Electric vehicle charging aggregation and adjustment prediction method and system based on integrated clustering

Abstract

The invention discloses an electric vehicle charging aggregation and adjustment prediction method and system based on integrated clustering, and relates to the technical field of electric vehicle and power grid interaction, and the method comprises the steps: collecting the charging demand data of an electric vehicle, carrying out the classification and clustering, and integrating different types of clusters; analyzing the charging demand data and screening outlier samples; establishing an electric vehicle charging load characteristic index model, and predicting a charging load curve of the electric vehicle; evaluating the theoretical potential of the electric vehicle participating in the demand response; and optimizing a power grid access mode based on the electric vehicle charging load characteristic index and the theoretical potential of the demand response. The method integrates the advantages of a plurality of clustering algorithms, and improves the accuracy and reliability of data analysis. By integrating different clustering algorithms, the respective defects can be complemented, so that the clustering precision is improved, the effective identification of the user behavior and the charging mode of the electric vehicle is ensured, and the stability of the overall clustering result is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of interaction between electric vehicles and power grids, and in particular to an electric vehicle charging aggregation and regulation prediction method and system based on integrated clustering. Background Art

[0002] As global attention to climate change and environmental pollution increases, emissions from traditional fuel vehicles are subject to strict regulation. As a low-emission or even zero-emission alternative means of transportation, electric vehicles can help reduce greenhouse gas emissions and improve air quality. Studying the aggregation and regulation of electric vehicles can better understand and realize their role in green travel and help promote sustainable development goals. With the rapid development of the electric vehicle (EV) industry, electric vehicles have become an important part of terminal energy consumption, which can replace fuel vehicles to alleviate the energy crisis and improve the environment. A series of promotion plans have been formulated around electric vehicles around the world to promote their widespread use.

[0003] As more and more electric vehicles are connected to the power grid, the load on the power grid increases dramatically, the load side begins to have power characteristics, and the difficulty of balancing the supply and demand of the power grid continues to increase. Especially during peak load periods, the charging behavior of electric vehicles will increase the burden on the power grid and reduce its operating efficiency, especially during peak charging periods, which may cause a sharp increase in the load on the power grid. Since the distribution of electric vehicles in time and space is uncertain, if they are not managed, it will bring huge problems to the dispatching and operation of the power system. In real life, most private cars only drive for 2-3 hours a day, and most of the rest of the time is parked. When the number of these idle private electric vehicles is large enough, their batteries can be used to realize the interaction between electric vehicles and the power grid (V2G). Therefore, this paper aims to effectively reduce the impact on the power grid and balance the power supply and demand by aggregating the charging needs of electric vehicles and reasonably adjusting the charging period. Formulating corresponding prediction methods can help power companies better manage loads and dispatch energy, and improve the stability and reliability of the power system.

[0004] Existing EV load aggregation methods mainly focus on classification methods, such as classification analysis based on industry characteristics, load importance and working characteristics, or cluster analysis based on load curve characteristics. However, these methods have certain limitations in processing large-scale data and complex multidimensional features. Therefore, there is an urgent need for an intelligent model that can effectively manage and optimize the interaction between EVs and the power grid.

[0005] Clustering-based integration technology plays an important role in the research of electric vehicle aggregation and regulation prediction. It can combine the advantages of multiple clustering algorithms and improve the accuracy and reliability of data analysis. Different clustering algorithms have their own advantages and applicable scenarios. For example, K-means is suitable for processing large-scale data, but is sensitive to outliers and noise; while GMM can adapt well to multimodal data distribution and can handle clusters of different shapes and sizes; but it is sensitive to initial parameters and may lead to local optimal solutions. By integrating different clustering algorithms, their respective shortcomings can be complemented, thereby improving clustering accuracy and ensuring effective identification of electric vehicle user behavior and charging patterns. By combining the results of multiple algorithms, the integrated clustering method can reduce the uncertainty and deviation brought by a single algorithm and improve the stability of the overall clustering results. Diversified clustering schemes can effectively reduce the fluctuation of clustering results caused by the sensitivity of specific algorithms to data distribution. Therefore, integration technology can provide more reliable user behavior patterns and charging demand predictions. Summary of the invention

[0006] In view of the above-mentioned problems, the present invention is proposed.

[0007] Therefore, the technical problem solved by the present invention is to optimize the interaction between electric vehicles and power grids by classifying and aggregating electric vehicle charging demand data.

[0008] In order to solve the above technical problems, the present invention provides the following technical solution: an electric vehicle charging aggregation and regulation prediction method based on integrated clustering, which comprises the following steps:

[0009] Collect charging demand data of electric vehicles, classify and cluster them, and use relabeling method and projection method to integrate clusters of different categories; use sliding time window method to analyze charging demand data and screen outlier samples; establish electric vehicle charging load characteristic indicator model to predict the charging load curve of electric vehicles in future time periods; evaluate the theoretical potential of electric vehicles to participate in demand response; optimize the way of accessing the power grid based on the characteristic indicators of electric vehicle charging load and the theoretical potential of demand response.

[0010] As a preferred solution of the electric vehicle charging aggregation and regulation prediction method based on integrated clustering described in the present invention, wherein: the re-labeling method includes defining a basic clustering ψ 1 , 2 ,,,ψ γ , expand the clustering according to the optimal clustering number N, and obtain the clustering results where i = 1, 2, .. γ, α 1 , β 2 , ..λ N It is clustering.

[0011] Relabel the clusters in each clustering result to obtain the same label, and unify the results into

[0012] Create a computational matrix for shared samples, expressed as,

[0013]

[0014] Among them, μ represents the number of references, ν represents the number to be labeled, and c and d represent the current clusters.

[0015] Determine similar cluster pairs, expressed as,

[0016] (u′,v′)=argmax (u,v)∈I ο(u,v)

[0017] Among them, I = {(μ, ν) | 1 ≤ μ, ν ≤ N}, (u', v') represents the cluster pair after optimal relabeling.

[0018] Integrate similar clustering and convert S′1, S′ in the results 2 ,,,S′ γ The N categories of similar clusters in are integrated into the set η 1 , η 2 , ....η N , expressed as,

[0019]

[0020] As a preferred solution of the electric vehicle charging aggregation and regulation prediction method based on integrated clustering described in the present invention, the projection method includes superimposing the visualized cluster point diagram, vertically projecting to obtain the final result, removing duplicate samples, and superimposing similar clusters η 1 , η 2 , ....η N The combined correction is η 1 ′,η′ 2 , ....η′ N , analyze and screen controversial samples, and divide the controversial samples into multiple clusters.

[0021] As a preferred solution of the electric vehicle charging aggregation and regulation prediction method based on integrated clustering described in the present invention, wherein: the sliding time window method includes analyzing the standard deviation of the electric vehicle charging demand data, expressed as,

[0022]

[0023] Among them, x' i Indicates the current sample standard value, Represents the sample mean.

[0024] Use a sliding window to calculate the fluctuation of adjacent samples and calculate the standard deviation of each window. If it is greater than ω, the outlier samples are identified and removed.

[0025] As a preferred solution of the electric vehicle charging aggregation and regulation prediction method based on integrated clustering described in the present invention, the electric vehicle charging load characteristic index includes the daily average load A 1 、Daily load rate A 2 , Daily peak-to-valley difference rate A 3 , peak load rate A 4 , Normal load rate A 5 , valley load rate A 6 , as shown below,

[0026]

[0027] A 4 =P av.peak / P av

[0028] A 5 =P av.sh / P av

[0029] A 6 =P av.val / P av

[0030] Among them, P sum Represents total power, P max , P min Represent the maximum and minimum power, P av.peak , P av.sh , P av.val Represents the average power during peak, middle and valley periods, P av Represents the average power for the whole day.

[0031] As a preferred solution of the electric vehicle charging aggregation and regulation prediction method based on integrated clustering described in the present invention, wherein: the theoretical potential of electric vehicles to participate in demand response is evaluated, and the load is predicted according to the typical daily load curve. Calculate the load value of the power system during peak and valley periods and theoretical potential w i , expressed as,

[0032]

[0033] Where h represents the peak and valley time of the power system, P i,max and P i,maxare the predicted maximum and minimum load values ​​of the i-th type of load, τ represents the duration of the power system in the peak and valley phases, η i represents the proportion of flexible load of the i-th load under the demand response measures, Indicates that the peak period potential increases, Indicates the potential for increase during valley period.

[0034] As a preferred solution of the electric vehicle charging aggregation and regulation prediction method based on integrated clustering described in the present invention, the method of optimizing access to the power grid includes adjusting the charging strategy of aggregated electric vehicles based on charging load characteristic indicators and demand response potential.

[0035] Another object of the present invention is to provide an electric vehicle charging aggregation and regulation prediction system based on integrated clustering, which can solve the problems of large grid load fluctuations and uneven distribution of charging resources in the prior art by efficiently aggregating and regulating the charging needs of electric vehicles.

[0036] To solve the above technical problems, the present invention provides the following technical solutions: an electric vehicle charging aggregation and regulation prediction system based on integrated clustering, including a data acquisition module, a clustering and marking integration module, an outlier sample analysis module and a grid access optimization module.

[0037] The data acquisition module is responsible for collecting the charging demand data of the electric vehicle.

[0038] The clustering and labeling integration module is responsible for integrating clusters of different categories using a relabeling method and a projection method.

[0039] The outlier sample analysis module is responsible for analyzing charging demand data using a sliding time window.

[0040] The grid access optimization module is responsible for optimizing the grid access mode based on charging load characteristic indicators and demand response potential.

[0041] A computer device includes a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of the electric vehicle charging aggregation and regulation prediction method based on integrated clustering are implemented as described above.

[0042] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the electric vehicle charging aggregation and regulation prediction method based on integrated clustering as described above.

[0043] Beneficial effects of the present invention: The present invention combines the advantages of multiple clustering algorithms to improve the accuracy and reliability of data analysis. Different clustering algorithms have their own advantages and applicable scenarios. For example, K-means is suitable for processing large-scale data, but is sensitive to outliers and noise; while GMM can adapt well to multimodal data distribution and can process clusters of different shapes and sizes; but it is sensitive to initial parameters and may lead to local optimal solutions. By integrating different clustering algorithms, their respective deficiencies can be complemented to improve clustering accuracy and ensure effective identification of electric vehicle user behavior and charging patterns. The integrated clustering method can reduce the uncertainty and deviation caused by a single algorithm and improve the stability of the overall clustering results by combining the results of multiple algorithms. Diversified clustering schemes can effectively reduce the fluctuations in clustering results caused by the sensitivity of specific algorithms to data distribution. Therefore, integrated technology can provide more reliable user behavior patterns and charging demand predictions. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0045] Figure 1 An overall flow chart of the electric vehicle charging aggregation and regulation prediction method based on integrated clustering provided in the first embodiment of the present invention.

[0046] Figure 2 The present invention provides an integrated process based on re-labeling and projection method in the electric vehicle charging aggregation and regulation prediction method based on integrated clustering according to the first embodiment of the present invention.

[0047] Figure 3 The present invention provides outlier day screening of a sliding time window in an electric vehicle charging aggregation and regulation prediction method based on integrated clustering in the first embodiment of the present invention.

[0048] Figure 4 The silhouette coefficients of Kmeans and GMM under different numbers of clusters in the electric vehicle charging aggregation and regulation prediction method based on integrated clustering provided in the first embodiment of the present invention.

[0049] Figure 5 The overall framework diagram of the electric vehicle charging aggregation and regulation prediction system based on integrated clustering provided in the second embodiment of the present invention. DETAILED DESCRIPTION

[0050] 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.

[0051] Example 1, reference Figure 1 to Figure 4 , which is an embodiment of the present invention, provides an electric vehicle charging aggregation and regulation prediction method based on integrated clustering, characterized in that:

[0052] S1: Collect the charging demand data of electric vehicles, classify and cluster them, and use the relabeling method and projection method to integrate clusters of different categories.

[0053] The charging demand data includes the start time, end time, charging power, total charging amount, and user category of each charging. The above data is preprocessed, including data cleaning and feature selection. The outliers in the data are processed and the time format is unified. The daily average charging power, daily charging times, and daily total charging time characteristics of each vehicle's charging data are extracted.

[0054] Classification and clustering include: using four clustering algorithms, K-means, GMM, AGNES, and hierarchical clustering, to preliminarily classify the data and determine the optimal number of clusters. Using relabeling and projection methods, multiple clustering results are integrated into the final typical clusters. Relabeling and integration of similar clusters, and eliminating overlapping samples through projection methods. Check the quality of clustering results through visualization tools to ensure the accuracy of classification.

[0055] Kmeans is a clustering algorithm based on partitioning. Each data category divided by Kmeans has a centroid vector. The category partitioning and centroid calculation are continuously iterated until convergence, so as to obtain the optimal clustering. Specifically, it is achieved by calculating the distance between the sample point and the cluster center and continuously updating the cluster center. The formula is as follows:

[0056] H ji =||x j -ν i || 2

[0057] In the formula, Euclidean distance is used, x j is the sample point, H ji represents the cluster center distance, ν iRepresents the mean vector of the ith cluster among the k clusters. When performing internal calculations, the goal of Kmeans is to divide the sample points into k clusters so that the sample points in the class are as similar as possible and the classes are far away from each other. The specific steps are analyzed as follows.

[0058] (a) Determine the number of cluster categories k and randomly create an initial cluster center vector ν = {ν k}.

[0059] (b) Based on the center vector created in step (a), the data is classified into different clusters. The division process is as described above. The distance from each point to the center point is calculated. If a point is closest to the cluster center, it is determined that the point belongs to the cluster represented by the cluster center.

[0060] (c) Based on the division results of step (b), calculate the centroids of different clusters, reset the center vector, and obtain a new center vector ν = {ν k For Kmeans clustering using Euclidean distance, the centroid is calculated using the sample mean, as follows: k Represents clustering cluster |C k | is the mean vector of, and x represents the sample.

[0061]

[0062] (d) If {ν k} and {ν′ k} reaches the standard, that is, the mean vector of two consecutive iterations does not change, then the iteration stops, the model is judged to be converged, and the clustering ends. Otherwise, repeat the above method of dividing the data set according to the center point, and continue to divide the data set according to the new center point.

[0063] Model-based clustering algorithm GMM. The GMM algorithm has strong data modeling capabilities, strong robustness to data noise, and can effectively improve the blindness of the clustering process. In the clustering integration model, the GMM algorithm can also make up for the vacancy that Kmeans, AGNES, and BIRCH cannot perform probabilistic modeling. In addition, the GMM algorithm can also obtain the confidence coefficient of the clustering cluster, which can not only cluster the existing data set, but also has the ability to summarize unknown data. GMM uses a probability model that obeys the Gaussian distribution to express clustering. The basic function is as follows.

[0064]

[0065] In the formula, x G represents the vector, ν G is the n-dimensional mean vector, and δ is the n×n covariance matrix. The main calculation process is as follows.

[0066] (a) Define the target number of clusters.

[0067] (b) Randomly initialize the initial clusters and determine their normal distribution parameters. Secondly, fit the distribution of the clusters and calculate the probability that each sample belongs to the cluster. The closer the sample is to the center of the cluster, the greater the possibility that it belongs to the cluster.

[0068] (c) Based on the calculated values ​​in step (b), update the normal distribution parameters of each cluster.

[0069] (d) Repeat steps (b) and (c) until the entire model converges.

[0070] AGNES, a hierarchical clustering algorithm. From the previous description, we can see that the partition-based Kmeans clustering algorithm requires the definite number of clusters, but for the hierarchical clustering algorithm, the number of clusters does not need to be determined in advance, which has strong interpretability. At the same time, the hierarchical clustering algorithm is also more flexible in the choice of distance method. This method utilizes the distance relationship between data, adopts the method of splitting and merging data, and realizes the construction of the clustering tree through hierarchical partitioning. For this algorithm, obtaining the formation process of the clustering tree is equivalent to obtaining the clustering process, and then results of different granularities can be obtained. AGNES is a bottom-up merged hierarchical clustering. The initial state is that each sample is a class. It merges by finding the class with the smallest distance and the most similarity, and iterates continuously until the number of clusters is met. The process diagram is shown below. Figure 2 . Since the operation of hierarchical clustering will generate a cluster tree, the amount of operation and the intensity of operation of hierarchical clustering are relatively large. In terms of complexity, the complexity of the AGNES algorithm is O(N2log(N)), and the complexity of the BIRCH algorithm is O(2N). At the same time, since hierarchical clustering uses a greedy algorithm, it can only guarantee the local optimal result. Among them, the bottom-up merged hierarchical clustering assumes that each sample is initially a separate data category, which is more complicated to calculate. Therefore, taking the AGNES algorithm as an example, its specific calculation steps are described as follows.

[0071] (a) Each sample in the initial state is established as a separate cluster. Assuming the number of samples is i, select an appropriate distance formula, calculate the distance between samples separately, obtain i(i-1) / 2 distances, and combine them into a distance set W1.

[0072] (b) Select the two samples with the closest distance from the distance set W1, merge them into a new sample, and take their center as the representative of the new sample. Repeat step (a) to update the distance set W1 to the second round set W2 and enter the next cycle.

[0073] (c) Repeat step (b) until the cluster meets the conditions or all samples are involved.

[0074] (4) Hierarchical clustering uses highly compressed ternary features to construct a feature tree. The distance between samples is an important basis for merging or splitting, so it is necessary to define the distance between samples. Compared with Kmeans clustering, hierarchical clustering algorithms AGNES and BIRCH are more flexible in the choice of distance methods. There are three commonly used inter-class clustering measurement methods, as shown in equations (1) to (4).

[0075]

[0076] Formula (1) to (3) correspond to the shortest distance H between classes respectively. min , the longest distance between classes H max And the average distance between classes H avg In the formula, C j and C i Represents different clusters, dist is the distance calculation function, and e and f are samples in the cluster.

[0077] like Figure 3 The integrated process of re-labeling and projection method is shown. The re-labeling method includes defining the basic clustering ψ 1 , 2 ,,,ψ γ , expand the clustering according to the optimal clustering number N, and obtain the clustering results where i = 1, 2, .. γ, α 1 , β 2 , ..λ N It is clustering.

[0078] Relabel the clusters in each clustering result to obtain the same label, and unify the results into

[0079] Create a computational matrix for shared samples, expressed as,

[0080]

[0081] Among them, μ represents the number of references, ν represents the number to be labeled, and c and d represent the current clusters.

[0082] Determine similar cluster pairs, expressed as,

[0083] (u′,v′)=argmax (u,v)∈I ο(u,v)

[0084] Among them, I = {(μ, ν) | 1 ≤ μ, ν ≤ N}, (u', v') represents the cluster pair after optimal relabeling.

[0085] Integrate similar clustering and convert the results into S′ 1 , S′ 2 ,,,S′ γ The N categories of similar clusters in are integrated into the set η 1 , η 2 , ....η N , expressed as,

[0086]

[0087] The projection method includes superimposing the visualized cluster point graph, vertically projecting to obtain the final result, removing duplicate samples, and superimposing similar clusters η 1 , η 2 , ....η N The combined correction is η 1 ′,η′ 2 , ....η′ N , analyze and screen controversial samples, and divide the controversial samples into multiple clusters.

[0088] S2: Use the sliding time window method to analyze the charging demand data and filter outlier samples.

[0089] Sliding window is a commonly used data analysis technique, especially suitable for processing time series data. In each cluster, for the charging behavior of each vehicle, a sliding window of 3 days is used, which means that data within 3 days is taken for analysis each time, such as from day 1 to day 3, then moving to day 2 to day 4, and so on, until all the data is analyzed. In this way, the changing trends and patterns in the time series can be effectively captured while maintaining a relatively fixed analysis granularity. The standard deviation τ of the electric vehicle charging load data, let xi represent the current sample, let x represent the mean, is as follows:

[0090]

[0091] Among them, x' i Indicates the current sample standard value, Represents the sample mean.

[0092] (2) Use sliding windows to calculate the fluctuation of adjacent samples. The length of each window is 3, and the standard deviation of each window is calculated. Calculating the standard deviation in each sliding window is a statistical method for detecting data fluctuations, which can reflect the degree of dispersion between data points and the average value:

[0093]

[0094] In the formula, x i represents the i-th sample value, μ represents the mean of the sample, and n represents the total number of samples.

[0095] (3) Set the standard deviation threshold ω>0.1 to identify and remove outlier samples.

[0096] Preferably, within each cluster, a sliding window analysis is performed on the charging behavior of each vehicle, with a window length of 3 days.

[0097] Clustering indicators can be divided into internal indicators and external indicators. The silhouette coefficient is one of the commonly used internal indicators, which can be evaluated by assessing the intrinsic properties of the target cluster. The value of the silhouette coefficient ranges from -1 to 1. The larger the value of the indicator, the higher the clustering quality. Figure 3 It can be seen that the optimal number of clusters for Kmeans and GMM clustering algorithms is 3, and the larger the number of clusters, the smaller the silhouette coefficient tends to be.

[0098] S3: Establish an electric vehicle charging load characteristic index model to predict the charging load curve of electric vehicles in the future time period.

[0099] The characteristic indicators of electric vehicle charging load include daily average load A 1 、Daily load rate A 2 , Daily peak-to-valley difference rate A 3 , peak load rate A 4 , Normal load rate A 5 , valley load rate A 6 , as shown below,

[0100]

[0101] A 4 =P av.peak / P av

[0102] A 5 =P av.sh / P av

[0103] A 6 =P av.val / P av

[0104] Among them, P sum Represents total power, P max , P min Represent the maximum and minimum power, P av.peak , P av.sh , P av.val Represents the average power during peak, middle and valley periods, P av Represents the average power for the whole day.

[0105] S4: Evaluate the theoretical potential of electric vehicles for demand response.

[0106] Assessing the theoretical potential of EVs for demand response involves forecasting load based on a typical daily load profile. Calculate the load value of the power system during peak and valley periods and theoretical potential w i , expressed as,

[0107]

[0108] Where h represents the peak and valley time of the power system, P i,max and P i,max are the predicted maximum and minimum load values ​​of the i-th type of load, τ represents the duration of the power system in the peak and valley phases, η i represents the proportion of flexible load of the i-th load under the demand response measures, Indicates that the peak period potential increases, Indicates the potential for increase during valley period.

[0109] S5: Optimize the access mode to the power grid based on the characteristic indicators of electric vehicle charging load and the theoretical potential of demand response.

[0110] Ways to optimize access to the grid include adjusting the charging strategy of aggregated electric vehicles based on charging load characteristic indicators and demand response potential.

[0111] Example 2, reference Figure 5 , which is an embodiment of the present invention, provides a system for an electric vehicle charging aggregation and regulation prediction method based on integrated clustering, characterized in that it includes a data acquisition module, a clustering and labeling integration module, an outlier sample analysis module and a grid access optimization module.

[0112] The data acquisition module is responsible for collecting the charging demand data of electric vehicles;

[0113] The clustering and labeling integration module is responsible for integrating clusters of different categories using relabeling and projection methods;

[0114] The outlier sample analysis module is responsible for analyzing charging demand data using a sliding time window;

[0115] The grid access optimization module is responsible for optimizing the grid access mode based on charging load characteristic indicators and demand response potential.

[0116] 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 method 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, etc., which can store program code.

[0117] 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.

[0118] 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.

[0119] 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.

[0120] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the 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. Electric vehicle charging aggregation and regulation prediction method based on integrated clustering, characterized in that: include: Collect the charging demand data of electric vehicles, classify and cluster them, and use the relabeling method and projection method to integrate clusters of different categories; The sliding time window method is used to analyze charging demand data and filter outlier samples; Establish an electric vehicle charging load characteristic index model to predict the charging load curve of electric vehicles in the future time period; Assess the theoretical potential of electric vehicles for demand response; Optimize the access mode to the power grid based on the characteristic indicators of electric vehicle charging load and the theoretical potential of demand response.

2. The electric vehicle charging aggregation and regulation prediction method based on integrated clustering as claimed in claim 1, characterized in that: The re-labeling method includes defining basic clusters ψ1, ψ2, ψ γ , expand the clustering according to the optimal clustering number N, and obtain the clustering results where i=1,2,..γ,α1,β2,..λ N It is clustering; Relabel the clusters in each clustering result to obtain the same label, and unify the results into Create a computational matrix for shared samples, expressed as, Among them, μ represents the number of references, ν represents the number to be marked, and c and d represent the current clusters; Determine similar cluster pairs, expressed as, (u',v')=argmax (u,v)∈I ο(u,v) Where I = {(μ, ν)|1≤μ, ν≤N}, (u', v') represents the cluster pair after optimal relabeling; Integrate similar clustering and convert S′1, S′2, S′ γ The N categories of similar clusters in are integrated into the set η1, η2, ....η N , expressed as, 3. The electric vehicle charging aggregation and regulation prediction method based on integrated clustering as claimed in claim 2, characterized in that: The projection method includes superimposing the visualized cluster point graph, vertically projecting to obtain the final result, removing duplicate samples, and superimposing similar clusters η1, η2, ....η N The combined correction is η′1, η′2, ....η′ N , analyze and screen controversial samples, and divide the controversial samples into multiple clusters.

4. The electric vehicle charging aggregation and regulation prediction method based on integrated clustering as claimed in claim 3 is characterized by: The sliding time window method includes analyzing the standard deviation of the charging demand data of the electric vehicle, expressed as, Among them, x' i Indicates the current sample standard value, represents the sample mean; Use a sliding window to calculate the fluctuation of adjacent samples and calculate the standard deviation of each window. If it is greater than ω, the outlier samples are identified and removed.

5. The electric vehicle charging aggregation and regulation prediction method based on integrated clustering as claimed in claim 4 is characterized by: The electric vehicle charging load characteristic indicators include daily average load A1, daily load rate A2, daily peak-to-valley difference rate A3, peak load rate A4, normal load rate A5, and valley load rate A6, as shown below: A4=P av.peak / P av A5=P av.sh / P av A6=P av.val / P av Among them, P sum Represents total power, P max , P min Represent the maximum and minimum power, P av.peak , P av.sh , P av.val Represents the average power during peak, middle and valley periods, P av Represents the average power for the whole day.

6. The electric vehicle charging aggregation and regulation prediction method based on integrated clustering as claimed in claim 5, characterized in that: The theoretical potential of EVs for demand response is evaluated by predicting load based on a typical daily load curve. Calculate the load value of the power system during peak and valley periods and theoretical potential w i , expressed as, Where h represents the peak and valley time of the power system, P i,max and P i,max are the predicted maximum and minimum load values ​​of the i-th type of load, τ represents the duration of the power system in the peak and valley phases, η i represents the proportion of flexible load of the i-th type of load under demand response measures, Indicates that the peak period potential increases, Indicates the potential for increase during valley period.

7. The electric vehicle charging aggregation and regulation prediction method based on integrated clustering as claimed in claim 6, characterized in that: The method of optimizing access to the power grid includes adjusting the charging strategy of aggregated electric vehicles based on charging load characteristic indicators and demand response potential.

8. A system using the electric vehicle charging aggregation and regulation prediction method based on integrated clustering as claimed in any one of claims 1 to 7, characterized in that: It includes data acquisition module, clustering and labeling integration module, outlier sample analysis module and grid access optimization module; The data acquisition module is responsible for collecting charging demand data of electric vehicles; The clustering and labeling integration module is responsible for integrating clusters of different categories using a relabeling method and a projection method; The outlier sample analysis module is responsible for analyzing charging demand data using a sliding time window; The grid access optimization module is responsible for optimizing the grid access mode based on charging load characteristic indicators and demand response potential.

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 electric vehicle charging aggregation and regulation prediction method based on integrated clustering 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 electric vehicle charging aggregation and regulation prediction method based on integrated clustering as described in any one of claims 1 to 7 are implemented.

Images