Charging station multi-feature portrait modeling method and system based on data driving
By building a multi-character image labeling system and cluster analysis of charging stations, the problem that traditional management methods cannot fully reflect the operating status of charging stations and participate in the interactive regulation of power grids is solved, the management efficiency and power grid coordination capabilities are improved, and the new energy vehicle industry has been promoted. development.
Patent Information
- Application Number
- CN202410917620.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-10
- Publication Date
- 2024-11-26
AI Technical Summary
Traditional charging station management methods are difficult to fully reflect the operating status, equipment performance and user needs of the charging station, and cannot effectively support the charging station to participate in the interactive regulation of the power grid. There is a lack of mature data-driven multi-character image modeling methods for charging stations.
By collecting and preprocessing the multi-dimensional data of the charging station, a multi-feature image labeling system is built, a clustering algorithm is used to divide the charging station into groups with specific common characteristics, and operating strategies and grid interaction regulation decisions are formulated based on the image analysis results.
It has improved the management level and operational efficiency of charging stations, promoted the interactive and coordinated development of charging stations and power grids, promoted the sustainable development of the new energy vehicle industry, and achieved peak-cutting and valley filling of the power grid.
Smart Images

Figure CN119026446A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of new energy vehicle charging stations, and in particular to a data-driven charging station multi-feature portrait modeling method and system. Background Art
[0002] With the rapid development of the electric vehicle industry, charging stations, as important infrastructure to support the operation of electric vehicles, are directly related to the popularity of electric vehicles and the charging experience of users. However, traditional charging station management methods are often limited to a single operating indicator, which makes it difficult to fully reflect the multi-dimensional characteristics of charging stations, such as operating status, equipment performance, and user needs, and cannot effectively support charging stations to participate in the interactive regulation of power grids. With the development of big data and artificial intelligence technologies, data-driven methods have been widely used in various fields. In the field of charging station operation, by collecting and analyzing multi-source information such as charging station operation data, user charging behavior data, and equipment status data, a comprehensive perception and accurate portrait of the operation of charging stations can be achieved. This data-driven charging station portrait modeling method can deeply explore the potential characteristics of charging stations and provide a scientific basis for the management and optimization of charging stations. However, there is currently no mature data-driven multi-feature portrait modeling method for charging stations. Existing charging portrait construction methods often focus on the analysis of electric vehicle user behavior characteristics, and lack in-depth mining and comprehensive analysis of accurate portraits of charging stations. At the same time, existing charging station operation analysis methods often lack consideration of interactive regulation of power grids, making it difficult to provide effective support for charging stations to participate in interactive regulation of power grids.
[0003] Therefore, the present invention aims to propose a data-driven multi-feature portrait modeling method for charging stations. By constructing a multi-feature portrait label system for charging stations, the clustering algorithm is used to divide charging stations into groups with specific common characteristics. The present invention can not only deeply explore the multi-dimensional characteristics of charging stations, but also support the subsequent charging stations to participate in the decision-making of grid interactive regulation based on the results of portrait analysis. On the one hand, it can improve the management level and operational efficiency of charging stations, and on the other hand, it can promote the interactive and coordinated development of charging stations and power grids, and promote the sustainable development of the new energy vehicle industry. Summary of the invention
[0004] In view of the above existing problems, a data-driven multi-feature portrait modeling method for charging stations is proposed. The multi-dimensional data such as points of interest around the charging station, number of orders, charging amount, charging frequency, and charging load are collected to construct a portrait label system that comprehensively reflects the operation of the charging station. The charging station group classification is achieved by extracting the label data features and clustering them. The portrait includes not only the basic attributes of the charging station, such as location and scale, but also dynamic features such as the operating status of the charging station, the user's charging behavior pattern, and the impact of user charging on the power grid. By constructing a multi-feature portrait, charging operators can comprehensively and accurately analyze the operation of the charging station and formulate more reasonable application strategies to improve the site operation efficiency; at the same time, they can also better guide the charging station to participate in the interactive regulation of the power grid according to the characteristics of the charging station portrait group, so as to achieve peak shaving and valley filling of the power grid.
[0005] In order to solve the above technical problems, a data-driven multi-feature portrait modeling method for charging stations is proposed, including:
[0006] The charging station data is collected and preprocessed, and a multi-feature portrait label system for charging stations is constructed based on portrait theory. The label system is processed and feature selection is performed through the exhaustive search method, and the charging station portrait labels are screened to determine the optimal subset of portrait labels. After classifying the charging stations, the k-means algorithm is used to construct the user charging behavior feature portrait and conduct cluster analysis on the charging stations.
[0007] As a preferred solution of the data-driven multi-feature portrait modeling method for charging stations described in the present invention, the charging station data collection includes collecting multi-dimensional data including charging station attribute data, charging station operation data and charging station external data.
[0008] The location of the station, the station application scenarios including public charging stations, bus charging stations, commercial complex charging stations and large park charging stations, station business hours, and station operator information.
[0009] The charging station operation data includes the number of charging piles at the site, daily charging capacity, average daily charging capacity of each pile, average charging capacity per order, single-gun charging time, average charging time per order, number of charging orders, single-pile charging frequency, daily order time period, and large-cycle orders.
[0010] The external data of the charging station includes POI data acquired by a map API, including data acquisition radius, POI name, POI category and POI coordinates, and power grid load data of the area to which the charging station belongs.
[0011] The preprocessing includes collecting charging station attribute data, operation data and external data to check data quality, removing duplicate data, cleaning the collected data, interpolating missing data, fitting and correcting outliers, retaining all data sources in data processing, and normalizing the cleaned data using the Z-Score method. When the data is unbalanced, sample weight calculation is performed to balance the deviation:
[0012]
[0013] Among them, w i is the sample weight of the i-th charging station, α, β and γ are adjustment coefficients to balance the impact of different features on the weight, R i is the usage frequency of the i-th charging station, is the average usage frequency of all charging stations, sat i is the user satisfaction score of the ith charging station, min(sat) and max(sat) are the minimum and maximum values of the satisfaction scores of all charging stations, respectively, Z is the total number of charging station types, z k is the number of samples of the k-th charging station.
[0014] As a preferred solution of the data-driven multi-feature portrait modeling method for charging stations described in the present invention, the construction of a multi-feature portrait label system for charging stations includes designing a data-driven portrait indicator system for charging stations based on preprocessed data of charging stations, and constructing a multi-level multi-feature portrait label system for charging stations including fact labels, model labels and feature labels.
[0015] The fact labels include dividing the processed raw data into fact labels according to their own attributes, charging needs and regulation potential, and building model labels based on the fact labels, including calculating portrait feature indicators including the number of charging station POIs, charging station utilization rate, average charging amount of a single pile, peak power consumption rate and peak time difference rate.
[0016] The number of charging station POIs includes collecting POIs within a radius of (100m, 500m) of the charging station and counting the number N of POIs near the charging station.
[0017] The charging station utilization rate includes one charging station with n normally used charging piles, and T j represents the charging time of the jth charging pile in one day, then the utilization rate of the jth charging pile R j Represented as R j =T j / 24, average utilization rate of charging stations R s It can be expressed as:
[0018] The average charging capacity of a single pile is:
[0019]
[0020] Among them, E tot,i is the total transaction power of the i-th charging station, n i is the number of charging piles at the i-th charging station. ave,i is the average charging capacity of the i-th charging station.
[0021] The peak power consumption rate is expressed as:
[0022]
[0023] Among them, E peak,a represents the peak power consumption rate of charging station i, T peak represents the time set of peak hours, P(t) represents the charging current at time t, Φ(t) represents the electricity price at time t, I(t) represents the charging current at time t, and E represents the charging efficiency.
[0024]
[0025] Wherein, t represents the time point during the charging process.
[0026] The peak time difference rate is expressed as:
[0027]
[0028] Among them, T p,i represents the peak time difference rate of charging station a, x represents the number of peak load periods of the power grid during the monitoring period, y represents the number of peak charging load periods of the charging station during the monitoring period, P(t) represents the load power data of the charging station at time t, t1 and t2 are the start and end times of the monitoring period, and J peak The number of time points during the peak period, J represents the number of peak periods.
[0029] As a preferred solution of the data-driven multi-feature portrait modeling method for charging stations described in the present invention, the multi-level multi-feature portrait label system of the feature labels of the charging stations includes determining the feature labels of each charging station through the portrait feature indicators of the model labels, including the location importance of the charging station, the charging demand of the charging station, and the transferability of the charging load, and constructing a multi-feature portrait label system of the charging station through the importance, demand and transferability to represent the multi-faceted operational characteristics of the charging station.
[0030] Determine the importance of the charging station location by the number of charging station POIs:
[0031]
[0032] The more POIs N there are, the more important the location of the charging station is, and the charging demand of the current charging station needs to be guaranteed first. Here, Z is the number of different types of charging stations, and R k is the usage frequency of the kth service type, A k is the information index of the kth service type including user evaluation and service speed, and erf(N) is the error function used to normalize the impact of the number of charging station POIs.
[0033] The charging demand of the charging station is determined by evaluating the charging station utilization rate and the charging capacity of a single pile:
[0034]
[0035] When the charging demand is between 0 and 1, it means that the charging station is not in use and the charging demand is low.
[0036] When the charging demand is much greater than 1, it means that the charging station utilization rate is high, the single pile charging capacity is large, and the charging demand is high. Among them, g(x) represents a nonlinear function that screens and weights data according to different characteristics of the charging station, h(r) represents the charging station service quality function, r represents the service quality index of the charging station, λ(E ave,i ) represents the probability density function of the charging capacity of a single pile. The charging capacity is standardized. and They represent the minimum and maximum values of the single pile charging capacity respectively, and sat' is the user satisfaction score.
[0037] Through the peak power consumption rate and peak coverage rate, the potential level of charging load transfer of the charging station is evaluated from the charging load and time, and the transferability of the charging load is determined:
[0038]
[0039] When the transferability is within the range of 0 to 0.5, it indicates that the transferability of the charging load is low, and when the transferability is within the range of 0.5 to 1, it indicates that the transferability of the charging load is high.
[0040] As a preferred solution of the data-driven multi-feature portrait modeling method for charging stations described in the present invention, the optimal subset of the portrait label includes performing feature combinations on the obtained charging station feature labels through MATLAB exhaustive search, combination one is location importance + charging demand, combination two is location importance + charging load transferability, combination three is charging demand + charging load transferability, combination four is location importance + charging demand + charging load transferability, evaluate the performance of each feature combination, and select the feature combination with the best performance:
[0041]
[0042] Where F(c) is the performance evaluation function, c is the feature combination, Ω is the data space, Y(c,δ) is the model prediction function corresponding to the feature combination c, g(δ) is the real data distribution function, δ represents the eigenvalue in the charging station data, ω(δ) is the normalization constant to make the integral result in the range [0,1], U is the number of data points, and r u is the true value of the u-th data point, is the model prediction value corresponding to the feature combination c, θ u is the standard deviation of the prediction error, ε is the cost factor, V is the number of features, and w v is the importance weight of the vth feature, t v (c) is the computation time of the vth feature in feature combination c.
[0043] Enter the four combinations into the formula in turn to calculate the evaluation value, and sort the evaluation values of the four combinations. The larger the value, the better the performance. Sort the combinations from high to low according to the evaluation value, and score the combinations from high to low with 4, 3, 2, and 1 points respectively.
[0044] While evaluating the performance of each feature combination, perform a redundancy check on the feature combination to remove duplicate and similar user features:
[0045]
[0046] Among them, ρ(X1,X2) Spearman correlation coefficient is used to evaluate the redundancy between features in the combination, and W scales the two features of the redundant test to the same range. Remove duplicate and similar information in feature 1, Remove the duplicate and similar information in feature 2, Φ weights the difference between feature 1 and feature 2, X1 and X2 are the data sets for comparing redundant features 1 and 2, and They represent the sample points in the feature 1 and feature 2 data sets respectively, U1 and U2 are the number of data points in the feature 1 and feature 2 data sets, and b and d represent the integration operation on the interval [b, d].
[0047]
[0048] Among them, Q is the frequency point of Fourier transform, L is the number of wavelet functions, e -j2πft is the kernel function of Fourier transform, ψ k (t) is the wavelet function.
[0049] When the ρ(X1, X2) correlation coefficient is greater than 0.8 and the ρ(X1, X2) correlation coefficient is less than -0.8, it is considered that there is a high correlation between the features. When the ρ(X1, X2) correlation coefficient is between (-0.8, 0.8), it is considered that the correlation between the features can be ignored.
[0050] When there is a high correlation between features, calculate the correlation ratio between feature combinations, rank the high correlation ratios in the combinations, rank the combinations from high to low according to the ratio values, score the combinations, assign 1, 2, 3, and 4 points from high to low, calculate the total score based on the evaluation value scores, select the combination with the highest score and remove the features with high correlation as the optimal subset of portrait labels, and when the total scores are the same, select the combination with the higher evaluation value score as the optimal subset of portrait labels.
[0051] As a preferred solution of the data-driven multi-feature portrait modeling method for charging stations described in the present invention, the cluster analysis of charging stations includes randomly selecting K initial cluster centers C from a sample data set size U calculated according to the label system. i (1≤i≤K), select the elbow method to determine the optimal number of clusters, calculate the distance between the sample data and the initial cluster center one by one, group them according to the class corresponding to the cluster center, and when the data in the group is consistent with the last grouping, the algorithm converges and outputs the result. When the data in the group is inconsistent with the last grouping, find the nearest cluster center and classify it into the current cluster, update the cluster center, calculate the mean of the data items in each cluster as the new cluster center, and iterate again to reduce the intra-cluster sum of squared errors SSE until the cluster center no longer changes or the objective function converges, and the iteration stops.
[0052] After clustering, the clustering effectiveness evaluation formula is:
[0053]
[0054] Among them, x i is the i-th sample data, μ k is the mean of the kth cluster center, σ k is the variance of the kth cluster center. When B(U,K,SSE)>0, the larger the value, the better the clustering effect and the higher the clustering validity. When B(U,K,SSE)≤0, it means that the clustering effect is no different from random assignment and the clustering is invalid.
[0055] The optimal subset of portrait labels with effective clustering is regulated in layers and grades.
[0056] As a preferred solution of the data-driven multi-feature portrait modeling method for charging stations described in the present invention, the hierarchical and graded regulation includes three levels of regulation:
[0057] The first-level control is that the system classifies the charging stations according to the optimal subset of portrait labels, and divides the charging stations into three charging demand levels: high, medium and low. The system sends control instructions to the charging stations in real time according to the grid load conditions, including power limitation and capacity increase. The charging stations adjust the output power of the charging piles according to the received control instructions to meet the grid control requirements, and the charging stations feed back the control results to the system.
[0058] The secondary control system analyzes the changing trend of charging demand at charging stations and predicts short-term charging demand based on the feedback results of the primary control. Based on the prediction results, the system sends real-time control instructions to charging stations through automatic scheduling pricing and demand response, and formulates control strategies including adjusting the output power of charging piles, optimizing charging station layout and pre-scheduling. The system sends the control strategy to the charging station, and the charging station makes corresponding adjustments based on the control strategy, and the charging station feeds back the adjustment results to the system.
[0059] The third-level control is that the system analyzes the long-term operation of the charging station and evaluates the overall performance of the charging station according to the feedback results of the second-level control. According to the evaluation results, the system formulates optimization strategies, including improving the utilization rate of charging piles and reducing the operating costs of the charging station. The system sends the optimization measures to the charging station, and the charging station makes adjustments based on the optimization measures. The charging station feeds back the adjustment results to the system. The system evaluates the optimization effect according to the feedback results and adjusts the optimization measures.
[0060] Another object of the present invention is to provide a data-driven multi-feature portrait modeling system for charging stations. The present invention constructs a comprehensive multi-level multi-feature portrait labeling system by collecting and preprocessing the attributes, operations and external data of the charging stations, and then uses feature selection and cluster analysis methods to perform refined classification and regulation of the charging stations. This system aims to optimize the operation and management of charging stations, improve the utilization efficiency of charging piles, reduce operating costs, and at the same time achieve optimal distribution of power grid loads, meet users' charging needs, improve user satisfaction, and support the long-term sustainable development of charging stations. Through such a closed-loop system and dynamic control strategy, charging stations can better adapt to market changes and power grid needs, and improve the performance and reliability of the overall charging network.
[0061] As a preferred solution of the data-driven multi-feature portrait modeling system of charging stations described in the present invention, it is characterized by including a data processing module, a feature selection and portrait construction module, and a clustering analysis and regulation module.
[0062] The data processing module collects attribute data, operation data and external data of the charging station, performs preprocessing operations such as cleaning, interpolation filling and outlier correction on the collected data, normalizes the data using the Z-Score method, calculates sample weights to balance data deviations, and constructs a multi-level and multi-feature portrait label system including fact labels, model labels and feature labels.
[0063] The feature selection and portrait construction module constructs a multi-feature portrait of the charging station through the feature labels of the location importance, charging demand and charging load transferability of the charging station, uses an exhaustive search method to process the portrait label system, screens out the optimal subset of portrait labels, and performs performance evaluation and redundancy test on the feature combination to determine the optimal feature combination.
[0064] The cluster analysis and control module uses the k-means algorithm to perform cluster analysis on the charging stations to identify different user charging behavior characteristics, and classifies and controls the charging stations according to the clustering results.
[0065] A computer device includes a memory and a processor, wherein the memory stores a computer program, and is characterized in that when the processor executes the computer program, the steps of a method described in a data-driven charging station multi-feature portrait modeling are implemented.
[0066] A computer-readable storage medium having a computer program stored thereon, characterized in that when the computer program is executed by a processor, the steps of a method described in a data-driven charging station multi-feature portrait modeling are implemented.
[0067] Beneficial effects of the present invention: At present, traditional charging station management is limited to a single operating indicator, which makes it difficult to fully reflect the multi-dimensional characteristics of the charging station, such as the operating status, equipment performance, and user needs, and it is also unable to effectively support the charging station's participation in the interactive regulation of the power grid. The present invention fully taps the value of charging station big data, and provides a method for constructing a charging station portrait feature label system by analyzing the historical operating data of the charging station, and using a clustering algorithm to implement a method for classifying charging station features. In addition, the present invention also utilizes multiple input features, including attribute features, charging demand features, and grid regulation participation features. In addition to considering the operating conditions of the charging station, it also considers the construction of a portrait of the charging station participating in the interactive regulation of the power grid, so as to support the subsequent formulation of hierarchical and graded strategies for the charging station's participation in the interactive regulation of the power grid and the refined management of the regulation process. Specifically, the beneficial effects of the present invention are mainly reflected in:
[0068] Multi-dimensional operation characteristic analysis: Based on the portrait theory, a charging station operation portrait label system is constructed from multiple dimensions such as the basic attributes of the charging station, energy demand, and regulation characteristics, which comprehensively characterizes the operation characteristics of the charging station.
[0069] Feature selection and cluster analysis: Label correlation analysis is used to determine the optimal subset of portrait labels, and then the k-means algorithm is used for cluster analysis to divide the charging stations into groups with specific common characteristics, which helps to more accurately identify and understand the operating characteristics of different charging stations.
[0070] Promoting hierarchical management of charging stations: The present invention divides N charging stations into groups with K characteristics by constructing a charging station portrait model. On the one hand, operators can formulate group-specific operation strategies based on the portrait data to improve the operation efficiency of the stations. On the other hand, the portrait data can be used to guide charging stations to participate in grid dispatching in an orderly manner in the form of groups to reduce the impact on the grid during peak hours.
[0071] Promoting the development of electric vehicle charging networks: By optimizing the operating efficiency of charging stations and their ability to participate in grid dispatching, the present invention helps promote the healthy development of electric vehicle charging networks and promote the popularization and application of electric vehicles. BRIEF DESCRIPTION OF THE DRAWINGS
[0072] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative work, among which:
[0073] Figure 1 An overall flow chart of a data-driven multi-feature portrait modeling method for a charging station provided in accordance with an embodiment of the present invention.
[0074] Figure 2 A flow chart of a clustering algorithm implementation of a data-driven charging station multi-feature portrait modeling method provided in accordance with an embodiment of the present invention.
[0075] Figure 3 An optimal subset of a portrait label system for a data-driven charging station multi-feature portrait modeling method is provided according to an embodiment of the present invention.
[0076] Figure 4 A data-driven charging station multi-feature portrait modeling method provided by an embodiment of the present invention uses the TSNE method to reduce the dimension and visualize the clustering distribution of 81 charging station samples.
[0077] Figure 5 A system solution flow chart of a data-driven charging station multi-feature portrait modeling system provided for one embodiment of the present invention. DETAILED DESCRIPTION
[0078] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.
[0079] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.
[0080] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor is it an embodiment that is mutually exclusive with other embodiments, either individually or selectively.
[0081] The present invention is described in detail with reference to schematic diagrams. When describing the embodiments of the present invention, for the sake of convenience, the cross-sectional diagrams showing the device structure will not be partially enlarged according to the general scale, and the schematic diagrams are only examples, which should not limit the scope of protection of the present invention. In addition, in actual production, the three-dimensional dimensions of length, width and depth should be included.
[0082] At the same time, in the description of the present invention, it should be noted that the directions or positional relationships indicated by the terms "upper, lower, inner and outer" are based on the directions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific direction, be constructed and operated in a specific direction, and therefore cannot be understood as limiting the present invention. In addition, the terms "first, second or third" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.
[0083] In the present invention, unless otherwise clearly specified and limited, the terms "install, connect, connect" should be understood in a broad sense, for example: it can be a fixed connection, a detachable connection or an integral connection; it can also be a mechanical connection, an electrical connection or a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.
[0084] Example 1
[0085] Reference Figure 1-Figure 2, which is the first embodiment of the present invention, and provides a data-driven multi-feature portrait modeling method for charging stations, comprising:
[0086] S1: Collect and pre-process charging station data, and build a multi-feature portrait label system for charging stations based on portrait theory.
[0087] Furthermore, multi-dimensional data is collected including charging station attribute data, charging station operation data and charging station external data.
[0088] The location of the station, the station application scenarios including public charging stations, bus charging stations, commercial complex charging stations and large park charging stations, station business hours, and station operator information.
[0089] The charging station operation data includes the number of charging piles at the site, daily charging capacity, average daily charging capacity of each pile, average charging capacity per order, single-gun charging time, average charging time per order, number of charging orders, single-pile charging frequency, daily order time period, and large-cycle orders.
[0090] The external data of the charging station includes POI data acquired by a map API, including data acquisition radius, POI name, POI category and POI coordinates, and power grid load data of the area to which the charging station belongs.
[0091] The preprocessing includes collecting charging station attribute data, operation data and external data to check data quality, removing duplicate data, cleaning the collected data, interpolating missing data, fitting and correcting outliers, retaining all data sources in data processing, and normalizing the cleaned data using the Z-Score method. When the data is unbalanced, sample weight calculation is performed to balance the deviation:
[0092]
[0093] Among them, w i is the sample weight of the i-th charging station, α, β and γ are adjustment coefficients to balance the impact of different features on the weight, R i is the usage frequency of the i-th charging station, is the average usage frequency of all charging stations, sat i is the user satisfaction score of the ith charging station, min(sat) and max(sat) are the minimum and maximum values of the satisfaction scores of all charging stations, respectively, Z is the total number of charging station types, z k is the number of samples of the k-th charging station.
[0094] It should be noted that based on the preprocessed data of charging stations, a data-driven portrait index system for charging stations is designed, and a multi-level and multi-feature portrait label system for charging stations is constructed, which includes fact labels, model labels and feature labels.
[0095] The fact labels include dividing the processed raw data into fact labels according to their own attributes, charging demand and regulation potential, and constructing model labels based on the fact labels, including calculating the portrait feature indicators including the number of charging station POIs, charging station utilization rate, average charging capacity of a single pile, peak power consumption rate and peak time difference rate; constructing the following fact labels:
[0096] Table 1 Multidimensional feature fact labels of charging stations
[0097]
[0098]
[0099] The number of charging station POIs includes collecting POIs within a radius of (100m, 500m) of the charging station and counting the number N of POIs near the charging station; the identification and classification of POIs near the station in the present invention is shown in Table 2.
[0100] Table 2 Classification of POIs near charging stations
[0101]
[0102]
[0103] The charging station utilization rate includes one charging station with n normally used charging piles, and T j represents the charging time of the jth charging pile in one day, then the utilization rate of the jth charging pile R j Represented as R j =T j / 24, average utilization rate of charging stations R s It can be expressed as:
[0104] The average charging capacity of a single pile is:
[0105]
[0106] Among them, E tot,i is the total transaction power of the i-th charging station, n i is the number of charging piles at the i-th charging station. ave,i is the average charging capacity of the i-th charging station.
[0107] The peak power consumption rate is expressed as:
[0108]
[0109] Among them, E peak,a represents the peak power consumption rate of charging station i, T peak represents the time set of peak hours, P(t) represents the charging current at time t, Φ(t) represents the electricity price at time t, I(t) represents the charging current at time t, and E represents the charging efficiency.
[0110]
[0111] Wherein, t represents the time point during the charging process.
[0112] The peak time difference rate is expressed as:
[0113]
[0114] Among them, T p,i represents the peak time difference rate of charging station a, x represents the number of peak load periods of the power grid during the monitoring period, y represents the number of peak charging load periods of the charging station during the monitoring period, P(t) represents the load power data of the charging station at time t, t1 and t2 are the start and end times of the monitoring period, and J peak The number of time points during the peak period, J represents the number of peak periods.
[0115] Preferably,
[0116]
[0117] Among them, Y(N,R s ,E ave,i ,E peak,i ,T p,i ) represents the comprehensive index of the multi-feature portrait of the charging station, N represents the number of charging station POIs, R s represents the utilization rate of charging stations, E ave,i Indicates the average charging capacity of a single pile, E peak,i Indicates the peak power consumption rate, T p,i Indicates the peak time difference rate, N0 and N t They represent the initial value and target value of the number of charging station POIs, n represents the number of charging piles, represents the utilization rate of the jth charging pile, represents the average charging capacity of the jth charging pile, and They represent the initial value and target value of the peak power consumption rate, G(E peak,i ) represents the normalized function of peak power consumption rate, H(T p,i ) represents the complex information filtering function of the peak time difference rate.
[0118] When the value of Y is larger, it means that the overall performance of the charging station is better, including excellent performance in indicators such as the number of charging stations, utilization rate, charging amount, power consumption rate and time difference rate.
[0119] When the value of Y is smaller, it means that the overall performance of the charging station is worse, and the above indicators need to be improved to improve the operating efficiency and service quality of the charging station.
[0120] It should also be noted that the characteristic labels of each charging station are determined by the portrait characteristic indicators of the model label, including the importance of the charging station location, the charging demand of the charging station, and the transferability of the charging load. The multi-feature portrait label system of the charging station is constructed through the importance, demand, and transferability to represent the multi-faceted operational characteristics of the charging station; the charging station portrait characteristic label system is shown in Table 3:
[0121] Table 3 Charging station multi-feature portrait label system
[0122]
[0123] Determine the importance of the charging station location by the number of charging station POIs:
[0124]
[0125] The more POIs N there are, the more important the location of the charging station is, and the charging demand of the current charging station needs to be guaranteed first. Here, Z is the number of different types of charging stations, and R k is the usage frequency of the kth service type, A k is the information index of the kth service type including user evaluation and service speed, and erf(N) is the error function used to normalize the impact of the number of charging station POIs.
[0126] The charging demand of the charging station is determined by evaluating the charging station utilization rate and the charging capacity of a single pile:
[0127]
[0128] When the charging demand is between 0 and 1, it means that the charging station is not in use and the charging demand is low.
[0129] When the charging demand is much greater than 1, it means that the charging station utilization rate is high, the single pile charging capacity is large, and the charging demand is high. Among them, g(x) represents a nonlinear function that screens and weights data according to different characteristics of the charging station, h(r) represents the charging station service quality function, r represents the service quality index of the charging station, λ(E ave,i ) represents the probability density function of the charging capacity of a single pile. The charging capacity is standardized. and They represent the minimum and maximum values of the single pile charging capacity respectively, and sat' is the user satisfaction score.
[0130] Preferably,
[0131]
[0132] Among them, x represents a characteristic value of the charging station, μ x and σ x are the mean and standard deviation of the eigenvalue x, which are used for standardization. B and D are the control slopes that affect the nonlinearity of weight distribution.
[0133]
[0134] Among them, r represents the service quality index of the charging station, r max is the maximum possible value of the service quality indicator, o is the number of service quality indicators, and ω is the weight of different service quality indicators.
[0135]
[0136] Among them, sat total Represents the user satisfaction score, is the average level of user satisfaction, and θ is a tuning parameter used to control the speed of exponential decay.
[0137] Through the peak power consumption rate and peak coverage rate, the potential level of charging load transfer of the charging station is evaluated from the charging load and time, and the transferability of the charging load is determined:
[0138]
[0139] When the transferability is within the range of 0 to 0.5, it indicates that the transferability of the charging load is low, and when the transferability is within the range of 0.5 to 1, it indicates that the transferability of the charging load is high.
[0140] S2: Process the label system and perform feature selection through exhaustive search method, filter the charging station portrait labels and determine the optimal subset of portrait labels.
[0141] Furthermore, the obtained charging station feature labels are combined through MATLAB exhaustive search. Combination 1 is location importance + charging demand, combination 2 is location importance + charging load transferability, combination 3 is charging demand + charging load transferability, and combination 4 is location importance + charging demand + charging load transferability. The performance of each feature combination is evaluated and the best performance feature combination is selected:
[0142]
[0143] Where F(c) is the performance evaluation function, c is the feature combination, Ω is the data space, Y(c,δ) is the model prediction function corresponding to the feature combination c, g(δ) is the real data distribution function, δ represents the eigenvalue in the charging station data, ω(δ) is the normalization constant to make the integral result in the range [0,1], U is the number of data points, and r u is the true value of the u-th data point, is the model prediction value corresponding to the feature combination c, θ u is the standard deviation of the prediction error, ε is the cost factor, V is the number of features, and w v is the importance weight of the vth feature, t v (c) is the computation time of the vth feature in feature combination c.
[0144] Enter the four combinations into the formula in turn to calculate the evaluation value, and sort the evaluation values of the four combinations. The larger the value, the better the performance. Sort the combinations from high to low according to the evaluation value, and score the combinations from high to low with 4, 3, 2, and 1 points respectively.
[0145] While evaluating the performance of each feature combination, perform a redundancy check on the feature combination to remove duplicate and similar user features:
[0146]
[0147] Among them, ρ(X1,X2) Spearman correlation coefficient is used to evaluate the redundancy between features in the combination, and W scales the two features of the redundant test to the same range. Remove duplicate and similar information in feature 1, Remove the duplicate and similar information in feature 2, Φ weights the difference between feature 1 and feature 2, X1 and X2 are the data sets for comparing redundant features 1 and 2, and They represent the sample points in the feature 1 and feature 2 data sets respectively, U1 and U2 are the number of data points in the feature 1 and feature 2 data sets, and b and d represent the integration operation on the interval [b, d].
[0148]
[0149] Among them, Q is the frequency point of Fourier transform, L is the number of wavelet functions, e -j2πft is the kernel function of Fourier transform, ψ k (t) is the wavelet function.
[0150] When the ρ(X1, X2) correlation coefficient is greater than 0.8 and the ρ(X1, X2) correlation coefficient is less than -0.8, it is considered that there is a high correlation between the features. When the ρ(X1, X2) correlation coefficient is between (-0.8, 0.8), it is considered that the correlation between the features can be ignored.
[0151] When there is a high correlation between features, calculate the correlation ratio between feature combinations, rank the high correlation ratios in the combinations, rank the combinations from high to low according to the ratio values, score the combinations, assign 1, 2, 3, and 4 points from high to low, calculate the total score based on the evaluation value scores, select the combination with the highest score and remove the features with high correlation as the optimal subset of portrait labels, and when the total scores are the same, select the combination with the higher evaluation value score as the optimal subset of portrait labels.
[0152] S3: After classifying the charging stations, the k-means algorithm is used to construct a user charging behavior feature portrait and conduct cluster analysis on the charging stations.
[0153] Furthermore, the sample data set size calculated according to the label system is U, and K initial cluster centers C are randomly selected from it. i (1≤i≤K), select the elbow method to determine the optimal number of clusters, calculate the distance between the sample data and the initial cluster center one by one, group them according to the class corresponding to the cluster center, and when the data in the group is consistent with the last grouping, the algorithm converges and outputs the result. When the data in the group is inconsistent with the last grouping, find the nearest cluster center and classify it into the current cluster, update the cluster center, calculate the mean of the data items in each cluster as the new cluster center, and iterate again to reduce the intra-cluster sum of squared errors SSE until the cluster center no longer changes or the objective function converges, and the iteration stops.
[0154] After clustering, the clustering effectiveness evaluation formula is:
[0155]
[0156] Among them, x i is the i-th sample data, μ k is the mean of the kth cluster center, σ k is the variance of the kth cluster center. When B(U,K,SSE)>0, the larger the value, the better the clustering effect and the higher the clustering validity. When B(U,K,SSE)≤0, it means that the clustering effect is no different from random assignment and the clustering is invalid.
[0157] The optimal subset of portrait labels with effective clustering is regulated in layers and grades.
[0158] It should be noted that hierarchical and graded regulation includes three levels of regulation:
[0159] The first-level control is that the system classifies the charging stations according to the optimal subset of portrait labels, and divides the charging stations into three charging demand levels: high, medium and low. The system sends control instructions to the charging stations in real time according to the grid load conditions, including power limitation and capacity increase. The charging stations adjust the output power of the charging piles according to the received control instructions to meet the grid control requirements, and the charging stations feed back the control results to the system.
[0160] The secondary control system analyzes the changing trend of charging demand at charging stations and predicts short-term charging demand based on the feedback results of the primary control. Based on the prediction results, the system sends real-time control instructions to charging stations through automatic scheduling pricing and demand response, and formulates control strategies including adjusting the output power of charging piles, optimizing charging station layout and pre-scheduling. The system sends the control strategy to the charging station, and the charging station makes corresponding adjustments based on the control strategy, and the charging station feeds back the adjustment results to the system.
[0161] The third-level control is that the system analyzes the long-term operation of the charging station and evaluates the overall performance of the charging station according to the feedback results of the second-level control. According to the evaluation results, the system formulates optimization strategies, including improving the utilization rate of charging piles and reducing the operating costs of the charging station. The system sends the optimization measures to the charging station, and the charging station makes adjustments based on the optimization measures. The charging station feeds back the adjustment results to the system. The system evaluates the optimization effect according to the feedback results and adjusts the optimization measures.
[0162] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.
[0163] Example 2
[0164] Reference Figure 3-Figure 4 , which is the second embodiment of the present invention, provides a data-driven multi-feature portrait modeling method for charging stations. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through experiments.
[0165] It is necessary to collect the operation data of electric vehicle charging stations in a specific area, including but not limited to the number of charging times, charging time, charging amount, number of charging orders, POI around the charging station, charging station type, charging time distribution, distribution network peak and valley time data, etc. Among them, the number of charging times, charging time, charging amount, number of charging orders, charging station type, charging time distribution, distribution network peak and valley time data of the charging station are obtained from the charging station operation platform; the POI data around the charging station is obtained from the API provided by Baidu Maps, and the geographic information data around the specific station is obtained, including the name and type of POI near the station. The present invention obtains POI data within a radius of 300 meters around the station.
[0166] The collected data is cleaned to remove outliers, duplicate records and incomplete data. Then data preprocessing, including data conversion and normalization, is performed to facilitate subsequent analysis.
[0167] Based on the portrait theory, a charging station operation portrait label system is constructed. First, fact labels are constructed based on the collected data; then, based on the fact labels, model labels that can reflect the operation characteristics and regulation characteristics of the charging station are defined and selected; finally, business application requirements are constructed based on the charging station portrait, and charging station portrait feature labels are constructed from the three dimensions of attributes, requirements, and regulation. The label system constructed by the present invention is shown in Table 3 above.
[0168] The Spearman correlation coefficient in SPSS26.0 software was used to test the redundancy of indicators, and the optimal subset was selected from the constructed label system for subsequent cluster analysis. In the present invention, the correlation coefficient between two indicators is greater than 0.8 or less than -0.8, which is considered to be the threshold value of high correlation or redundancy between indicators. The specific results of correlation analysis of label indicators are shown in Table 4:
[0169] Table 4 Label correlation analysis results
[0170]
[0171]
[0172] From the correlation results, it can be seen that the label systems constructed by the present invention are positively correlated with each other, and the correlation between the charging station utilization rate and the single pile charging amount is high (greater than 0.8). From the correlation results of the two labels with other labels, the single pile charging amount has a stronger correlation with other labels (larger correlation coefficient). Therefore, the single pile charging amount label is removed here. The final optimal charging station portrait label system is obtained as follows: Figure 3 shown.
[0173] The improved k-means algorithm is used to perform cluster analysis on the selected features and classify the charging stations into different groups, each of which represents a collection of charging stations with similar operating characteristics.
[0174] In this invention, a total of 81 charging stations in area A are selected as samples, and the optimal cluster k value is determined using the elbow rule. The calculation results show that the turning point of the selected sample SSE decreases significantly is 4, so 4 is the optimal cluster number in the sample of this invention. In order to more intuitively display the cluster distribution of the sample charging stations, the t-distributed stochastic neighbor embedding (TSNE) dimensionality reduction method is used to visualize the clustering results. The visualized clustering results are shown in Figure 2. Figure 4 shown.
[0175] Based on the calculation results of the optimal subset label feature value of the charging station portrait and the number of charging station clusters determined by the elbow rule, the 81 charging stations are divided into four categories. The specific clustering results are shown in Table 5.
[0176] Table 5 Charging station clustering results data
[0177]
[0178] Further analysis of the above clustering results can identify the characteristics and needs of different groups. Based on this information, corresponding operation strategies and grid interaction control strategies can be formulated for each group.
[0179] Establish a feedback mechanism, collect feedback from operators and users, and continuously iterate and optimize models and regulation strategies to adapt to market changes and user needs.
[0180] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.
[0181] Example 3
[0182] The third embodiment of the present invention is different from the first two embodiments in that:
[0183] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, and other media that can store program codes.
[0184] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.
[0185] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.
[0186] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.
[0187] Example 4
[0188] Reference Figure 5 , which is the fourth embodiment of the present invention, provides a data-driven multi-feature portrait modeling system for charging stations, including a data processing module, a feature selection and portrait construction module, and a clustering analysis and regulation module.
[0189] The data processing module collects the attribute data, operation data and external data of the charging station, performs preprocessing operations such as cleaning, interpolation filling and outlier correction on the collected data, normalizes the data using the Z-Score method, and calculates sample weights to balance data deviations, and constructs a multi-level and multi-feature portrait label system that includes fact labels, model labels and feature labels.
[0190] The feature selection and portrait construction module constructs a multi-feature portrait of the charging station through the feature labels of the charging station's location importance, charging demand, and charging load transferability. It uses an exhaustive search method to process the portrait label system, screens out the optimal subset of portrait labels, and performs performance evaluation and redundancy test on the feature combination to determine the optimal feature combination.
[0191] The clustering analysis and control module uses the k-means algorithm to perform cluster analysis on charging stations to identify different user charging behavior characteristics, and classifies and controls the charging stations based on the clustering results.
[0192] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.
Claims
1. A data-driven multi-feature portrait modeling method for charging stations, characterized by: include, Collect and pre-process charging station data, and build a multi-feature portrait label system for charging stations based on portrait theory; Through the exhaustive search method, the label system is processed and feature selection is performed to filter the charging station portrait labels and determine the optimal subset of portrait labels; After classifying the charging stations, the k-means algorithm is used to construct a user charging behavior feature portrait and conduct cluster analysis on the charging stations.
2. The data-driven multi-feature portrait modeling method for a charging station according to claim 1, characterized in that: The charging station data collection includes collecting multi-dimensional data including charging station attribute data, charging station operation data and charging station external data; The location of the station, the station application scenarios including public charging stations, bus charging stations, commercial complex charging stations and large park charging stations, station business hours, and station operator information; The charging station operation data includes the number of charging piles at the station, daily charging capacity, daily average charging capacity of each pile, average charging capacity per order, single-gun charging time, average charging time per order, number of charging orders, single-pile charging frequency, daily order time period, and large-cycle orders; The charging station external data includes POI data acquired by the map API, including data acquisition radius, POI name, POI category and POI coordinates, and power grid load data of the area to which the charging station belongs; The preprocessing includes collecting charging station attribute data, operation data and external data to check data quality, removing duplicate data, cleaning the collected data, interpolating missing data, fitting and correcting outliers, retaining all data sources in data processing, and normalizing the cleaned data using the Z-Score method. When the data is unbalanced, sample weight calculation is performed to balance the deviation: Among them, w i is the sample weight of the i-th charging station, α, β and γ are adjustment coefficients to balance the impact of different features on the weight, R i is the usage frequency of the i-th charging station, is the average usage frequency of all charging stations, sat i is the user satisfaction score of the ith charging station, min(sat) and max(sat) are the minimum and maximum values of the satisfaction scores of all charging stations, respectively, Z is the total number of charging station types, z k is the number of samples of the k-th charging station.
3. The data-driven multi-feature portrait modeling method for charging stations according to claim 2, characterized in that: The construction of the multi-feature portrait label system for charging stations includes designing a data-driven portrait indicator system for charging stations based on preprocessed data of charging stations, and constructing a multi-level multi-feature portrait label system for charging stations including fact labels, model labels, and feature labels; The fact label includes dividing the processed raw data into fact labels according to its own attributes, charging demand and regulation potential, and building a model label based on the fact label, including calculating the portrait feature indicators including the number of charging station POIs, charging station utilization rate, average charging capacity of a single pile, peak power consumption rate and peak time difference rate; The number of charging station POIs includes collecting POIs within a radius of (100m, 500m) of the charging station and counting the number of POIs N near the charging station; The charging station utilization rate includes one charging station with n normally used charging piles, and T j represents the charging time of the jth charging pile in one day, then the utilization rate of the jth charging pile R j Represented as R j =T j / 24, average utilization rate of charging stations R s It can be expressed as: The average charging capacity of a single pile is: Among them, E tot,i is the total transaction power of the i-th charging station, n i is the number of charging piles at the i-th charging station; E ave,i is the average charging capacity of the i-th charging station; The peak power consumption rate is expressed as: Among them, E peak,a represents the peak power consumption rate of charging station i, T peak represents the time set of the peak period, P(t) represents the charging current at time t, Φ(t) represents the electricity price at time t, I(t) represents the charging current at time t, and E represents the charging efficiency; Wherein, t represents the time point during the charging process; The peak time difference rate is expressed as: Among them, T p,i represents the peak time difference rate of charging station a, x represents the number of peak load periods of the power grid during the monitoring period, y represents the number of peak charging load periods of the charging station during the monitoring period, P(t) represents the load power data of the charging station at time t, t1 and t2 are the start and end times of the monitoring period, and J peak The number of time points during the peak period, J represents the number of peak periods.
4. The data-driven multi-feature portrait modeling method for a charging station according to claim 3, characterized in that: The multi-level and multi-feature portrait label system of the feature label of the charging station includes determining the feature label of each charging station through the portrait feature index of the model label, including the importance of the charging station location, the charging demand of the charging station, and the transferability of the charging load, and constructing the multi-feature portrait label system of the charging station through the importance, demand and transferability to represent the multi-faceted operation characteristics of the charging station; Determine the importance of the charging station location by the number of charging station POIs: The more POIs N there are, the more important the location of the charging station is, and the charging demand of the current charging station needs to be prioritized. Here, Z is the number of different types of charging stations, and R k is the usage frequency of the kth service type, A k is the information index of the kth service type including user evaluation and service speed, erf(N) is the error function used to normalize the impact of the number of charging station POIs; The charging demand of the charging station is determined by evaluating the charging station utilization rate and the charging capacity of a single pile: When the charging demand is between 0 and 1, it means that the charging station is not used and the charging demand is low; When the charging demand is much greater than 1, it means that the charging station utilization rate is high, the single pile charging capacity is large, and the charging demand is high. Among them, g(x) represents a nonlinear function that screens and weights data according to different characteristics of the charging station, h(r) represents the charging station service quality function, r represents the service quality index of the charging station, λ(E ave,i ) represents the probability density function of the charging capacity of a single pile. The charging capacity is standardized. and They represent the minimum and maximum values of the single pile charging capacity, respectively, and sat' is the user satisfaction score; Through the peak power consumption rate and peak coverage rate, the potential level of charging load transfer of the charging station is evaluated from the charging load and time, and the transferability of the charging load is determined: When the transferability is within the range of 0 to 0.5, it indicates that the transferability of the charging load is low, and when the transferability is within the range of 0.5 to 1, it indicates that the transferability of the charging load is high.
5. The data-driven multi-feature portrait modeling method for a charging station according to claim 4, characterized in that: The optimal subset of the portrait label includes performing feature combinations on the obtained charging station feature labels through MATLAB exhaustive search, where combination one is location importance + charging demand, combination two is location importance + charging load transferability, combination three is charging demand + charging load transferability, and combination four is location importance + charging demand + charging load transferability, evaluating the performance of each feature combination and selecting the feature combination with the best performance: Where F(c) is the performance evaluation function, c is the feature combination, Ω is the data space, Y(c,δ) is the model prediction function corresponding to the feature combination c, g(δ) is the real data distribution function, δ represents the eigenvalue in the charging station data, ω(δ) is the normalization constant to make the integral result in the range [0,1], U is the number of data points, and r u is the true value of the u-th data point, is the model prediction value corresponding to the feature combination c, θ u is the standard deviation of the prediction error, ε is the cost factor, V is the number of features, and w v is the importance weight of the vth feature, t v (c) is the computation time of the vth feature in feature combination c; Enter the four combinations into the formula in turn to calculate the evaluation value, sort the evaluation values of the four combinations, the larger the value, the better the performance, sort the combinations from high to low according to the evaluation value, and score the combinations from high to low with 4, 3, 2, and 1 points respectively; While evaluating the performance of each feature combination, perform a redundancy check on the feature combination to remove duplicate and similar user features: Among them, ρ(X1,X2) Spearman correlation coefficient is used to evaluate the redundancy between features in the combination, and W scales the two features of the redundant test to the same range. Remove duplicate and similar information in feature 1, Remove the duplicate and similar information in feature 2, Φ weights the difference between feature 1 and feature 2, X1 and X2 are the data sets for comparing redundant features 1 and 2, and Represent the sample points in the feature 1 and feature 2 data sets respectively, U1 and U2 are the number of data points in the feature 1 and feature 2 data sets, b and d represent the integral operation on the interval [b, d]; Among them, Q is the frequency point of Fourier transform, L is the number of wavelet functions, e -j2πft is the kernel function of Fourier transform, ψ k (t) is the wavelet function; When the ρ(X1, X2) correlation coefficient is greater than 0.8 and the ρ(X1, X2) correlation coefficient is less than -0.8, it is considered that there is a high correlation between the features. When the ρ(X1, X2) correlation coefficient is between (-0.8, 0.8), it is considered that the correlation between the features can be ignored. When there is a high correlation between features, calculate the correlation ratio between feature combinations, rank the high correlation ratios in the combinations, rank the combinations from high to low according to the ratio values, score the combinations, assign 1, 2, 3, and 4 points from high to low, calculate the total score based on the evaluation value scores, select the combination with the highest score and remove the features with high correlation as the optimal subset of portrait labels, and when the total scores are the same, select the combination with the higher evaluation value score as the optimal subset of portrait labels.
6. A data-driven multi-feature portrait modeling method for a charging station as claimed in claim 5, characterized in that: The cluster analysis of charging stations includes randomly selecting K initial cluster centers C from a sample data set of size U calculated based on the label system. i (1≤i≤K), select the elbow method to determine the optimal number of clusters, calculate the distance between the sample data and the initial cluster center one by one, group them according to the class corresponding to the cluster center, and when the data in the group is consistent with the last grouping, the algorithm converges and outputs the result. When the data in the group is inconsistent with the last grouping, find the cluster center with the nearest distance and classify it into the current cluster, update the cluster center, calculate the mean of the data items in each cluster as the new cluster center, iterate again, reduce the sum of squared errors SSE within the cluster, until the cluster center no longer changes or the objective function converges, and the iteration stops; After clustering, the clustering effectiveness evaluation formula is: Among them, x i is the i-th sample data, μ k is the mean of the kth cluster center, σ k is the variance of the kth cluster center. When B(U,K,SSE)>0, the larger the value, the better the clustering effect and the higher the clustering validity. When B(U,K,SSE)≤0, it means that the clustering effect is the same as random assignment and the clustering is invalid. The optimal subset of portrait labels with effective clustering is regulated in layers and grades.
7. A data-driven multi-feature portrait modeling method for a charging station according to claim 6, characterized in that: The hierarchical regulation includes three levels of regulation: The first-level control is that the system classifies charging stations according to the optimal subset of portrait labels, and divides charging stations into three charging demand levels: high, medium, and low. The system sends control instructions to charging stations in real time according to the load of the power grid, including power restriction and capacity increase. The charging stations adjust the output power of the charging piles according to the received control instructions to meet the control requirements of the power grid, and the charging stations feed back the control results to the system. Secondary control is the system that analyzes the charging demand change trend of the charging station based on the feedback results of primary control and predicts the short-term charging demand. Based on the prediction results, the system sends real-time control instructions to the charging station through automatic scheduling pricing and demand response, and formulates control strategies including adjusting the output power of the charging pile, optimizing the layout of the charging station and pre-scheduling. The system sends the control strategy to the charging station, and the charging station makes corresponding adjustments based on the control strategy, and the charging station feeds back the adjustment results to the system; The third-level control is that the system analyzes the long-term operation of the charging station and evaluates the overall performance of the charging station according to the feedback results of the second-level control. According to the evaluation results, the system formulates optimization strategies, including improving the utilization rate of charging piles and reducing the operating costs of the charging station. The system sends the optimization measures to the charging station, and the charging station makes adjustments based on the optimization measures. The charging station feeds back the adjustment results to the system. The system evaluates the optimization effect according to the feedback results and adjusts the optimization measures.
8. A system using a data-driven charging station multi-feature portrait modeling method as claimed in any one of claims 1 to 7, characterized in that: It includes data processing module, feature selection and image building module, and cluster analysis and regulation module; The data processing module collects attribute data, operation data and external data of the charging station, performs preprocessing operations such as cleaning, interpolation filling and outlier correction on the collected data, normalizes the data using the Z-Score method, calculates sample weights to balance data deviations, and constructs a multi-level and multi-feature portrait label system including fact labels, model labels and feature labels; The feature selection and portrait construction module constructs a multi-feature portrait of the charging station through the feature labels of the location importance, charging demand and charging load transferability of the charging station, uses an exhaustive search method to process the portrait label system, screens out the optimal subset of the portrait labels, and performs performance evaluation and redundancy test on the feature combination to determine the optimal feature combination; The cluster analysis and control module uses the k-means algorithm to perform cluster analysis on the charging stations to identify different user charging behavior characteristics, and classifies and controls the charging stations according to the clustering results.
9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.
Citation Information
Cited By
Enterprise data element treatment method and device based on big data
CN119557739A
Charger charging current control system and method based on feedback analysis
CN120320460A
Charger charging current control system and method based on feedback analysis
CN120320460B
Charging station charging amount prediction method based on user charging behavior portrait
CN120542672A
Distribution network electric vehicle charging load bearing capacity evaluation method considering space-time portrait
CN121682349A
