Power distribution system power failure monitoring and early warning method based on line topology structure

Abstract

The present application relates to the technical field of smart grid, and particularly relates to a power distribution system power failure monitoring and early warning method based on line topology structure, which solves the technical problem of limited monitoring precision and generalization ability of the prior art. The method comprises: constructing a line topology model of a power distribution network, the line topology model being composed of a node topology graph, nodes in the node topology graph being used to represent busbars in the power distribution network, and edges between the nodes being used to represent lines having a connection relationship between the busbars; identifying key nodes in the node topology graph based on monitoring data of each node and edges between the nodes in the node topology graph; and performing differential hierarchical power failure monitoring and early warning for the key nodes and non-key nodes in the node topology graph respectively.

Description

Technical Field

[0001] This application relates to the field of smart grid technology for power grid monitoring and early warning, specifically a method for power outage monitoring and early warning of distribution systems based on line topology. Background Technology

[0002] In smart grids, as power system structures become increasingly complex and grid scale continues to expand, intelligent monitoring and early warning of power outages in distribution systems are becoming crucial for ensuring the safe and stable operation of the power grid. Distribution network systems rely on the grid's line topology, high-precision sensors, and data analysis platforms to collect parameters such as voltage, current, frequency, and power factor in real time. Through real-time monitoring, fault prediction, and rapid response, they minimize the frequency of power outages and their impact on the distribution system.

[0003] In existing technologies, power outage monitoring and early warning in power distribution systems mostly employ static models or classic machine learning algorithms (such as support vector machines and random forests) for power outage risk prediction. While these methods can achieve basic early warning functions in specific scenarios, they suffer from limited monitoring accuracy and generalization ability overall, leading to biased power outage risk predictions and making it difficult to meet the monitoring and early warning needs of large-scale and complex power distribution networks. Summary of the Invention

[0004] To address the limitations of existing technologies in terms of monitoring accuracy and generalization ability, this invention aims to provide a power outage monitoring and early warning method for power distribution systems based on line topology. The specific technical solution adopted is as follows: Construct a line topology model of the power distribution network. The line topology model consists of a node topology graph. The nodes in the node topology graph represent the buses in the power distribution network, and the edges between the nodes represent the lines that are connected to the buses. Based on the monitoring data of each node and the edges between each node in the node topology graph, identify the key nodes in the node topology graph. Differentiated, tiered power outage monitoring and early warning are implemented for critical and non-critical nodes in the node topology graph.

[0005] In one possible implementation, the node monitoring data includes at least one of node voltage deviation rate, power margin, and voltage margin; the edge monitoring data includes node parameter data and environmental factor data; the node parameter data includes at least one of line load rate, transformer load rate, and conductor overload rate; and the environmental factor data includes at least one of wind force, pressure, temperature, and humidity.

[0006] In one possible implementation, the method includes: Based on the monitoring data of each node and the edges between each node in the node topology graph, the feature value sequence of each node and the feature value sequence of each edge are determined respectively. Topological identification of the node topology graph is performed based on the feature value sequence of each node and the feature value sequence of each edge, and key nodes in the node topology graph are identified.

[0007] In one possible implementation, the method includes: For each node in the node topology graph, the stability evaluation parameters of the node are calculated based on the feature value sequence of the node and its nearest neighbor nodes at each monitoring time. For each node in the node topology graph, the topological criticality evaluation parameters of the node are calculated based on the feature value sequence of the edges connecting the node to each nearest neighbor node at each monitoring time. Key nodes in the node topology graph are identified based on the stability evaluation parameters and topology criticality evaluation parameters for each node.

[0008] In one possible implementation, the method includes: For each node in the node topology graph, positive and negative correlation elements of node features are determined from the feature value sequences of the node and its nearest neighbors at each monitoring time. Based on these elements, the node distribution fault probability weights between the node and its nearest neighbors are determined. Positive correlation elements are those elements in the node's feature value sequence that are positively correlated with the distribution fault probability; negative correlation elements are those elements in the node's feature value sequence that are negatively correlated with the distribution fault probability. Calculate the node sequence similarity between the feature value sequences of the calculated node and its nearest neighbor nodes at each monitoring time. The stability evaluation parameters of a node are calculated based on the node power distribution fault probability weight and the node sequence similarity.

[0009] In one possible implementation, the method includes: For each node in the node topology graph, positive and negative correlation elements of the edge features are determined from the feature value sequence of the edges connecting the node to any two nearest neighbors at each monitoring time. Then, the edge distribution fault probability weight of the edges connecting the node to any two nearest neighbors is determined based on these elements. Positive correlation elements are those elements in the feature value sequence that are positively correlated with the distribution fault probability; negative correlation elements are those elements in the feature value sequence that are negatively correlated with the distribution fault probability. Calculate the edge sequence similarity between the feature value sequences of the edges connecting a node to any two nearest neighbor nodes at each monitoring time. The stability evaluation parameters of nodes are calculated based on the edge distribution fault probability weight and edge sequence similarity.

[0010] In one possible implementation, the method includes: For each node in the node topology graph, the stability evaluation parameters and topology criticality evaluation parameters of the node are weighted, and the weighted degree of the node is generated by combining the degree of the node. Based on the weighted degree of each node in the node topology graph, a hierarchical core decomposition is performed on the node topology graph to select key nodes.

[0011] In one possible implementation, the method further includes: For each monitoring time, clustering is performed based on the difference distance between the feature value sequences of key nodes and non-key nodes, with key nodes as the center in the node topology graph. Dynamic topology reconstruction of the node topology graph is then performed based on the clustering results.

[0012] In one possible implementation, the method includes: High-frequency monitoring and early warning are performed on key nodes. The high-frequency monitoring and early warning scans the status of key nodes at a first preset time interval. Low-frequency monitoring and early warning are performed on non-critical nodes. The low-frequency monitoring and early warning scans the status of non-critical nodes at a second preset time interval. The first preset time interval is less than the second preset time interval.

[0013] In one possible implementation, the method further includes: In response to the detection of abnormal operating status data on any node in the node topology graph and the identification of a broken connection between the nodes through topology identification, a power outage fault warning message is generated for the node, and the power outage fault warning message is marked on the corresponding node in the node topology graph.

[0014] The present invention has the following beneficial effects: Based on the above technical solution, this application constructs a line topology model with buses in the power distribution network as nodes and connected lines as edges to accurately reconstruct the physical connection relationships of the power distribution network. Based on monitoring data of each node and the edges between nodes in the node topology model, key nodes in the node topology model are identified. Thus, this application can perform differentiated and hierarchical power outage monitoring and early warning for key and non-key nodes in the node topology model. Compared to existing technologies that only use static models or classic machine learning algorithms for power outage risk prediction, this application can build an effective modeling topology structure, avoiding identification biases in topology locations. Furthermore, through differentiated hierarchical early warning for key and non-key nodes, it can ensure real-time monitoring of key nodes, optimize system resource allocation, and overall improve the monitoring accuracy and generalization capability of power outage monitoring and early warning in the power distribution system. Attached Figure Description

[0015] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a flowchart illustrating a power outage monitoring and early warning method for a power distribution system based on line topology, provided in one embodiment of the present invention. Figure 2 This is a schematic diagram of the node topology of a power distribution network provided in one embodiment of the present invention; Figure 3 This is a flowchart illustrating another power outage monitoring and early warning method for a power distribution system based on line topology, provided in one embodiment of the present invention. Figure 4 This is a flowchart illustrating another power outage monitoring and early warning method for a power distribution system based on line topology, provided as an embodiment of the present invention. Detailed Implementation

[0017] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a power outage monitoring and early warning method for a power distribution system based on line topology proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0018] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0019] As power system structures become increasingly complex and power grids continue to expand, intelligent monitoring and early warning of power outages in distribution systems are becoming crucial for ensuring the safe and stable operation of the power grid. Distribution network systems rely on the grid's line topology, high-precision sensors, and data analysis platforms to collect parameters such as voltage, current, frequency, and power factor in real time. Through real-time monitoring, fault prediction, and rapid response, they minimize the frequency of power outages and their impact on the distribution system.

[0020] Traditional methods use static models or classic machine learning algorithms (such as support vector machines and random forests) to predict power outage risks, and the system immediately triggers an early warning mechanism. While these methods have some practicality under specific conditions, they generally suffer from the following shortcomings: they cannot effectively model the complex topological structures and global relationships between power equipment; and when faced with high-dimensional, multi-source, and heterogeneous data, they cannot accurately identify topological relationships, resulting in limited accuracy and generalization ability in power outage monitoring.

[0021] The following description, in conjunction with the accompanying drawings, details the specific scheme of a power outage monitoring and early warning method for a power distribution system based on line topology provided by the present invention.

[0022] Please see Figure 1 The diagram illustrates a flowchart of a power outage monitoring and early warning method for a power distribution system based on line topology, according to an embodiment of the present invention. The method includes the following steps: Step 101: Construct the line topology model of the power distribution network.

[0023] The line topology model consists of a node topology graph. Nodes in the node topology graph represent buses in the power distribution network, and edges between nodes represent lines that are connected to buses.

[0024] In the distribution network, the busbar is the key connection point for power distribution, which can connect power equipment such as distribution transformers and load centers. Therefore, using the busbar as a node in the node topology diagram can accurately reflect the core connection relationship of the distribution network. Lines that are connected between the busbars (such as connecting lines, cables, overhead lines, transformers and other electrical components, including the resistance, reactance and susceptance of the conductors themselves) are used as edges between nodes, which can reflect the physical connection logic of the distribution network.

[0025] For example, the node system used in a power distribution network can be a circuit system defined by the Institute of Electrical and Electronics Engineers (IEEE), including but not limited to IEEE 33, IEEE 85, and IEEE 137, such as... Figure 2 The diagram shown is a node topology diagram of the power distribution network provided in this application embodiment. The node topology diagram contains 33 nodes and 32 branches, where solid lines are connected transmission lines and dashed lines are standby connected transmission lines.

[0026] Based on the above examples, this application can use the buses in the power distribution network as nodes in the node topology graph, the lines with connectivity between buses as edges between nodes in the node topology graph, and the power supply node at the beginning of the power distribution network as the reference node. The resulting node topology graph can be represented as follows: ,in, Represents a set of nodes. , Let represent the i-th node; E represents the set of edges. , Let $N$ represent the edge between the $i$-th node and the $j$-th node; $N$ represents the total number of nodes in the node topology graph, for example, $\mathbf{i}$. Figure 2 The node topology graph in the graph has N = 33.

[0027] It should be noted that this application can select an appropriate node system based on the complexity and scale of the power distribution system. The higher the complexity and the larger the scale, the more complex the abstracted topology becomes, so as to accurately represent the network structure of the power distribution system. Therefore, a node topology with a large number of nodes should be selected. Conversely, the lower the complexity and the smaller the scale, the more accurately the topology with a small number of nodes can be represented.

[0028] Step 102: Based on the monitoring data of each node and the edges between each node in the node topology graph, identify the key nodes in the node topology graph.

[0029] Critical nodes refer to nodes in the distribution network that are sensitive to power outages and play a core role in topology identification. Their operational status directly affects the safety and stability of the entire distribution network. In existing technologies, when identifying the node topology of a distribution network, the relevant measurement data of the nodes, such as electrical parameters like current, voltage, and power, are usually treated as time-series data, ignoring the nodes' connectivity within the distribution network. This can lead to identification errors in similar topologies, resulting in low accuracy and subsequent errors in power outage warnings.

[0030] This application can identify key nodes through monitoring data of nodes and edges, avoiding identification errors caused by relying solely on topological distances (such as the positions of central and edge nodes), and improving the accuracy of key node identification.

[0031] In some embodiments, the monitoring data includes electrical parameters of the power distribution system itself (such as node voltage and line load rate) and environmental data (such as humidity and wind speed). This data can be collected in real time by monitoring equipment and transmitted to data processing equipment for analysis. For example, the data processing equipment can receive monitoring data every 5 minutes as basic data for identifying key nodes. That is to say, the above-mentioned monitoring data can be monitoring data from one or more monitoring moments.

[0032] Step 103: Perform differentiated hierarchical power outage monitoring and early warning for critical nodes and non-critical nodes in the node topology diagram.

[0033] Among them, differentiated hierarchical monitoring and early warning refers to using different monitoring frequencies according to the importance of nodes. For example, a higher monitoring frequency (high-frequency early warning) is used for critical nodes to quickly capture their abnormal status; a lower monitoring frequency (low-frequency early warning) is used for non-critical nodes to balance monitoring accuracy and system resource consumption.

[0034] In one possible implementation, this application can perform high-frequency monitoring and early warning on critical nodes, with the high-frequency monitoring and early warning scanning the state of critical nodes at a first preset time interval; and perform low-frequency monitoring and early warning on non-critical nodes, with the low-frequency monitoring and early warning scanning the state of non-critical nodes at a second preset time interval.

[0035] The first preset time interval is shorter than the second preset time interval. The core of tiered early warning lies in differentiation, that is, achieving rapid response by focusing on key nodes while avoiding resource waste caused by over-monitoring of non-critical nodes. For example, the first preset time interval can be set to 10-15 minutes, and the second preset time interval can be set to 45-60 minutes. For instance, if a node is a bus node (critical node) in the core load center of an city, its operating status can be scanned every 10-15 minutes; if a node is a low-load bus node (non-critical node) in a remote area, it only needs to be scanned every 45-60 minutes.

[0036] In some embodiments, this application may generate a power outage fault warning message for a node in response to detecting abnormal operating status data on any node in the node topology graph and identifying a disconnection of the node's connection relationship through topology identification, and mark the power outage fault warning message on the corresponding node in the node topology graph.

[0037] For example, this application can deploy sensors (such as synchronous phasor measurement units (PMUs) and remote terminal units (RTUs)) on nodes to collect data reflecting the node's operating status in real time, such as voltage, current, and power. The monitoring data is analyzed to initially determine whether a power outage fault has occurred, such as a sudden drop in voltage to zero, a sudden increase in current, or a sudden drop in active power to zero. Secondly, real-time topology identification is performed on the node topology map to determine whether the line connection at the node is broken. When synchronous monitoring data shows a power outage anomaly and the topology identification indicates a broken node connection, a power outage fault is considered to have occurred, and an early warning message is generated. This real-time early warning message is accurately marked on the corresponding node in the topology map.

[0038] Based on the above technical solution, this application constructs a line topology model with buses in the power distribution network as nodes and connected lines as edges to accurately reconstruct the physical connection relationships of the power distribution network. Based on monitoring data of each node and the edges between nodes in the node topology model, key nodes in the node topology model are identified. Thus, this application can perform differentiated and hierarchical power outage monitoring and early warning for key and non-key nodes in the node topology model. Compared to existing technologies that only use static models or classic machine learning algorithms for power outage risk prediction, this application can build an effective modeling topology structure, avoiding identification biases in topology locations. Furthermore, through differentiated hierarchical early warning for key and non-key nodes, it can ensure real-time monitoring of key nodes, optimize system resource allocation, and overall improve the monitoring accuracy and generalization capability of power outage monitoring and early warning in the power distribution system.

[0039] In addition, this application can collect corresponding monitoring data for each node in the node topology graph and the edges between each node. The monitoring data of the nodes and the monitoring data of the edges are described below.

[0040] As one possible embodiment of this application, the node monitoring data includes at least one of node voltage deviation rate, power margin, and voltage margin.

[0041] Among them, the voltage deviation rate is the ratio of the absolute value of the difference between the actual voltage and the rated voltage on the bus to the rated voltage. The larger the value of the voltage deviation rate, the greater the degree to which it exceeds the allowable range, and the higher the probability of power outage faults in the distribution network.

[0042] Power margin is the ratio of the difference between the system's maximum load power and the load power on the bus to the maximum load power. The larger the power margin, the further the operating power is from the allowable power of system failure, the higher the system's safety, and the lower the probability of power outages.

[0043] Similarly, voltage margin is the ratio of the difference between the system's maximum load voltage and the load voltage on the bus to the maximum load voltage. A larger voltage margin means the operating voltage is further from the system's allowable failure voltage, resulting in higher system safety and a lower probability of power outages.

[0044] The monitoring data includes node parameter data and environmental factor data. Node parameter data includes at least one of line load rate, transformer load rate, and conductor overload rate. Environmental factor data includes at least one of wind force, pressure, temperature, and humidity.

[0045] Here, the edges can be electrical components between nodes, such as cables, connecting lines, transformers, etc. The line load rate is the ratio of the transmitted power on the power grid line to the maximum transmitted power. The larger this ratio is, the greater the probability of a power grid outage.

[0046] The transformer load factor is the ratio of the transformer's output active power to its rated power. The transformer load factor is directly proportional to the probability of a power outage in the power grid. The higher the value of the transformer load factor, the higher the power grid load factor, the more likely power outages will occur, affecting the normal operation of the power grid.

[0047] The conductor overload rate is the ratio of the current flowing through the conductor to the conductor's maximum allowable current carrying capacity. The higher the conductor overload rate, the greater the risk of overload and the higher the probability of a power outage.

[0048] Secondly, since environmental factors such as temperature, humidity, pressure, and wind directly affect electrical components—for example, excessive humidity makes breakdown discharge more likely between energized conductors and between conductors and grounded components; and reduced pressure can lead to corona discharge breakdown—the monitoring data in this application may also include environmental factor data. This environmental factor data may include the difference between each environmental factor and the upper limit allowed by the power distribution network, with the ratio of this difference to the corresponding upper limit used as the deviation ratio for each environmental factor.

[0049] This application provides clear data support for the subsequent determination of feature value sequences and identification of key nodes by collecting monitoring data of nodes and edges. Compared with traditional methods, this application not only adds monitoring of edge-related data, but also refines the monitoring data into parameters directly related to the probability of power outages (such as voltage deviation rate and line load rate), which can more accurately reflect the operating status of nodes and edges, thereby improving the accuracy of key node identification and laying a data foundation for subsequent graded early warning.

[0050] The following section provides a further description of the scheme for identifying key nodes in this application.

[0051] As one possible embodiment of this application, combined with Figure 1 ,like Figure 3 As shown, step 102 above can be achieved through the following steps: Step 301: Based on the monitoring data of each node and the edges between each node in the node topology graph, determine the feature value sequence of each node and the feature value sequence of each edge.

[0052] The eigenvalue sequence is an ordered data set formed after normalizing the monitoring data, used to quantify the operational status of nodes and edges. Since different monitoring data have different dimensions (e.g., voltage deviation rate is a percentage, power margin is a percentage, and temperature in environmental data is in degrees Celsius), normalization is required to eliminate the influence of dimensions. For example, normalization can be performed through deviation standardization, standard deviation standardization, etc., and this application does not limit the specific methods used.

[0053] It should be noted that at each monitoring time, the feature value sequence of each node consists of the normalized results of all monitoring data for that node, and the feature value sequence of each edge consists of the normalized results of all monitoring data for that edge. For example, node The monitoring data includes voltage deviation rate, power margin, and voltage margin. Then, in the... At each monitoring time, this node eigenvalue sequence , This is the normalized result of the voltage deviation rate. This is the result of power margin normalization. This is the voltage margin normalization result; the monitoring data for the edges are similar.

[0054] Step 302: Perform topological identification of the node topology graph based on the feature value sequence of each node and the feature value sequence of each edge, and identify the key nodes in the node topology graph.

[0055] When a power distribution fault occurs in the power distribution network, the change of certain switches will inevitably cause the branches in the node topology diagram to be disconnected or connected. This will cause a change in the topology of the node topology diagram. During the change, the connection relationship of some nodes remains stable. Moreover, through these nodes, the topology of the power distribution network can be more easily identified. These nodes are the key nodes in the power distribution network.

[0056] However, since the connections between buses in a power distribution network can vary, the connectivity of different nodes in the node topology graph differs. Judging a node's criticality solely by its distance from the topology center ignores nodes with strong connectivity at the edges. Therefore, this application determines node importance indicators based on the stability of monitoring data for both nodes and edges to identify critical nodes in the node topology graph.

[0057] In some embodiments, this application can perform node topology identification by using the feature value sequences of each node and its nearest neighbor nodes, thereby obtaining the stability evaluation parameters of each node. It can also perform edge topology identification by using the edges connected to each node, thereby obtaining the topological criticality evaluation parameters of each node. Then, based on the two types of parameters, namely the stability evaluation parameters and the topological criticality evaluation parameters, the importance of the nodes can be evaluated, and the key nodes can be selected.

[0058] For example, the more active power a node transmits and the more stable the monitoring data at the node, the lower the probability of power outages during the operation of the distribution network, and the higher its importance in the distribution network. For nodes with strong connectivity, the Kolmogorov-Smirnov (KS) values ​​of the surrounding nodes should generally be large and similar, and the distribution of monitoring data on the connection edges between the node and the surrounding nodes should also be relatively stable, resulting in a lower probability of power outages.

[0059] The following provides a further explanation of the key node identification method provided in this application.

[0060] As one possible embodiment of this application, combined with Figure 3 ,like Figure 4 As shown, step 302 above can be achieved through the following steps: Step 401: For each node in the node topology graph, calculate the stability evaluation parameters of the node based on the feature value sequence of the node and its nearest neighbor nodes at each monitoring time.

[0061] Among them, the nearest neighbor node refers to the node that is directly connected to the node through an edge (i.e., adjacent node). The stability evaluation parameter is used to characterize the consistency and stability of the operating state between the node and the nearest neighbor node. If the similarity of the feature value sequence of the node and the nearest neighbor node is high and stable, the stability evaluation parameter of the node is high, indicating that its operating state is less affected by the surrounding nodes and is more stable in the topology.

[0062] In one possible implementation, for each node in the node topology graph, this application determines the positive and negative correlation elements of node features from the feature value sequences of the node and its neighboring nodes at each monitoring time, and determines the node distribution fault probability weights of the node and its neighboring nodes based on the positive and negative correlation elements of node features. Then, it calculates the node sequence similarity between the feature value sequences of the node and its neighboring nodes at each monitoring time, and calculates the node stability evaluation parameters based on the node distribution fault probability weights and the node sequence similarity.

[0063] Among them, the positive correlation element of node features is the element in the feature value sequence of the node that is positively correlated with the probability of power distribution fault. That is, the larger the value of the positive correlation element of node features, the higher the probability of power distribution fault. For example, the normalized result of node voltage deviation rate (the larger the voltage deviation rate, the higher the probability of fault).

[0064] The negative correlation element of a node feature is the element in the node's feature value sequence that is negatively correlated with the probability of power distribution failure. That is, the larger the value of the negative correlation element of the node feature, the lower the probability of power distribution failure. For example, the normalized result of the node power margin (the larger the power margin, the lower the probability of failure).

[0065] It should be noted that positive correlation means that when one variable increases, the other variable also increases, and the two variables change in the same direction. When one variable changes from large to small or from small to large, the other variable also changes from large to small or from small to large. The specific relationship is determined by the actual application, and this application does not impose any special restrictions.

[0066] For example, the probability weights of node power distribution faults satisfy the following formula:

[0067] in, Represents a node and neighboring nodes In the The monitoring data at each monitoring time point reflects the probability weight of node power distribution faults. , They are nodes Nearest neighbor nodes In the The sum of positively correlated elements of node features in the feature value sequence at each monitoring time point. , They are nodes Nearest neighbor nodes In the The sum of the negatively correlated elements of the node features in the feature value sequence at each monitoring time.

[0068] The stability evaluation parameters satisfy the following formula:

[0069] in, Represents a node Stability evaluation parameters, It is a node The number of nearest neighbors, To monitor the number of time points, , Representing nodes respectively Nearest neighbor nodes In the The sequence of characteristic values ​​at each monitoring time point Represents a sequence and Similarity between them, for example, similarity evaluation metrics include, but are not limited to, cosine similarity and Pearson correlation coefficient.

[0070] This application improves the sensitivity of identified key nodes to monitoring data of power outage faults by weighting the similarity of subsequent sequences with node power distribution fault probability weights.

[0071] Step 402: For each node in the node topology graph, calculate the topological criticality evaluation parameters of the node based on the feature value sequence of the edges connecting the node to each nearest neighbor node at each monitoring time.

[0072] Among them, the topological criticality evaluation parameter is used to characterize the coreness of a node in the topology. If the similarity of the feature value sequence of the edges between a node and multiple neighboring nodes is high and stable, it indicates that the node plays a pivotal role in the power distribution network, its topological criticality evaluation parameter is high, and it has a great impact on the connectivity of the entire network.

[0073] In one possible implementation, for each node in the node topology graph, this application determines the positive and negative correlation elements of the edge features from the feature value sequence of the edges connecting the node to any two nearest neighbor nodes at each monitoring time, and determines the edge distribution fault probability weight of the edges connecting the node to any two nearest neighbor nodes based on the positive and negative correlation elements of the edge features. Then, it calculates the edge sequence similarity between the feature value sequences of the edges connecting the node to any two nearest neighbor nodes at each monitoring time, and calculates the topological criticality evaluation parameters of the node based on the edge distribution fault probability weight and the edge sequence similarity.

[0074] Among them, the positively correlated edge features are the elements in the feature value sequence that are positively correlated with the probability of power distribution faults, such as the normalized results of line load rate and humidity (the higher the line load rate and the greater the humidity, the higher the probability of faults).

[0075] Negatively correlated edge features are elements in the feature value sequence that are negatively correlated with the probability of power distribution faults, such as the normalized result of pressure (the higher the pressure, the lower the probability of faults).

[0076] For example, the probability weights of side distribution faults satisfy the following formula:

[0077] in, Represents a node and neighboring nodes Nearest neighbor nodes The edges connecting them are in the th order. The monitoring data at each monitoring time reflects the probability weight of side distribution faults; For nodes and neighboring nodes The edges connecting them are in the th order. The sum of positively correlated side features in the feature value sequence at each monitoring time point. For nodes and neighboring nodes The edges connecting them are in the th order. The sum of positively correlated edge features in the feature value sequence at each monitoring time point. For nodes and neighboring nodes The edges connecting them are in the th order. The sum of negatively correlated edge features in the feature value sequence at each monitoring time point. For nodes and neighboring nodes The edges connecting them are in the th order. The sum of negatively correlated edge features in the feature value sequence at each monitoring time.

[0078] For example, the topology criticality evaluation parameters satisfy the following formula:

[0079] in, Represents a node Topological criticality evaluation parameters, express A set consisting of nearest neighbor nodes To monitor the number of time points, Represents a node and neighboring nodes The edges connecting them are in the th order. The sequence of characteristic values ​​at each monitoring time point Represents a node and neighboring nodes The edges connecting them are in the th order. The sequence of characteristic values ​​at each monitoring time point Represents a sequence and The similarity between them.

[0080] Step 403: Identify key nodes in the node topology graph based on the stability evaluation parameters and topology criticality evaluation parameters of each node.

[0081] Among them, the stability evaluation parameter reflects the operational stability of the node, while the topology criticality evaluation parameter reflects the degree of topological coreness of the node. Combining the two can comprehensively measure the importance of the node and avoid misjudging critical nodes due to relying on only a single parameter. The node importance evaluation parameter is composed of the stability evaluation parameter and the topology criticality evaluation parameter, and this importance evaluation parameter is positively correlated with the stability evaluation parameter and the topology criticality evaluation parameter, respectively.

[0082] In one possible implementation, for each node in the node topology graph, this application generates a weighted degree of the node based on the node's stability evaluation parameters and topology criticality evaluation parameters, combined with the node's degree. Then, based on the weighted degree of each node in the node topology graph, a hierarchical core decomposition is performed on the node topology graph to select key nodes from the node topology graph.

[0083] For example, the ratio of the linear weighted result of a node's stability evaluation parameter and topology criticality evaluation parameter to the sum of the linear weighted results of all nodes can be used as the importance evaluation parameter of that node. Reasonable weight values ​​can be set for the stability evaluation parameter and the topology criticality evaluation parameter, such as setting the weight values ​​of the stability evaluation parameter and the topology criticality evaluation parameter to 0.4 and 0.6 respectively, and the sum of the products of each stability evaluation parameter and the topology criticality evaluation parameter with their respective weight values ​​is calculated to obtain the linear weighted result of the node's stability evaluation parameter and topology criticality evaluation parameter.

[0084] Next, the weighted degree of a node is obtained by multiplying its importance evaluation parameter by its degree, and key nodes in the node topology are identified using hierarchical core decomposition. For example, hierarchical core decomposition can be achieved using the k-shell decomposition algorithm, the specific process of which will not be elaborated here.

[0085] It should be noted that the number of edges from a node to other nodes is the out-degree, and the number of edges from other nodes to a node is the in-degree. The degree of a node is the sum of its in-degree and out-degree. In this application, the out-degree and in-degree are determined by the direction of active power transmission in the distribution network. For example, active power from node... Input node Then the node The out-degree of the node increases by 1. The in-degree increases by 1.

[0086] In some embodiments, this application may also perform clustering based on the difference distance between the feature value sequences of the key nodes and non-key nodes, with the key nodes in the node topology graph as the center at each monitoring time, and perform dynamic topology reconstruction of the node topology graph based on the clustering results.

[0087] For example, for each monitoring moment in power outage monitoring and early warning, density clustering is used. Centered on a key node, the dissimilarity distance between the feature value sequences of the remaining nodes and the key node at the same monitoring moment is calculated as the metric distance during the clustering process. The remaining nodes are then assigned to the key node with the smallest metric distance, and connections are established between the nodes. This performs dynamic topology reconstruction of the node topology graph. The specific process of density clustering is not detailed here. The dissimilarity distance includes, but is not limited to, Euclidean distance and dynamic time warping (DTW) distance.

[0088] This embodiment quantifies node importance by calculating stability evaluation parameters and topological criticality evaluation parameters. The introduction of distribution fault probability weights allows the evaluation parameters to focus on nodes and edges with high fault risk, while the introduction of sequence similarity ensures that the evaluation parameters reflect the consistency of operating status. The evaluation parameters calculated by combining these two methods can accurately capture the characteristics of critical nodes, providing a reliable quantitative basis for subsequent critical node selection and further improving the accuracy of critical node identification.

[0089] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0090] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A power outage monitoring and early warning method for a power distribution system based on line topology, characterized in that, The method includes: A line topology model of a power distribution network is constructed. The line topology model consists of a node topology graph. The nodes in the node topology graph are used to represent the buses in the power distribution network, and the edges between the nodes are used to represent the lines that are connected to the buses. Based on the monitoring data of each node and the edges between each node in the node topology graph, the key nodes in the node topology graph are identified. Differentiated, tiered power outage monitoring and early warning are implemented for critical and non-critical nodes in the node topology diagram. The identification of key nodes in the node topology graph based on monitoring data of each node and the edges between nodes includes: Based on the monitoring data of each node and the edges between each node in the node topology graph, the feature value sequence of each node and the feature value sequence of each edge are determined respectively. Based on the feature value sequence of each node and the feature value sequence of each edge, topological identification of the node topology graph is performed to identify key nodes in the node topology graph; The topological identification of the node topology graph based on the feature value sequence of each node and the feature value sequence of each edge, and the identification of key nodes in the node topology graph, includes: For each node in the node topology graph, the stability evaluation parameters of the node are calculated based on the feature value sequence of the node and its nearest neighbors at each monitoring time. For each node in the node topology graph, the topological criticality evaluation parameters of the node are calculated based on the feature value sequence of the edges connecting the node to each nearest neighbor node at each monitoring time. Key nodes in the node topology graph are identified based on the stability evaluation parameters and topology criticality evaluation parameters for each node. For each node in the node topology graph, based on the feature value sequence of the node and its nearest neighbors at each monitoring time, the stability evaluation parameters of the node are calculated, including: For each node in the node topology graph, positively correlated and negatively correlated node features are determined from the feature value sequences of the node and its nearest neighbors at each monitoring time. Then, the node distribution fault probability weights between the node and its nearest neighbors are determined based on these positively correlated and negatively correlated node features. The positively correlated node features are the elements in the node's feature value sequence that are positively correlated with the distribution fault probability; the negatively correlated node features are the elements in the node's feature value sequence that are negatively correlated with the distribution fault probability. Calculate the node sequence similarity between the node and its nearest neighbor nodes at each monitoring time; The stability evaluation parameters of the node are calculated based on the node power distribution fault probability weight and the node sequence similarity. For each node in the node topology graph, based on the feature value sequence of the edges connecting the node to each nearest neighbor node at each monitoring time, the topological criticality evaluation parameters of the node are calculated, including: For each node in the node topology graph, positively correlated and negatively correlated edge features are determined from the feature value sequence of the edges connecting the node to any two nearest neighbors at each monitoring time. Then, the edge distribution fault probability weight of the edges connecting the node to any two nearest neighbors is determined based on these positively correlated and negatively correlated edge features. The positively correlated edge features are elements in the feature value sequence that are positively correlated with the distribution fault probability; the negatively correlated edge features are elements in the feature value sequence that are negatively correlated with the distribution fault probability. Calculate the edge sequence similarity between the feature value sequences of the edges connecting the node to any two nearest neighbor nodes at each monitoring time; The topological criticality evaluation parameters of the node are calculated based on the edge distribution fault probability weight and the edge sequence similarity. The identification of key nodes in the node topology graph based on stability evaluation parameters and topology criticality evaluation parameters for each node includes: For each node in the node topology graph, a weighted degree of the node is generated based on the stability evaluation parameters and the topology criticality evaluation parameters of the node, combined with the degree of the node. Based on the weighted degree of each node in the node topology graph, a hierarchical core decomposition is performed on the node topology graph to select key nodes from the node topology graph.

2. The method according to claim 1, characterized in that, The monitoring data of the node includes at least one of node voltage deviation rate, power margin, and voltage margin; the monitoring data of the edge includes node parameter data and environmental factor data; the node parameter data includes at least one of line load rate, transformer load rate, and conductor overload rate; the environmental factor data includes at least one of wind force, pressure, temperature, and humidity.

3. The method according to claim 1, characterized in that, The method further includes: For each monitoring time, clustering is performed based on the difference distance between the feature value sequences of the key nodes and non-key nodes in the node topology graph, with the key nodes as the center, and dynamic topology reconstruction of the node topology graph is performed based on the clustering results.

4. The method according to any one of claims 1-3, characterized in that, The differentiated hierarchical power outage monitoring and early warning system, targeting critical and non-critical nodes in the node topology graph, includes: High-frequency monitoring and early warning are performed on the key nodes, and the high-frequency monitoring and early warning scans the status of the key nodes at a first preset time interval; Low-frequency monitoring and early warning are performed on the non-critical nodes, and the low-frequency monitoring and early warning scans the status of the non-critical nodes at a second preset time interval; Wherein, the first preset time interval is less than the second preset time interval.

5. The method according to claim 4, characterized in that, The method further includes: In response to the detection of abnormal operating status data on any node in the node topology graph and the identification of a disconnected connection of the node through topology identification, a power outage fault warning message is generated for the node, and the power outage fault warning message is marked on the corresponding node in the node topology graph.

Images