Power dynamic scheduling method and system based on big data

Abstract

The application provides a power dynamic scheduling method and system based on big data, and relates to the technical field of power scheduling. The method comprises the following steps: acquiring information of each power equipment in a scheduling area, and constructing a power grid topology structure; updating the power grid topology structure in real time; starting from a root node, sequentially selecting an adjacent node as a first node; extracting an adjacent node after the first node as a second node; acquiring historical power data of the second node; performing power grading on the second node set according to the historical power data; acquiring historical environmental data of each second node in the same power grade, learning the relationship between the historical environmental data and the historical power data through a deep learning model, and obtaining a power-environment correlation function of the second node; and determining a power scheduling strategy of the second node through environmental data of a current time point of the second node and the power-environment correlation function. The distribution mode of layer-by-layer progression is adopted to solve the defect of insufficient power resources of the previous node.

Description

Technical Field

[0001] This invention relates to the field of power dispatching technology, and more specifically to a power dynamic dispatching method and system based on big data. Background Technology

[0002] In the field of modern power system dispatching, the power resource allocation strategy directly affects the reliability and economy of power grid operation. Traditional power resource allocation methods usually adopt a reverse dispatching mode of "allocation starting from the end node". Although this mode meets the basic dispatching requirements to a certain extent, it has gradually revealed serious systemic defects in actual operation.

[0003] Traditional methods typically start by calculating power demand from the end of the power grid (load nodes), and then trace back upstream level by level, requiring the upstream nodes to meet the power demand of the downstream nodes, and finally summarizing it to the power source nodes to form a global scheduling scheme. However, when the load demand at the end of the grid accumulates to the upstream nodes, the capacity of the power source nodes may be insufficient, causing the global scheduling scheme to fail. Summary of the Invention

[0004] The purpose of this invention is to provide a power dynamic dispatching method and system based on big data, and the technical problem to be solved is how to dynamically adjust the power dispatching strategy.

[0005] This invention is achieved through the following technical solution:

[0006] The first aspect provides a power dynamic dispatching method based on big data, including the following steps:

[0007] Information on each power device in the scheduling area is obtained, and a power grid topology is constructed based on the information of the power devices; the power grid topology is updated in real time; wherein, each power device is a node, and the connection relationship between power devices is an edge.

[0008] Starting from the root node, select adjacent nodes in sequence as the first node; extract the adjacent nodes after the first node as the second node to obtain the second node set;

[0009] Obtain the historical power data of each second node in the aforementioned second node set;

[0010] Based on the historical power data of the second node, the power level of the second node set is classified to obtain a power level subset;

[0011] Historical environmental data of each second node in the same power level is obtained, and the relationship between historical environmental data and historical power data is learned through a deep learning model to obtain the power-environment correlation function of the second node.

[0012] The power dispatch strategy for the second node is determined by using the environmental data at the current time point and the power-environment correlation function.

[0013] This invention proposes a real-time topology update mechanism based on node communication. When power equipment (nodes) connect or disconnect, the power grid structure is dynamically adjusted through a request-response mechanism to ensure that the dispatching system always makes decisions based on the latest network status. A deep learning model is used to analyze the correlation between historical power data and environmental data, constructing a power-environment correlation function to improve the accuracy of predicting power demand and supply fluctuations. A dynamic grading method based on historical power data, combined with time segmentation, identifies the corresponding power level according to the time segment of the input data and allocates power resources in conjunction with current environmental data, thereby optimizing the dispatching strategy.

[0014] By integrating multi-source information such as power grid topology, historical power data, and environmental data, and using deep learning models to generate optimal scheduling strategies, data-driven dynamic optimization scheduling is achieved, thereby improving the stability and economy of the power grid.

[0015] The second aspect provides a big data-based dynamic power dispatching system, which includes:

[0016] The information acquisition module is used to collect information from various power equipment in the dispatch area.

[0017] A network construction module is connected to the information acquisition module. The network construction module is used to construct a power grid topology based on information from power equipment. Each power device is considered a node, and the connections between power devices are considered edges.

[0018] The scheduling strategy generation module is connected to the network construction module and the information acquisition module; the scheduling strategy generation module is used to perform the following steps:

[0019] Starting from the root node, select adjacent nodes in sequence as the first node;

[0020] Extract the adjacent nodes after the first node as the second node to obtain the second node set;

[0021] Obtain the historical power data of each second node in the aforementioned second node set;

[0022] Based on the historical power data of the second node, the power level of the second node set is classified to obtain a power level subset;

[0023] Historical environmental data of each second node in the same power level is obtained, and the relationship between historical environmental data and historical power data is learned through a deep learning model to obtain the power-environment correlation function of the second node.

[0024] The power dispatch strategy for the second node is determined by using the environmental data at the current time point and the power-environment correlation function.

[0025] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0026] Real-time updates of the power grid topology enhance the power grid's dynamic perception and adaptive adjustment capabilities in real-time conditions, solving the problem of existing technologies being unable to adapt to dynamic changes in power load, resulting in excessively low or high load voltage drop.

[0027] The root node, as a power generation or energy storage device, is used to allocate power resources. After the root node, there are multiple branch nodes (called branch node 1), and after branch node 1, there are branch nodes (called branch node 2). Power resources are allocated to branch node 1 first, and then to branch node 2, realizing a layer-by-layer progressive allocation method, which solves the defect of allocating power resources from the last node, while the previous node has insufficient power resources.

[0028] The impact of environmental factors on the power resources required by power equipment has been taken into account, which improves the accuracy of power resource dispatch. Attached Figure Description

[0029] To more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly described below. It should be understood that the following drawings only show some embodiments of the present invention and should not be considered as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort. In the drawings:

[0030] Figure 1 Main flowchart;

[0031] Figure 2 The power grid topology provided for the fourth embodiment. Detailed Implementation

[0032] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the embodiments and accompanying drawings. The illustrative embodiments and descriptions of the present invention are only used to explain the present invention and are not intended to limit the present invention.

[0033] First embodiment:

[0034] Combination Figure 1 The power dynamic dispatching method based on big data includes the following steps:

[0035] Information on each power device in the scheduling area is obtained, and a power grid topology is constructed based on the information of the power devices; the power grid topology is updated in real time; wherein, each power device is a node, and the connection relationship between power devices is an edge.

[0036] Starting from the root node, select adjacent nodes in sequence as the first node; extract the adjacent nodes after the first node as the second node to obtain the second node set;

[0037] Obtain the historical power data of each second node in the aforementioned second node set; wherein, the aforementioned historical power data is the power value flowing through the second node at each time point;

[0038] Based on the historical power data of the second node, the power level of the second node set is classified to obtain a power level subset;

[0039] Historical environmental data (meteorological parameters, economic indicators, and user behavior) of each second node in the same power level are obtained. Meteorological parameters include temperature and humidity, economic indicators include per capita annual income, industrial chain, and enterprise annual income in the dispatch area, and user behavior includes work and rest time and community activities. The relationship between historical environmental data and historical power data is learned through a deep learning model to obtain the power-environment correlation function of the second node.

[0040] The power dispatch strategy for the second node is determined by using the environmental data at the current time point and the power-environment correlation function.

[0041] Real-time updates of the power grid topology enhance the power grid's dynamic perception and adaptive adjustment capabilities in real-time conditions, solving the problem of existing technologies being unable to adapt to dynamic changes in power load, resulting in excessively low or high load voltage drop.

[0042] The root node, as a power generation or energy storage device, is used to allocate power resources. After the root node, there are multiple branch nodes (called branch node 1), and after branch node 1, there are branch nodes (called branch node 2). Power resources are allocated to branch node 1 first, and then to branch node 2, realizing a layer-by-layer progressive allocation method, which solves the defect of allocating power resources from the last node, while the previous node has insufficient power resources.

[0043] The impact of environmental factors on the power resources required by power equipment has been taken into account, which improves the accuracy of power resource dispatch.

[0044] Second embodiment:

[0045] Based on the first embodiment, the power classification of the second node set is performed according to the historical power data of the second node. The specific steps include:

[0046] Sort the power values ​​of each second node in the second node set at the same time point in order of magnitude.

[0047] Call the preset power level set of the second node mentioned above and the power threshold range corresponding to each power level;

[0048] Extract the minimum and maximum power values ​​after sorting;

[0049] Based on the aforementioned minimum power value, maximum power value, and power threshold range corresponding to the power level, the power level subset corresponding to the second node set is determined;

[0050] By using the power values ​​of each second node in the second node set and the power threshold ranges corresponding to each power level in the power level subset, the power level corresponding to each second node in the second node set is determined.

[0051] By comparing the power values ​​of all second-level nodes at the same time point, the differences in real-time supply and demand between nodes are reflected, providing data support for hierarchical resource allocation and avoiding the problem of resource surplus for low-load nodes or insufficient allocation for high-load nodes. Higher-level nodes (i.e., high-load nodes) are given priority in obtaining resources to ensure the stability of the main line. When the power grid topology is updated, the power values ​​are used to classify the connected or interrupted power equipment, enabling the system to quickly adapt to the new node relationships and forming a closed loop of "topology awareness → dynamic classification → allocation".

[0052] Furthermore, since nodes within the same power level have similar load characteristics, deep learning models can train power-environment correlation functions for different levels, thereby achieving data dimensionality reduction.

[0053] Third embodiment:

[0054] Based on the second embodiment, the time segments of the second nodes are divided according to the power levels of each second node in the second node set at each time point in a time period, and the correspondence between time segments and power levels is obtained.

[0055] The second embodiment achieves node power classification at a single time point, but power load exhibits significant temporal fluctuations (such as diurnal differences and seasonal variations). A time-dimensional aggregation analysis is introduced, dividing time periods and establishing a time-segment-power level correspondence to elevate the classification results from discrete time points to load characteristics with temporal continuity.

[0056] One possible use case is to divide the day into morning peak time periods, midday off-peak time periods, and evening peak time periods. The power levels at nodes within each time period are relatively stable. Variance analysis is performed on the power levels at different times within the same time period to ensure that the fluctuations in power levels within the time period are within a preset tolerance range, avoiding excessive segmentation. Different power-environment correlation functions are used for different time periods. For example, the morning peak time period focuses on user behavior, while the midday off-peak time period focuses on temperature, thereby improving the prediction accuracy of power dispatching strategies.

[0057] Fourth embodiment:

[0058] Based on the third embodiment, historical environmental data of each second node in the same power level within the same time period is obtained, and the relationship between historical environmental data and historical power data is learned through a deep learning model to obtain the power-environment correlation function of the second node.

[0059] One possible use case is a power grid in a coastal industrial area that supplies power through a substation. During the evening peak hours, devices A, B, C, and D are directly connected to the substation. Devices a through d are also connected after device A. The power grid topology is constructed based on the substation, devices A through D, and devices a through d (e.g., ...). Figure 2 The equipment in the evening peak time period is classified into different levels, and equipment A is the highest power level (called power level 1), equipment B to D are power level 2, and equipment a to d are power level 1.

[0060] Taking device A as an example with power level 1, the training is carried out by inputting the average current per minute of device A during the evening peak hours of nearly one year and the corresponding environmental data (temperature and humidity during the same period obtained through the meteorological bureau API; daily cargo tonnage of the port obtained through the customs database; and night shift schedule of the factory obtained through the enterprise management system) into the LSTM neural network to learn the relationship between environmental data and power load, and obtain the trained LSTM model of device A (i.e., power-environment correlation function).

[0061] Obtain the environmental data of device A at the current moment, input the environmental data into the trained LSTM model of device A, and predict the power required by device A. When the temperature rises, it causes a surge in air conditioning load; the arrival of a new cargo ship at a port increases cargo volume; and the temporary addition of night shifts at a factory alters user behavior.

[0062] Fifth embodiment:

[0063] Based on any of the above embodiments, the power grid topology is updated in real time, and the updating steps include:

[0064] When the aforementioned power equipment connects to the first node as a second node, the power equipment sends an access request to the first node;

[0065] Starting from the first node mentioned above, access requests are transmitted to the root node sequentially through adjacent nodes. After receiving the access request, the root node responds by transmitting the access response sequentially through adjacent nodes to the first node.

[0066] When the root node issues an access response, the power grid topology is updated.

[0067] In a specific embodiment, the step of updating the above-mentioned power grid topology in real time further includes:

[0068] When the aforementioned power equipment, acting as the second node, interrupts its connection to the first node, the power equipment sends an interruption request to the first node.

[0069] Starting from the first node mentioned above, interrupt requests are transmitted to the root node sequentially through adjacent nodes. Upon receiving an interrupt request, the root node responds by transmitting an interrupt response sequentially through adjacent nodes to the first node.

[0070] When the root node issues an interrupt response, update the power grid topology.

[0071] Traditional power dispatching methods primarily rely on static grid topology and pre-defined load forecasting models, lacking the ability to dynamically perceive and adaptively adjust to the real-time state of the grid. This makes them ill-suited to adapting to dynamic changes in power load, and power dispatching strategies based on static grid topology are prone to issues such as excessively low or high load voltage drops. By updating the grid topology in real time, the dynamic perception and adaptive adjustment capabilities of the grid to its real-time state can be improved, addressing the problem of existing technologies failing to adapt to dynamic changes in power load and resulting in excessively low or high load voltage drops.

[0072] Sixth embodiment:

[0073] A big data-based dynamic power dispatching system is used to implement the aforementioned dynamic power dispatching method.

[0074] The power dynamic dispatch system includes:

[0075] The information acquisition module is used to collect information from various power equipment in the dispatch area.

[0076] A network construction module is connected to the information acquisition module. The network construction module is used to construct a power grid topology based on information from power equipment. Each power device is considered a node, and the connections between power devices are considered edges.

[0077] The scheduling strategy generation module is connected to the network construction module and the information acquisition module; the scheduling strategy generation module is used to perform the following steps:

[0078] Starting from the root node, select adjacent nodes in sequence as the first node;

[0079] Extract the adjacent nodes after the first node as the second node to obtain the second node set;

[0080] Obtain historical power data for each second node in the aforementioned second node set;

[0081] Based on the historical power data of the second node, the power level of the second node set is classified to obtain a power level subset;

[0082] Historical environmental data of each second node in the same power level is obtained, and the relationship between historical environmental data and historical power data is learned through a deep learning model to obtain the power-environment correlation function of the second node.

[0083] The power dispatch strategy for the second node is determined by using the environmental data at the current time point and the power-environment correlation function.

[0084] In a specific embodiment, the network construction module is also used to update the power grid topology in real time, and the update steps include:

[0085] When the aforementioned power equipment connects to the first node as a second node, the power equipment sends an access request to the first node;

[0086] Starting from the first node mentioned above, access requests are transmitted to the root node sequentially through adjacent nodes. After receiving the access request, the root node responds by transmitting the access response sequentially through adjacent nodes to the first node.

[0087] When the root node issues an access response, the power grid topology is updated.

[0088] When the aforementioned power equipment, acting as a second node, interrupts its connection to the first node, the power equipment sends an interruption request to the first node.

[0089] Starting from the first node mentioned above, interrupt requests are transmitted to the root node sequentially through adjacent nodes. Upon receiving an interrupt request, the root node responds by transmitting an interrupt response sequentially through adjacent nodes to the first node.

[0090] When the root node issues an interrupt response, update the power grid topology.

[0091] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A power dynamic dispatching method based on big data, characterized in that, Includes the following steps: Obtain information on each power device in the dispatch area, and construct a power grid topology based on the information of the power devices; The power grid topology is updated in real time; wherein, each power device is a node, and the connection relationship between power devices is an edge. Starting from the root node, select adjacent nodes in sequence as the first node; extract the adjacent nodes after the first node as the second node to obtain the second node set; wherein, the root node is a power generation or energy storage device; Obtain historical power data for each second node in the second node set; Based on the historical power data of the second node, the power level of the second node set is classified to obtain a power level subset; Historical environmental data of each second node in the same power level is obtained, and the relationship between historical environmental data and historical power data is learned through a deep learning model to obtain the power-environment correlation function of the second node. The power dispatch strategy for the second node is determined by using the environmental data at the current time point and the power-environment correlation function. The historical power data refers to the power values ​​flowing through the second node at each point in time; Based on the historical power data of the second node, the power classification of the second node set is performed, and the specific steps include: Sort the power values ​​of each second node in the second node set at the same time point in order of magnitude; Call the preset power level set of the second node and the power threshold range corresponding to each power level; Extract the minimum and maximum power values ​​after sorting; The power level subset corresponding to the second node set is determined by the minimum power value, the maximum power value, and the power threshold range corresponding to the power level. The power level corresponding to each second node in the second node set is determined by the power value of each second node in the second node set and the power threshold range corresponding to each power level in the power level subset.

2. The power dynamic dispatching method according to claim 1, characterized in that, Based on the power levels of each second node in the second node set at each time point in a time period, the time segments of the second nodes are divided to obtain the correspondence between time segments and power levels.

3. The power dynamic dispatching method according to claim 2, characterized in that, Historical environmental data of each second node in the same power level within the same time period are obtained. The relationship between historical environmental data and historical power data is learned through a deep learning model to obtain the power-environment correlation function of the second node.

4. The power dynamic dispatching method according to claim 1, characterized in that, The power grid topology is updated in real time, and the update steps include: When the power equipment connects to the first node as a second node, the power equipment sends an access request to the first node; Starting from the first node, access requests are transmitted sequentially to the root node through adjacent nodes. After receiving the access request, the root node responds by transmitting the access response sequentially to the first node through adjacent nodes. When the root node issues an access response, the power grid topology is updated.

5. The power dynamic dispatching method according to claim 2, characterized in that, The step of updating the power grid topology in real time also includes: When the power equipment, acting as the second node, interrupts its connection to the first node, the power equipment sends an interruption request to the first node. Starting from the first node, interrupt requests are transmitted to the root node sequentially through adjacent nodes. Upon receiving an interrupt request, the root node responds by transmitting an interrupt response sequentially through adjacent nodes to the first node. When the root node issues an interrupt response, the power grid topology is updated.

6. A power dynamic dispatching system based on big data, characterized in that, The power dynamic dispatching system is used to implement the power dynamic dispatching method according to any one of claims 1 to 5; The power dynamic dispatch system includes: Information acquisition module, which is used to collect information of various power equipment in the dispatch area; A network construction module is connected to an information acquisition module; the network construction module is used to construct a power grid topology based on information from power equipment; wherein, each power equipment is considered a node, and the connections between power equipment are considered edges. A scheduling strategy generation module, connected to the network construction module and the information acquisition module, is used to perform the following steps: Starting from the root node, select adjacent nodes in sequence as the first node; Extract the adjacent nodes after the first node as the second node to obtain the second node set; Obtain historical power data for each second node in the second node set; Based on the historical power data of the second node, the power level of the second node set is classified to obtain a power level subset; Historical environmental data of each second node in the same power level is obtained, and the relationship between historical environmental data and historical power data is learned through a deep learning model to obtain the power-environment correlation function of the second node. The power dispatch strategy for the second node is determined by using the environmental data at the current time point and the power-environment correlation function.

7. The power dynamic dispatching system according to claim 6, characterized in that, The network construction module is also used to update the power grid topology in real time, and the update steps include: When the power equipment connects to the first node as a second node, the power equipment sends an access request to the first node; Starting from the first node, access requests are transmitted sequentially to the root node through adjacent nodes. After receiving the access request, the root node responds by transmitting the access response sequentially to the first node through adjacent nodes. When the root node issues an access response, the power grid topology is updated.

8. The power dynamic dispatching system according to claim 7, characterized in that, The steps for updating the power grid topology by the network construction module also include: When the power equipment, acting as the second node, interrupts its connection to the first node, the power equipment sends an interruption request to the first node. Starting from the first node, interrupt requests are transmitted to the root node sequentially through adjacent nodes. Upon receiving an interrupt request, the root node responds by transmitting an interrupt response sequentially through adjacent nodes to the first node. When the root node issues an interrupt response, the power grid topology is updated.

Images