Source-oriented grid intelligent decision method, system, device, medium and computer program product for low-carbon operation of grid load storage

Abstract

The application relates to the technical field of intelligent power grids, and particularly discloses a power grid intelligent decision method, system, equipment, medium and computer program product for source-grid-load-storage low-carbon operation, wherein the method comprises the following steps: dividing an agent cluster applicable to a target power grid; establishing an equivalent model of the agent cluster and generating a decision model of an independent agent; a target agent cluster receives a power regulation instruction issued by a power grid coordination center, searches for a matched candidate agent cluster in a regional level, splits the power regulation instruction into sub-instructions, and issues the sub-instructions, wherein, when processing the received sub-instructions, an independent agent determines a matched target agent and adjusts power according to energy output information of the target agent in a preset historical time length. The technical scheme provided by the application can improve the overall scheduling flexibility of a power grid system through an intelligent power grid system.

Description

Technical Field

[0001] This application relates to the field of smart grid technology, and in particular to a smart grid decision-making method, system, equipment, medium and computer program product for low-carbon operation of power generation, grid, load and storage. Background Technology

[0002] In the current coal-fired power system, power balance and generation sufficiency are primarily determined by deterministic approaches, relying on a "source follows load" balancing model to ensure power supply and demand balance, with renewable energy output serving only as a supplement. As the proportion of new energy sources continues to increase, the power structure of the new power system will shift from being dominated by controllable and continuous coal-fired power generation to being dominated by new energy generation with strong uncertainty and weak controllability. Both the supply and demand sides of the power system will face strong uncertainty, and renewable energy and flexible loads will bear some responsibility for power balance. The power balance mechanism will shift towards a probabilistic, multi-regional, multi-entity, source-grid-load-storage collaborative balancing model. The existing power system structure and balancing mechanism cannot support a higher proportion of new energy grid integration and consumption, necessitating the construction of a flexible balancing system adapted to the new power system.

[0003] However, facing the vast number of distributed adjustable resources on both the supply and demand sides of the new power system, characterized by small adjustable capacity, geographical dispersion, and strong heterogeneity, traditional centralized single-layer operation and control methods are difficult to achieve efficient integration and utilization, and are also not conducive to the expansion of distributed adjustable resources. Therefore, conducting theoretical research on the distributed cluster intelligent collaborative control architecture of source-load adjustable resources is of great significance for achieving a comprehensive improvement in the flexibility of full-domain control.

[0004] Under the current power grid decision-making system, the flexibility of adjustable resources on the supply and demand sides is not fully utilized, and the source-load interaction capability is insufficient. In addition, the regional distribution of adjustable resources is uneven, and cross-regional coordination is difficult, resulting in low flexibility and efficiency in power grid decision-making. Therefore, there is an urgent need for a power grid decision-making scheme with a high degree of intelligence and flexibility. Summary of the Invention

[0005] This application provides a method, system, device, medium, and computer program product for intelligent power grid decision-making for low-carbon operation of power generation, grid, load, and storage, which can improve the overall dispatch flexibility of the power grid system through intelligent power grid system.

[0006] To achieve the above objectives, the main technical solutions adopted in this application include: Firstly, this application provides a smart grid decision-making method for low-carbon operation of power generation, grid, load, and storage systems, the method comprising: The new energy fluctuation characteristics of the target power grid are statistically analyzed. Based on the new energy fluctuation characteristics, the supply and demand adjustable complementary resources in the target power grid are determined. Based on the supply and demand adjustable complementary resources, intelligent agent clusters suitable for the target power grid are divided. The hierarchy of the intelligent agent clusters includes at least the regional level, the local level, and the unit level. For any agent cluster within the aforementioned regional level, an equivalent model of the agent cluster is established. For any independent agent within the local and unit levels of the region to which the agent cluster belongs, the equivalent model is differentiated based on the attribute characteristics of the independent agent to generate a decision model for the independent agent. The target intelligent agent cluster within the regional level receives power control instructions issued by the power grid coordination center, and searches for matching candidate intelligent agent clusters within the regional level based on its own equivalent model and the power exchange matrix represented by the power control instructions. The target intelligent agent cluster breaks down the power regulation command into sub-commands and issues the sub-commands to the local and unit levels within its region. For any independent intelligent agent in the local and unit levels within its region, when processing the received sub-command, the independent intelligent agent, based on its own decision model, determines the matching target intelligent agent in the local and unit levels within the region to which the candidate intelligent agent cluster belongs, and adjusts the power output based on the energy output information of the target intelligent agent within a preset historical time period.

[0007] In this embodiment, based on the renewable energy fluctuation characteristics of the target power grid, the target power grid can be divided into intelligent agent clusters. The resulting intelligent agent clusters can possess complementary characteristics from the perspective of supply and demand side resources, thus providing a data foundation for subsequent intelligent decision-making. Intelligent agent clusters can be divided hierarchically, including regional, local, and unit levels. For the regional level, a corresponding equivalent model can be constructed, which can characterize the general characteristics of the regional level intelligent agent clusters. Subsequently, for the independent intelligent agents at the local and unit levels within the region, decision models for each independent intelligent agent can be obtained through differential processing.

[0008] After obtaining the equivalent model and decision model, power dispatching can be performed layer by layer on the target power grid based on these models. Specifically, after receiving power control instructions from the power grid coordination center, the target agent cluster at the regional level can break down the instructions and distribute the resulting sub-instructions to the local and unit levels. The target agent cluster can then search for candidate agent clusters within the target power grid that match its own characteristics. When processing sub-instructions, independent agents at the local and unit levels can further search for target agents with matching characteristics within the candidate agent clusters. Subsequently, during functional adjustments, functional adjustment strategies suitable for the current independent agents can be quickly generated based on the target agents' historical energy output information.

[0009] In this way, despite the large scale of the target power grid, by establishing equivalent models and decision models, it is possible to find intelligent agent clusters and agents with similar scheduling logic. These intelligent agent clusters or agents with similar characteristics can follow similar power adjustment rules, thereby avoiding a large number of repetitive decision generation processes. This layer-by-layer matching and regionally autonomous adjustment method can greatly improve the overall scheduling flexibility of the target power grid.

[0010] In one implementation, the new energy fluctuation characteristics of the target power grid include: For any grid node in the target power grid, determine the renewable energy output time series of the grid node within a specified time period, and identify the upper and lower envelopes of the renewable energy output time series; The fluctuation area formed between the upper envelope and the lower envelope is determined, and the fluctuation area is normalized according to the installed capacity and the sequence length of the new energy output time series to obtain the fluctuation area ratio. Generate the fluctuation median line of the new energy output time series, determine the mean amplitude of the fluctuation median line, and determine the ratio of the mean amplitude to the installed capacity as the fluctuation amplitude ratio; Based on the fluctuation area ratio and the fluctuation amplitude ratio, the new energy fluctuation characteristics of the power grid node are determined, and the new energy fluctuation characteristics of each power grid node in the target power grid are summarized to form the new energy fluctuation characteristics of the target power grid.

[0011] In this embodiment, the volatility of the renewable energy output time series consists of two parts: volatility in high-frequency information and volatility in low-frequency information. The volatility in high-frequency information can be characterized by the volatility area ratio, and the volatility in low-frequency information can be characterized by the volatility amplitude ratio. By combining the volatility area ratio and the volatility amplitude ratio, the renewable energy volatility characteristics of the grid node can be determined. These renewable energy volatility characteristics can accurately characterize the renewable energy volatility of the grid node from both high-frequency and low-frequency dimensions, providing an accurate data foundation for the subsequent intelligent decision-making process.

[0012] In one implementation, determining the supply-demand-side adjustable complementary resources in the target power grid based on the fluctuation characteristics of the new energy source includes: For any grid node in the target power grid, the source-load adjustable resources of the grid node are obtained, and the source-load adjustable resources are divided into multiple time-period adjustable resources according to the new energy fluctuation characteristics of the grid node in different time periods. Traverse the time-adjustable resources of each power grid node in the target power grid, and for any first time-adjustable resource and second time-adjustable resource, calculate the Kendall correlation coefficient between the first time-adjustable resource and the second time-adjustable resource; Calculate the Kendall correlation coefficient between adjustable resources in any two time periods in the target power grid, and determine the supply-demand side adjustable complementary resources in the target power grid based on the calculated Kendall correlation coefficient.

[0013] In this embodiment, the Kendall correlation coefficient can capture the dependence relationship of adjustable resources in two time periods. Moreover, the Kendall correlation coefficient does not have strict requirements on the characteristics of the variables themselves, making it more suitable for adjustable resource complementarity studies with a longer time scale. This enables the accurate identification of adjustable complementary resources on the supply and demand sides of the target power grid, providing an accurate data foundation for the subsequent intelligent decision-making process.

[0014] In one implementation, determining the supply-demand side adjustable complementary resources in the target power grid based on the statistical Kendall correlation coefficient includes: For any time-period adjustable resource in the target power grid, query the target Kendall correlation coefficient that is related to the time-period adjustable resource and has a negative value from the statistical Kendall correlation coefficients; Calculate the absolute value of the Kendall correlation coefficient for each target and select the target with the largest absolute value. The two time-period adjustable resources characterized by the target Kendall correlation coefficient with the largest absolute value are identified as a pair of supply-demand side adjustable complementary resources in the target power grid.

[0015] In this implementation, the Kendall correlation coefficient ranges from -1 to 1. A positive value indicates a positive correlation between the variables, while a negative value indicates a negative correlation. Regardless of sign, a larger absolute value indicates a stronger correlation. Complementarity requires that the changes between variables exhibit opposite trends as much as possible, using the high value of one variable to fill the low value of another, thus achieving complementarity. Therefore, complementarity corresponds to negative correlation, and a larger absolute value indicates stronger complementarity between variables. Thus, when studying the complementarity of source-load adjustable resources, different adjustable resources can be selected, the correlation between different adjustable resources can be calculated, and the adjustable resource with the best complementarity can be selected based on the correlation coefficient.

[0016] In one implementation, dividing the smart agent clusters suitable for the target power grid based on the adjustable complementary resources on the supply and demand sides includes: The target power grid is divided into multiple iterative partitions. In each iterative partition, the resulting agent clusters are obtained, and multiple partition reference time periods are determined. For any partitioned intelligent agent cluster, count the number of power grid nodes in the intelligent agent cluster that are mutually adjustable and complementary resources on the supply and demand sides during each partitioning reference time period. Based on the statistically obtained quantity, calculate the balance index of the agent cluster, and determine the comprehensive index of the current iteration partition based on the balance index of each agent cluster. The iterative partition with the highest comprehensive index is selected, and the partitioning result of the iterative partition with the highest comprehensive index is determined as the intelligent agent cluster suitable for the target power grid.

[0017] In this embodiment, during the partitioning of agent clusters, the agent cluster to which a grid node belongs can be continuously changed. Based on the resource complementarity of grid nodes within the agent cluster, a comprehensive index for the current iterative partitioning can be determined. A higher comprehensive index indicates a higher degree of resource complementarity among grid nodes within the agent cluster, resulting in a more reasonable partitioning outcome. This comprehensive index prioritizes resource complementarity among grid nodes over traditional similarity during agent cluster partitioning. Through coordination among grid nodes, it optimizes cluster complementarity while also mitigating uncertainties related to renewable energy to some extent.

[0018] In one implementation, establishing an equivalent model of the agent cluster includes: For a single agent in the agent cluster, determine the grid node number to which the single agent belongs, and determine the resource type identifier and resource number of the flexible resources accessed under the grid node; Within a preset time period, a power consumption model is constructed that includes the power grid node number, the resource type identifier, and the resource number. The power consumption model includes instantaneous power and cumulative power consumption. Determine the upper limit and lower limit of the instantaneous power model of the power consumption model within the preset time period, and determine the upper limit and lower limit of the cumulative power consumption model of the power consumption model within the preset time period; The power consumption model with an upper and lower bound is determined as the equivalent model of the single agent. The equivalent models of each individual agent within the agent cluster are summarized, and the summarized results are used as the equivalent model of the agent cluster.

[0019] In this embodiment, the equivalent model of the agent cluster can be obtained by aggregating the equivalent models of individual agents within the cluster. The equivalent model of a single agent can be generated based on a power consumption model, which has an upper and lower bound. Ultimately, the power consumption model with the upper and lower bounds can be determined as the equivalent model of a single agent. This equivalent model can be used to describe various typical types of flexible resources. Different types of flexible resources only affect the parameters in this equivalent model, not its form. By providing a general equivalent model, the complexity of data processing can be greatly simplified, and the efficiency of data processing can be improved.

[0020] In one embodiment, the attribute characteristics include at least one of energy storage resource characteristics, distributed generation characteristics, steady-state load characteristics, and industrial electrolysis load characteristics; Accordingly, the equivalent model is subjected to difference processing to generate the decision model for the independent agent, including: Determine the feature type of the attribute feature representation of the independent intelligent agent, and generate actual values ​​of the upper limit and lower limit of the model that match the feature type at different time periods; The upper and lower bounds of the model with actual values ​​are substituted into the equivalent model to generate the decision model of the independent intelligent agent.

[0021] In this embodiment, different independent agents may possess different attribute characteristics. To make accurate decisions for each independent agent, actual values ​​for the upper and lower bounds of the model can be generated based on the feature types represented by the attribute characteristics of the independent agent. In this way, by assigning values ​​to the upper and lower bounds of the model respectively, a personalized decision model for the independent agent can be generated. Subsequent scheduling decisions based on this decision model can improve the accuracy of the decision.

[0022] In one implementation, searching for a matching cluster of candidate agents within the regional level, based on its own equivalent model and the power exchange matrix represented by the power control command, includes: The power exchange matrix represented by the power control command is analyzed to identify several intelligent agent clusters that have power exchanges with themselves. Based on the number of power grid nodes, resource types, and resource categories contained in each resource type represented by its own equivalent model, a model vector matching its own equivalent model is constructed. The model vector of the agent is compared with the model vector of each agent cluster in the plurality of agent clusters. The agent clusters with similarity higher than a preset first threshold are identified as matching candidate agent clusters.

[0023] In this embodiment, by vectorizing the number of power grid nodes, resource types, and resource categories contained in each resource type as represented by the equivalent model, a model vector of the equivalent model can be obtained. This model vector accurately represents the characteristics of the equivalent model. Subsequently, similarity matching based on the model vector can query candidate agent clusters at the regional level with similar scheduling logic in the vast target power grid. The significance of this process is that for agent clusters with similar scheduling logic, the same or similar scheduling strategies can be adopted, thereby greatly improving the efficiency of scheduling strategy generation and providing a data foundation for subsequent refined power adjustment of independent agents.

[0024] In one implementation, determining the matching target agent within the local and unit levels of the region to which the candidate agent cluster belongs, based on its own decision-making model, includes: Identify the time-period characteristics of your own decision model, whereby the time-period characteristics are used to characterize the actual values ​​assigned to the upper limit and lower limit of the model that match your own attribute characteristics in different time periods. For any time period characterized by the time period feature, a time period sequence is constructed based on the time period number corresponding to the time period, as well as the actual values ​​of the upper limit and lower limit of the model within the time period. Integrate the time series from various time periods to generate a decision vector composed of the time series from each time period; The similarity calculation is performed between the decision vector of the agent and the decision vectors of each agent in the local and unit levels within the region to which the candidate agent cluster belongs, and the agent with a similarity higher than a preset second threshold is identified as the matching target agent.

[0025] In this implementation, a vectorized representation of the decision model can be obtained based on the time-segment sequence of different time periods. By calculating the similarity between vectors, a matching target agent can be found in the candidate agent cluster. This provides an accurate data foundation for subsequent power adjustment, ensuring that the final power adjustment can be executed precisely.

[0026] In one implementation, the energy output information of the target intelligent agent within a preset historical time period is represented as the power fluctuation of the target intelligent agent between two consecutive completed global optimization instructions; Power adjustment based on the target intelligent agent's energy output information within a preset historical time period includes: Determine the current instruction adjustment time zone, and within the preset historical time period, query the target time zone that matches the instruction adjustment time zone; Determine the power fluctuation of the target agent within the target time zone, and generate the power change curve of the independent agent in a form complementary to the power fluctuation; Adjust the power according to the power change curve.

[0027] In this embodiment, within similar instruction adjustment time zones, the power adjustment processes of two agents with related scheduling logic should also be the same or similar. Based on this principle, the power fluctuations of the target agent within the target time zone can be determined, and power change curves for independent agents can be generated in a form complementary to these power fluctuations. The purpose of this approach is that since there is currently power exchange between the independent agents and the target agent, complementary power change curves can be used to adjust the power of the independent agents, thereby making the power exchange process more consistent and improving the efficiency and accuracy of power adjustment.

[0028] Secondly, this application also provides a smart grid decision-making system for low-carbon operation of power generation, grid, load, and storage, the system comprising: A cluster partitioning unit is used to statistically analyze the new energy fluctuation characteristics of the target power grid, determine the supply and demand-side adjustable complementary resources in the target power grid based on the new energy fluctuation characteristics, and partition intelligent agent clusters suitable for the target power grid according to the supply and demand-side adjustable complementary resources. The hierarchy of the intelligent agent clusters includes at least regional level, local level and unit level. The model building unit is used to build an equivalent model of any intelligent agent cluster within the region level, and to perform difference processing on the equivalent model for any independent intelligent agent in the local level and unit level within the region to which the intelligent agent cluster belongs, based on the attribute characteristics of the independent intelligent agent, so as to generate the decision model of the independent intelligent agent. The cluster search unit is used to control the target intelligent agent cluster within the regional level to receive the power regulation instructions issued by the power grid coordination center, and to enable the target intelligent agent cluster to search for matching candidate intelligent agent clusters within the regional level based on its own equivalent model and the power exchange matrix represented by the power regulation instructions. The instruction processing unit is used to control the target intelligent agent cluster to break down the power regulation instruction into sub-instructions and issue the sub-instructions to the local level and unit level within its region. Specifically, for any independent intelligent agent in the local level and unit level within the region, when processing the received sub-instruction, the independent intelligent agent determines the matching target intelligent agent in the local level and unit level within the region of the candidate intelligent agent cluster according to its own decision model, and adjusts the power according to the energy output information of the target intelligent agent within a preset historical time period.

[0029] Thirdly, this application also provides a computer device, which includes a memory and a processor. The memory is used to store a computer program, and when the computer program is executed by the processor, it implements the above-mentioned intelligent power grid decision-making method for low-carbon operation of source-grid-load-storage.

[0030] Fourthly, this application also provides a computer-readable storage medium for storing a computer program, which, when executed by a processor, implements the above-described intelligent power grid decision-making method for low-carbon operation of source-grid-load-storage systems.

[0031] Fifthly, this application also provides a computer program product, which includes a computer program that, when executed by a processor, implements the above-mentioned intelligent power grid decision-making method for low-carbon operation of source-grid-load-storage systems. Attached Figure Description

[0032] To more clearly illustrate the technical solutions in the specific embodiments of this application or the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0033] Figure 1 A schematic diagram illustrating the steps of a smart grid decision-making method for low-carbon operation of source-grid-load-storage provided in an embodiment of this application; Figure 2 (a) to Figure 2 (c) A schematic diagram of fluctuations in different time series provided in the embodiments of this application; Figure 3A schematic diagram illustrating the steps for determining the fluctuation characteristics of new energy sources provided in this application embodiment; Figure 4 This is a schematic diagram illustrating the division of intelligent agents at different levels as provided in the embodiments of this application; Figure 5 A schematic diagram illustrating the steps of dividing an intelligent agent cluster according to an embodiment of this application; Figure 6 A schematic diagram of the functional modules of a smart grid decision-making system for low-carbon operation of power generation, grid, load and storage provided in an embodiment of this application; Figure 7 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0034] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0035] One embodiment of this application provides a smart grid decision-making method for low-carbon operation of power generation, grid, load, and storage. Please refer to [link to relevant documentation]. Figure 1 This method may include the following steps.

[0036] S1: Statistically analyze the new energy fluctuation characteristics of the target power grid, determine the supply and demand adjustable complementary resources in the target power grid based on the new energy fluctuation characteristics, and divide the intelligent agent clusters applicable to the target power grid according to the supply and demand adjustable complementary resources. The hierarchy of the intelligent agent clusters includes at least the regional level, the local level, and the unit level.

[0037] In this embodiment, existing methods, when statistically analyzing the fluctuation characteristics of new energy sources, essentially only consider the fluctuation information contained in the high-frequency information of the new energy output time series. In reality, the volatility of a new energy output time series should consist of two parts: high-frequency volatility (volatility in high-frequency information) and trend volatility (volatility in low-frequency information). Defining new energy fluctuation characteristics by only considering high-frequency volatility is incomplete. Unlike existing methods, this application, during its implementation, observed that the more volatile the time series, the larger the area within its upper and lower envelopes. Therefore, it attempts to utilize this area to define the high-frequency volatility of distributed generation output time series.

[0038] like Figure 2 (a) to Figure 2As shown in (c), straight lines with the same time length (200 min, 1 min interval) are examined respectively. Small-volatility time series and large-volatility time series (maximum amplitude 49.5). From Figure 2 As can be seen, the more volatile the time series, the larger the area within the envelope.

[0039] In addition, from Figure 2 It can also be seen that, apart from the different areas within the envelope, Figure 2 The trend of time series changes in (b) is also relatively... Figure 2 The time series in (c) is flatter. Figure 2 The fluctuations in the time series trend in (c) are more pronounced. However, under certain special circumstances... Figure 2 (c) The time series curve becomes smoother, and it is possible that its area under the envelope is higher than that of the curve. Figure 2 (b) The area within the envelope may be smaller. Furthermore, as... Figure 2 As shown in (c), the area within the black ellipse is almost zero, and this portion precisely represents the period of intense fluctuation in the distributed generation power curve, a fluctuation that cannot be reflected in high-frequency information. Therefore, considering only the high-frequency fluctuations of the renewable energy output curve is biased. It is worth noting that this shortcoming is also one of the shortcomings of existing methods.

[0040] Based on the above intuitive analysis, when statistically analyzing the fluctuation characteristics of new energy sources, the volatility in both high-frequency and low-frequency information within the new energy output time series should be considered simultaneously. Therefore, this application, utilizing the envelope of the time series and Lebesgue integral theory, considers the amplitude changes of the time series within a fixed time window and defines a novel new characteristic for measuring the fluctuation of distributed generation power time series.

[0041] Definition of indicators for measuring high-frequency volatility 1) Definition of envelope First, we define the upper and lower envelopes of the new energy output time series to calculate the fluctuation area of ​​high-frequency information.

[0042] Define the time length as The time series of new energy power output is , , .

[0043] Divide the time series into lengths of The equal-length subintervals are defined as time windows, where Number of time windows ; Then, linear interpolation is used to interpolate the maximum and minimum value sequences to a length of... time series , , respectively serving as the upper and lower envelopes of the new energy output time series; Finally, the area of ​​the upper and lower envelopes is defined as the time window length. The area of ​​fluctuation is denoted as Note that the upper and lower envelopes, like the new energy output time series, are discrete time series. Therefore, Riemann integral theory cannot be used to calculate the fluctuation area because discrete time series do not meet the Riemann integrability condition. Therefore, this application uses Lebesgue integral theory to solve for this area.

[0044] It is worth mentioning that the size of the fluctuation area is related to the length of the subinterval. The selection is related. With As the value increases, the area of ​​fluctuation will tend to increase. When If the obtained upper and lower envelopes are consistent with the original time series, then... Therefore, this application assumes... Furthermore, for a time series without fluctuations (a straight line), regardless of the value taken, there is... .

[0045] 2) Lebesgue integral Lebesgue integral ( Unlike the Riemann integral, the integral (or lattice integral) expands the range of integrable functions, making many highly discontinuous functions integrable as well. Therefore, it is an invaluable theoretical tool for solving the problem of integrals at points of discontinuity.

[0046] The integral is defined as: if the function measurable set interval A bounded measurable function on . Let . upper and lower bounds , The interval formed by the upper and lower bounds. Divided into Sub-intervals ,in ,remember , , So, each All are measurable sets.

[0047] definition Yamato , Xiaohe They are respectively: in, for The measure of . Then, for a bounded measurable function There must be Lebesgue defines a bounded measurable function Within the measurable interval On Integral, denoted as 3) Definition of fluctuation area ratio The area of ​​fluctuation is defined as the area between the upper and lower envelopes, which can be used... The length of the subinterval is then obtained by integration. The area of ​​fluctuation is Then, the defined fluctuation area is normalized using installed capacity and time series length, serving as an indicator to characterize the high-frequency fluctuations in new energy output—the fluctuation area ratio. in, The installed capacity of distributed generation power. Let be the duration of time. From the definition (2-3), we know that... This is equivalent to the time series of new energy output in terms of time length. The area within the upper and lower envelopes and the time length under full power output of new energy sources The ratio of the total internal area. This definition eliminates differences in time length and installed capacity, allowing for comparison of the output fluctuations of any two new energy sources.

[0048] Definition of indicators for measuring the volatility of trends 1) Definition of the time series trend of new energy power output The changing trend of the time series of new energy output can be reflected by the new energy output curve after removing high-frequency information ("glitch"). For example... Figure 2 As shown, the median of the upper and lower envelopes of the time series can reflect the changing trend of the distributed generation power time series to a certain extent. For simplicity, this application uses the median of the upper and lower envelopes of the new energy output time series to reflect its trend. The median is defined as follows: 2) Definition of fluctuation amplitude ratio To quantitatively characterize the volatility in low-frequency information of new energy output, especially reflecting... Figure 2 (c) contains fluctuation information. This application uses the ratio of the mean amplitude of the median line within a fixed time window to the installed capacity to characterize the magnitude of the fluctuation, named the amplitude ratio, which is defined as follows: in, , , The total number of time windows. This refers to the installed capacity. It should be noted that the length of the time window here can be the same as or different from the time window used to calculate the envelope. For simplicity, this application uses the same time window length as when calculating the envelope.

[0049] 3) Definition of volatility By combining the fluctuation area ratio and fluctuation amplitude ratio, and treating fluctuations in high-frequency and low-frequency information equally, a new energy fluctuation characteristic is defined to measure the volatility of new energy output. as follows: This new energy fluctuation characteristic has the following advantages in quantitatively characterizing the volatility of new energy output: ① Characteristics of new energy fluctuations The definition takes into full account the high-frequency and low-frequency fluctuations in renewable energy output, and can more comprehensively and accurately reflect the fluctuations in renewable energy output.

[0050] ② Characteristics of New Energy Fluctuations It allows for comparison of the volatility of new energy output time series with different installed capacities and durations.

[0051] ③ Characteristics of new energy fluctuations The definition is more intuitive, and the calculation process is simple. This provides a new perspective for understanding the volatility of new energy output. It is worth mentioning that the volatility measurement index proposed in this application is designed based on the characteristics of time series and can theoretically be used to measure the volatility of any time series.

[0052] As can be seen from the above description, when statistically analyzing the renewable energy fluctuation characteristics of a target power grid, the following can be performed: Figure 3 The following steps are shown: S11: For any grid node in the target power grid, determine the renewable energy output time series of the grid node within a specified time period, and identify the upper and lower envelopes of the renewable energy output time series.

[0053] S12: Determine the fluctuation area formed between the upper envelope and the lower envelope, and normalize the fluctuation area according to the installed capacity and the sequence length of the new energy output time series to obtain the fluctuation area ratio.

[0054] S13: Generate the fluctuation median line of the new energy output time series, determine the mean amplitude of the fluctuation median line, and determine the ratio of the mean amplitude to the installed capacity as the fluctuation amplitude ratio.

[0055] S14: Based on the fluctuation area ratio and the fluctuation amplitude ratio, determine the new energy fluctuation characteristics of the power grid node, and summarize the new energy fluctuation characteristics of each power grid node in the target power grid to form the new energy fluctuation characteristics of the target power grid.

[0056] In this embodiment, the volatility of the renewable energy output time series consists of two parts: volatility in high-frequency information and volatility in low-frequency information. The volatility in high-frequency information can be characterized by the volatility area ratio, and the volatility in low-frequency information can be characterized by the volatility amplitude ratio. By combining the volatility area ratio and the volatility amplitude ratio, the renewable energy volatility characteristics of the grid node can be determined. These renewable energy volatility characteristics can accurately characterize the renewable energy volatility of the grid node from both high-frequency and low-frequency dimensions, providing an accurate data foundation for the subsequent intelligent decision-making process.

[0057] In one implementation, after statistically obtaining the renewable energy fluctuation characteristics of the target power grid, adjustable complementary resources on the supply and demand sides of the target power grid can be determined based on these characteristics. Considering that the Kendall correlation coefficient can capture the dependencies between variables and does not have strict requirements on the characteristics of the variables themselves, it is more suitable for research on the complementarity of source-load adjustable resources over a longer time scale. Therefore, in this implementation, for any power grid node in the target power grid, the source-load adjustable resources of that node can be obtained, and based on the renewable energy fluctuation characteristics of the node at different time periods, these resources can be divided into multiple time-period adjustable resources. Then, the time-period adjustable resources of each power grid node in the target power grid can be traversed, and for any first-time-period adjustable resource and any second-time-period adjustable resource, the Kendall correlation coefficient between the first-time-period adjustable resource and the second-time-period adjustable resource can be calculated. Finally, the Kendall correlation coefficient between any two time-period adjustable resources in the target power grid can be statistically analyzed, and based on the statistically analyzed Kendall correlation coefficient, the adjustable complementary resources on the supply and demand sides of the target power grid can be determined.

[0058] Specifically, the Kendall correlation coefficient can be calculated as follows: In the formula, The Debye function has the following form: in The parameters in the Copula function are solved using the following method: Based on the available resources in the first period Second time period available resources The marginal distribution function of the sample can be obtained. and The sample marginal probability density functions are respectively and Then the sample The joint distribution function is: in, For unknown parameters in the Copula function, the sample The joint density function is: Therefore, the likelihood function of the sample is: The log-likelihood function is obtained as follows: By finding the maximum point of the log-likelihood function, we can obtain the unknown parameters in the Copula function. ,Right now: The solution obtained It can then be approximated Substituting this value into formula (2-29), we can obtain the Kendall correlation coefficient between any two adjustable resources at different time periods.

[0059] In this embodiment, the Kendall correlation coefficient can capture the dependence relationship of adjustable resources in two time periods. Moreover, the Kendall correlation coefficient does not have strict requirements on the characteristics of the variables themselves, making it more suitable for adjustable resource complementarity studies with a longer time scale. This enables the accurate identification of adjustable complementary resources on the supply and demand sides of the target power grid, providing an accurate data foundation for the subsequent intelligent decision-making process.

[0060] The Kendall correlation coefficient is used to describe the degree of complementarity. The Kendall correlation coefficient ranges from -1 to 1. A positive value indicates a positive correlation between the variables; a negative value indicates a negative correlation; if the variables change independently, the correlation coefficient is 0. Regardless of whether it is positive or negative, the larger the absolute value, the stronger the correlation. Complementarity requires that the changes between the variables be as opposite as possible, using the higher values ​​of one variable to fill the lower values ​​of another, thus achieving complementarity. Therefore, complementarity corresponds to negative correlation, and the larger the absolute value, the stronger the complementarity between the variables.

[0061] When conducting research on the complementarity of source-load adjustable resources, different adjustable resources can be selected, the correlation between different adjustable resources can be calculated, and the adjustable resource with the best complementarity can be selected based on the correlation coefficient.

[0062] Specifically, for any time-period adjustable resource in the target power grid, the target Kendall correlation coefficients that are correlated with the time-period adjustable resource and have negative values ​​are retrieved from the statistical Kendall correlation coefficients. Then, the absolute values ​​of each target Kendall correlation coefficient are calculated, and the target Kendall correlation coefficient with the largest absolute value is selected. Finally, the two time-period adjustable resources represented by the target Kendall correlation coefficient with the largest absolute value can be identified as a pair of supply-demand side adjustable complementary resources in the target power grid.

[0063] After identifying the adjustable and complementary resources on the supply and demand sides of the target power grid, intelligent agent clusters suitable for the target power grid can be formed based on these resources. In practical applications, since the target power grid covers a large area, a hierarchical division can be used to represent the intelligent agent clusters for ease of control. These hierarchies can include regional, local, and unit levels. Figure 4 As shown, in a specific application example, the target power grid's agent cluster can be divided into regional agent clusters, local agent clusters, and unit agent clusters. In this specific application example, the unit agent cluster can be represented as the power agent layer. Under the premise of network security constraints, each region realizes load or energy transfer under the unified arrangement of the dispatch center, while monitoring and displaying the load and energy status of each region. In this dispatch mode, a variety of control measures can be executed, and information such as the forecast of energy supply and use, customer energy consumption, and control strategies for each region can be globally shared and policy-based dispatched across energy routers. Figure 4 In this system, power can be exchanged between Region 1, Region 2, and Region 3 through an active distribution network.

[0064] In one implementation, it can be in accordance with Figure 5 The steps shown are as follows to divide the agent clusters applicable to the target power grid.

[0065] S111: Perform multiple rounds of iterative partitioning on the target power grid. In each round of iterative partitioning, obtain the partitioned intelligent agent clusters and determine multiple partitioning reference time periods.

[0066] S112: For any partitioned intelligent agent cluster, count the number of power grid nodes in the intelligent agent cluster that are mutually adjustable and complementary resources on the supply and demand sides during each partition reference time period.

[0067] S113: Calculate the balance index of the agent cluster based on the statistically obtained quantity, and determine the comprehensive index of the current iteration partition based on the balance index of each agent cluster.

[0068] S114: Select the iterative partition with the highest comprehensive index, and determine the partition result of the iterative partition with the highest comprehensive index as the intelligent agent cluster suitable for the target power grid.

[0069] In this implementation, the cluster to which the power grid nodes belong can be continuously changed. In each round of iterative partitioning, a temporary agent cluster can be generated. By using quantitative indicators to evaluate the rationality of the agent clusters in each round of iterative partitioning, the agent cluster partitioning method that is most suitable for the target power grid can be finally evaluated.

[0070] In each iteration of partitioning, a temporary agent cluster can be generated, along with multiple partitioning reference time periods, each with a duration. In the preceding steps, the supply and demand-side adjustable complementary resources of the grid nodes have been paired for the target grid. Within the temporarily partitioned agent clusters, the number of grid nodes that are mutually supply and demand-side adjustable complementary resources within each partitioning reference time period can be counted first. Then, based on the counted numbers, the balance index of each agent cluster and the comprehensive index of the current iteration's partitioning are calculated sequentially. Finally, the partitioning result with the highest comprehensive index is determined as the agent cluster suitable for the target grid.

[0071] The balance index of the agent cluster and the comprehensive index of the current iteration partition can be calculated using the following formula: In the formula, Let be the comprehensive index for the P-th round, and c be the total number of agent clusters obtained within the P-th round. For the first The balance index of an intelligent agent cluster. The total number of time periods for dividing the reference time period, For the first The intelligent agent cluster in the th The number of grid nodes that are mutually adjustable and complementary resources on the supply and demand sides within a given reference time period.

[0072] In this embodiment, during the partitioning of agent clusters, the agent cluster to which a grid node belongs can be continuously changed. Based on the resource complementarity of grid nodes within the agent cluster, a comprehensive index for the current iterative partitioning can be determined. A higher comprehensive index indicates a higher degree of resource complementarity among grid nodes within the agent cluster, resulting in a more reasonable partitioning outcome. This comprehensive index prioritizes resource complementarity among grid nodes over traditional similarity during agent cluster partitioning. Through coordination among grid nodes, it optimizes cluster complementarity while also mitigating uncertainties related to renewable energy to some extent.

[0073] S2: For any agent cluster within the region level, establish an equivalent model for the agent cluster. For any independent agent in the local and unit levels within the region to which the agent cluster belongs, perform difference processing on the equivalent model based on the attribute characteristics of the independent agent to generate a decision model for the independent agent.

[0074] In this embodiment, considering that the target power grid is typically large in scale and the number of agent clusters obtained is relatively large, in order to reduce the complexity of subsequent data processing, a corresponding equivalent model can be constructed for the agent clusters at the regional level. This equivalent model can be a general model that can characterize the general characteristics of the agent clusters at the regional level.

[0075] In one implementation, taking a single agent in the agent cluster as an example, the grid node number to which the single agent belongs can be determined, and the resource type identifier and resource number of the flexible resources connected under the grid node can be determined. Within a preset time period, a power consumption model including the grid node number, the resource type identifier, and the resource number can be constructed. The grid node number can be a globally unique identifier for the grid node within the target grid. Flexible resources connected under the grid node can include conventional loads (loads with unchangeable power consumption, such as lighting loads), electric vehicle charging loads, distributed new energy resources, industrial electrolytic aluminum loads, and conventional generating units. These flexible resources can each have their own resource type identifier. To facilitate subsequent quantitative calculations, the resource type identifier can be represented by numerical features. For example, the five types of flexible resources mentioned above can be represented by five numerical features from 0 to 4. Multiple resources can exist under the same type of flexible resource, and these multiple resources can be represented by resource numbers. Thus, the combination of the resource type identifier and the resource number can uniquely characterize a specific flexible resource in the target grid.

[0076] In this embodiment, a single intelligent agent may exhibit different characteristics at different times. Therefore, when constructing the equivalent model, a day, a quarter, or even a year can be divided into time periods to obtain multiple preset time periods. Within each preset time period, a power consumption model containing the grid node number, the resource type identifier, and the resource number can be constructed. Specifically, the power consumption model of a single intelligent agent within a preset time period can be constructed as follows: A single agent in a time period The electrical characteristics can be represented by a quadruple ( , , , Description. The parameters of this quadruple represent the characteristics of the individual agent during the preset time period. The upper and lower limits of the instantaneous power model and the upper and lower limits of the cumulative power consumption model, i.e. in, Indicates the grid node number is In the power grid nodes, the first Flexible Resources (numbered as follows) Flexible resources in the first The power consumption model for each time period, where δ represents the unit duration.

[0077] As can be seen from the above, the upper and lower limits of the instantaneous power consumption model within the preset time period can be determined, as well as the upper and lower limits of the cumulative power consumption model within the preset time period. Thus, the power consumption model with upper and lower limits can be determined as the equivalent model for the individual agent. By summarizing the equivalent models of each individual agent within the agent cluster, the summarized result can be used as the equivalent model for the agent cluster.

[0078] In this embodiment, the equivalent model of the agent cluster can be obtained by aggregating the equivalent models of individual agents within the cluster. The equivalent model of a single agent can be generated based on a power consumption model, which has an upper and lower bound. Ultimately, the power consumption model with the upper and lower bounds can be determined as the equivalent model of a single agent. This equivalent model can be used to describe various typical types of flexible resources. Different types of flexible resources only affect the parameters in this equivalent model, not its form. By providing a general equivalent model, the complexity of data processing can be greatly simplified, and the efficiency of data processing can be improved.

[0079] In this embodiment, the equivalent model of the agent cluster reflects the general characteristics of the regional level. However, independent agents within the local and unit levels of the region will exhibit certain differences due to the influence of their own attribute characteristics. To more accurately represent these differences, the equivalent model can be processed according to the attribute characteristics of the independent agents to generate decision models for the independent agents.

[0080] Specifically, in one embodiment, the attribute features include at least one of energy storage resource features, distributed generation features, steady-state load features, and industrial electrolysis load features. When performing differential processing on the equivalent model, the feature type represented by the attribute features of the independent intelligent agent can be determined first, and actual values ​​of the upper limit and lower limit of the model matching the feature type can be generated in different time periods. Taking distributed generation features as an example, for distributed new energy sources connected to demand-side users, such as rooftop photovoltaic power generation or small wind turbines, they can be regarded as negative flexible loads. Assuming the maximum output of new energy sources for each time period of the next day is predicted to be... And assuming it is possible to go from 0 to If the output of new energy sources is continuously adjusted, the actual values ​​of the upper and lower limits of the model for instantaneous power and cumulative power consumption can be expressed as: Subsequently, by substituting the upper and lower bounds of the model with actual assigned values ​​into the equivalent model mentioned above, a decision model that conforms to the individual differences of independent intelligent agents can be generated.

[0081] In this embodiment, different independent agents may possess different attribute characteristics. To make accurate decisions for each independent agent, actual values ​​for the upper and lower bounds of the model can be generated based on the feature types represented by the attribute characteristics of the independent agent. In this way, by assigning values ​​to the upper and lower bounds of the model respectively, a personalized decision model for the independent agent can be generated. Subsequent scheduling decisions based on this decision model can improve the accuracy of the decision.

[0082] S3: The target intelligent agent cluster within the regional level receives the power control instructions issued by the power grid coordination center, and searches for matching candidate intelligent agent clusters within the regional level based on its own equivalent model and the power exchange matrix represented by the power control instructions.

[0083] In this embodiment, the power grid coordination center can issue power control commands to the intelligent agent clusters within each regional level, based on the overall operating status of the current target power grid. These power control commands can represent the power exchange matrix. ,in This indicates a cluster of intelligent agents at the regional level. Towards intelligent agent clusters The amount of power transferred.

[0084] In this embodiment, after a target intelligent agent cluster within a region receives a power control instruction issued by the power grid coordination center, it needs to respond to the power control instruction by issuing corresponding sub-instructions to the local and unit levels within its region, thereby completing the power transfer amount required by the power control instruction.

[0085] In this process, to improve the efficiency of power regulation, the target agent cluster can query candidate agent clusters that match its own characteristics within the target power grid. Thus, despite the large scale of the target power grid, the equivalent model of the agent cluster can identify candidate agent clusters with similar scheduling logic. These agent clusters with similar characteristics can adopt similar power adjustment rules, thereby avoiding a large number of repetitive decision-making processes.

[0086] In one implementation, the target agent cluster can parse the power exchange matrix represented by the power control command to identify several agent clusters with which it has power exchange. These agent clusters can then be identified by their cluster numbers in the power exchange matrix.

[0087] Based on its equivalent model, the target intelligent agent cluster can construct a model vector that matches its equivalent model, representing the number of power grid nodes, resource types, and resource categories within each resource type. Specifically, the number of power grid nodes, resource types, and resource categories within each resource type, as reflected in the target intelligent agent cluster's equivalent model, can all be represented by quantification. For example, in the aforementioned steps, resource types can be represented by numerical resource type identifiers, and resource categories can be represented by numerical numbers. Furthermore, to more accurately represent the characteristics of the target intelligent agent cluster, numerical features such as instantaneous power limits and cumulative power consumption limits for different time periods can also be considered. These numerical features are then used as vector elements and arranged in a preset order to form a model vector that matches its equivalent model. Generally, the richer the number of vector elements in the model vector, the more accurately the model vector can represent the overall characteristics of the target intelligent agent cluster.

[0088] In addition, agent clusters at other regional levels within the target power grid can also generate their respective model vectors in a similar manner. Then, the target agent cluster calculates the similarity between its own model vector and the model vectors of each of the aforementioned agent clusters. Agent clusters with similarity scores higher than a preset first threshold are identified as matching candidate agent clusters. The preset first threshold can be a manually set value or a value set according to a fixed percentage; no limitation is made here.

[0089] In this embodiment, by vectorizing the number of power grid nodes, resource types, and resource categories contained in each resource type as represented by the equivalent model, a model vector of the equivalent model can be obtained. This model vector accurately represents the characteristics of the equivalent model. Subsequently, similarity matching based on the model vector can query candidate agent clusters at the regional level with similar scheduling logic in the vast target power grid. The significance of this process is that for agent clusters with similar scheduling logic, the same or similar scheduling strategies can be adopted, thereby greatly improving the efficiency of scheduling strategy generation and providing a data foundation for subsequent refined power adjustment of independent agents.

[0090] S4: The target intelligent agent cluster breaks down the power regulation command into sub-commands and sends the sub-commands to the local level and unit level within its region. For any independent intelligent agent in the local level and unit level within its region, when processing the received sub-command, the independent intelligent agent determines the matching target intelligent agent in the local level and unit level within the region of the candidate intelligent agent cluster according to its own decision model, and adjusts the power according to the energy output information of the target intelligent agent within a preset historical time period.

[0091] In this embodiment, the target agent cluster at the region level can further break down the power control command into several sub-commands, which can then be processed in a refined manner by agents at the local and unit levels. For example, in... Figure 4 In the process, after receiving the power control command, the intelligent agent cluster in region 2 can break down the power control command according to the current working status of the local level and the unit level in its region, thereby obtaining sub-commands for controlling each independent intelligent agent in the local level and the unit level. For example, the sub-command can require the photovoltaic intelligent agent under the unit level to output a certain amount of power.

[0092] For any independent agent at the local or unit level, when processing received sub-instructions, it can prioritize power exchange with agents within the region of the candidate agent cluster. This is because the candidate agent cluster and the target agent cluster share similar characteristics, and power exchange with agents within the candidate agent cluster's region offers higher power compatibility. Furthermore, agents within the candidate agent cluster's region also exhibit their own renewable energy fluctuation characteristics. If these characteristics complement those of the independent agent, the power exchange process will be more compatible, with relatively smaller short-term power fluctuations, thus preventing significant power fluctuations for the entire target power grid. Therefore, an independent agent can use its decision-making model to find a target agent within the candidate agent cluster's region that matches its own characteristics, and then perform complementary power adjustments based on the target agent's energy output information over a preset historical timeframe.

[0093] Specifically, an independent agent can identify the time-period characteristics of its own decision-making model. These time-period characteristics are used to characterize the actual values ​​of the upper and lower bounds of the model that match the agent's own attribute characteristics in different time periods. For example, in the aforementioned steps, the actual values ​​of the upper and lower bounds of the model that match the distributed generation characteristics can be represented by formula (3-5).

[0094] Then, for any time period characterized by the time period features, a time period sequence can be constructed based on the time period number corresponding to the time period, and the actual values ​​of the upper and lower bounds of the model within the time period. This time period sequence can be represented, for example, as (t1, pd, pu, ed, eu), where t1 is the time period number, pu is the actual value of the upper bound of the instantaneous power model within the time period, pd is the actual value of the lower bound of the instantaneous power model within the time period, ed is the actual value of the lower bound of the cumulative power consumption model within the time period, and eu is the actual value of the upper bound of the cumulative power consumption model within the time period. Such a time period sequence can be generated for each time period. Then, by treating each time period sequence as a vector element, the time period sequences of each time period can be integrated to generate a decision vector composed of the time period sequences of each time period. Finally, also based on vector matching, the similarity between the decision vector of an independent intelligent agent and the decision vectors of each intelligent agent in the local and unit levels within the region to which the candidate intelligent agent cluster belongs can be calculated, and intelligent agents with similarity higher than a preset second threshold are identified as matching target intelligent agents. The preset second threshold can be a value set manually or a value set according to a fixed percentage; there is no limitation here.

[0095] In this implementation, a vectorized representation of the decision model can be obtained based on the time-segment sequence of different time periods. By calculating the similarity between vectors, a matching target agent can be found in the candidate agent cluster. This provides an accurate data foundation for subsequent power adjustment, ensuring that the final power adjustment can be executed precisely.

[0096] In one implementation, after finding a target agent with matching characteristics for an independent agent, the power of the independent agent can be adjusted in a power complementary manner. Specifically, the energy output information of the target agent within a preset historical time period can characterize the power fluctuation of the target agent between two consecutive completed global optimization commands. When adjusting the power based on the energy output information of the target agent within the preset historical time period, the command adjustment time zone at the current moment can be determined, and a target time zone matching the command adjustment time zone can be queried within the preset historical time period. Then, based on historical information, the power fluctuation of the target agent within the target time zone can be determined, and a power change curve of the independent agent can be generated in a form complementary to the power fluctuation. Here, the complementary form means that if the power change curve of the independent agent is superimposed with the power fluctuation of the target agent within the target time zone, a relatively smooth power curve can be obtained. In this way, when adjusting the power according to the power change curve, the power exchange process will be more compatible, and the power change in the short term will be relatively small, which can avoid large power fluctuations for the entire target power grid.

[0097] In this embodiment, within similar instruction adjustment time zones, the power adjustment processes of two agents with related scheduling logic should also be the same or similar. Based on this principle, the power fluctuations of the target agent within the target time zone can be determined, and power change curves for independent agents can be generated in a form complementary to these power fluctuations. The purpose of this approach is that since there is currently power exchange between the independent agents and the target agent, complementary power change curves can be used to adjust the power of the independent agents, thereby making the power exchange process more consistent and improving the efficiency and accuracy of power adjustment.

[0098] Please see Figure 6 This application also provides a smart grid decision-making system for low-carbon operation of power generation, grid, load, and storage, the system comprising: Cluster partitioning unit 100 is used to statistically analyze the new energy fluctuation characteristics of the target power grid, determine the supply and demand side adjustable complementary resources in the target power grid based on the new energy fluctuation characteristics, and partition intelligent agent clusters applicable to the target power grid according to the supply and demand side adjustable complementary resources. The hierarchy of the intelligent agent clusters includes at least regional level, local level and unit level. The model building unit 200 is used to build an equivalent model of any intelligent agent cluster within the region level, and to perform difference processing on the equivalent model for any independent intelligent agent in the local level and unit level within the region to which the intelligent agent cluster belongs, based on the attribute characteristics of the independent intelligent agent, so as to generate a decision model for the independent intelligent agent. The cluster search unit 300 is used to control the target intelligent agent cluster within the regional level to receive the power regulation instructions issued by the power grid coordination center, and to enable the target intelligent agent cluster to search for matching candidate intelligent agent clusters within the regional level based on its own equivalent model and the power exchange matrix represented by the power regulation instructions. The instruction processing unit 400 is used to control the target intelligent agent cluster to split the power regulation instruction into sub-instructions and send the sub-instructions to the local level and unit level within its region. Specifically, for any independent intelligent agent in the local level and unit level within the region, when processing the received sub-instruction, the independent intelligent agent determines the matching target intelligent agent in the local level and unit level within the region of the candidate intelligent agent cluster according to its own decision model, and adjusts the power according to the energy output information of the target intelligent agent within a preset historical time period.

[0099] The further functional descriptions of each of the above units are the same as those of the corresponding method embodiments described above, and will not be repeated here.

[0100] In this embodiment, the smart grid decision-making system for low-carbon operation of source-grid-load-storage is presented in the form of functional units. Here, a unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that execute one or more software or fixed programs, and / or other devices that can provide the above functions.

[0101] Please see Figure 7 , Figure 7 This application provides a schematic diagram of the structure of a computer device, as shown in the embodiment of the present application. Figure 7As shown, the computer device includes one or more processors 10, memory 20, and interfaces for connecting the components, including high-speed interfaces and low-speed interfaces. The components communicate with each other via different buses and can be mounted on a common motherboard or otherwise installed as needed. The processors can process instructions executed within the computer device, including instructions stored in or on memory to display graphical information of a GUI on external input / output devices (such as display devices coupled to the interfaces). In some alternative implementations, multiple processors and / or multiple buses can be used with multiple memories and multiple memory modules, if desired. Similarly, multiple computer devices can be connected, each providing some of the necessary operations (e.g., as a server array, a group of blade servers, or a multiprocessor system). Figure 7 Take a processor 10 as an example.

[0102] Processor 10 may be a central processing unit, a network processor, or a combination thereof. Processor 10 may further include a hardware chip. The hardware chip may be an application-specific integrated circuit (ASIC), a programmable logic device (PLD), or a combination thereof. The programmable logic device may be a complex programmable logic device (CAMP), a field-programmable gate array (FPGA), a general-purpose array logic (GDA), or any combination thereof.

[0103] The memory 20 stores instructions executable by at least one processor 10 to enable the at least one processor 10 to execute the smart grid decision-making method for low-carbon operation of source-grid-load-storage as shown in the above embodiments.

[0104] The memory 20 may include a program storage area and a data storage area. The program storage area may store the operating system and applications required for at least one function; the data storage area may store data created based on the use of the computer device. Furthermore, the memory 20 may include high-speed random access memory and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some alternative embodiments, the memory 20 may optionally include memory remotely located relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0105] The memory 20 may include volatile memory, such as random access memory; the memory may also include non-volatile memory, such as flash memory, hard disk or solid-state drive; the memory 20 may also include a combination of the above types of memory.

[0106] The computer device also includes a communication interface 30 for communicating with other devices or communication networks.

[0107] This application also provides a computer-readable storage medium. The methods described in this application can be implemented in hardware or firmware, or implemented as computer code that can be recorded on a storage medium, or implemented as computer code downloaded via a network and originally stored on a remote storage medium or a non-transitory machine-readable storage medium and then stored on a local storage medium. Thus, the methods described herein can be processed by software stored on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. The storage medium can be a magnetic disk, optical disk, read-only memory, random access memory, flash memory, hard disk, or solid-state drive, etc.; further, the storage medium can also include combinations of the above types of memory. It is understood that computers, processors, microprocessor controllers, or programmable hardware include storage components capable of storing or receiving software or computer code. When the software or computer code is accessed and executed by the computer, processor, or hardware, the smart grid decision-making method for low-carbon operation of source-grid-load-storage shown in the above embodiments is implemented.

[0108] This application also provides a computer program product, which includes a computer program. When the computer program is executed by a processor, it implements the above-mentioned intelligent power grid decision-making method for low-carbon operation of source-grid-load-storage.

[0109] The systems and units described in the above embodiments can be implemented by computer chips or physical entities, or by products with certain functions. A typical implementation device is a computer. Specifically, a computer can be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smartphone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or any combination of these devices.

[0110] For ease of description, the above system is described by dividing it into various functional units. Of course, in implementing this application, the functions of each unit can be implemented in one or more software and / or hardware.

[0111] Those skilled in the art will understand that this application may take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application may take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0112] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatuses, and devices according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0113] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0114] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0115] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0116] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the system, device, medium, and computer program product embodiments are basically similar to the method embodiments, so the descriptions are relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0117] The above description is merely an embodiment of this application and is not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

[0118] Although embodiments of this application have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of this application, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A smart grid decision-making method for low-carbon operation of power generation, grid, load, and storage, characterized in that, The method includes: The new energy fluctuation characteristics of the target power grid are statistically analyzed. Based on the new energy fluctuation characteristics, the supply and demand adjustable complementary resources in the target power grid are determined. Based on the supply and demand adjustable complementary resources, intelligent agent clusters suitable for the target power grid are divided. The hierarchy of the intelligent agent clusters includes at least the regional level, the local level, and the unit level. For any agent cluster within the aforementioned regional level, an equivalent model of the agent cluster is established. For any independent agent within the local and unit levels of the region to which the agent cluster belongs, the equivalent model is differentiated based on the attribute characteristics of the independent agent to generate a decision model for the independent agent. The target intelligent agent cluster within the regional level receives power control instructions issued by the power grid coordination center, and searches for matching candidate intelligent agent clusters within the regional level based on its own equivalent model and the power exchange matrix represented by the power control instructions. The target intelligent agent cluster breaks down the power regulation command into sub-commands and issues the sub-commands to the local and unit levels within its region. For any independent intelligent agent in the local and unit levels within its region, when processing the received sub-command, the independent intelligent agent, based on its own decision model, determines the matching target intelligent agent in the local and unit levels within the region to which the candidate intelligent agent cluster belongs, and adjusts the power output based on the energy output information of the target intelligent agent within a preset historical time period.

2. The method according to claim 1, characterized in that, The characteristics of renewable energy fluctuations in the target power grid include: For any grid node in the target power grid, determine the renewable energy output time series of the grid node within a specified time period, and identify the upper and lower envelopes of the renewable energy output time series; The fluctuation area formed between the upper envelope and the lower envelope is determined, and the fluctuation area is normalized according to the installed capacity and the sequence length of the new energy output time series to obtain the fluctuation area ratio. Generate the fluctuation median line of the new energy output time series, determine the mean amplitude of the fluctuation median line, and determine the ratio of the mean amplitude to the installed capacity as the fluctuation amplitude ratio; Based on the fluctuation area ratio and the fluctuation amplitude ratio, the new energy fluctuation characteristics of the power grid node are determined, and the new energy fluctuation characteristics of each power grid node in the target power grid are summarized to form the new energy fluctuation characteristics of the target power grid.

3. The method according to claim 1 or 2, characterized in that, Based on the aforementioned new energy fluctuation characteristics, the adjustable complementary resources on the supply and demand sides of the target power grid include: For any grid node in the target power grid, the source-load adjustable resources of the grid node are obtained, and the source-load adjustable resources are divided into multiple time-period adjustable resources according to the new energy fluctuation characteristics of the grid node in different time periods. Traverse the time-adjustable resources of each power grid node in the target power grid, and for any first time-adjustable resource and second time-adjustable resource, calculate the Kendall correlation coefficient between the first time-adjustable resource and the second time-adjustable resource; Calculate the Kendall correlation coefficient between adjustable resources in any two time periods in the target power grid, and determine the supply-demand side adjustable complementary resources in the target power grid based on the calculated Kendall correlation coefficient.

4. The method according to claim 3, characterized in that, Based on the statistical Kendall correlation coefficient, the supply-demand side adjustable complementary resources in the target power grid are determined to include: For any time-period adjustable resource in the target power grid, query the target Kendall correlation coefficient that is related to the time-period adjustable resource and has a negative value from the statistical Kendall correlation coefficients; Calculate the absolute value of the Kendall correlation coefficient for each target and select the target with the largest absolute value. The two time-period adjustable resources characterized by the target Kendall correlation coefficient with the largest absolute value are identified as a pair of supply-demand side adjustable complementary resources in the target power grid.

5. The method according to claim 1, characterized in that, Based on the adjustable and complementary resources on the supply and demand sides, the smart agent clusters suitable for the target power grid are divided as follows: The target power grid is divided into multiple iterative partitions. In each iterative partition, the resulting agent clusters are obtained, and multiple partition reference time periods are determined. For any partitioned intelligent agent cluster, count the number of power grid nodes in the intelligent agent cluster that are mutually adjustable and complementary resources on the supply and demand sides during each partitioning reference time period. Based on the statistically obtained quantity, calculate the balance index of the agent cluster, and determine the comprehensive index of the current iteration partition based on the balance index of each agent cluster. The iterative partition with the highest comprehensive index is selected, and the partitioning result of the iterative partition with the highest comprehensive index is determined as the intelligent agent cluster suitable for the target power grid.

6. The method according to claim 1, characterized in that, Establishing the equivalent model of the aforementioned agent cluster includes: For a single agent in the agent cluster, determine the grid node number to which the single agent belongs, and determine the resource type identifier and resource number of the flexible resources accessed under the grid node; Within a preset time period, a power consumption model is constructed that includes the power grid node number, the resource type identifier, and the resource number. The power consumption model includes instantaneous power and cumulative power consumption. Determine the upper limit and lower limit of the instantaneous power model of the power consumption model within the preset time period, and determine the upper limit and lower limit of the cumulative power consumption model of the power consumption model within the preset time period; The power consumption model with an upper and lower bound is determined as the equivalent model of the single agent. The equivalent models of each individual agent within the agent cluster are summarized, and the summarized results are used as the equivalent model of the agent cluster.

7. The method according to claim 1 or 6, characterized in that, The attribute characteristics include at least one of the following: energy storage resource characteristics, distributed generation characteristics, steady-state load characteristics, and industrial electrolysis load characteristics; Accordingly, the equivalent model is subjected to difference processing to generate the decision model for the independent agent, including: Determine the feature type of the attribute feature representation of the independent intelligent agent, and generate actual values ​​of the upper limit and lower limit of the model that match the feature type at different time periods; The upper and lower bounds of the model with actual values ​​are substituted into the equivalent model to generate the decision model of the independent intelligent agent.

8. The method according to claim 1 or 6, characterized in that, Based on its own equivalent model and the power exchange matrix represented by the power control command, the search for matching candidate agent clusters within the regional level includes: The power exchange matrix represented by the power control command is analyzed to identify several intelligent agent clusters that have power exchanges with themselves. Based on the number of power grid nodes, resource types, and resource categories contained in each resource type represented by its own equivalent model, a model vector matching its own equivalent model is constructed. The model vector of the agent is compared with the model vector of each agent cluster in the plurality of agent clusters. The agent clusters with similarity higher than a preset first threshold are identified as matching candidate agent clusters.

9. The method according to claim 1, characterized in that, Based on its own decision-making model, the matching target agents are determined at the local and unit levels within the region to which the candidate agent cluster belongs, including: Identify the time-period characteristics of your own decision model, whereby the time-period characteristics are used to characterize the actual values ​​assigned to the upper limit and lower limit of the model that match your own attribute characteristics in different time periods. For any time period characterized by the time period feature, a time period sequence is constructed based on the time period number corresponding to the time period, as well as the actual values ​​of the upper limit and lower limit of the model within the time period. Integrate the time series from various time periods to generate a decision vector composed of the time series from each time period; The similarity calculation is performed between the decision vector of the agent and the decision vectors of each agent in the local and unit levels within the region to which the candidate agent cluster belongs, and the agent with a similarity higher than a preset second threshold is identified as the matching target agent.

10. The method according to claim 1, characterized in that, The energy output information of the target intelligent agent within a preset historical time period is represented by the power fluctuation of the target intelligent agent between two consecutive completed global optimization instructions; Power adjustment based on the target intelligent agent's energy output information within a preset historical time period includes: Determine the current instruction adjustment time zone, and within the preset historical time period, query the target time zone that matches the instruction adjustment time zone; Determine the power fluctuation of the target agent within the target time zone, and generate the power change curve of the independent agent in a form complementary to the power fluctuation; Adjust the power according to the power change curve.

11. A smart grid decision-making system for low-carbon operation of power generation, grid, load, and storage, characterized in that: The system includes: A cluster partitioning unit is used to statistically analyze the new energy fluctuation characteristics of the target power grid, determine the supply and demand-side adjustable complementary resources in the target power grid based on the new energy fluctuation characteristics, and partition intelligent agent clusters suitable for the target power grid according to the supply and demand-side adjustable complementary resources. The hierarchy of the intelligent agent clusters includes at least regional level, local level and unit level. The model building unit is used to build an equivalent model of any intelligent agent cluster within the region level, and to perform difference processing on the equivalent model for any independent intelligent agent in the local level and unit level within the region to which the intelligent agent cluster belongs, based on the attribute characteristics of the independent intelligent agent, so as to generate the decision model of the independent intelligent agent. The cluster search unit is used to control the target intelligent agent cluster within the regional level to receive the power regulation instructions issued by the power grid coordination center, and to enable the target intelligent agent cluster to search for matching candidate intelligent agent clusters within the regional level based on its own equivalent model and the power exchange matrix represented by the power regulation instructions. The instruction processing unit is used to control the target intelligent agent cluster to break down the power regulation instruction into sub-instructions and issue the sub-instructions to the local level and unit level within its region. Specifically, for any independent intelligent agent in the local level and unit level within the region, when processing the received sub-instruction, the independent intelligent agent determines the matching target intelligent agent in the local level and unit level within the region of the candidate intelligent agent cluster according to its own decision model, and adjusts the power according to the energy output information of the target intelligent agent within a preset historical time period.

12. A computer device, characterized in that, The computer device includes a memory and a processor. The memory is used to store a computer program. When the computer program is executed by the processor, it implements the smart grid decision-making method for low-carbon operation of source-grid-load-storage as described in any one of claims 1 to 10.

13. A computer-readable storage medium, characterized in that, The computer-readable storage medium is used to store a computer program, which, when executed by a processor, implements the smart grid decision-making method for low-carbon operation of source-grid-load-storage as described in any one of claims 1 to 10.

14. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the smart grid decision-making method for low-carbon operation of source-grid-load-storage as described in any one of claims 1 to 10.

Images