Data management method and device of power system and storage medium

Abstract

The invention provides a data management method and device for an electric power system and a storage medium, and the method comprises the steps: carrying out the data classification of collected electric power data of a first electric power system, and obtaining structured data, semi-structured data and unstructured data in the electric power data; respectively constructing a first tree branch, a second tree branch and a third tree branch by taking the first power system as a root node and taking the structured data, the semi-structured data and the non-structured data as child nodes; integrating the first tree branch, the second tree branch and the third tree branch to construct a first data management tree of the first power system; matching a second power system according to the system information of the first power system; and integrating the first data management tree with a second data management tree of the second power system to obtain a target data management tree of the first power system. The power system data are stored in a tree structure mode, and the storage flexibility and query efficiency of the power system data are improved.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to a data management method, device and storage medium for an electric power system. Background Art

[0002] In modern power systems, with the rapid development of smart grid technology, power data has shown explosive growth. These data cover multiple aspects such as the operating status of power equipment, grid topology, power load forecasting, fault diagnosis and recording, and have significant characteristics such as massive, multi-source, and heterogeneous.

[0003] Traditional power system data storage methods mostly use relational databases. For example, grid equipment information is stored in two-dimensional tables, and data in different tables are linked through foreign keys to represent the relationship between equipment and the logical connection between data. However, the data model of a relational database requires a fixed table structure to be defined in advance, and information such as field type and length is determined when the table is created. When a new data type is added to the power system or the data structure changes, such as the introduction of new smart power equipment to bring new monitoring data indicators, the database table structure needs to be modified, which not only involves complex SQL statement operations, but may also affect existing data queries, storage procedures, and application programming interfaces, and data storage flexibility is poor. At the same time, when performing complex associative queries in relational databases, since it is based on table connection operations, when the amount of data is large and the association relationship is complex, the query performance will drop sharply. Summary of the invention

[0004] The present invention provides a data management method, device and storage medium for an electric power system, which are used to improve the storage flexibility and query efficiency of electric power system data.

[0005] In a first aspect, the present invention provides a data management method for a power system, comprising:

[0006] Classifying the collected power data of the first power system to obtain structured data, semi-structured data, and unstructured data in the power data;

[0007] Taking the first power system as a root node, taking the structured data, the semi-structured data and the unstructured data as child nodes, respectively constructing a first tree branch, a second tree branch and a third tree branch;

[0008] Integrate the first tree branch, the second tree branch and the third tree branch to construct a first data management tree for the first power system;

[0009] Matching a second power system according to the system information of the first power system;

[0010] The first data management tree is integrated with the second data management tree of the second power system to obtain a target data management tree of the first power system.

[0011] In a second aspect, the present invention also provides a data management device for a power system, which is applied to the data management method for a power system as described in the first aspect; the data management device for a power system comprises:

[0012] A data classification module, used to classify the collected power data of the first power system to obtain structured data, semi-structured data and unstructured data in the power data;

[0013] A tree branch construction module, used to take the first power system as a root node, the structured data, the semi-structured data and the unstructured data as child nodes, and respectively construct a first tree branch, a second tree branch and a third tree branch;

[0014] A tree branch integration module, used for integrating the first tree branch, the second tree branch and the third tree branch to construct a first data management tree for the first power system;

[0015] A matching module, configured to match a second power system according to the system information of the first power system;

[0016] A management tree integration module is used to integrate the first data management tree with the second data management tree of the second power system to obtain a target data management tree of the first power system.

[0017] In a third aspect, the present invention further provides an electronic device, comprising: a memory for storing a computer software program; a processor for reading and executing the computer software program, thereby implementing any of the above-mentioned methods for data management of the power system.

[0018] In a fourth aspect, the present invention further provides a non-transitory computer-readable storage medium, wherein the storage medium stores a computer software program, and when the computer software program is executed by a processor, the data management method for the power system as described in any one of the above is implemented.

[0019] In a fifth aspect, the present invention further provides a computer program product, comprising a computer program, which, when executed by a processor, implements any of the above-described methods for managing data in a power system.

[0020] The data management method of the power system provided by the embodiment of the present invention, on the one hand, stores the power system data of the power system in a tree structure, so its nodes and tree branches can be flexibly adjusted according to the changes of the power system data. When new equipment or parameters are added, it is only necessary to add new child nodes under the corresponding parent nodes, without large-scale modification of the entire table structure like a relational database, which greatly improves the adaptability to changes in the power system data structure and improves the storage flexibility of the power system data. On the other hand, in the tree structure, data query can quickly locate the target node and obtain the data in the target node by traversing the child nodes under it and the data in its child nodes, and the time complexity is O (n) (n is the number of child nodes), avoiding the need for multiple table connections and screening operations in relational databases, thereby improving the query efficiency of power system data. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 is a flow chart of a data management method for a power system provided by an embodiment of the present invention;

[0022] Figure 2 is a schematic diagram of the structure of a first tree branch provided by an embodiment of the present invention;

[0023] Figure 3 is a schematic diagram of the structure of a second tree branch provided by an embodiment of the present invention;

[0024] Figure 4 is a schematic diagram of the structure of a third tree branch provided by an embodiment of the present invention;

[0025] Figure 5 is a schematic diagram of the structure of a target data management tree provided by an embodiment of the present invention;

[0026] Figure 6 is a structural schematic diagram of a data management device for a power system provided by an embodiment of the present invention;

[0027] Figure 7 An embodiment diagram of an electronic device provided by an embodiment of the present invention;

[0028] Figure 8 An embodiment diagram of a computer-readable storage medium provided for an embodiment of the present invention. DETAILED DESCRIPTION

[0029] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0030] In the description of the present invention, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0031] In the description of the present invention, the term "for example" is used to mean "used as an example, illustration or explanation". Any embodiment described as "for example" in the present invention is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is given to enable any technician in the field to implement and use the present invention. In the following description, details are listed for the purpose of explanation. It should be understood that a person of ordinary skill in the art can recognize that the present invention can be implemented without using these specific details. In other examples, well-known structures and processes will not be elaborated in detail to avoid obscuring the description of the present invention with unnecessary details. Therefore, the present invention is not intended to be limited to the embodiments shown, but is consistent with the widest scope consistent with the principles and features disclosed in the present invention.

[0032] See also Figure 1 , Figure 1 : is a flow chart of the data management method for the power system provided by the present invention. The execution subject of the data management method for the power system in the embodiment of the present invention is a data management device. Therefore, the data management method for the power system includes:

[0033] Step 10: classify the collected power data of the first power system to obtain structured data, semi-structured data and unstructured data in the power data.

[0034] Specifically, the data management device collects power data from any first power system, performs feature extraction on the collected power data, converts each data point in the power data into a corresponding feature vector, and obtains a feature vector set V = {v1, v2, ..., v m}, where m represents the number of data records. For example, for meter readings in structured data, the feature vector may contain dimensions such as timestamp, reading value, and meter number; for equipment maintenance records in semi-structured data, the feature vector may contain keyword occurrence frequency (such as fault type keywords, maintenance action keywords), text length, etc.; for infrared images in unstructured data, the feature vector may be image texture features, color features (converted to numerical representations such as grayscale), shape features (such as geometric features of equipment outlines), and other dimensions.

[0035] Furthermore, the data management device calculates each feature vector v according to the density calculation formula i The density ρ i , we get the density vector set ρ={ρ1,ρ2,...,ρ m}, where the density calculation formula is as follows:

[0036]

[0037] Among them, ρ i Represents data point x i The density of the data point x i is the number of data points in the neighborhood with a radius of r as the center; n is the total number of data points; χ(x) represents the indicator function, when x ≥ 0, χ(x) = 1, otherwise χ(x) = 0; ||x i -x j || represents the Euclidean distance between two eigenvectors.

[0038] Furthermore, for each feature vector v i , the data management device calculates its minimum distance δ to the more dense feature vector i , we get the distance vector set δ={δ1,δ2,...,δ m}, where for each data point x i , if the data point x i is one of the points with the highest density, then otherwise, At the same time, the data management device determines the density peak point set P = {p1, p2, ..., p k}, where k is the number of density peak points. Density peak points are points that have a high density and are far away from other high-density points. They can be determined by setting density thresholds and distance thresholds. The thresholds can be determined based on statistical analysis of the data and multiple experiments.

[0039] Further, starting from the density peak point set P, the data management device associates with each density peak point and satisfies the distance condition (such as δ i The data points with a distance less than the preset distance threshold are divided into the same class, and the divided classes are marked as structured data class C according to the typical feature distribution of each type of data. s , Semi-structured data type C ss and unstructured data class C ns .

[0040] Step 20, taking the first power system as the root node, and taking the structured data, semi-structured data and unstructured data as child nodes, respectively construct a first tree branch, a second tree branch and a third tree branch.

[0041] Furthermore, the data management device uses the first power system as the root node, and uses the equipment hierarchy and data logical relationship in the power system as the hierarchical constraints of the structured data. For example, power generation equipment, transmission equipment, substation equipment, power equipment, power users, etc. are used as upper nodes, and the lower nodes are subdivided according to specific types and parameters. In the process of constructing nodes, the information entropy of each node is calculated to determine the node splitting and merging strategy to improve the rationality of the tree structure and the query efficiency. When the information entropy is large, it means that the data distribution of the node is relatively scattered and the node needs to be further split; when the information entropy is small, it means that the data distribution of the node is relatively concentrated, and it is considered to merge the node or stop splitting to construct the first tree branch, as described in steps 201 to 204.

[0042] Furthermore, the data management device constructs a tree structure according to labels or tag information in the semi-structured data (such as element names in XML or JSON) and data semantic relationships.

[0043] For example, for equipment maintenance records, the maintenance event is taken as the upper node, and the maintenance date, maintenance personnel, maintenance equipment and other information are taken as the lower nodes. Then, the tree branch is constructed according to the fusion degree between each node in the lower node, that is, the nodes with higher fusion degree are constructed as parent-child nodes in the tree structure to obtain the second tree branch, as described in steps 205 to 208.

[0044] Furthermore, the data management device performs feature clustering on the unstructured data, firstly determines the initial value of the number of clusters k, which can be estimated according to the scale and feature distribution of the data, and then adjusted, using a density-adaptive clustering algorithm to cluster data with similar features into one category, thereby dividing the unstructured data into k clusters. Taking the root node "first power system" as the starting point, the vector clustering result is added to the root node as a first-layer child node set, and for each cluster child node, the key feature information is extracted from the data in the cluster to construct the second-layer child node, and the third tree branch is constructed, as described in steps 209 to 213. For example, for infrared images, images with similar temperature distribution and hot spot features are clustered into one category, and a tree structure is constructed based on the vector clustering result. For example, the upper node is the cluster category, and the lower node is the key information extracted from the image features, such as the hot spot location range and temperature range.

[0045] Step 30: Integrate the first tree branch, the second tree branch and the third tree branch to construct a first data management tree for the first power system.

[0046] Furthermore, the data management device takes the first power system as a root node, integrates the first tree branch, the second tree branch and the third tree branch into one tree, and constructs a first data management tree of the first power system.

[0047] Step 40: Match a second power system according to the system information of the first power system.

[0048] It should be noted that the system information in the embodiment of the present invention includes jurisdictional area information and power system information, which can be matched according to the distributed hash table DHT and semantic similarity.

[0049] Therefore, the data management device converts the jurisdiction area information and power system information of the first power system into corresponding hash values ​​through a preset hash function, and matches the candidate power systems according to the hash values.

[0050] Furthermore, the data management device converts the jurisdiction area information and power system information of the first power system into corresponding semantic vectors through semantic coding, and matches the second power system from the candidate power systems according to the semantic vectors, as described in steps 401 to 405.

[0051] Step 50: Integrate the first data management tree with the second data management tree of the second power system to obtain a target data management tree of the first power system.

[0052] Furthermore, the data management device traverses all nodes in the first data management tree and the second data management tree, and for any first node in the first data management tree and any second node in the second data management tree, connects the first node and the second node according to the correlation between the first node and the second node, and the positions of the first node and the second node in the first data management tree and the second data management tree respectively, to obtain the target data management tree of the first power system, as specifically described in steps 501 to 504.

[0053] The embodiment of the present invention stores the power system data of the power system in a tree structure, so its nodes and tree branches can be flexibly adjusted according to the changes in the power system data. When new equipment or parameters are added, it is only necessary to add new child nodes under the corresponding parent nodes, without the need to make large-scale modifications to the entire table structure like a relational database, which greatly improves the adaptability to changes in the power system data structure and the storage flexibility of the power system data. On the other hand, in a tree structure, data query can quickly locate the target node and obtain the data in the target node by traversing the child nodes under it and the data in its child nodes, and the time complexity is O(n), which avoids the need for multiple table connections and screening operations in a relational database, thereby improving the query efficiency of the power system data.

[0054] In one embodiment, the description of steps 201 to 204 is as follows:

[0055] Step 201: input structured data into a first data parsing model, parse the structured data based on the first data parsing model according to the device hierarchy and data logical relationship, and obtain the first layer of structured data output by the first data parsing model.

[0056] It should be noted that the data management device of the embodiment of the present invention is embedded with a pre-trained first data analysis model, wherein the first data analysis model is obtained by training the pre-trained model based on the sample structured data and its corresponding equipment hierarchy and data logical relationship labels, and the pre-trained model can be a forward feedback neural network, a backward feedback neural network, etc. Therefore, the data management device inputs the structured data into the first data analysis model, and the first data analysis model analyzes the structured data according to the equipment hierarchy and data logical relationship, and outputs the first layer of structured data of the structured data, thereby obtaining the first layer of structured data output by the first data analysis model, wherein the first layer of structured data includes power generation equipment, power transmission equipment, power transformation equipment, power consumption equipment and power users.

[0057] Step 202, taking power generation equipment, power transmission equipment, power transformation equipment, power consumption equipment and power users as the first-layer child nodes of the root node.

[0058] Furthermore, the data management device sets the first power system as the root node r1, and the first-layer child nodes L1={E 发电 ,E 输电 E 变电 ,E 用电设备 ,E 用电用户}, that is, add power generation equipment, transmission equipment, substation equipment, power consumption equipment and power users as child nodes under the root node.

[0059] Step 203, traverse the second layer structured data of each node in the first layer subnodes, and use the second layer structured data of each node in the first layer subnodes as the child nodes of each node to construct the second layer child nodes of the first layer child nodes.

[0060] Furthermore, for each node E in the first-layer child nodes i , the data management device traverses the node E i The structured data of the class is obtained according to the specific type and parameters of each node E i The second layer of structured data, and each node E in the first layer of child nodes i The second layer of structured data is used as each node E i 's child nodes, construct the second-layer child nodes of the first-layer child nodes.

[0061] Step 204, split each node according to the information entropy of each node in the second layer of child nodes, construct the child nodes of each node in the second layer of child nodes, and obtain the first tree branch.

[0062] Further, the data management device splits each node according to the information entropy of each node in the second-layer sub-nodes, constructs the sub-nodes of each node in the second-layer sub-nodes, and obtains the first tree branch. The specific analysis is as follows: for each node in the second-layer sub-nodes, if the information entropy of each node is greater than or equal to the preset threshold, the data management device splits each node to obtain the third-layer sub-nodes of the second-layer sub-nodes. Further, the data management device splits according to the information entropy of each node in the third-layer sub-nodes until the information entropy of all nodes is less than the preset threshold, and the first tree branch is constructed, wherein the preset threshold is set according to the actual situation, such as 1.5, 3, 5, etc., and the information entropy settlement formula is:

[0063]

[0064] Among them, H(N) represents the information entropy of node N, n represents the number of child nodes belonging to node N, and p i Indicates the probability that the data belongs to each child node in node N. In one embodiment, the power generation device node E 发电 For example, we start to traverse the structured data belonging to the power generation equipment class in the power generation equipment database table. The power generation equipment database table contains parameter data of different types of generators (such as thermal generators, hydroelectric generators, etc.), such as generator number, rated power, rated voltage, energy type and other fields.

[0065] For E 发电 The second-level sub-nodes under the node are constructed, and preliminary division is performed according to the generator type, such as creating a "thermal generator" sub-node fire N 火发 and the "Hydroelectric Generator" child node water N 水发 Calculate N 火发 Information entropy of a node: In the embodiment of the present invention, N 火发 There are m thermal power generator data records under the node. For each data record, its feature vector can be expressed as v i =(w i ,u i ,t i ), where w i Indicates the rated power value, u i Indicates the rated voltage value, t iRepresents the energy type, which is fixed as "firepower" here. Calculate the probability that the data belongs to different characteristic value ranges. For example, divide the rated power into several intervals (such as the low power interval [0, W1], the medium power interval (W1, W2], and the high power interval (W2, +∞]), and count the proportion of the number of data in each interval to the total number of data m, which are recorded as p1, p2, and p3 respectively (that is, the probability of belonging to the low, medium, and high power intervals). Similarly, the rated voltage is similarly divided and the probabilities q1, q2, and q3 are calculated.

[0066] Furthermore, according to the information entropy formula Calculate N 火发 The information entropy of the node H(N 火发 ), If H(N 火发 ) is greater than or equal to the preset threshold, then further splitting N 火发 node.

[0067] Furthermore, N is further divided according to the rated power range. 火发 Node. Create a "low-power thermal generator" child node N 低火发 、“Medium power thermal generator” child node N 中火发 and the "high power thermal generator" child node N 高火发 . Calculate the information entropy of these three new child nodes again. For example, for N 低火发 Node, re-count the probability distribution of its data in different intervals of rated voltage, and calculate H(N 低火发 ). If H(N 低火发 ) is less than the preset threshold, then stop N 低火发 Node splitting; if H(N 低火发 ) is still greater than or equal to the preset threshold, continue to split according to other features (such as generator manufacturer, etc.) until the information entropy of all child nodes is less than the preset threshold, then stop splitting.

[0068] For other first-layer child nodes (such as E 输电 E 变电 ,E 用电设备 ,E 用电用户 ) and its lower-level child nodes are constructed, information entropy is calculated, and child node splitting operations are performed according to the same process to construct the first tree branch.

[0069] In a specific embodiment, part of the structured data of the first power system is as follows:

[0070]

[0071]

[0072] First, build the root node "First Power System" and add the first-level child node "Generation Equipment" E 发电 .

[0073] For E 发电 Node, build the second-layer child node "thermal generator" N 火发 and "Hydroelectric Generator" N 水发 Calculate N 火发 Information entropy of the node: The rated power is divided into the low power interval [0,200], the medium power interval (200,400], and the high power interval (400,+∞]. There are 2 data in the low power interval. 1 data in the medium power range 0 data is in the high power interval (probability p3 = 0). The rated voltage is divided into the low voltage interval [0,150], the medium voltage interval (150,250], and the high voltage interval (250,+∞]. There is 1 data in the low voltage interval 2 data in the medium voltage range 0 data are in the high voltage interval (probability q3=0).

[0074]

[0075] The preset threshold in the embodiment of the present invention is T s =1.5, due to H(N 火发 )>T s , then N is further divided according to the rated power range 火发 The node is "low power thermal generator" N 低火发 、"Medium power thermal generator"N 中火发 and "High Power Thermal Generator" N 高火发 . Calculate the information entropy of the new child node: N 低火发 There is one data (number 003) under the node, and its rated voltage is in the low voltage range, so the probability p1 = 1, q1 = 1, and the probability of other intervals is 0. Therefore, H(N 低火发 )=-(1*1log2(1*1)+1*0log2(1*0)+0*1log2(0*1)+0*0log2(0*0))=0. 火发 ) <T s , stop N 低火发 Node splitting. Calculate N in the same way 中火发 Node and N 高火发 The information entropy of the node determines whether to continue splitting, and so on to construct the tree branch structure of the entire power generation equipment.

[0076] Further, similar operations are performed on the power transmission equipment, power transformation equipment, power consumption equipment, power users, etc., and finally the construction of the first tree branch is completed, such as Figure 2 shown.

[0077] The embodiment of the present invention stores the structured data of the power system in a tree structure. Therefore, during the storage process, it can be flexibly adjusted according to the changes in the structured data. When new structured data is added, it is only necessary to add a new child node under the corresponding parent node. This greatly improves the adaptability to changes in the power system data structure and the storage flexibility of the power system data.

[0078] In one embodiment, the description of steps 205 to 208 is as follows:

[0079] Step 205 , input the semi-structured data into the second data parsing model, and parse the semi-structured data according to the data semantic relationship based on the second data parsing model to obtain the first layer of semi-structured data output by the second data parsing model.

[0080] It should be noted that the data management device of the embodiment of the present invention is embedded with a pre-trained second data analysis model, wherein the second data analysis model is obtained by training the pre-trained model based on the sample semi-structured data and its corresponding data semantic relationship labels. Therefore, the data management device inputs the semi-structured data into the second data analysis model, and the second data analysis model performs data analysis on the semi-structured data according to the data semantic relationship, and outputs the first layer of semi-structured data, wherein the first layer of semi-structured data includes equipment maintenance records, equipment inspection reports, and power system event logs.

[0081] Step 206, taking the equipment maintenance record, equipment inspection report and power system event log as the first-level child nodes of the root node.

[0082] Furthermore, the data management device takes the equipment maintenance record, equipment inspection report and power system event log as the first-layer sub-node L2 = {S1, S2, S3} of the root node, wherein S1 represents the equipment maintenance record node, S2 represents the equipment inspection report node, and S3 represents the power system event log node, that is, the equipment maintenance record, equipment inspection report, and power system event log are added to the root node.

[0083] Step 207: perform semantic analysis on the text information of each node in the first-layer child nodes, and construct a second-layer child node for each node in the first-layer child nodes.

[0084] Furthermore, the data management device performs semantic analysis on the text information of each node in the first-layer sub-nodes to obtain the text of each node in the first-layer sub-nodes, and constructs the second-layer sub-nodes of each node in the first-layer sub-nodes based on the text of each node in the first-layer sub-nodes.

[0085] In one embodiment, the JSON text information of the equipment maintenance record node is: "record_id": "R001", "equipment_id": "EQ001", "date": "2024-05-10", "personnel": "JohnDoe", "description": "It was found that the transformer oil temperature was too high, the cooling fan was replaced, and the heat sink was cleaned."

[0086] The JSON text information of the equipment maintenance record node is semantically analyzed, and the second-level child nodes of the equipment maintenance record node are constructed according to the obtained text as follows:

[0087] Maintenance date node: The node name is "date" and the node value is "2024-05-10". Maintenance personnel node: The node name is "personnel" and the node value is "JohnDoe". Maintenance equipment node: The node name is "equipment_id" and the node value is "EQ001". Maintenance description node: The node name is "description", the fault type is "Oil temperature is too high", create a "Fault type" child node, and the node value is "Oil temperature is too high"; the maintenance measures are "Replace cooling fan, clean heat sink", create a "Maintenance measures" child node, and the node value is "Replace cooling fan, clean heat sink".

[0088] The XML text information of the equipment inspection report node is:<inspection_report>

[0089] <report_id> I001< / report_id>

[0090] <equipment> EQ002< / equipment>

[0091] <inspector> AliceSmith< / inspector>

[0092] <date> 2024-06-05< / date>

[0093] <result> There is no abnormality in the appearance of the equipment, the insulators are clean, and the connections are firm.< / result>

[0094] < / inspection_report>

[0095] Perform semantic analysis on the XML text information of the equipment inspection report node, and construct the second-level child nodes of the equipment inspection report node based on the obtained text as follows: Inspection date node: the node name is "date", and the node value is "2024-06-05". Inspection personnel node: the node name is "inspector", and the node value is "AliceSmith". Inspection equipment node: the node name is "equipment", and the node value is "EQ002". Inspection result node: the node name is "result", create the "Equipment Appearance Status" child node, and the node value is "No abnormality"; create the "Insulator Status" child node, and the node value is "Clean"; create the "Connection Part Status" child node, and the node value is "Firm".

[0096] The text information of the power system event log node is: "2024-07-15-10:30:00, the 10kV busbar of substation B experienced a voltage drop for 5 seconds. The cause is suspected to be an external line fault."

[0097] The text information of the power system event log node is semantically analyzed, and the second-level sub-nodes of the power system event log node are constructed based on the obtained text as follows: Event date and time node: the node name is "datetime" and the node value is "2024-07-15-10:30:00". Event location node: the node name is "location" and the node value is "substation B". Event type node: the node name is "event_type" and the node value is "voltage sag". Event duration node: the node name is "duration" and the node value is "5 seconds". Event cause node: the node name is "reason" and the node value is "external line failure.

[0098] Step 208, constructing a node relationship between the second-layer child nodes of each node in the first-layer child nodes according to the fusion degree between the second-layer child nodes of each node in the first-layer child nodes, and obtaining a second tree branch.

[0099] Furthermore, the data management device determines the degree of fusion between the second-layer subnodes of each node in the first-layer subnodes, and constructs the node relationship between the second-layer subnodes of each node in the first-layer subnodes based on the degree of fusion between the second-layer subnodes of each node in the first-layer subnodes to obtain a second tree branch, as specifically described in steps 2081 to 2085.

[0100] The embodiment of the present invention stores the semi-structured data of the power system in a tree structure. Therefore, during the storage process, it can be flexibly adjusted according to the changes of the semi-structured data. When new semi-structured data is added, it is only necessary to add a new child node under the corresponding parent node. This greatly improves the adaptability to changes in the power system data structure and the storage flexibility of the power system data.

[0101] In one embodiment, the description of steps 2081 to 2085 is as follows:

[0102] Step 2081, convert each word in the text of each node in the second layer of child nodes into a vector representation through a preset word vector model.

[0103] Optionally, the data management device converts each word in the text of each node in the second-layer child nodes into a vector representation based on a preset word vector model trained using natural language processing technology, such as a Word2Vec or GloVe model.

[0104] In one embodiment, for the equipment maintenance record node, for nodes A and B in its second-layer child nodes, the text corresponding to node A is converted by the word vector model to obtain a vector representation set that can be represented as a={a1, a2, ..., a m}, the text corresponding to node B is converted into a vector representation set after the word vector model is transformed, which can be expressed as b = {b1, b2, ..., b n}.

[0105] Step 2082, based on the vector representation of any two nodes in the second layer of child nodes, determine the semantic association between any two nodes in the second layer of child nodes.

[0106] Further, the data management device determines the semantic association between any two nodes in the second-layer child nodes according to the vector representation of any two nodes in the second-layer child nodes. In one embodiment, continuing the above embodiment, for nodes A and B in the second-layer child nodes of the equipment maintenance record node, the specific calculation formula of the semantic association between nodes A and B is as follows:

[0107]

[0108] Among them, S(A,B) represents the semantic association between node A and node B, a i The i-th vector representation of node A, b j represents the jth vector representation of node B; cos(a i ,b j ) represents the vector representation a i and vector b jThe cosine similarity between them indicates the similarity between the two vector representations in the semantic space; m represents the number of vector representations of node A, and n represents the number of vector representations of node B.

[0109] Step 2083, based on the event probabilities of any two nodes in the second-layer sub-nodes, determine the data dependency between any two nodes in the second-layer sub-nodes.

[0110] Furthermore, the data management device analyzes the dependency between any two nodes in the second-layer sub-nodes according to the event probability of any two nodes in the second-layer sub-nodes, and determines the data dependency between any two nodes in the second-layer sub-nodes. The specific analysis is as follows: Continuing with the above embodiment, in the equipment maintenance record node, the type of maintenance equipment may determine the range of fault types and possible types of maintenance measures. The data dependency between node A and node B is represented by D(A,B).

[0111] Therefore, for discrete data, the data dependency is determined by calculating the conditional probability. In one embodiment, if it is known that the probability of a specific fault type B occurring under a certain device type A is P(B|A), and the probability of fault type B occurring under all device types is P(B), then the data dependency between node A and node B is D(A,B)=P(B|A) / P(B). If D(A,B)>1, it means that node B has a strong dependency on node A; if D(A,B)=1, it means that node B and node A are independent of each other; if D(A,B)<1, it means that node A has a certain inhibitory effect or weak dependency on node B.

[0112] Furthermore, for continuous data (such as the change values ​​of certain equipment operating parameters in the maintenance records), linear regression or other correlation analysis methods can be used to calculate the dependence.

[0113] For example, the equipment running time A and equipment wear parameter B are linearly regressed to obtain the regression equation B = αA + β. According to the regression coefficient α and the goodness of fit R 2 The data dependency is determined by comprehensively considering indicators such as

[0114] Step 2084, based on the semantic association and data dependency between any two nodes in the second layer of sub-nodes, determine the fusion degree between any two nodes in the second layer of sub-nodes.

[0115] Furthermore, the data management device determines the fusion degree F(A, B) between any two nodes in the second-layer sub-nodes according to the semantic association and data dependency between any two nodes in the second-layer sub-nodes, wherein the specific calculation formula of the fusion degree F(A, B) is:

[0116] F(A,B)=ωS(A,B)+(1-ω)D(A,B);

[0117] Among them, ω represents the preset weight coefficient, which can be determined according to the actual data characteristics and analysis focus. For example, if more attention is paid to the impact of semantic information on the tree structure, ω can be set to a larger value; if more attention is paid to the intrinsic dependency between data, (1-ω) can be set to a larger value.

[0118] Step 2085: If the fusion degree between any two nodes in the second layer of child nodes is greater than or equal to a preset fusion degree threshold, a parent-child node relationship between any two nodes in the second layer of child nodes is constructed to obtain a second tree branch.

[0119] Further, the data management device determines whether the fusion degree between any two nodes in the second-layer child nodes is greater than or equal to a preset fusion degree threshold, wherein the preset fusion degree threshold is set according to actual conditions, such as 0.7, 0.8, 0.9, etc. If the fusion degree between any two nodes in the second-layer child nodes is greater than or equal to the preset fusion degree threshold, the data management device constructs a parent-child node relationship between any two nodes in the second-layer child nodes to obtain a second tree branch.

[0120] In a specific embodiment, taking the equipment maintenance record node S1 of the first-layer child node in the above embodiment as an example, the parent-child node relationship between any two nodes in the second-layer child node of the equipment maintenance record node S1 is constructed as follows: Calculate the semantic association between the "maintenance equipment type" node and the "fault type" node. For the "Transformer" transformer equipment type word vector and the "oil temperature was too high" oil temperature fault type word vector set, it is calculated by the semantic association formula. After the word vector model conversion, the "Transformer" vector is t, the "oil" vector is o, the "temperature" vector is p, the "too" vector is q, and the "high" vector is h. Then sum and average the cosine similarities of cos(t,o), cos(t,p), cos(t,q), and cos(t,h) to obtain the semantic association S(maintenance equipment type, fault type) = 0.57.

[0121] In the historical maintenance data, it is found that the probability of oil temperature over-high fault in transformer type equipment P(oil temperature over-high, Transformer) = 0.3, and the probability of oil temperature over-high fault in all equipment types P(oil temperature over-high) = 0.1, so the data dependence D(maintenance equipment type, fault type) = 0.3 / 0.1 = 3.

[0122] In the embodiment of the present invention, the preset fusion threshold is 2, and the preset weight coefficient is 0.4. Therefore, F(maintenance equipment type, fault type) = 0.57*0.4+0.6*3 = 2.028. Since 2.028>2, the "maintenance equipment type" node is used as the parent node, and the "fault type" node is used as the child node of the "maintenance equipment type" node. Similarly, the fusion degree between other nodes is calculated, such as the fusion degree between "fault type" and "maintenance measures", and the tree structure relationship between the complete second-layer child nodes is constructed based on the results, and then the complete second tree branch is constructed. For example, if it is found that the fusion degree between a certain fault type and a specific maintenance measure is high, the maintenance measure node is used as a child node of the fault type node, so that the relevant maintenance measure information can be quickly obtained when querying a specific fault, as shown in the following example. Figure 3 shown.

[0123] The embodiment of the present invention stores the semi-structured data of the power system in a tree structure. Therefore, during the storage process, it can be flexibly adjusted according to the changes of the semi-structured data. When new semi-structured data is added, it is only necessary to add a new child node under the corresponding parent node. This greatly improves the adaptability to changes in the power system data structure and the storage flexibility of the power system data.

[0124] In one embodiment, the description of steps 209 to 213 is as follows:

[0125] Step 209: input the unstructured data into a feature extraction model, extract features from the unstructured data based on the feature extraction model, and output a feature vector of the unstructured data.

[0126] It should be noted that the data management device of the embodiment of the present invention is embedded with a pre-trained feature extraction model, wherein the feature extraction model is obtained by training the pre-trained model based on the sample unstructured data and its corresponding feature vector label results. Therefore, the data management device inputs the unstructured data into the feature extraction model, and the feature extraction model extracts features from the unstructured data to obtain the feature vector of the unstructured data, thereby obtaining the feature vector of the unstructured data output by the feature extraction model.

[0127] In one embodiment, the unstructured data includes N infrared images. For each infrared image, the feature extraction model uses the temperature value of each pixel in the image as basic data, constructs the temperature matrix of the image, and calculates the statistical features of the image, such as the average temperature. Temperature variance Number of hotspot pixels n h (the number of pixels whose temperature exceeds a certain threshold), etc., combining statistical features into a feature vector i=1,2,...,N represents the i-th infrared image.

[0128] Step 210: Determine the local density of each feature vector in the unstructured data and the minimum distance between each feature vector in the unstructured data and the target feature vector based on the local density.

[0129] Furthermore, for each eigenvector F i The data management device determines each feature vector F in the unstructured data according to the local density algorithm. i The local density f i , each eigenvector F i The local density f i Indicates that in each eigenvector F i The radius is d c The number of eigenvectors in the neighborhood of is calculated as follows:

[0130]

[0131] Where n represents the total number of eigenvectors; χ(x) represents the indicator function, when x ≥ 0, χ(x) = 1, otherwise, χ(x) = 0; d ij Denotes the eigenvector F i With other eigenvectors F j The distance between.

[0132] In one embodiment, when calculating the local density of the feature vector F1, all other feature vectors F j ,j=2,3,...,N, calculate the eigenvector F1 and other eigenvectors F j The Euclidean distance d between 1j , if d 1j ≤d c , then f1 is increased by 1. The local density f of all image feature vectors is calculated in sequence i ,i=1,2,..,N。

[0133] Furthermore, the data management device traverses the local density ratio feature vector F i All target feature vectors with high local density F j , get the feature vector set S i ={F j |f j >f i}, calculate the eigenvector F i To the feature vector set S i Each target feature vector F j The distance d ij , take the minimum value as the eigenvector F i and the target feature vector Fj The minimum distance It should be noted that for the eigenvector with the highest local density, its minimum distance Set to the maximum distance to all other eigenvectors.

[0134] Step 211 : determining a vector clustering result of each feature vector in the unstructured data based on the local density and the minimum distance of each feature vector in the unstructured data.

[0135] Furthermore, the data management device calculates the local density f of each feature vector according to the local density f of each feature vector. i and minimum distance The decision diagram is drawn with the value of local density f i is the horizontal axis, the minimum distance is the vertical axis).

[0136] In the decision graph, the local density f i and minimum distance The eigenvectors with larger eigenvectors are usually candidate eigenvectors of cluster centers. By observing the distribution of eigenvectors in the decision diagram, combined with the actual operation knowledge and data characteristics of the power system, the initial value of the number of clusters k is preliminarily determined. For example, k=3 is determined, and the three cluster center eigenvectors C1, C2, and C3 corresponding to each eigenvector are determined.

[0137] Furthermore, the data management device starts from the cluster center feature vector of each feature vector and divides the density-connected feature vectors in its neighborhood into corresponding clusters. For example, for the cluster center feature vector C1, the semi-structured data corresponding to the feature vectors in its neighborhood and density-connected (i.e., there is a path) are first divided into cluster C1. This process is repeated until all feature vectors of the semi-structured data are divided into a cluster, and three cluster sets C are obtained. cluster1 ,C cluster2 ,C cluster3 .

[0138] Step 212: The vector clustering result of each feature vector in the unstructured data is used as the first-layer child node of the root node.

[0139] Furthermore, the data management device uses the vector clustering result of each feature vector in the unstructured data as the first-level child node of the root node. In one embodiment, for the feature vector of the infrared image in the semi-structured data, the three clustering sets C of the feature vector of the infrared image are cluster1 ,C cluster2 ,C cluster3 They are added as first-layer child nodes under the root node and named “Cluster 1 infrared image data”, “Cluster 2 infrared image data” and “Cluster 3 infrared image data” respectively.

[0140] Step 213, extract the key feature information of each node in the first layer of child nodes, and use the key feature information of each node in the first layer of child nodes as the child nodes of each node in the first layer of child nodes, construct the second layer of child nodes of each node in the first layer of child nodes, and obtain the third tree branch.

[0141] Furthermore, the data management device extracts key feature information of each node in the first layer of sub-nodes, and uses the key feature information of each node in the first layer of sub-nodes as the sub-node of each node in the first layer of sub-nodes, constructs the second layer of sub-nodes of each node in the first layer of sub-nodes, and obtains the third tree branch. Continuing with the above embodiment, for the infrared image data in the "Cluster 1 Infrared Image Data" sub-node, extract key feature information to construct the second layer of sub-nodes, such as calculating the hotspot position range of all images in the cluster, traversing each infrared image in the cluster, finding the coordinate range of all hotspot pixels, and obtaining the x coordinate range of the hotspot through statistical calculation as [x min ,x max ], the y coordinate range is [y min ,y max ], create a "hotspot location range" child node, the attribute value is "x coordinate range: [x min ,x max ], y coordinate range: [y min ,y max ]". Calculate the temperature range of the images in the cluster, count the highest and lowest temperatures of all images in the cluster, and get the temperature range as [T min ,T max ], create a "Temperature Range" child node with the attribute value "[T min ,T max ]". Other key feature information can also be calculated, such as the average hotspot area (the average of the hotspot areas of all images in the cluster), creating an "average hotspot area" child node and setting the corresponding attribute values.

[0142] In the same way, the corresponding second-layer child nodes are constructed for the "Cluster 2 infrared image data" and "Cluster 3 infrared image data" child nodes, thereby completing the construction of the third tree branch. Figure 4 shown.

[0143] The embodiment of the present invention stores the unstructured data of the power system in a tree structure. Therefore, during the storage process, it can be flexibly adjusted according to the changes of the unstructured data. When new unstructured data is added, it is only necessary to add a new child node under the corresponding parent node. This greatly improves the adaptability to changes in the power system data structure and the storage flexibility of the power system data.

[0144] In one embodiment, the description of steps 401 to 405 is as follows:

[0145] Step 401: Integrate the jurisdiction area information and power system information of the first power system to obtain an information string of the first power system.

[0146] Specifically, for the first power system Z1, the data management device sends its jurisdiction area information I 1r (such as geographical location information expressed in latitude and longitude range, voltage level coverage, etc.) and power system information I 1s (such as power generation capacity, total length of transmission lines, number of substations, etc.) 1r Expressed as [lon 1min ,lon 1max ,lat 1min ,lat 1max ,V 1min ,V 1max ](longitude range, latitude range, voltage level range), power system information I 1s Expressed as [P 1s ,L 1s ,N 1s ](generation capacity, transmission line length, number of substations), connect each information element in a specific order to form an information string Str1=concat(I 1r ,I 1s ).

[0147] Step 402: Perform a hash calculation on the information string based on a hash function to obtain a first hash value.

[0148] Furthermore, the data management device performs hash calculation on the information string Str1 through the SHA-256 hash function H to obtain a hash value h1=H(Str1) of the first power system Z1.

[0149] Step 403: Obtain a candidate power system based on matching the first hash value in a preset hash table.

[0150] In the preset hash table, the hash value h2=H(Str2) of each power system Z2 is also in accordance with the process of the first power system, and the jurisdiction area information I 2r =[lon 2min ,lon 2max ,lat 2min ,lat 2max ,V 2min ,V 2max ] and power system information 2s =[P 2s ,L 2s,N 2s ] is combined into a message string Str2 = concat(I 2r ,I 2s ), and then use the SHA-256 hash function H to calculate the hash value h2=H(Str2) of the information string Str2.

[0151] Furthermore, the data management device searches the hash value h1 and the hash value h2 and the corresponding power system identifiers (such as Z1 and Z2) in a preset hash table to calculate the difference value Diff between the hash value h1 and the hash value h2. h = hammingdistance(h1,h2) (Hamming distance and other difference measurement methods). If the difference value Diff h Less than or equal to the hash difference threshold T h , then the power system Z2 is used as a candidate power system for the first power system Z1 and enters the next step of semantic similarity matching analysis; otherwise, it is determined that the power system Z2 has a low matching degree with the first power system Z1 and is not considered for integration. Therefore, it can be understood that the difference between the second hash value h2 and the first hash value h1 of the candidate power system is less than or equal to the hash difference threshold.

[0152] Step 404 , semantically encode the jurisdictional area information and the power system information of the first power system to obtain a first comprehensive semantic vector of the first power system.

[0153] Furthermore, for the jurisdiction area information I of the first power system Z1 1r The data management device constructs a semantic vector using the semantic coding method of the geographic information system (GIS). For example, the longitude and latitude range is mapped to a two-dimensional semantic space, the longitude range is mapped to the x-axis coordinate interval, and the latitude range is mapped to the y-axis coordinate interval to obtain the vector V 1r =[x 1rmin ,x 1rmax ,y 1rmin ,y 1rmax ]. For the voltage level range [V 1min ,V 1max ], it can be converted into a discrete semantic coding vector V according to the voltage level semantic classification table in the power field 1v (For example, the codes corresponding to low voltage, medium voltage, high voltage, etc.) 1r and the semantic encoding vector V 1v Combined into the jurisdiction area information I of the first power system Z1 1r The semantic vector SV 1r =combine(V 1r ,V 1v ).

[0154] Furthermore, for the power system information I of the first power system Z1 1s , construct semantic vectors based on the power domain ontology. For example, the power generation capacity P 1s It can be converted into semantic coding according to the power system scale classification (small, medium, large, etc.); the transmission line length L 1s and the number of substations N 1s It is also converted into code according to the corresponding semantic rules of the power field and combined into the power system information I of the first power system Z1 1s The semantic vector SV 1s Finally, the comprehensive semantic vector SV1 of the first power system Z1 is obtained = concat(SV 1r ,SV 1s ).

[0155] It should be noted that the comprehensive semantic vector SV2 of each power system in the candidate power systems is also obtained according to the above process.

[0156] Step 405: Matching the candidate power systems based on the first comprehensive semantic vector to obtain a second power system in the candidate power systems.

[0157] Furthermore, the data management device calculates the semantic similarity score S of the comprehensive semantic vector SV1 and the comprehensive semantic vector SV2 by using the cosine similarity formula. s , the specific calculation formula is as follows:

[0158]

[0159] Among them, a i and b i They respectively represent the corresponding dimensional elements of the comprehensive semantic vector SV1 and the comprehensive semantic vector SV2, and n represents the vector dimension.

[0160] Furthermore, if the semantic similarity score S is determined s Greater than or equal to a preset semantic similarity threshold, that is, it is determined that the power system in the candidate power system has a higher semantic compatibility with the first power system Z1, thereby obtaining the second power system of the first power system Z1. Therefore, it can be understood that the semantic similarity score between the first comprehensive semantic vector of the first power system and the second comprehensive semantic vector of the second power system is greater than or equal to the preset semantic similarity threshold.

[0161] The embodiment of the present invention matches the second power system with a high correlation with the first power system through the jurisdiction area information and the power system information. After the subsequent integration of the first power system and the second power system, better power dispatch coordination can be achieved. At the same time, the power system in the region can be evaluated from a more macro perspective, which is helpful to analyze the power supply and demand balance in the entire region and provide comprehensive data support for future plans such as power generation facility construction and power grid upgrades.

[0162] In one embodiment, the description of steps 501 to 504 is as follows:

[0163] Step 501: traverse all nodes in the first data management tree and the second data management tree, and for any first node in the first data management tree and any second node in the second data management tree, determine the association degree between the first node and the second node.

[0164] Optionally, the data management device traverses all nodes in the first data management tree and the second data management tree, and for a node pair in the first data management tree Tree1 of the first power system and the second data management tree Tree2 of the second power system, that is, for any first node in the first data management tree and any second node in the second data management tree, calculates the association between the first node and the second node, and considers the semantic compatibility and functional complementarity across systems through the association. For example, for the power generation equipment node E in the first data management tree Tree1 1g and the power transmission equipment node E in the second data management tree Tree2 2t In terms of semantic compatibility, check whether the voltage output range of the power generation equipment matches the voltage receiving range of the transmission equipment (based on the semantic knowledge of the power field). In terms of functional complementarity, consider whether the power generation capacity of the power generation equipment can be adapted to the transmission capacity of the transmission equipment (according to the operation logic of the power system).

[0165] Therefore, the data management device determines the cross-system data type matching score T between the first node X and the second node Y according to the power field equipment type matching rule. XY (XY), determine the cross-system semantic association score S between the first node X and the second node Y according to the semantic compatibility check XY (XY), determine the cross-system functional correlation score F between the first node X and the second node Y based on functional complementarity analysis XY (XY), determine the correlation R(X,Y) between the first node X and the second node Y according to the correlation calculation formula, and obtain the correlation matrix M R , where the correlation matrix M R The matrix element m in ij =R(N i ,N j), N i and N j Represents nodes in different data management trees, and the correlation calculation formula is:

[0166] R(X,Y)=αT XY (XY)+βS XY (XY)+γF XY (XY);

[0167] Among them, α, β, and γ represent weight coefficients, which are determined based on expert knowledge.

[0168] Step 502: determine a first path from a root node of the first data management tree to a first node based on the first data management tree, and determine a second path from a root node of the second data management tree to a second node based on the second data management tree.

[0169] Further, the data management device determines a path Path from the root node of the first data management tree to the first node X according to the first data management tree. X =[r1, n1, n2, ..., X], and determining a path Path of a second path Y from the root node of the second data management tree to the second node according to the second data management tree Y =[r2,m1,m2,...,Y], r1 represents the root node of the first data management tree Tree1, n1,n2,... represent the intermediate nodes of the first data management tree Tree1, r2 represents the root node of the second data management tree Tree2, m1,m2,... represent the intermediate nodes of the second data management tree Tree2.

[0170] Step 503: determine a path weight between the first node and the second node based on the first path and the second path.

[0171] Further, the data management device according to the first path Path X and the second path Path Y , calculate the path weight W(X,Y) between the first node and the second node, and get the path weight matrix M W , where the path weight matrix M W The matrix element m in ij =W(N i ,N j ), where the path weight W(X,Y) represents the combined influence of node importance and path length on the path from each root node to two associated nodes. The calculation formula of the path weight W(X,Y) is as follows:

[0172]

[0173] Among them, l X and lY Respectively represent the path Path X and Path Y The length of (number of nodes); and They represent the importance weights of the nodes on the path, which are determined according to the hierarchy and functional importance of the nodes in the data management tree of the respective power systems. For example, the root nodes and key equipment nodes have higher weights; λ represents the penalty coefficient for path length difference.

[0174] Step 504 : Connect the first node and the second node based on the correlation degree and the path weight to obtain a target data management tree for the first power system.

[0175] Furthermore, the data management device connects the first node and the second node whose correlation degree in the correlation matrix is ​​greater than or equal to the preset correlation degree threshold and whose path weight in the path weight matrix is ​​less than or equal to the preset path weight threshold, wherein the preset correlation degree threshold and the preset path weight threshold are set according to actual conditions. Repeat the above process until all node pairs that meet the conditions are processed, and obtain the target data management tree of the first power system, the target data management tree is as follows: Figure 5 shown.

[0176] The embodiment of the present invention integrates the data management trees of the first power system and the second power system. Therefore, the integrated target data management tree can provide a more comprehensive perspective. Taking into account the mutual influence between multiple systems, the possibility and scope of cross-system fault propagation can be better predicted by integrating data, and response measures can be formulated in advance to ensure the safe and stable operation of the regional power system.

[0177] Further, the data management device for the power system provided by the present invention is described below, and the data management device for the power system described below and the data management method for the power system described above can be referred to each other. Figure 6 , Figure 6 It is a structural schematic diagram of a data management device for an electric power system provided by the present invention. The data management device for an electric power system includes:

[0178] A data classification module 610 is used to classify the collected power data of the first power system to obtain structured data, semi-structured data and unstructured data in the power data;

[0179] A tree branch construction module 620 is used to construct a first tree branch, a second tree branch and a third tree branch respectively with the first power system as a root node and structured data, semi-structured data and unstructured data as child nodes;

[0180] A tree branch integration module 630, used to integrate the first tree branch, the second tree branch and the third tree branch to construct a first data management tree for the first power system;

[0181] A matching module 640, configured to match a second power system according to system information of the first power system;

[0182] The management tree integration module 650 is used to integrate the first data management tree with the second data management tree of the second power system to obtain a target data management tree of the first power system.

[0183] The embodiment of the present invention stores the power system data of the power system in a tree structure, so its nodes and tree branches can be flexibly adjusted according to the changes in the power system data. When new equipment or parameters are added, it is only necessary to add new child nodes under the corresponding parent nodes, without the need to make large-scale modifications to the entire table structure like a relational database, which greatly improves the adaptability to changes in the power system data structure and the storage flexibility of the power system data. On the other hand, in a tree structure, data query can quickly locate the target node and obtain the data in the target node by traversing the child nodes under it and the data in its child nodes, and the time complexity is O(n), which avoids the need for multiple table connections and screening operations in a relational database, thereby improving the query efficiency of the power system data.

[0184] See also Figure 7 , Figure 7 FIG. 1 is an embodiment diagram of an electronic device provided by an embodiment of the present invention. Figure 7 As shown, an embodiment of the present invention provides an electronic device 700, including a memory 710, a processor 720, and a computer program 711 stored in the memory 710 and executable on the processor 720. When the processor 720 executes the computer program 711, the following steps are implemented:

[0185] Classifying the collected power data of the first power system to obtain structured data, semi-structured data, and unstructured data in the power data;

[0186] Taking the first power system as a root node, and taking structured data, semi-structured data, and unstructured data as child nodes, respectively construct a first tree branch, a second tree branch, and a third tree branch;

[0187] Integrate the first tree branch, the second tree branch and the third tree branch to construct a first data management tree for the first power system;

[0188] Matching a second power system according to system information of the first power system;

[0189] The first data management tree is integrated with the second data management tree of the second power system to obtain a target data management tree of the first power system.

[0190] See also Figure 8 , Figure 8 Detailed description of an embodiment of a computer-readable storage medium provided by an embodiment of the present invention. Figure 8 As shown, this embodiment provides a computer-readable storage medium 800, on which a computer program 711 is stored. When the computer program 711 is executed by a processor, the following steps are implemented:

[0191] Classifying the collected power data of the first power system to obtain structured data, semi-structured data, and unstructured data in the power data;

[0192] Taking the first power system as a root node, and taking structured data, semi-structured data, and unstructured data as child nodes, respectively construct a first tree branch, a second tree branch, and a third tree branch;

[0193] Integrate the first tree branch, the second tree branch and the third tree branch to construct a first data management tree for the first power system;

[0194] Matching a second power system according to system information of the first power system;

[0195] The first data management tree is integrated with the second data management tree of the second power system to obtain a target data management tree of the first power system.

[0196] On the other hand, the present invention further provides a computer program product, the computer program product includes a computer program, the computer program can be stored in a non-transitory computer-readable storage medium, when the computer program is executed by a processor, the computer can execute the data management method of the power system provided by the above methods, the method includes:

[0197] Classifying the collected power data of the first power system to obtain structured data, semi-structured data, and unstructured data in the power data;

[0198] Taking the first power system as a root node, and taking structured data, semi-structured data, and unstructured data as child nodes, respectively construct a first tree branch, a second tree branch, and a third tree branch;

[0199] Integrate the first tree branch, the second tree branch and the third tree branch to construct a first data management tree for the first power system;

[0200] Matching a second power system according to system information of the first power system;

[0201] The first data management tree is integrated with the second data management tree of the second power system to obtain a target data management tree of the first power system.

[0202] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0203] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0204] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A data management method for a power system, characterized in that: include: Classifying the collected power data of the first power system to obtain structured data, semi-structured data, and unstructured data in the power data; Taking the first power system as a root node, taking the structured data, the semi-structured data and the unstructured data as child nodes, respectively constructing a first tree branch, a second tree branch and a third tree branch; Integrate the first tree branch, the second tree branch and the third tree branch to construct a first data management tree for the first power system; Matching a second power system according to the system information of the first power system; The first data management tree is integrated with the second data management tree of the second power system to obtain a target data management tree of the first power system.

2. The data management method for the power system according to claim 1, characterized in that: The steps of constructing a first tree branch with the first power system as the root node and the structured data as the child node include: The structured data is input into a first data parsing model, and the structured data is parsed based on the first data parsing model according to the equipment hierarchy and data logical relationship to obtain the first layer of structured data output by the first data parsing model; the first layer of structured data includes power generation equipment, power transmission equipment, power transformation equipment, power consumption equipment and power users; the first data parsing model is obtained by label training based on sample structured data and its corresponding equipment hierarchy and data logical relationship; The power generation equipment, the power transmission equipment, the power transformation equipment, the power consumption equipment and the power user are the first-layer child nodes of the root node; Traversing the second-layer structured data of each node in the first-layer child nodes, and taking the second-layer structured data of each node in the first-layer child nodes as the child nodes of each node, to construct the second-layer child nodes of the first-layer child nodes; Each node is split according to the information entropy of each node in the second-layer child nodes, and the child nodes of each node in the second-layer child nodes are constructed to obtain the first tree branch.

3. The data management method for the power system according to claim 2, characterized in that: The step of splitting each node according to the information entropy of each node in the second-layer child nodes, constructing a child node of each node in the second-layer child nodes, and obtaining the first tree branch includes: For each node in the second-layer sub-nodes, if the information entropy of each node is greater than or equal to the preset threshold, each node is split to obtain the third-layer sub-node of the second-layer sub-node; split according to the information entropy of each node in the third-layer sub-node until the information entropy of all nodes is less than the preset threshold, and the first tree branch is constructed; the calculation formula of the information entropy is: Among them, H(N) represents the information entropy of node N, n represents the number of child nodes belonging to node N, and p i Represents the probability that the data belongs to each child node in node N.

4. The data management method for the power system according to claim 1, characterized in that: Taking the first power system as the root node and the semi-structured data as the child node, the steps of constructing the second tree branch include: The semi-structured data is input into a second data analysis model, and the semi-structured data is analyzed according to the data semantic relationship based on the second data analysis model to obtain the first layer of semi-structured data output by the second data analysis model; the first layer of semi-structured data includes equipment maintenance records, equipment inspection reports and power system event logs; the second data analysis model is trained based on sample semi-structured data and its corresponding data semantic relationship labels; The equipment maintenance record, the equipment inspection report and the power system event log are the first-layer child nodes of the root node; Performing semantic analysis on text information of each node in the first-layer sub-nodes to construct a second-layer sub-node for each node in the first-layer sub-nodes; According to the fusion degree between the second-layer child nodes of each node in the first-layer child nodes, a node relationship between the second-layer child nodes of each node in the first-layer child nodes is constructed to obtain a second tree branch.

5. The data management method for the power system according to claim 4, characterized in that: The step of constructing a node relationship between the second-layer child nodes of each node in the first-layer child nodes according to the fusion degree between the second-layer child nodes of each node in the first-layer child nodes to obtain a second tree branch includes: Convert each word in the text of each node in the second-layer sub-node into a vector representation through a preset word vector model; Determine the semantic association between any two nodes in the second layer of child nodes based on the vector representations of any two nodes in the second layer of child nodes; Determine the data dependency between any two nodes in the second layer of sub-nodes based on the event probabilities of any two nodes in the second layer of sub-nodes; Determining the degree of fusion between any two nodes in the second layer of sub-nodes based on the semantic association and data dependency between any two nodes in the second layer of sub-nodes; If the fusion degree between any two nodes in the second layer of child nodes is greater than or equal to a preset fusion degree threshold, a parent-child node relationship between any two nodes in the second layer of child nodes is constructed to obtain the second tree branch.

6. The data management method for the power system according to claim 1, characterized in that: Taking the first power system as the root node and the unstructured data as the child node, the steps of constructing the third tree branch include: The unstructured data is input into a feature extraction model, and features of the unstructured data are extracted based on the feature extraction model, and a feature vector of the unstructured data is output; the feature extraction model is trained based on sample unstructured data and its corresponding feature vector label results; Determining the local density of each feature vector in the unstructured data and the minimum distance between each feature vector in the unstructured data and a target feature vector based on the local density; Determining a vector clustering result of each feature vector in the unstructured data based on the local density and the minimum distance of each feature vector in the unstructured data; Taking the vector clustering result of each feature vector in the unstructured data as the first-layer child node of the root node; Extract key feature information of each node in the first layer of child nodes, and use the key feature information of each node in the first layer of child nodes as the child nodes of each node in the first layer of child nodes, construct the second layer of child nodes of each node in the first layer of child nodes, and obtain the third tree branch.

7. The data management method for a power system according to any one of claims 1 to 6, characterized in that: The system information includes jurisdiction information and power system information; and matching the second power system according to the system information of the first power system includes: Integrate the jurisdiction area information and power system information of the first power system to obtain an information string of the first power system; Performing a hash calculation on the information string based on a preset hash function to obtain a first hash value; Based on the matching of the first hash value in the preset hash table, a candidate power system is obtained; the difference between the second hash value of the candidate power system and the first hash value is less than or equal to a hash difference threshold; Semantically encoding the jurisdiction area information and the power system information of the first power system to obtain a first comprehensive semantic vector of the first power system; Based on the first comprehensive semantic vector, matching is performed on the candidate power system to obtain a second power system among the candidate power systems; a semantic similarity score between the first comprehensive semantic vector and the second comprehensive semantic vector of the second power system is greater than or equal to a preset semantic similarity threshold.

8. The data management method for the power system according to claim 7, characterized in that: The step of integrating the first data management tree with the second data management tree of the second power system to obtain a target data management tree of the first power system includes: Traversing all nodes in the first data management tree and the second data management tree, and determining, for any first node in the first data management tree and any second node in the second data management tree, a degree of association between the first node and the second node; Determine a first path from a root node of the first data management tree to the first node based on the first data management tree, and determine a second path from a root node of the second data management tree to the second node based on the second data management tree; Determining a path weight between the first node and the second node based on the first path and the second path; The first node and the second node are connected based on the association degree and the path weight to obtain a target data management tree for the first power system.

9. A data management device for a power system, characterized in that: A data management method for a power system as claimed in any one of claims 1 to 8; The data management device of the power system comprises: A data classification module, used to classify the collected power data of the first power system to obtain structured data, semi-structured data and unstructured data in the power data; A tree branch construction module, used to take the first power system as a root node, the structured data, the semi-structured data and the unstructured data as child nodes, and respectively construct a first tree branch, a second tree branch and a third tree branch; A tree branch integration module, used for integrating the first tree branch, the second tree branch and the third tree branch to construct a first data management tree for the first power system; A matching module, configured to match a second power system according to the system information of the first power system; A management tree integration module is used to integrate the first data management tree with the second data management tree of the second power system to obtain a target data management tree of the first power system.

10. A non-transitory computer-readable storage medium having a computer software program stored therein, characterized in that: When the computer software program is executed by a processor, the data management method for the power system as claimed in any one of claims 1 to 8 is implemented.

Images