Electric vehicle charging load characteristics modeling method based on knowledge graph

Through the knowledge graph-based method, a knowledge graph library and typical feature model of tram charging load is constructed, and the impact of electric vehicle charging load on the power grid is solved, the stability and reliability of the power grid is improved, and scientific support is provided for the operation of the power grid.

CN118674175BActive Publication Date: 2024-12-06HEFEI POWER SUPPLY COMPANY OF STATE GRID ANHUI ELECTRIC POWER +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411155601.X
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-22
Publication Date
2024-12-06
Estimated Expiration
2044-08-22

AI Technical Summary

Technical Problem

The disorder, transient and imbalance of the charging load of electric vehicles has an impact on the power grid, and the existing technology has failed to effectively analyze and solve these problems.

Method used

Using a knowledge graph-based method, by obtaining historical data of tram charging load, building a tram charging load knowledge graph library, knowledge extraction and data clustering, and automatically generating a typical tram charging feature model for modeling and predicting charging load characteristics.

Benefits of technology

It realizes the refined mining of electric vehicle charging load and multi-dimensional independent and controllable typical load characteristics modeling, improves the stability and reliability of the power grid, and provides scientific reference and decision-making support for grid operation and management.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118674175B_ABST
    Figure CN118674175B_ABST
Patent Text Reader

Abstract

The present invention discloses a method for modeling electric vehicle charging load characteristics based on a knowledge graph, and relates to the technical field of modeling electric vehicle charging load characteristics, comprising the following steps: obtaining historical data of electric vehicle charging load, constructing a knowledge graph library of electric vehicle charging load, and performing knowledge extraction to construct an uncertain electric vehicle charging load knowledge extraction model, performing data clustering based on the uncertain electric vehicle charging load knowledge extraction model, and constructing an automatic generation model for typical characteristics of electric vehicle charging. Compared with the existing technology, the method for modeling electric vehicle charging load characteristics based on a knowledge graph proposed in this application realizes refined mining of electric vehicle charging load characteristics, and establishes a typical load characteristic model of electric vehicle charging load in a multi-dimensional and autonomous manner, laying a safety foundation for the safety and stability of the power grid, and providing a technical reference for power grid operation managers.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of electric vehicle charging load characteristic modeling, and more specifically, to an electric vehicle charging load characteristic modeling method based on a knowledge graph. Background Art

[0002] The limited rationality of electric vehicles makes it easy for them to make irrational decisions. Their behaviors, including charging decisions and route selection, are difficult to characterize. The charging load of electric vehicles will have a significant impact on the power grid.

[0003] For example, the invention patent with publication number: CN114581180A discloses a charging station recommendation method, a charging pile status determination method and device, which determine candidate charging stations for target vehicles; for each candidate charging station, determine the estimated arrival time of the target vehicle, and obtain the estimated status of the charging pile. The estimated status of the charging pile is determined based on the current status of the charging pile and the scheduled charging information. The scheduled charging information includes: the estimated charging time and the estimated charging duration of the vehicle to be charged that has made an appointment to go to the candidate charging station; according to the estimated status of the charging pile, the target charging station is determined. Compared with the method of making recommendations based only on the current status of the charging pile of the candidate charging station, it is more reasonable, and can reduce unnecessary waiting in line for electric vehicle owners as much as possible, reduce the time cost of vehicle charging, and optimize the travel experience of electric vehicle owners.

[0004] For example, the invention patent publication No. CN117993581A discloses a method for predicting charging load of electric vehicles taking weather influence into consideration.

[0005] The above disclosed technical solutions have at least the following technical problems: the random uncertainty characteristics of the charging load of electric vehicles have not yet been clearly analyzed, and the charging load of electric vehicles is disorderly, instantaneous and unbalanced, which will cause impact on the power grid.

[0006] In view of the above problems, the present invention proposes a solution. Summary of the invention

[0007] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides an electric vehicle charging load characteristic modeling method based on a knowledge graph. By modeling the electric vehicle charging load characteristics, the problem that the electric vehicle charging load is disorderly, instantaneous and unbalanced, which may cause an impact on the power grid, is solved.

[0008] To achieve the above object, the present invention provides the following technical solutions:

[0009] The method for modeling the characteristics of electric vehicle charging load based on knowledge graph includes the following steps: obtaining historical data of electric vehicle charging load, constructing a knowledge graph library of electric vehicle charging load, and performing knowledge extraction to construct an uncertain electric vehicle charging load knowledge extraction model; performing data clustering based on the uncertain electric vehicle charging load knowledge extraction model, and constructing an automatic generation model of typical characteristics of electric vehicle charging.

[0010] In a preferred embodiment, the method for acquiring the electric vehicle charging load knowledge graph library is as follows: acquiring the electric vehicle charging load historical data of the power grid ledger, and constructing the electric vehicle charging load knowledge graph library based on the RDFS triple and the Neo4j graph database; the knowledge extraction to construct the uncertain electric vehicle charging load knowledge extraction model is specifically as follows: defining multiple scenarios, extracting knowledge from the electric vehicle charging load knowledge graph library based on multiple scenarios, and obtaining the multi-scenario charging pile load curve; constructing the uncertain electric vehicle charging load knowledge extraction model based on the multi-scenario charging pile load curve; constructing the electric vehicle charging load typical feature automatic generation model is specifically as follows: automatically obtaining the multi-scenario charging pile load curve based on the uncertain electric vehicle charging load knowledge extraction model, normalizing the multi-scenario charging pile load curves respectively, and obtaining the corresponding charging pile load normalization curve; and fitting the multi-scenario charging pile load normalization curve based on the kmeans clustering algorithm to obtain the electric vehicle charging load typical feature curve; constructing the electric vehicle charging typical feature automatic generation model based on the electric vehicle charging load typical feature curve.

[0011] In a preferred embodiment, the electric vehicle charging load history data includes charging time, charging pile type, charging address and power grid data.

[0012] In a preferred embodiment, the electric vehicle charging load knowledge graph library is constructed based on RDFS triples and Neo4j graph database, specifically as follows: entities and relationships are defined, specifically as follows: the electric vehicle charging load knowledge graph library takes electric vehicle charging piles as entities, and includes three classes, the three classes include electric piles, roads, and power grids, each class includes attribute descriptions, and relationships are defined based on the characteristics of classes and attributes; based on entities and relationships, the class is used as the subject, the relationship is used as the predicate, and the attribute is used as the object to construct several types of RDFS triples: <class, relationship, attribute>; several RDFS triples are converted into nodes and relationships of the Neo4j graph database for storage; each node corresponds to a electric vehicle charging pile entity, and each relationship corresponds to the predicate of the triple; RDFS triples are automatically generated and stored as the electric vehicle charging load knowledge graph library through a conversion script according to the electric vehicle charging load historical data.

[0013] In a preferred embodiment, the multiple scenarios are defined, and knowledge is extracted from the electric vehicle charging load knowledge graph library based on the multiple scenarios to obtain the multi-scenario charging pile load curve, specifically: one type of several types of RDFS triples is selected as the load characteristic analysis requirement, and multiple scenarios are defined according to the load characteristic analysis requirements; the multiple scenarios include: entity-based multiple scenarios, relationship-based multiple scenarios and time-based multiple scenarios; the extracted charging load data is integrated, and the charging pile load curve for each scenario is generated based on smoothing and mean calculation; based on the generated charging pile load curve for each scenario, an uncertain electric vehicle charging load knowledge extraction model is constructed.

[0014] In a preferred embodiment, the multi-scenario charging pile load curves are normalized separately, specifically: the maximum and minimum values ​​of the electric load data of the charging pile load curve in the entire time series under each scenario are calculated, and the electric load data of the charging pile load curve at each time is normalized based on the maximum and minimum value normalization calculation formula, and a charging pile load normalization curve is constructed.

[0015] In a preferred embodiment, the kmeans clustering algorithm is used to fit the normalized curve of charging pile loads in multiple scenarios to obtain the typical characteristic curve of electric vehicle charging loads, specifically: the number of categories of electric piles in the same scenario is set to the number of clusters; each time point of the normalized curve of charging pile loads is input into the k-means clustering model as a feature vector; the same number of centroids as the number of categories of electric piles are randomly selected as the center of each cluster; the Euclidean distance from the electric load value at each time point to the centroid is calculated; the electric load value at each time point is assigned to the cluster to which the centroid closest to it belongs; a new centroid is calculated based on the average value within each cluster; the change value between the new centroid and the old centroid is calculated, and when the change value is greater than a preset iteration threshold, the centroid iteration process is performed again, otherwise the electric load value at each time point in each cluster is output, and the typical characteristic curve of electric vehicle charging load is constructed based on the time series order; the process of automatically obtaining the multi-scenario charging pile load curves based on the uncertain electric vehicle charging load knowledge extraction model and processing to obtain the typical characteristic curve of electric vehicle charging load is used as the automatic generation model of electric vehicle charging typical characteristics.

[0016] The technical effects and advantages of the electric vehicle charging load characteristics modeling method based on knowledge graph of the present invention are as follows:

[0017] Compared with the existing technology, the present invention proposes a knowledge graph-based electric vehicle charging load characteristic modeling method, which realizes the refined mining of electric vehicle charging load characteristics, and establishes a multi-dimensional autonomous and controllable typical load characteristic model of electric vehicle charging load, laying a safety foundation for the safety and stability of the power grid, and providing technical reference for power grid operation managers. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a structural schematic diagram of the electric vehicle charging load characteristics modeling method based on the knowledge graph of the present invention. DETAILED DESCRIPTION

[0019] 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 ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0020] Embodiment 1, Figure 1 The present invention provides a method for modeling electric vehicle charging load characteristics based on a knowledge graph, comprising the following steps:

[0021] Obtain the historical data of tram charging load, build a knowledge graph library of tram charging load, and perform knowledge extraction to build an uncertain tram charging load knowledge extraction model. Perform data clustering based on the uncertain tram charging load knowledge extraction model to build an automatic generation model of typical features of tram charging.

[0022] With the popularization of electric vehicles and the rapid growth of charging demand, the power grid is facing uncertainty and impact of charging load; in order to solve these problems, the steps of the present invention include obtaining historical data of electric vehicle charging load, building a knowledge graph library of electric vehicle charging load, performing knowledge extraction and data clustering, so as to construct an automatic generation model of typical characteristics of electric vehicle charging.

[0023] This embodiment systematically analyzes and processes the randomness and uncertainty of electric vehicle charging loads. It can effectively deal with the impact of charging loads on the power grid and improve the stability and reliability of the power grid by acquiring and analyzing charging load data, building a knowledge graph library, extracting key information, clustering data, and automatically generating typical characteristic curves. These technical solutions help to achieve optimal scheduling of power grid loads and provide scientific basis and decision support for power grid operation.

[0024] In this embodiment, the method for acquiring the electric vehicle charging load knowledge graph library is as follows:

[0025] Obtain the historical data of electric vehicle charging load from the power grid ledger, and build a knowledge graph library of electric vehicle charging load based on RDFS triples and Neo4j graph database;

[0026] It should be noted that knowledge graph technology has the ability to analyze multi-label big data reasoning knowledge and is currently widely used in the field of power systems. Due to the huge amount of data involving uncertain electric vehicle charging piles, the flexible changes in spatial and power grid topology, and the big data retrieval of charging pile charging loads in different scenarios, the amount of calculation is huge and the scenarios are varied; knowledge graph technology can use graph models to retrieve big data multi-scenario charging pile charging load data, which is more suitable for the generation of typical electric vehicle charging loads in the present invention.

[0027] Neo4j is an open source graph database storage software. A large number of practical applications at home and abroad have enabled Neo4j software to achieve new development both at home and abroad. Neo4j can also be regarded as a high-performance graph engine with all the characteristics of a mature database, which provides a basis for realizing the electric vehicle charging load characteristic modeling method based on knowledge graph of the present invention.

[0028] In this embodiment, the electric vehicle charging load history data includes charging time, charging pile type, charging address and power grid data.

[0029] In this embodiment, the electric vehicle charging load knowledge graph library is constructed based on the RDFS triples and the Neo4j graph database, specifically:

[0030] Define entities and relationships, specifically:

[0031] The electric vehicle charging load knowledge graph library takes the electric vehicle charging pile as the entity and includes three classes, namely, charging pile, road, and power grid. Each class includes attribute description and defines the relationship based on the characteristics of the class and the attribute.

[0032] Based on entities and relationships, we take classes as subjects, relationships as predicates, and attributes as objects to construct several types of RDFS triples: <class, relationship, attribute>;

[0033] Convert several RDFS triples into nodes and relationships of Neo4j graph database for storage;

[0034] Each node corresponds to a vehicle charging station entity, and each relationship corresponds to a predicate of a triple;

[0035] According to the historical data of the electric vehicle charging load, the RDFS triples are automatically generated and stored as the electric vehicle charging load knowledge graph library through the conversion script.

[0036] It should be noted that the charging piles include community charging piles, taxi charging piles, bus charging piles, company charging piles, shopping mall charging piles, highway charging piles, and gas station charging piles;

[0037] The road includes latitude and longitude, province, city, and road name;

[0038] The power grid includes distribution transformers, voltage levels, main grid substations, load curves, total capacity, charging modes, and charging utilization rates.

[0039] At present, the power grid account contains a large amount of historical data on electric vehicle charging load. Based on the above historical data, a knowledge graph of electric vehicle charging load is constructed; the knowledge graph of electric vehicle charging load is described in the form of triples, and each piece of electric vehicle charging load knowledge can be decomposed into the following form: (subject, predicate, object); the subject is an individual, the individual is an instance of a class, and the predicate connects two individuals. The object is an individual or an instance of a data type; the carrier of electric vehicle charging load is the electric vehicle charging pile, so the electric vehicle charging pile is an individual in the knowledge graph of electric vehicle charging load. In order to describe uncertainty in detail, the class and attribute description are defined for each electric vehicle charging pile individual, which can provide a detailed data model basis for further analysis.

[0040] It should be noted that the above steps, by constructing a knowledge graph library of electric vehicle charging load and using RDFS triples and Neo4j graph database for storage and management, can help solve the problems of disorder, instantaneity and imbalance of electric vehicle charging load:

[0041] Define entities and relationships: Electric vehicle charging piles, roads, and power grids are taken as three main classes, and specific attributes are defined for each class. Through this systematic modeling, various types of information about charging piles and their relationships can be clearly described and recorded. Defining clear attributes and relationships for each entity helps to fully understand and analyze the characteristics and influencing factors of charging loads.

[0042] Help: This systematic modeling approach can structure the complexity and diversity of charging loads, thereby better performing data analysis and prediction and reducing uncertainty in data processing.

[0043] Triple description: Use RDFS triples (<class, relationship, attribute>) to describe entities such as charging piles, power grids, roads, and the relationships between them, making the data more structured and semantic;

[0044] Help: This description method can intuitively express the relationship between data and provide clear data query and reasoning capabilities, which helps to discover the regularity and potential patterns of charging loads, thereby reducing the disorder of the system.

[0045] Nodes and relationships: Convert RDFS triples into nodes and relationships in the Neo4j graph database, so that data is stored and managed in the form of a graph in the graph database; each node represents a charging pile entity, and each relationship represents the association between entities;

[0046] Help: The storage method of graph database is very suitable for processing complex relationship and network structure data. It can quickly perform relationship query and graph analysis, which helps to identify and analyze the instantaneous changes and spatial distribution of charging load, thereby coping with the imbalance problem of charging load.

[0047] Automated processing: Automatically generate and store RDFS triples through conversion scripts to achieve automated data processing and dynamic updating of graphs;

[0048] Help: Automated processing can improve data processing efficiency, reduce manual intervention, and ensure timely data updates and accuracy, thereby better adapting to changes and uncertainties in charging loads.

[0049] Detailed characterization: By defining classes and attributes for each EV charging station, a detailed data model foundation is provided; this foundation can help further analyze the characteristics and impact of EV charging load;

[0050] Help: Building a detailed data model foundation can more accurately describe the characteristics and behavior patterns of charging loads, support deeper analysis and prediction, and effectively manage the impact of charging loads on the power grid.

[0051] Disorder: Through structured data modeling and knowledge graph construction, charging load data can be systematically managed and analyzed to reduce the disorder in data processing;

[0052] Instantaneity: By utilizing the fast query and analysis capabilities of the graph database, it is possible to track and analyze instantaneous changes in charging load in real time, and respond and adjust the grid operation strategy in a timely manner.

[0053] Imbalance: By characterizing and analyzing the load characteristics of different charging piles in detail, it is possible to identify and solve load imbalance problems and optimize grid load distribution and scheduling strategies.

[0054] Through the above steps, the electric vehicle charging load data can be systematically managed and analyzed, thus providing a comprehensive and structured data model and analysis platform; these measures can effectively deal with the disorder, instantaneous and unbalanced problems of electric vehicle charging load, improve the stability and reliability of the power grid, and provide scientific data support and decision-making basis for power grid operators.

[0055] In this embodiment, the knowledge extraction is performed to construct an uncertain electric vehicle charging load knowledge extraction model, specifically:

[0056] Define multiple scenarios, extract knowledge from the electric vehicle charging load knowledge graph based on multiple scenarios, and obtain the charging pile load curves for multiple scenarios;

[0057] A knowledge extraction model for uncertain electric vehicle charging load is constructed based on multi-scenario charging pile load curves.

[0058] Furthermore, the multi-scenario definition extracts knowledge from the electric vehicle charging load knowledge graph library based on the multi-scenario knowledge graph to obtain the multi-scenario charging pile load curve, specifically:

[0059] Select one type of several types of RDFS triples as the load characteristic analysis requirement, and define multiple scenarios according to the load characteristic analysis requirement;

[0060] Multi-scenario includes: multi-scenario based on entity, multi-scenario based on relationship and multi-scenario based on time dimension;

[0061] Integrate the extracted charging load data and generate the charging pile load curve for each scenario based on smoothing and averaging calculation;

[0062] Based on generating the charging pile load curve in each scenario, a knowledge extraction model for uncertain electric vehicle charging load is constructed.

[0063] The specific calculation formula for generating the charging pile load curve in each scenario based on smoothing and averaging is as follows:

[0064]

[0065] In the formula, is the electric load value at time i after smoothing, n is the size of the smoothing window, is the charging load data at time j.

[0066] Through the above, it is the typical electric vehicle charging load in this scenario. According to the knowledge graph library of electric vehicle charging uncertainty of physical electric vehicle charging, the present invention can realize the rapid retrieval of electric vehicle charging load historical data in different scenarios, laying a data foundation for generating a typical electric vehicle charging load curve.

[0067] It should be noted that based on the above-mentioned uncertainty electric vehicle charging load knowledge graph library, knowledge extraction is performed according to categories and attributes, so that the extraction of multi-scenario uncertainty electric vehicle charging load knowledge graphs can be realized; the electric vehicle charging load presents the characteristics of many typical scenarios and large fluctuations depending on the location, time, and type of the charging pile. Establishing a typical model of the electric vehicle charging load in a single scenario cannot clarify the characteristics of the electric load. The present invention uses the electric vehicle charging load knowledge graph library to establish a multi-scenario uncertainty electric vehicle typical charging load extraction model according to different categories.

[0068] It should be noted that entity-based multi-scenario uses the query function of the graph database to filter entities and relationships that match the defined scenario, such as the bus charging piles on Huangshan Road in Hefei;

[0069] Relationship-based multi-scenario is to analyze the relationship between entities and extract key data related to multi-scenario coupling, such as analyzing the load relationship between bus charging piles and distribution network transformers;

[0070] The multi-scenario based on the time dimension is a scenario in which the time dimension is extracted to extract the charging load curve data within a specific time period, such as the charging load data within 24 hours.

[0071] It should be noted that the above steps, by defining multiple scenarios and extracting knowledge, specify the complex characteristics of electric vehicle charging loads, significantly improving the ability to understand and manage charging loads; through multi-scenario analysis based on entities, relationships, and time dimensions, it is possible to fully capture the characteristics of charging loads in different scenarios, thereby effectively managing disorder, instantaneity, and imbalance. By generating charging pile load curves through smoothing and mean calculation, and building an uncertain electric vehicle charging load knowledge extraction model, these steps help reduce the volatility and uncertainty of data, enabling power grid operators to more accurately predict and adjust power grid loads, thereby reducing the impact of charging loads on the power grid and improving the stability and reliability of the power grid.

[0072] In this implementation, the construction of the automatic generation model of typical characteristics of electric vehicle charging is specifically as follows:

[0073] Based on the uncertain electric vehicle charging load knowledge extraction model, the multi-scenario charging pile load curves are automatically obtained, and the multi-scenario charging pile load curves are normalized respectively to obtain the corresponding charging pile load normalization curves;

[0074] Based on the kmeans clustering algorithm, the normalized curves of charging pile loads in multiple scenarios are fitted to obtain the typical characteristic curve of electric vehicle charging load;

[0075] Based on the typical characteristic curve of tram charging load, a model for automatically generating typical characteristics of tram charging is constructed.

[0076] A typical load characteristic is constructed based on historical data. Since the rated capacities of different charging piles are different, in order to generate a typical electric vehicle charging load curve, the present invention first performs normalization processing based on historical data.

[0077] Furthermore, the load curves of charging piles in multiple scenarios are normalized respectively, specifically:

[0078] Calculate the maximum and minimum values ​​of the electric load data of the charging pile load curve in the entire time series under each scenario, and normalize the electric load data of the charging pile load curve at each time based on the maximum and minimum value normalization calculation formula, and construct the charging pile load normalization curve.

[0079] Furthermore, the maximum and minimum value normalization calculation formula is specifically:

[0080]

[0081] Where X is the charging pile load data to be normalized, min(X) is the minimum value of the charging pile load data to be normalized, max(X) is the maximum value of the charging pile load data to be normalized, and Y is the normalized charging pile load data.

[0082] Furthermore, the normalized curve of charging pile load in multiple scenarios is fitted based on the kmeans clustering algorithm to obtain the typical characteristic curve of electric vehicle charging load, which is specifically:

[0083] The number of categories of charging piles in the same scenario is set to the number of clusters;

[0084] Each time point of the normalized curve of the charging pile load is input into the k-means clustering model as a feature vector;

[0085] Randomly select the same number of centroids as the number of categories of electric poles as the center of each cluster;

[0086] Calculate the Euclidean distance from the electric load value to the centroid at each time point;

[0087] Assign the electric load value at each time point to the cluster to which the centroid closest to it belongs;

[0088] Calculate new centroids based on the mean within each cluster;

[0089] Calculate the change between the new centroid and the old centroid. When the change is greater than the preset iteration threshold, perform the centroid iteration process again. Otherwise, output the electric load value at each time point in each cluster and construct a typical characteristic curve of the electric vehicle charging load based on the time sequence.

[0090] The process of automatically acquiring multi-scenario charging pile load curves based on the uncertainty electric vehicle charging load knowledge extraction model and processing to obtain the typical characteristic curve of the electric vehicle charging load is used as the automatic generation model of the typical characteristics of electric vehicle charging.

[0091] The specific formula of the Euclidean distance is as follows:

[0092]

[0093] In the formula, is the Euclidean distance, q is the scene number, m is the total number of scenes, is the electric load value at time point p in scenario q, is the electrical load value of the kth centroid in scene q.

[0094] It should be noted that the present invention adopts kmeans clustering algorithm to fit the typical characteristic curve of electric vehicle charging load in different scenarios. The kmeans clustering algorithm can cluster the electric vehicle charging load historical data similar to each other at each moment into different clusters, and the centroid of the cluster is represented by the different set K values ​​of the cluster center pile vehicle type. In this scenario, the electric vehicle charging load historical data retrieved contains: bus charging piles, shopping mall charging piles, the K value is set to 2, and so on; finally, the centroid of each cluster at each moment forms the typical characteristic curve of the electric vehicle charging load, providing data reference for power grid operators.

[0095] The electric vehicle charging load characteristic modeling method based on knowledge graph proposed in the present invention realizes the refined mining of electric vehicle charging load characteristics compared with the existing technology, and establishes a typical load characteristic model of electric vehicle charging load in a multi-dimensional and autonomous manner, laying a safety foundation for the safety and stability of the power grid and providing technical reference for power grid operation managers.

[0096] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0097] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented by software, the above embodiments may be implemented in whole or in part in the form of a computer program product.

[0098] Those of ordinary skill in the art will appreciate that the modules and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0099] In addition, each functional module in each embodiment of the present application may be integrated into one processing module, or each module may exist physically separately, or two or more modules may be integrated into one module.

[0100] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0101] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. The electric vehicle charging load characteristics modeling method based on knowledge graph is characterized by: The steps include: Obtain the historical data of the electric vehicle charging load and build a knowledge graph library of the electric vehicle charging load. The knowledge graph library of the electric vehicle charging load is obtained by obtaining the historical data of the electric vehicle charging load from the power grid ledger and is built based on the RDFS triples and the Neo4j graph database; Extract knowledge based on the electric vehicle charging load knowledge graph library and build an uncertain electric vehicle charging load knowledge extraction model; The knowledge extraction is performed to construct an uncertain electric vehicle charging load knowledge extraction model, specifically: Define multiple scenarios, extract knowledge from the electric vehicle charging load knowledge graph based on multiple scenarios, and obtain the charging pile load curves for multiple scenarios; The method of defining multiple scenarios and extracting knowledge from the electric vehicle charging load knowledge graph library based on multiple scenarios to obtain a multi-scenario charging pile load curve is specifically as follows: selecting one type of several types of RDFS triples as a load characteristic analysis requirement, and defining multiple scenarios respectively according to the load characteristic analysis requirements; Integrate the extracted charging load data and generate the charging pile load curve for each scenario based on smoothing and averaging calculation; Based on the generation of charging pile load curves in each scenario, a knowledge extraction model for uncertain electric vehicle charging load is constructed; Construct an uncertain electric vehicle charging load knowledge extraction model based on multi-scenario charging pile load curves; The construction of the automatic generation model of typical characteristics of electric vehicle charging is specifically as follows: Based on the uncertain electric vehicle charging load knowledge extraction model, the multi-scenario charging pile load curves are automatically obtained, and the multi-scenario charging pile load curves are normalized respectively to obtain the corresponding charging pile load normalized curves; Based on the kmeans clustering algorithm, the normalized curves of charging pile loads in multiple scenarios are fitted to obtain the typical characteristic curve of electric vehicle charging load; Based on the typical characteristic curve of electric vehicle charging load, a model for automatically generating typical characteristics of electric vehicle charging is constructed; Data clustering is performed based on the uncertain tram charging load knowledge extraction model, and an automatic generation model of typical features of tram charging is constructed.

2. The electric vehicle charging load characteristic modeling method based on knowledge graph according to claim 1 is characterized in that: The electric vehicle charging load historical data includes charging time, charging pile type, charging address and power grid data.

3. The electric vehicle charging load characteristic modeling method based on knowledge graph according to claim 2 is characterized in that: The electric vehicle charging load knowledge graph library is constructed by obtaining the historical data of electric vehicle charging load from the power grid ledger and based on RDFS triples and Neo4j graph database, specifically: The electric vehicle charging load knowledge graph library takes the electric vehicle charging pile as an entity and includes three classes, namely, charging pile, road, and power grid. Each class includes attribute description and defines the relationship based on the characteristics of the class and the attribute. Based on entities and relationships, construct several types of RDFS triples; Convert several RDFS triples into nodes and relationships of Neo4j graph database for storage; Define the categories and triples corresponding to each node and each relationship; According to the historical data of the electric vehicle charging load, the RDFS triples are automatically generated and stored as the electric vehicle charging load knowledge graph library through the conversion script.

4. The electric vehicle charging load characteristic modeling method based on knowledge graph according to claim 3 is characterized in that: The normalization process of the load curves of charging piles in multiple scenarios is as follows: Calculate the maximum and minimum values ​​of the electric load data of the charging pile load curve in the entire time series under each scenario; Based on the maximum and minimum normalization calculation formula, the electric load data at each time of the charging pile load curve is normalized and a charging pile load normalization curve is constructed.

5. The electric vehicle charging load characteristic modeling method based on knowledge graph according to claim 4 is characterized in that: The kmeans clustering algorithm is used to fit the normalized curve of charging pile load in multiple scenarios to obtain the typical characteristic curve of electric vehicle charging load, which is specifically: The number of categories of charging piles in the same scenario is set to the number of clusters; Each time point of the normalized curve of the charging pile load is input into the k-means clustering model as a feature vector; Randomly select the same number of centroids as the number of categories of electric poles as the center of each cluster; Calculate the Euclidean distance from the electric load value to the centroid at each time point; Assign the electric load value at each time point to the cluster to which the centroid closest to it belongs; Calculate new centroids based on the mean within each cluster; Calculate the change between the new centroid and the old centroid. When the change is greater than the preset iteration threshold, perform the centroid iteration process again. Otherwise, output the electric load value at each time point in each cluster and construct a typical characteristic curve of the electric vehicle charging load based on the time sequence. The process of automatically acquiring multi-scenario charging pile load curves based on the uncertainty electric vehicle charging load knowledge extraction model and processing to obtain the typical characteristic curve of the electric vehicle charging load is used as the automatic generation model of the typical characteristics of electric vehicle charging.

6. The electric vehicle charging load characteristic modeling method based on knowledge graph according to claim 5 is characterized in that: The specific calculation formula for generating the charging pile load curve in each scenario based on smoothing and averaging is as follows: In the formula, is the electric load value at time i after smoothing, n is the size of the smoothing window, is the charging load data at time j.

7. The electric vehicle charging load characteristic modeling method based on knowledge graph according to claim 6 is characterized in that: The maximum and minimum value normalization calculation formula is specifically: Where X is the charging pile load data to be normalized, min(X) is the minimum value of the charging pile load data to be normalized, max(X) is the maximum value of the charging pile load data to be normalized, and Y is the normalized charging pile load data.

8. The electric vehicle charging load characteristic modeling method based on knowledge graph according to claim 7 is characterized in that: The specific formula of the Euclidean distance is as follows: In the formula, is the Euclidean distance, q is the scene number, m is the total number of scenes, is the electric load value at time point p in scenario q, is the electrical load value of the kth centroid in scene q.