alternating upper and lower layers is used to solve the bi-layer planning model: the upper layer first solves the integer planning to obtain the feeder investment scheme, and calls the lower layer optimal power flow module to verify the operation safety constraints; if the lower layer verification finds that the limit is exceeded, an infeasible cutting plane is generated and fed back to the upper layer, and the upper layer adds constraints and solves again; the iteration continues until the upper and lower layers converge simultaneously to obtain the distribution network planning scheme for each grid.
[0048] After the solution is completed, preferably, the net load peak value and time series curves for all typical scenarios in each grid are extracted to form initial power supply demand information, which is then uploaded to the main grid planning module. The main grid layer receives the initial power supply demand information uploaded by the distribution network and constructs a substation site selection and capacity-determined hybrid integer programming model with the goal of minimizing the substation's life cycle cost. This model decides whether to construct candidate sites and selects the capacity level, while simultaneously performing an N-1 safety check to ensure that the system can still meet all power supply demands through transfer after a single main transformer failure. After the solution is completed, the initial power supply facility layout scheme is obtained, and the remaining power supply capacity of the substation is allocated to each grid according to the demand ratio, generating power supply margin parameters for each grid. Based on the power supply margin and net power supply demand of each grid, a global supply-demand matching balance index (the range of margin rates of each grid) is calculated and compared with a preset convergence threshold (typically 5%, where 5% corresponds to the acceptable deviation level of the remaining power supply capacity after N-1 verification or an engineering experience statistical value). If the balance index is less than or equal to the preset convergence threshold, the current power supply facility planning scheme is determined to meet the global supply-demand balance requirements, the iteration terminates, and the final scheme is output. If the balance index is greater than the preset convergence threshold, the supply-demand matching is determined to be unbalanced, triggering demand constraint updates and the next round of optimization. Next, an adaptive step-size mechanism is used to update the upper bound constraint of power supply demand for each grid, that is, to dynamically adjust the demand constraint boundary in the power supply demand information: for grids with a margin rate lower than the global average (tight power supply), their net power supply demand upper bound is tightened, forcing the distribution network layer to reduce peak demand by increasing elastic resources for local consumption and optimizing energy storage scheduling; for grids with a margin rate higher than the global average (loose power supply), their net power supply demand upper bound is relaxed, allowing them to take on more load. The step size is dynamically adjusted based on the equilibrium trend of previous iterations: if the equilibrium index continues to decline, the step size remains unchanged to maintain the convergence speed; if the equilibrium index stagnates or rises, the step size is halved to avoid overshoot correction; if the index rises instead of falling for several consecutive rounds, divergence detection is triggered, and the step size is forcibly reset to the minimum value to stabilize the iteration process. The updated power demand constraint boundary is fed back to the planning model as a new rigid constraint on the distribution network side, and the distribution network planning solution is re-executed to obtain a new round of power demand information; it is then fed back into the main grid layer to re-optimize the power supply facility layout and margin allocation, and calculate a new round of equilibrium index. This process is repeated until the equilibrium index meets the convergence condition or reaches the maximum number of iterations, and finally, the optimized multi-level power grid power supply facility planning scheme is output.
[0049] This step incorporates the adjustable range of power resources into the planning constraints. By optimizing the scheduling of flexible resources, the model can reduce peak net demand and decrease the rigid investment scale of feeder expansion and substation capacity increase. At the same time, the two-layer architecture coordinates investment costs and operating costs, achieving global optimization of life-cycle costs. Compared with traditional rigid planning schemes, it can significantly reduce overall planning investment. Furthermore, a two-way feedback channel is established between the main and distribution layers. By dynamically correcting the power supply demand boundary, it guides the rebalancing of supply and demand, effectively solving the structural supply and demand imbalance caused by unidirectional serial planning. The power supply redundancy of each grid tends to be consistent, and the overall power supply reliability and asset utilization rate are improved simultaneously.
[0050] Another embodiment of the present invention provides a multi-level power grid collaborative planning system. For details, please refer to [link to relevant documentation]. Figure 2 , Figure 2 The diagram shown illustrates the structure of a multi-level power grid collaborative planning system according to one embodiment of the present invention. The system includes: The grid partitioning module 11 is used to divide the multi-level power grid into various power grid grids based on the power grid topology information; Data acquisition module 12 is used to acquire historical operating data of power equipment within each of the power grid grids; Cluster analysis module 13 is used to perform cluster analysis on each of the historical operating data to obtain the historical load curve corresponding to each of the power grid grids; Data analysis module 14 is used to determine the adjustable range of power resources within each grid grid based on the operating status data of all energy storage devices that communicate and interact with each grid grid. The curve analysis module 15 is used to obtain the net load curve of each grid grid based on the historical load curve and the adjustable range of power resources, and to determine the balance index used to characterize the changes in power supply and demand within the grid grid based on the changing pattern of the net load curve. The model solving module 16 is used to construct a planning model with the goal of minimizing the power grid planning cost, and to solve the planning model with the adjustable range of power resources as a constraint to obtain the power supply facility planning scheme of the multi-level power grid. Based on the balance index, the power supply facility planning scheme is optimized to obtain the optimized power supply facility planning scheme.
[0051] Furthermore, in the above embodiments, the clustering analysis module includes: The feature analysis unit is used to extract the operating load data from each of the historical operating data and perform feature analysis on the operating load data to obtain time-series load features; The clustering analysis unit is used to process the time-series load characteristics based on the K-means clustering algorithm to obtain historical load scenarios; The scenario analysis unit is used to analyze the historical load scenarios and obtain the historical load curve corresponding to each power grid grid.
[0052] Furthermore, in the above embodiments, the curve analysis module includes: The scheduling unit is used to analyze the adjustable range of the power resources to obtain the optimized scheduling quantity of the power resources; The net load curve unit is used to process the historical load curve based on the electricity demand of each of the power grid grids and the power resource optimization scheduling amount, so as to obtain the net load curve of each of the power grid grids.
[0053] Furthermore, in the above embodiments, the curve analysis module includes: The pattern analysis unit is used to analyze the variation pattern of the net load curve and determine the net power supply demand proposed by each grid grid to the main grid layer based on the analysis results. A power supply margin unit is used to determine the power supply margin allocated by the main grid layer to each of the power grid grids; The balance index unit is used to analyze the net power supply demand and the power supply margin to obtain the balance index used to characterize the changes in power supply and demand within the power grid.
[0054] Furthermore, in the above embodiments, the model solving module includes: A power supply demand information unit is used to determine the power supply demand information of each of the power grid grids in the power supply facility planning scheme; The scheme optimization unit is used to optimize the power supply demand information based on the balance index to obtain the optimized power supply facility planning scheme.
[0055] Compared with the prior art, the beneficial effects of the embodiments of the present invention are at least one of the following: (1) This invention divides the multi-level power grid into various grid grids based on the power grid topology information, decomposes the global planning problem into local planning problems of multiple grid units, reduces the variable scale and computational complexity of a single optimization solution, and provides a computable basic architecture for subsequent hierarchical collaborative planning; (2) This invention obtains the adjustable range of power resources corresponding to each grid grid by aggregating and analyzing the operating status data of each energy storage device. It realizes the unified quantitative representation of multiple types of heterogeneous resources such as energy storage devices, distributed power sources, and controllable loads in the microgrid layer. It abstracts the complex source-storage-load coupling characteristics inside the microgrid into equivalent elastic parameters that can be directly called by the distribution network layer, so that the planning model can make full use of the adjustment capability of distributed resources and improve the adaptability and economy of the planning scheme to the power system. (3) This invention obtains the net load curve of each power grid grid based on historical load curves and adjustable power resource range, providing accurate demand basis for the capacity configuration of the main grid substation, and avoiding the deviation in power supply capacity configuration caused by information asymmetry. (4) This invention constructs a planning model with the minimum power grid planning cost as the objective function and solves the planning model with the adjustable range of power resources as the constraint condition, so that the planning scheme achieves the optimal life cycle cost under the premise of meeting the safety constraints and the flexible resource adjustment capability. (5) This invention optimizes the power supply facility planning scheme based on the balance index, realizes bidirectional information interaction and closed-loop feedback between the main grid and the distribution network, so that the planning results of the distribution network layer can transmit demand information upward to affect the layout of the main grid power supply facilities, and the power supply margin information of the main grid layer can also be fed back downward to guide the distribution network to adjust the planning boundary. This overcomes the technical defects of information fragmentation at each level in the unidirectional serial planning mode in the prior art, and makes the multi-level power grid planning results approach the global optimal supply and demand matching state.
[0056] The above embodiments are preferred embodiments of the present invention. It should be noted that those skilled in the art can make several improvements and modifications without departing from the principle of the present invention, and these improvements and modifications are also considered to be within the scope of protection of the present invention.
Claims
1. A multi-level power grid collaborative planning method, characterized in that, include: Based on power grid topology information, the multi-level power grid is divided into various power grid grids; Obtain historical operating data of power equipment within each of the aforementioned power grid grids; Cluster analysis is performed on each of the historical operating data to obtain the historical load curve corresponding to each of the power grid grids; Based on the operational status data of all energy storage devices that communicate with each grid grid, the adjustable range of power resources within each grid grid is determined. Based on the historical load curves and the adjustable range of power resources, the net load curves of each power grid grid are obtained, and based on the changing patterns of the net load curves, an equilibrium index for characterizing the changes in power supply and demand within the power grid grid is determined. The determination of the equilibrium index for characterizing the changes in power supply and demand within the power grid based on the variation pattern of the net load curve includes: The variation pattern of the net load curve is analyzed, and the net power supply demand proposed by each grid grid to the main grid layer is determined based on the analysis results. Determine the power supply margin allocated by the main grid layer to each of the power grid grids; The net power supply demand and the power supply margin are analyzed to obtain the balance index used to characterize the changes in power supply and demand within the power grid. A planning model is constructed with the goal of minimizing the power grid planning cost. The adjustable range of power resources is used as a constraint condition. The planning model is solved to obtain the power supply facility planning scheme of the multi-level power grid. The power supply facility planning scheme is then optimized based on the balance index to obtain the optimized power supply facility planning scheme.
2. The multi-level power grid collaborative planning method as described in claim 1, characterized in that, The step of performing cluster analysis on each of the historical operating data to obtain the historical load curve corresponding to each power grid grid includes: Extract the operating load data from each of the historical operating data sets, and perform feature analysis on the operating load data to obtain time-series load characteristics; Based on the K-means clustering algorithm, the time-series load characteristics are processed to obtain historical load scenarios; The historical load scenarios are analyzed to obtain the historical load curves corresponding to each power grid grid.
3. The multi-level power grid collaborative planning method as described in claim 1, characterized in that, The process of obtaining the net load curve for each of the power grid grids based on the historical load curve and the adjustable range of power resources includes: The adjustable range of the power resources is analyzed to obtain the optimal power resource scheduling quantity; Based on the electricity demand of each of the power grid grids and the power resource optimization scheduling, the historical load curves are processed to obtain the net load curves of each of the power grid grids.
4. The multi-level power grid collaborative planning method as described in claim 1, characterized in that, The optimization of the power supply facility planning scheme based on the balance index to obtain the optimized power supply facility planning scheme includes: Determine the power supply demand information for each of the power grid grids in the power supply facility planning scheme; Based on the balance index, the power supply demand information is optimized to obtain the optimized power supply facility planning scheme.
5. A multi-level power grid collaborative planning system, characterized in that, include: The grid partitioning module is used to divide a multi-level power grid into various grid grids based on power grid topology information; The data acquisition module is used to acquire historical operating data of power equipment within each of the power grid grids; The clustering analysis module is used to perform clustering analysis on each of the historical operating data to obtain the historical load curve corresponding to each of the power grid grids; The data analysis module is used to determine the adjustable range of power resources within each grid grid based on the operating status data of all energy storage devices that communicate and interact with each grid grid. The curve analysis module is used to obtain the net load curve of each grid grid based on the historical load curve and the adjustable range of power resources, and to determine the balance index used to characterize the changes in power supply and demand within the grid grid based on the changing pattern of the net load curve. The determination of the equilibrium index for characterizing the changes in power supply and demand within the power grid based on the variation pattern of the net load curve includes: The variation pattern of the net load curve is analyzed, and the net power supply demand proposed by each grid grid to the main grid layer is determined based on the analysis results. Determine the power supply margin allocated by the main grid layer to each of the power grid grids; The net power supply demand and the power supply margin are analyzed to obtain the balance index used to characterize the changes in power supply and demand within the power grid. The model solving module is used to construct a planning model with the goal of minimizing the power grid planning cost, and to solve the planning model with the adjustable range of power resources as a constraint to obtain the power supply facility planning scheme of the multi-level power grid. Based on the balance index, the power supply facility planning scheme is optimized to obtain the optimized power supply facility planning scheme.
6. A multi-level power grid collaborative planning system as described in claim 5, characterized in that, The clustering analysis module includes: The feature analysis unit is used to extract the operating load data from each of the historical operating data, and to perform feature analysis on the operating load data to obtain time-series load features; The clustering analysis unit is used to process the time-series load characteristics based on the K-means clustering algorithm to obtain historical load scenarios; The scenario analysis unit is used to analyze the historical load scenarios and obtain the historical load curve corresponding to each power grid grid.
7. A multi-level power grid collaborative planning system as described in claim 5, characterized in that, The curve analysis module includes: The scheduling unit is used to analyze the adjustable range of the power resources to obtain the optimized scheduling quantity of the power resources; The net load curve unit is used to process the historical load curve based on the electricity demand of each of the power grid grids and the power resource optimization scheduling amount, so as to obtain the net load curve of each of the power grid grids.
8. A multi-level power grid collaborative planning system as described in claim 5, characterized in that, The model solving module includes: A power supply demand information unit is used to determine the power supply demand information of each of the power grid grids in the power supply facility planning scheme; The scheme optimization unit is used to optimize the power supply demand information based on the balance index to obtain the optimized power supply facility planning scheme.
