CN-121981525-A - Power grid fault risk positioning method, device, equipment and medium based on energy margin

CN121981525ACN 121981525 ACN121981525 ACN 121981525ACN-121981525-A

Abstract

The application discloses a power grid fault risk positioning method, device, equipment and medium based on energy margin, wherein the method comprises the steps of determining basic energy margin of each node in a power grid according to power grid operation data; the method comprises the steps of correcting a basic energy margin based on a hybrid neural network of a long-term memory network and a reverse propagation network to obtain a corrected dynamic energy margin, constructing a graph neural network according to a power grid topology, determining node characteristics of the graph neural network based on the dynamic energy margin to generate vulnerability scores of all elements in the power grid based on the graph neural network, constructing a two-dimensional risk assessment matrix according to the dynamic energy margin and the vulnerability scores, and judging risk grades of all elements in the power grid to perform risk positioning. The method and the system can improve the instantaneity and accuracy of power grid fault risk positioning.

Inventors

ZHOU XIN
ZHOU YUMIN
YAO HAICHENG
YANG XINYI
XIAO LIANG
MAO ZHENYU
WANG WEI
YUAN QUAN
DU XU
ZHOU YONGCAN
ZHANG QIANG

Assignees

中国南方电网有限责任公司

Dates

Publication Date: 20260505
Application Date: 20251229

Claims (10)

1. The utility model provides a power grid fault risk positioning method based on energy margin, which is characterized by comprising the following steps: Acquiring power grid operation data in a preset time period; According to the power grid operation data, determining basic energy margin of each node in the power grid; Based on a mixed neural network of a long-term memory network and a back propagation network, correcting the basic energy margin to obtain a corrected dynamic energy margin; Constructing a graph neural network according to a power grid topology, and determining node characteristics of the graph neural network based on the dynamic energy margin to generate vulnerability scores of all elements in the power grid based on the graph neural network, wherein the elements comprise nodes and edges for connecting the nodes; And constructing a two-dimensional risk assessment matrix according to the dynamic energy margin and the vulnerability score, and judging the risk level of each element in the power grid so as to perform risk positioning.
2. The energy margin-based power grid fault risk location method of claim 1, wherein determining the base energy margin for each node in the power grid from the power grid operation data comprises: according to the power grid operation data, obtaining actual power and critical power of each node in the power grid; Determining energy margin of each node according to the difference value between the actual power and the critical power; and generating a time sequence energy margin matrix based on the energy margins corresponding to the time periods, and taking the time sequence energy margin matrix as a basic energy margin.
3. The method for positioning a power grid fault risk based on an energy margin according to claim 2, wherein the method for acquiring the critical power specifically comprises: Extracting static data of a single period according to the power grid operation data and a preset period, wherein the static data comprises power grid topology parameters, operation state parameters and equipment rated parameters; The node power balance in the power flow calculation is taken as a core, an active power equation and a reactive power equation of each node are constructed, the load is gradually increased by adopting a continuous power flow method, the process that the system approaches to the voltage collapse critical point is simulated, and the critical power is obtained through iterative solution, wherein the critical power comprises an active power critical value and a reactive power critical value.
4. The energy margin based grid fault risk location method of claim 2, wherein the energy margin comprises a node level functional margin, a node level nonfunctional margin and a system level integrated energy margin; The step of determining the energy margin of each node according to the difference between the actual power and the critical power comprises the following steps: according to the difference value between the actual active power corresponding to the node and the active power critical value, calculating to obtain a node-level functional capacity margin corresponding to the node; According to the difference value between the actual reactive power and the reactive power critical value corresponding to the node, calculating to obtain a node level nonfunctional margin corresponding to the node; And determining a system-level comprehensive energy margin according to the node-level functional margin and the node-level nonfunctional margin corresponding to all nodes in the system.
5. The method for positioning a power grid fault risk based on an energy margin according to claim 1, wherein the modifying the basic energy margin by using the hybrid neural network based on the long-short-term memory network and the back propagation network to obtain the modified dynamic energy margin comprises: Extracting disturbance influence characteristics and equipment state characteristics according to the power grid operation data, wherein the disturbance influence characteristics comprise at least one of new energy output fluctuation quantity, load mutation rate and meteorological extreme index, and the equipment state characteristics comprise at least one of line loss rate, transformer aging coefficient and reactive compensation equipment response state; Inputting the basic energy margin, the disturbance influence characteristic and the equipment state characteristic into the hybrid neural network as input characteristics, extracting the characteristics through a long-short-period memory network, and mapping and outputting through a back propagation network to obtain a correction coefficient; and calculating the corrected dynamic energy margin according to the product of the basic energy margin and the correction coefficient.
6. The energy margin-based grid fault risk localization method of claim 1, wherein constructing a graph neural network from a grid topology and determining node characteristics of the graph neural network based on the dynamic energy margin comprises: abstracting a physical entity in a power grid as a node, abstracting a connecting device connected with the physical entity as an edge, and constructing a graph neural network according to the power grid topology; Acquiring the dynamic energy margin, node voltage amplitude, active power and reactive power corresponding to the node as state characteristics; acquiring an aging coefficient, a historical failure rate and a latest maintenance time interval of equipment associated with the node as equipment characteristics; Extracting the medium centrality and the degree centrality of the nodes based on the power grid topology, and taking the medium centrality and the degree centrality as topological characteristics; and taking the state characteristics, the equipment characteristics and the topology characteristics as node characteristics of the nodes.
7. The method for positioning a risk of a power grid fault based on an energy margin according to claim 1, wherein constructing a two-dimensional risk assessment matrix according to the dynamic energy margin and the vulnerability score, and determining risk levels of elements in the power grid to perform risk positioning comprises: Respectively grading the dynamic energy margin and the vulnerability score to obtain dynamic energy margin grading and vulnerability score grading; constructing a two-dimensional risk assessment matrix based on the dynamic energy margin grading and the vulnerability grading, and calculating a risk index of each element in the two-dimensional risk assessment matrix, wherein the risk index is used for quantifying the risk degree; according to the dynamic energy margin grading and vulnerability grading corresponding to each element, matching with the two-dimensional risk assessment matrix to obtain a risk index corresponding to each element; and determining a corresponding risk level according to the risk index so as to position the element with risk according to the risk level.
8. An energy margin-based power grid fault risk positioning device, comprising: The data acquisition module is used for acquiring power grid operation data in a preset time period; The energy margin calculation module is used for determining the basic energy margin of each node in the power grid according to the power grid operation data; The energy margin correction module is used for correcting the basic energy margin based on a hybrid neural network of the long-term memory network and the back propagation network to obtain a corrected dynamic energy margin; The vulnerability assessment module is used for constructing a graph neural network according to the power grid topology, determining node characteristics of the graph neural network based on the dynamic energy margin, and generating vulnerability scores of all elements in the power grid based on the graph neural network, wherein the elements comprise nodes and edges for connecting the nodes; And the risk positioning module is used for constructing a two-dimensional risk assessment matrix according to the dynamic energy margin and the vulnerability score, and judging the risk level of each element in the power grid so as to perform risk positioning.
9. A computer device comprising a processor and a memory, the memory having a computer program stored therein and the computer program being configured to be executed by the processor, when executing the computer program, implementing the energy margin based grid fault risk localization method of any one of claims 1 to 7.
10. A computer readable storage medium, wherein the computer readable storage medium stores a computer program, and wherein the computer program is executed by a device in which the computer readable storage medium is located, to implement the power grid fault risk positioning method based on energy margin according to any one of claims 1 to 7.

Description

Power grid fault risk positioning method, device, equipment and medium based on energy margin Technical Field The application relates to the technical field of power systems, in particular to a power grid fault risk positioning method, device, equipment and medium based on energy margin. Background With the continuous advancement of global energy transformation, the large-scale access of new energy enables the power grid operation environment to gradually present high fluctuation, strong coupling and dynamic characteristics, and the traditional risk assessment system faces significant challenges. On one hand, wind power and photovoltaic output have obvious randomness and time sequence fluctuation, so that key safety indexes such as voltage stability allowance, power transmission allowance and the like are changed rapidly along with time, actual running risks are difficult to reflect in time by a traditional method, on the other hand, the flexible access of ultra-high voltage interconnection and distributed energy storage enables the power grid topology to evolve into a complex dynamic network structure from a relatively simple radiation structure, element faults are easy to cause chain reaction through tide transfer, the traditional complex network medium number method only evaluates element states in an isolated mode, and the vulnerability evaluation accuracy of the power grid is low. The prior power system risk assessment technology can be divided into two types, namely physical model driving and single neural network, the prior physical model driving method generally depends on typical daily data and fixed topology, and has the defects of stationarity, conservation and slow response, while the single neural network method focuses on single task more, lacks coupling with the physical characteristics of a power grid, has insufficient interpretation, and is difficult to provide reliable local risk positioning for scheduling decisions. Disclosure of Invention In order to solve the technical problems, the application provides a power grid fault risk positioning method, device, equipment and medium based on energy margin, which can improve the real-time performance and accuracy of power grid fault risk positioning. The application provides a power grid fault risk positioning method based on energy margin, which comprises the following steps: Acquiring power grid operation data in a preset time period; According to the power grid operation data, determining basic energy margin of each node in the power grid; Based on a mixed neural network of a long-term memory network and a back propagation network, correcting the basic energy margin to obtain a corrected dynamic energy margin; Constructing a graph neural network according to a power grid topology, and determining node characteristics of the graph neural network based on the dynamic energy margin to generate vulnerability scores of all elements in the power grid based on the graph neural network, wherein the elements comprise nodes and edges for connecting the nodes; And constructing a two-dimensional risk assessment matrix according to the dynamic energy margin and the vulnerability score, and judging the risk level of each element in the power grid so as to perform risk positioning. As an improvement of the above solution, the determining, according to the power grid operation data, a basic energy margin of each node in the power grid includes: according to the power grid operation data, obtaining actual power and critical power of each node in the power grid; Determining energy margin of each node according to the difference value between the actual power and the critical power; and generating a time sequence energy margin matrix based on the energy margins corresponding to the time periods, and taking the time sequence energy margin matrix as a basic energy margin. As an improvement of the above solution, the method for obtaining the critical power specifically includes: Extracting static data of a single period according to the power grid operation data and a preset period, wherein the static data comprises power grid topology parameters, operation state parameters and equipment rated parameters; The node power balance in the power flow calculation is taken as a core, an active power equation and a reactive power equation of each node are constructed, the load is gradually increased by adopting a continuous power flow method, the process that the system approaches to the voltage collapse critical point is simulated, and the critical power is obtained through iterative solution, wherein the critical power comprises an active power critical value and a reactive power critical value. As an improvement of the scheme, the energy margin comprises a node level functional margin, a node level nonfunctional margin and a system level comprehensive energy margin; The step of determining the energy margin of each node according to the difference between the actual po