Data processing method and device of charging station, storage medium and electronic equipment

Abstract

The application provides a data processing method and device of a charging station, a storage medium and an electronic device, the method comprising: constructing a knowledge graph of the charging station; obtaining feature data of the charging station based on the knowledge graph; obtaining each historical sample data of the charging station; and applying the feature data to filter each historical sample data to obtain target sample data. The feature data and influence factors of the charging station are extracted through the constructed knowledge graph, and then the historical sample data is filtered based on the feature data, the historical sample data that is greatly different from the feature of the charging station and is invalid is filtered out, and the historical sample data that is highly similar to the feature of the charging station is retained. Thus, the volume of the data used for predicting the charging station can be effectively reduced, the accuracy of the applied data is improved, the required time for prediction is shortened, and the accuracy of the prediction is improved.

Description

Technical Field

[0001] This invention relates to the field of charging station technology, and in particular to a data processing method and apparatus, storage medium and electronic equipment for charging stations. Background Technology

[0002] As the number of electric vehicles increases, the number of charging stations, which serve as supporting infrastructure, is also constantly increasing. Forecasting charging load is fundamental to the planning and scheduling of charging stations. Currently, load forecasting methods that rely on large amounts of historical data are commonly used for charging station load forecasting.

[0003] Traditional load forecasting methods rely on a large amount of historical data, and the scale and quality of this historical data directly affect the accuracy of the forecast results. Currently, when forecasting the load of charging stations, the historical data used is very large, and much of it differs significantly from the characteristics of charging stations and is not meaningful for reference. This results in long forecasting times and low accuracy of the forecast results. Summary of the Invention

[0004] In view of this, embodiments of the present invention provide a data processing method and apparatus for charging stations, a storage medium and an electronic device. By applying the present invention, historical sample data of charging stations can be filtered to obtain data with high accuracy and small volume. Using this data to predict the load of charging stations can shorten the prediction time and improve the accuracy of the prediction results.

[0005] To achieve the above objectives, the embodiments of the present invention provide the following technical solutions:

[0006] A data processing method for a charging station includes:

[0007] Constructing a knowledge graph for charging stations;

[0008] The feature data of the charging station are obtained based on the knowledge graph.

[0009] Obtain historical sample data for each of the charging stations;

[0010] The feature data is used to filter each of the historical sample data to obtain target sample data; wherein, the filtering of each of the historical sample data refers to selecting or removing at least one historical sample data based on the correlation information between each of the historical sample data and the feature data.

[0011] Optionally, the above method, wherein constructing the knowledge graph of the charging station includes:

[0012] Acquire the semi-structured data and application layer data of the charging station;

[0013] Extract load influencing factor data from the semi-structured data;

[0014] A knowledge graph is constructed using the load influencing factor data and the application layer data.

[0015] Optionally, in the above method, the step of applying the feature data to filter each of the historical sample data to obtain the target sample data includes:

[0016] Based on the feature data, obtain the first screening flag bit for each of the historical sample data;

[0017] The historical sample data belonging to the first screening flag that meets the preset first screening condition is determined as the initial screening sample data;

[0018] The target sample data is determined based on the data from each of the initial screening samples.

[0019] Optionally, in the above method, determining the target sample data based on each of the initial screening sample data includes:

[0020] Obtain the influencing factors related to the feature data;

[0021] Based on the feature data and the influence factor, obtain the second screening flag bit for each of the initial screening sample data;

[0022] The preliminary screening sample data that meet the preset second screening conditions and belong to the second screening flag are determined as the target sample data.

[0023] Optionally, in the above method, obtaining the first screening flag bit for each historical sample data based on the feature data includes:

[0024] Obtain the filtering coefficient between each of the historical sample data and the feature data;

[0025] Based on the preset screening threshold and the screening coefficient of each historical sample data, a first screening flag bit is determined for each historical sample data.

[0026] Optionally, in the above method, obtaining the second screening flag bit for each of the initial screening sample data based on the feature data and the influence factor includes:

[0027] Based on the feature data and the influence factor, the feature similarity coefficient of each of the initial screening sample data is obtained;

[0028] Based on the preset feature similarity coefficient threshold and the feature similarity coefficient of each of the initial screening sample data, a second screening flag bit is determined for each of the initial screening sample data.

[0029] Optionally, in the above method, obtaining the feature similarity coefficient of each of the initial screening sample data based on the feature data and the influence factor includes:

[0030] Based on the feature data and the influence factors, the similarity coefficient of each of the initial screening sample data in each feature dimension is determined;

[0031] For each of the initial screening sample data, the similarity coefficients of the initial screening sample data are summed to obtain the feature similarity coefficient of the initial screening sample data.

[0032] Optionally, in the above method, determining the similarity coefficient of each of the initial screening sample data in each feature dimension based on the feature data and the influence factor includes:

[0033] Determine the parameter set for each of the initial screening sample data for each feature dimension. The parameter set includes a first feature value in the initial screening sample data corresponding to the feature dimension, a second feature value in the feature data corresponding to the feature dimension, and a factor value in the influence factor corresponding to the feature dimension.

[0034] For each of the initial screening sample data, the first feature value, the second feature value, and the factor value in each parameter set of the initial screening sample data are calculated to obtain the similarity coefficient of the feature dimension corresponding to each parameter set.

[0035] The above methods may also include:

[0036] The target sample data is used to train the preset prediction model, and the trained prediction model is used as the charging station load prediction model.

[0037] The operating data of the charging station is input into the charging station load prediction model to obtain the load prediction result of the charging station.

[0038] A sample data screening device for charging stations, comprising:

[0039] Building units are used to construct a knowledge graph of charging stations;

[0040] The first acquisition unit is used to acquire feature data of the charging station based on the knowledge graph.

[0041] The second acquisition unit is used to acquire various historical sample data of the charging station;

[0042] The filtering unit is used to apply the feature data to filter each of the historical sample data to obtain target sample data; wherein, the filtering of each of the historical sample data refers to selecting or removing at least one historical sample data based on the correlation information between each of the historical sample data and the feature data.

[0043] Optionally, in the aforementioned apparatus, the construction unit performs the process of constructing a knowledge graph for the charging station, including:

[0044] Acquire the semi-structured data and application layer data of the charging station;

[0045] Extract load influencing factor data from the semi-structured data;

[0046] A knowledge graph is constructed using the load influencing factor data and the application layer data.

[0047] Optionally, in the aforementioned apparatus, the filtering unit performs the process of applying the feature data to filter each of the historical sample data to obtain the target sample data, including:

[0048] Based on the feature data, obtain the first screening flag bit for each of the historical sample data;

[0049] The historical sample data belonging to the first screening flag that meets the preset first screening condition is determined as the initial screening sample data;

[0050] The target sample data is determined based on the data from each of the initial screening samples.

[0051] Optionally, in the aforementioned apparatus, the screening unit performs the process of determining the target sample data based on each of the initial screening sample data, including:

[0052] Obtain the influencing factors related to the feature data;

[0053] Based on the feature data and the influence factor, obtain the second screening flag bit for each of the initial screening sample data;

[0054] The preliminary screening sample data that meet the preset second screening conditions and belong to the second screening flag are determined as the target sample data.

[0055] Optionally, in the aforementioned apparatus, the filtering unit performs the process of obtaining a first filtering flag bit for each of the historical sample data based on the feature data, including:

[0056] Obtain the filtering coefficient between each of the historical sample data and the feature data;

[0057] Based on the preset screening threshold and the screening coefficient of each historical sample data, a first screening flag bit is determined for each historical sample data.

[0058] Optionally, in the aforementioned apparatus, the screening unit performs a process of obtaining a second screening flag bit for each of the initial screening sample data based on the feature data and the influence factor, including:

[0059] Based on the feature data and the influence factor, the feature similarity coefficient of each of the initial screening sample data is obtained;

[0060] Based on the preset feature similarity coefficient threshold and the feature similarity coefficient of each of the initial screening sample data, a second screening flag bit is determined for each of the initial screening sample data.

[0061] Optionally, in the aforementioned apparatus, the screening unit performs a process of obtaining the feature similarity coefficient of each of the initial screening sample data based on the feature data and the influence factor, including:

[0062] Based on the feature data and the influence factors, the similarity coefficient of each of the initial screening sample data in each feature dimension is determined;

[0063] For each of the initial screening sample data, the similarity coefficients of the initial screening sample data are summed to obtain the feature similarity coefficient of the initial screening sample data.

[0064] Optionally, in the aforementioned apparatus, the screening unit performs a process of determining the similarity coefficient of each of the initial screening sample data in each feature dimension based on the feature data and the influence factor, including:

[0065] Determine the parameter set for each of the initial screening sample data for each feature dimension. The parameter set includes a first feature value in the initial screening sample data corresponding to the feature dimension, a second feature value in the feature data corresponding to the feature dimension, and a factor value in the influence factor corresponding to the feature dimension.

[0066] For each of the initial screening sample data, the first feature value, the second feature value, and the factor value in each parameter set of the initial screening sample data are calculated to obtain the similarity coefficient of the feature dimension corresponding to each parameter set.

[0067] The aforementioned apparatus may optionally further include:

[0068] The training unit is used to train the preset prediction model using the target sample data, and to use the trained prediction model as the charging station load prediction model.

[0069] The input unit is used to input the operating data of the charging station into the charging station load prediction model to obtain the load prediction result of the charging station.

[0070] Compared with the prior art, the present invention has the following advantages:

[0071] This invention provides a data processing method, apparatus, storage medium, and electronic device for charging stations. The method includes: constructing a knowledge graph of the charging station; acquiring feature data of the charging station based on the knowledge graph; acquiring historical sample data of the charging station; and applying the feature data to filter the historical sample data to obtain target sample data. By extracting feature data of the charging station from the constructed knowledge graph, and then filtering historical sample data based on the feature data, invalid historical sample data with large differences from the features of the charging station are filtered out, while historical sample data with high similarity to the features of the charging station are retained. This effectively reduces the amount of data used in predicting charging stations, improves the accuracy of the applied data, shortens the prediction time, and increases prediction accuracy. Attached Figure Description

[0072] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0073] Figure 1 A flowchart of a data processing method for a charging station provided in an embodiment of the present invention;

[0074] Figure 2 A flowchart illustrating a method for constructing a knowledge graph for charging stations, provided in an embodiment of the present invention.

[0075] Figure 3 Example diagrams provided for constructing knowledge graphs and deriving feature data and influence factors from knowledge graphs in embodiments of the present invention;

[0076] Figure 4 A flowchart illustrating a method for obtaining target sample data from various historical sample data, provided in an embodiment of the present invention.

[0077] Figure 5 Another flowchart of a data processing method for a charging station provided in an embodiment of the present invention;

[0078] Figure 6 This is a schematic diagram of the structure of a data processing device for a charging station provided in an embodiment of the present invention;

[0079] Figure 7This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0080] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0081] In this application, the terms "comprising," "including," or any other variations thereof are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0082] As the number of electric vehicles increases, the number of charging stations, as supporting infrastructure, is also constantly increasing. Predicting charging load is fundamental to charging station planning and scheduling. The charging load of a single vehicle is influenced by multiple factors, including user driving habits, battery characteristics, charging facility characteristics, and charging method preferences. The total charging load of a charging station is affected by the number of electric vehicles and charging locations. The combined effect of these factors results in the randomness, intermittency, and volatility of the vehicle charging load in its temporal and spatial distribution.

[0083] Currently, there are two main methods for charging station load forecasting: model-driven and data-driven. Model-driven forecasting methods are mostly based on user travel characteristics and involve multiple influencing factors such as traffic and road networks to establish a load forecasting model. Data-driven forecasting methods are based on historical statistical data. Although model-driven methods do not require a large amount of historical data, the accuracy of the modeling directly affects the accuracy of the final forecast. They also have poor flexibility, as key influencing factors such as the current number of electric vehicles and urban traffic networks are rapidly evolving, resulting in low modeling accuracy and large forecasting errors. Data-driven forecasting methods are highly flexible and their accuracy increases over time. However, they require a large amount of historical data, and the computation time increases with the amount of data, significantly impacting the performance of the load forecasting module and making the forecast results highly inaccurate.

[0084] To address the aforementioned issues, this invention provides a data processing scheme for charging stations. This scheme can filter historical sample data of charging stations, removing invalid historical sample data that differs significantly from the characteristics of the charging stations, and retaining data with high similarity to the characteristics of the charging stations. This effectively reduces the amount of data and improves the accuracy of prediction.

[0085] This invention can be used in a wide variety of general-purpose or special-purpose computing environments or configurations. Examples include: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor devices, distributed computing environments including any of the above devices, etc. For instance, the method provided by this invention is applied to a charging station prediction system.

[0086] Reference Figure 1 The following is a flowchart of a data processing method for a charging station provided by an embodiment of the present invention, which is described in detail below:

[0087] S101. Construct a knowledge graph for charging stations.

[0088] In the method provided by this invention, a knowledge graph of the charging station is constructed using semi-structured data and application layer data of the charging station.

[0089] Furthermore, the knowledge graph includes various characteristics of charging stations related to loads, as well as various factors that affect loads.

[0090] Reference Figure 2 The flowchart below shows a method for constructing a knowledge graph for charging stations according to an embodiment of the present invention. The details are as follows:

[0091] S201. Obtain semi-structured data and application layer data of the charging station.

[0092] The semi-structured data of charging stations includes, but is not limited to, the number of charging piles, battery capacity, number of electric vehicles in the area, charging time, regional traffic conditions, driving habits of vehicle drivers, charging prices, and number of parking spaces.

[0093] Application layer data includes data from the feature layer and data layer of the charging station, including but not limited to date type, time period type, holiday type, city type, season type, region type, weather type, and road network type.

[0094] S202. Extract load influencing factor data from semi-structured data.

[0095] S203. Construct a knowledge graph using data on factors influencing application load and application layer data.

[0096] By mapping and fusing the data from charging stations, a knowledge graph of the charging stations can be constructed. The knowledge graph can reflect the mapping relationship between various data of the charging stations and fully reflect the correlation between various data, which facilitates the subsequent extraction of feature data and influencing factors of the charging stations.

[0097] S102. Obtain feature data of charging stations based on knowledge graphs.

[0098] The feature data of charging stations is derived from the knowledge graph. It should be noted that the feature data contains feature values ​​across N feature dimensions. Furthermore, when acquiring the feature data, influencing factors related to the feature data can also be obtained. These influencing factors can also be used to filter historical sample data; each influencing factor contains factor values ​​across N feature dimensions, where N is a positive integer. Preferably, the influencing factors derived from the knowledge graph reflect the differences in the coupling degree between different features and charging station load prediction.

[0099] Reference Figure 3 This is an example diagram illustrating the construction of a knowledge graph and the derivation of feature data and influencing factors from the knowledge graph, provided by embodiments of the present invention. By extracting and fusing information from the semi-structured data of charging stations, load data samples such as charging station information and charging vehicle information are obtained (i.e., the load influencing factor data mentioned above). Using the load influencing factor data, along with data from the feature layer and data layer, the dominant load features of the charging station (i.e., the feature data mentioned above) and participating factors (i.e., the influencing factors mentioned above) are constructed through knowledge reasoning and matching.

[0100] like Figure 3 As shown, semi-structured data includes, but is not limited to, the number of charging piles, battery capacity, number of electric vehicles in the area, charging time, regional traffic conditions, driving habits of vehicle drivers, charging prices, and the number of parking spaces. The feature layer and data layer constitute the application layer data, which includes data from the feature layer and data layer of the charging station, including but not limited to date type, time period type, holiday type, city type, season type, region type, weather type, and road network type.

[0101] S103. Obtain historical sample data of each charging station.

[0102] Preferably, the historical sample data of the charging station can be data within a specified time range, such as the historical data of the charging station within one year.

[0103] The historical sample data is actually the historical load characteristic data of the charging station, which includes the load characteristics of the charging station during operation.

[0104] S104. Using the feature data, filter the historical sample data to obtain the target sample data.

[0105] It should be noted that filtering historical sample data refers to selecting or removing at least one historical sample data based on the correlation information between each historical sample data and the feature data.

[0106] It should be noted that feature data is used to filter each historical sample data, retaining sample data with high feature similarity and high correlation with charging stations, and filtering out sample data with large feature differences from charging stations, thereby reducing the volume of sample data and improving the accuracy of sample data.

[0107] Preferably, multiple target sample data sets are obtained. After obtaining the target sample data, the preset prediction model is trained using the target sample data, and the trained prediction model is used as the charging station load prediction model. The charging station's operating data is input into the charging station load prediction model to obtain the charging station's load prediction result. It should be noted that the prediction model can be a model built using a neural network, and the charging station's operating data can be data generated by the charging station during operation within a preset time range, including but not limited to the number of charging piles, the number of normally operating charging piles, the number of malfunctioning charging piles, and the charging data of the charging piles.

[0108] In the embodiments provided by this invention, a knowledge graph of charging stations is constructed; feature data of charging stations is obtained based on the knowledge graph; historical sample data of charging stations are obtained; and the feature data is applied to filter the historical sample data to obtain target sample data. By extracting feature data of charging stations from the constructed knowledge graph, and then filtering historical sample data based on the feature data, invalid historical sample data with large differences from the features of charging stations are filtered out, while historical sample data with high similarity to the features of charging stations are retained. This effectively reduces the amount of data used in predicting charging stations, improves the accuracy of the applied data, shortens the prediction time, and improves the prediction accuracy.

[0109] Reference Figure 4 The following is a flowchart of a method for obtaining target sample data from various historical sample data, provided by an embodiment of the present invention. The specific details are as follows:

[0110] S301. Based on feature data, obtain the first screening flag bit for each historical sample data.

[0111] Preferably, the historical sample data can be referred to as the charging station load sample characteristics. Furthermore, the historical sample data contains feature values ​​of N feature dimensions.

[0112] Preferably, the data on the charging station load characteristics involved in this invention can be F = [f1, f2, ..., f N ] indicates that, among which, fN This represents the feature value of the Nth feature dimension of the charging station. Preferably, the feature data of the charging station can use F... g This indicates that historical sample data can be used with F. yi This indicates that the feature data of the charging station also contains feature values ​​of N feature dimensions, where N is a positive integer.

[0113] Preferably, the process of obtaining the first screening flag for each historical sample data is as follows: obtaining the screening coefficient between each historical sample data and the feature data; and determining the first screening flag for each historical sample data based on the preset screening threshold and the screening coefficient of each historical sample data.

[0114] Preferably, the screening coefficient can be the Jaccard coefficient. Each historical sample data and feature data can be treated as a group of data, and the Jaccard coefficient of each group of data can be obtained by performing a calculation on each group of data according to a preset coefficient calculation method.

[0115] For example, the preset coefficient calculation method is as follows:

[0116] Among them, J i F represents the Jaccard coefficient of the historical sample data numbered i. g F represents the characteristic data of the charging station. yi Let i represent the historical sample data with number i, where i = 1, 2, ..., m, and m is a positive integer.

[0117] For example, the characteristic data F of the charging station g The historical sample data can be represented as:

[0118]

[0119] Among them, F g For the characteristic data of the charging station; F y1 F y2 F y3 ......F ym All data are historical samples, f gN For feature data F g The eigenvalue of the Nth feature dimension; f ymN For the m-th historical sample data F ym The feature value of the Nth feature dimension.

[0120] Furthermore, based on the preset screening threshold and the screening coefficient for each historical sample data, the method for determining the first screening flag for each historical sample data is as follows:

[0121]

[0122] Where K1 is the preset screening threshold, T i This is the first filtering flag for the i-th historical sample data.

[0123] For example, when the screening coefficient J of historical sample data i i When the value exceeds the screening threshold, the first screening flag T of the historical sample data is... i The screening coefficient J is 1; when the historical sample data i has a screening coefficient of 1, the screening coefficient J is 1. i When the value is not greater than the screening threshold, the first screening flag T of historical sample data i is... i The threshold is 0; furthermore, the screening threshold can be set according to actual needs.

[0124] S302. The historical sample data belonging to the first screening flag that meets the preset first screening conditions is determined as the initial screening sample data.

[0125] Continuing with the explanation of the first filtering flag bit in S301, a first filtering flag bit with a value of 1 satisfies the preset first filtering condition, while a first filtering flag bit with a value of 0 does not satisfy the first filtering condition.

[0126] Preferably, the historical sample data corresponding to the first screening flag that does not meet the first screening condition is deleted; the historical sample data corresponding to the first screening flag that meets the first screening condition is determined as the initial screening sample data.

[0127] S303. Determine the target sample data based on the data of each preliminary screening sample.

[0128] The target sample data can be determined based on the data from each initial screening sample. For example, each initial screening sample data can be determined as the target sample data; or the initial screening sample data can be further filtered to obtain the target sample data.

[0129] The process of further filtering the initial screening sample data to obtain the target sample data is as follows:

[0130] Obtain the influencing factors related to the feature data;

[0131] Based on the feature data and the influence factor, obtain the second screening flag bit for each initial screening sample data;

[0132] The preliminary screening sample data that meet the preset second screening conditions and belong to the second screening flag are determined as the target sample data.

[0133] Furthermore, the details of obtaining the influencing factors related to the feature data are as described in step S102, and will not be repeated here.

[0134] By using feature data and influence factors, historical sample data can be filtered multiple times, thereby improving the accuracy of the obtained data.

[0135] It should be noted that the process of determining the second screening flag for each initial screening sample data is as follows: based on feature data and influencing factors, obtain the feature similarity coefficient of each initial screening sample data; based on the preset feature similarity coefficient threshold and the feature similarity coefficient of each initial screening sample data, determine the second screening flag for each initial screening sample data.

[0136] When determining the feature similarity coefficient of the initial screening sample data, first determine the similarity coefficient of the initial screening sample data in each feature dimension.

[0137] The process of determining the similarity coefficient of the initial screening sample data for each feature dimension is as follows: determine the parameter set of each initial screening sample data for each feature dimension. The parameter set includes the first feature value in the initial screening sample data corresponding to the feature dimension, the second feature value in the feature data corresponding to the feature dimension, and the factor value in the influencing factors corresponding to the feature dimension. For each initial screening sample data, calculate the first feature value, the second feature value, and the factor value in each parameter set of the initial screening sample data to obtain the similarity coefficient of the feature dimension corresponding to each parameter set.

[0138] Preferably, the initial screening sample data F yi and feature data F g Taking this as an example, we determine the parameter set Y of the initial screening sample data in feature dimension j. j Where the maximum value of j is N. Parameter set Y j Contains f yij f gj and k j ;f yij f is the first feature value corresponding to feature dimension j in the initial screening sample data. gj k is the second eigenvalue in the feature data corresponding to feature dimension j; j This refers to the factor value corresponding to feature dimension j in the influencing factors.

[0139] It should be noted that the initial screening sample data contains feature values ​​of N feature dimensions, which can yield N parameter sets.

[0140] With parameter set Y j The corresponding similarity coefficient H = k j (f gj ⊙f yij H represents the similarity of the initial screening sample data in feature dimension j.

[0141] Therefore, by adding up the various similarity coefficients of the initial screening sample data, the feature similarity coefficient of the initial screening sample data can be obtained.

[0142] Furthermore, the process of obtaining the feature similarity coefficients of the initial screening sample data can also be described as follows:

[0143] S(F g F yi )=∑ j k j (f gj ⊙f yij );

[0144] Among them, S(F g F yi F represents the characteristic data and historical sample data of the charging station. yi The feature similarity coefficient can be understood as the historical sample data F yi Feature similarity coefficient, S(F g F yi It can also be represented as S i ;k j f is the factor value corresponding to the j-th feature dimension in the influence factor; yij f is the first eigenvalue corresponding to the j-th feature dimension in the feature data; gj Let f be the second eigenvalue corresponding to the j-th feature dimension in the feature data; further, f gj ⊙f yij f is the j-th feature value in the data features. gj and the j-th feature value f in the initial screening sample data yij The XOR operation value.

[0145] Furthermore, after obtaining the feature similarity coefficient of each initial screening sample data, the second screening flag bit of each initial screening sample data is determined based on a preset feature similarity coefficient threshold, as follows:

[0146]

[0147] Where K2 is the feature similarity coefficient threshold, R i This is the second filtering flag for the i-th historical sample data.

[0148] For example, when the screening coefficient S of historical sample data i i When the value exceeds the screening threshold, the first screening flag R of the historical sample data is activated. i The screening coefficient S is 1; when the historical sample data i is selected... i When the value is not greater than the screening threshold, the second screening flag R of historical sample data i is... iThe threshold value is 0; furthermore, the feature similarity coefficient threshold can be set according to actual needs.

[0149] Continuing with the explanation of the second filtering flag, a second filtering flag with a value of 1 satisfies the preset second filtering condition, while a second filtering flag with a value of 0 does not satisfy the second filtering condition.

[0150] Reference Figure 5 This is another flowchart of a data processing method for a charging station provided by an embodiment of the present invention. The specific process is as follows:

[0151] Step S1: Extract and integrate the information of the charging station to construct a knowledge graph; then use the knowledge graph to deduce the dominant features and participating factors of the charging station; the dominant features here are the feature data mentioned above, and the participating factors here are the influencing factors mentioned above.

[0152] Step S2: Calculate the Jaccard coefficient between each historical sample data and the dominant features of the charging station based on the dominant features of the charging station.

[0153] Step S3: Perform initial screening of samples based on the calculated Jaccard coefficients to obtain an initial screening training sample set; the initial screening training sample set contains historical sample data that meet multiple Jaccard coefficient criteria.

[0154] Step S4: Based on the participation factor, calculate the feature similarity coefficient of each historical sample data in the initial screening training sample set, and perform a second screening of each historical sample data in the initial screening training sample set according to the calculated feature similarity coefficient, so as to obtain the final training sample set for prediction.

[0155] By applying this invention, the characteristics of historical load data samples can be fully explored. The prediction sample set obtained through double screening can better ensure consistency with the prediction target features and significantly improve the accuracy of the prediction model. It can also filter out many weakly similar samples, which can greatly reduce the data scale used for calculation and speed up the training process. Especially for application scenarios with a large number of historical data samples, the load prediction speed of this method is significantly improved.

[0156] This invention also provides a data processing device for a charging station, which is installed in a charging station prediction system and can support... Figure 1 The specific implementation of the method shown.

[0157] Reference Figure 6 The following is a schematic diagram of the structure of a data processing device for a charging station provided in an embodiment of the present invention, and is described in detail below:

[0158] Building unit 601 is used to build a knowledge graph of charging stations;

[0159] The first acquisition unit 602 is used to acquire feature data of the charging station based on the knowledge graph.

[0160] The second acquisition unit 603 is used to acquire various historical sample data of the charging station;

[0161] The filtering unit 604 is used to apply the feature data to filter each of the historical sample data to obtain target sample data; wherein, the filtering of each of the historical sample data refers to selecting or removing at least one historical sample data based on the correlation information between each of the historical sample data and the feature data.

[0162] In the embodiments provided by this invention, a knowledge graph of charging stations is constructed; feature data of charging stations is obtained based on the knowledge graph; historical sample data of charging stations are obtained; and the feature data is applied to filter the historical sample data to obtain target sample data. By extracting feature data of charging stations from the constructed knowledge graph, and then filtering historical sample data based on the feature data, invalid historical sample data with large differences from the features of charging stations are filtered out, while historical sample data with high similarity to the features of charging stations are retained. This effectively reduces the amount of data used in predicting charging stations, improves the accuracy of the applied data, shortens the prediction time, and improves the prediction accuracy.

[0163] In another embodiment provided by the present invention, the construction unit of the device performs the process of constructing a knowledge graph of the charging station, including:

[0164] Acquire the semi-structured data and application layer data of the charging station;

[0165] Extract load influencing factor data from the semi-structured data;

[0166] A knowledge graph is constructed using the load influencing factor data and the application layer data.

[0167] In another embodiment of the present invention, the filtering unit of the device performs a process of applying the feature data to filter each of the historical sample data to obtain target sample data, including:

[0168] Based on the feature data, obtain the first screening flag bit for each of the historical sample data;

[0169] The historical sample data belonging to the first screening flag that meets the preset first screening condition is determined as the initial screening sample data;

[0170] The target sample data is determined based on the data from each of the initial screening samples.

[0171] In another embodiment of the present invention, the screening unit of the device performs a process of determining the target sample data based on each of the initial screening sample data, including:

[0172] Obtain the influencing factors related to the feature data;

[0173] Based on the feature data and the influence factor, obtain the second screening flag bit for each of the initial screening sample data;

[0174] The preliminary screening sample data that meet the preset second screening conditions and belong to the second screening flag are determined as the target sample data.

[0175] In another embodiment provided by the present invention, the filtering unit of the device performs a process of obtaining a first filtering flag bit for each of the historical sample data based on the feature data, including:

[0176] Obtain the filtering coefficient between each of the historical sample data and the feature data;

[0177] Based on the preset screening threshold and the screening coefficient of each historical sample data, a first screening flag bit is determined for each historical sample data.

[0178] In another embodiment of the present invention, the screening unit of the device performs a process of obtaining a second screening flag bit for each of the initial screening sample data based on the feature data and the influence factor, including:

[0179] Based on the feature data and the influence factor, the feature similarity coefficient of each of the initial screening sample data is obtained;

[0180] Based on the preset feature similarity coefficient threshold and the feature similarity coefficient of each of the initial screening sample data, a second screening flag bit is determined for each of the initial screening sample data.

[0181] In another embodiment of the present invention, the screening unit of the device performs a process of obtaining the feature similarity coefficient of each of the initial screening sample data based on the feature data and the influence factor, including:

[0182] Based on the feature data and the influence factors, the similarity coefficient of each of the initial screening sample data in each feature dimension is determined;

[0183] For each of the initial screening sample data, the similarity coefficients of the initial screening sample data are summed to obtain the feature similarity coefficient of the initial screening sample data.

[0184] In another embodiment of the present invention, the screening unit of the device performs a process of determining the similarity coefficient of each of the initial screening sample data in each feature dimension based on the feature data and the influence factor, including:

[0185] Determine the parameter set for each of the initial screening sample data for each feature dimension. The parameter set includes a first feature value in the initial screening sample data corresponding to the feature dimension, a second feature value in the feature data corresponding to the feature dimension, and a factor value in the influence factor corresponding to the feature dimension.

[0186] For each of the initial screening sample data, the first feature value, the second feature value, and the factor value in each parameter set of the initial screening sample data are calculated to obtain the similarity coefficient of the feature dimension corresponding to each parameter set.

[0187] In another embodiment provided by the present invention, the device further includes:

[0188] The training unit is used to train the preset prediction model using the target sample data, and to use the trained prediction model as the charging station load prediction model.

[0189] The input unit is used to input the operating data of the charging station into the charging station load prediction model to obtain the load prediction result of the charging station.

[0190] This invention also provides a storage medium, which includes stored instructions, wherein when the instructions are executed, the device containing the storage medium is controlled to perform the above-described data processing method for the charging station.

[0191] This invention also provides an electronic device, the structural schematic of which is shown below. Figure 7 As shown, it specifically includes a memory 701 and one or more instructions 702, wherein one or more instructions 702 are stored in the memory 701 and configured to be executed by one or more processors 703 to perform the above-mentioned data processing method for the charging station.

[0192] The specific implementation processes and derivative methods of the above embodiments are all within the protection scope of this invention; the information, data and other content used in this application are all legal content.

[0193] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, for system or system embodiments, since they are basically similar to method embodiments, the description is relatively simple, and relevant parts can be referred to the descriptions in the method embodiments. The systems and system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0194] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0195] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A data processing method for a charging station, characterized in that, include: Constructing a knowledge graph for charging stations; wherein the process of constructing the knowledge graph for charging stations includes: acquiring semi-structured data and application layer data of the charging stations; extracting load influencing factor data from the semi-structured data; and constructing a knowledge graph using the load influencing factor data and the application layer data; Based on the knowledge graph, reasoning and matching are performed to obtain the feature data and influencing factors of the charging station; Obtain historical sample data for each of the charging stations; The feature data is used to filter the historical sample data to obtain the target sample data; wherein, filtering the historical sample data means selecting or removing at least one historical sample data based on the correlation information between the historical sample data and the feature data. The process of applying the feature data to filter each of the historical sample data to obtain target sample data includes: obtaining a first screening flag bit for each of the historical sample data based on the feature data; determining the historical sample data to which the first screening flag bit that meets the preset first screening condition belongs as the initial screening sample data; and determining the target sample data based on each of the initial screening sample data. Determining the target sample data based on each of the initial screening sample data includes: obtaining a second screening flag bit for each of the initial screening sample data based on the feature data and the influence factor; and determining the initial screening sample data to which the second screening flag bit that meets the preset second screening conditions belongs as the target sample data.

2. The method according to claim 1, characterized in that, Based on the feature data, obtain the first screening flag bit for each of the historical sample data, including: Obtain the filtering coefficient between each of the historical sample data and the feature data; Based on the preset screening threshold and the screening coefficient of each historical sample data, a first screening flag bit is determined for each historical sample data.

3. The method according to claim 1, characterized in that, Based on the feature data and the influence factor, a second screening flag is obtained for each of the initial screening sample data, including: Based on the feature data and the influence factor, the feature similarity coefficient of each of the initial screening sample data is obtained; Based on the preset feature similarity coefficient threshold and the feature similarity coefficient of each of the initial screening sample data, a second screening flag bit is determined for each of the initial screening sample data.

4. The method according to claim 3, characterized in that, Based on the feature data and the influence factor, the feature similarity coefficient of each of the initial screening sample data is obtained, including: Based on the feature data and the influence factors, the similarity coefficient of each of the initial screening sample data in each feature dimension is determined; For each of the initial screening sample data, the similarity coefficients of the initial screening sample data are summed to obtain the feature similarity coefficient of the initial screening sample data.

5. The method according to claim 4, characterized in that, Based on the feature data and the influence factor, the similarity coefficient of each of the initial screening sample data in each feature dimension is determined, including: Determine the parameter set for each of the initial screening sample data for each feature dimension. The parameter set includes a first feature value in the initial screening sample data corresponding to the feature dimension, a second feature value in the feature data corresponding to the feature dimension, and a factor value in the influence factor corresponding to the feature dimension. For each of the initial screening sample data, the first feature value, the second feature value, and the factor value in each parameter set of the initial screening sample data are calculated to obtain the similarity coefficient of the feature dimension corresponding to each parameter set.

6. The method according to claim 1, characterized in that, Also includes: The target sample data is used to train the preset prediction model, and the trained prediction model is used as the charging station load prediction model. The operating data of the charging station is input into the charging station load prediction model to obtain the load prediction result of the charging station.

7. A data processing device for a charging station, characterized in that, include: A construction unit is used to construct a knowledge graph of a charging station; wherein, the process of constructing the knowledge graph of a charging station includes: acquiring semi-structured data and application layer data of the charging station; extracting load influencing factor data from the semi-structured data; and constructing a knowledge graph using the load influencing factor data and the application layer data. The first acquisition unit is used to perform reasoning and matching based on the knowledge graph to acquire the feature data and influencing factors of the charging station. The second acquisition unit is used to acquire various historical sample data of the charging station; The filtering unit is used to apply the feature data to filter each of the historical sample data to obtain target sample data; wherein, filtering each of the historical sample data means selecting or removing at least one historical sample data based on the correlation information between each of the historical sample data and the feature data. The filtering unit is specifically used for: obtaining a first filtering flag bit for each of the historical sample data based on the feature data; determining the historical sample data to which the first filtering flag bit that meets the preset first filtering condition belongs as the initial screening sample data; and determining the target sample data based on each of the initial screening sample data. The determination of the target sample data based on each of the initial screening sample data includes: obtaining a second screening flag bit for each of the initial screening sample data based on the feature data and the influence factor; and determining the initial screening sample data to which the second screening flag bit that meets the preset second screening conditions belongs as the target sample data.

8. A storage medium, characterized in that, The storage medium includes stored instructions, wherein, when the instructions are executed, the device containing the storage medium is controlled to perform the data processing method for the charging station as described in any one of claims 1-6.

9. An electronic device, characterized in that, It includes a memory and one or more instructions, wherein one or more instructions are stored in the memory and configured to be executed by one or more processors as described in any one of claims 1-6.

Images