Electric vehicle cluster distributed charging control method and device based on Gaussian mixture

By calculating the charging behavior and capacity margin of the electric vehicle cluster based on hybrid Gaussian method, determining the charging start time, solving the problem of grid instability caused by disorderly charging of the electric vehicle group, and achieving orderly charging of the electric vehicle and grid load balance , reduces the risk of transformer overload.

CN119078550APending Publication Date: 2024-12-06STATE GRID ZHEJIANG ELECTRIC POWER COMPANY TAIZHOU POWER SUPPLY
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411277622.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-09-12
Publication Date
2024-12-06

AI Technical Summary

Technical Problem

The prior art has failed to effectively solve the grid voltage offset and transformer overload caused by disorderly charging of electric vehicles, resulting in unstable grid operation and increased charging costs.

Method used

The distributed charging control method for electric vehicle clusters based on hybrid Gaussian is used to calculate the cluster baseline and capacity margin, and the charging start time probability of the electric vehicle at each period is determined, and the start charging command is generated to realize orderly charging of the electric vehicle.

Benefits of technology

It significantly reduces the probability of transformer overload, improves the charging control efficiency of electric vehicle clusters, ensures grid load balance, and improves grid operation efficiency.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119078550A_ABST
    Figure CN119078550A_ABST
Patent Text Reader

Abstract

The invention discloses an electric vehicle cluster distributed charging control method and device based on Gaussian mixture, and relates to the technical field of electric vehicle cluster charging control, and the method comprises the steps: S1, calculating a cluster baseline based on an electric vehicle group charging behavior, and calculating a capacity margin; s2, acquiring the charging demand of the electric vehicle in response to the capacity margin; s3, calculating a first starting time probability of obtaining the charging starting authority of the electric vehicle in each time period according to the charging demand and the capacity margin of the electric vehicle; s4, the first starting time probability is sampled, the charging starting time of the electric vehicle is determined, and a charging starting instruction is generated; s5, executing an electric vehicle charging decision based on the charging starting instruction, and completing electric vehicle cluster charging control; the problem that the transformer overload is uncontrollable due to the lack of comprehensive consideration of the electric vehicle group base line and calculation of the time probability of starting charging in each time period in the prior art is solved, and the distributed charging control efficiency of the electric vehicle cluster is remarkably improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of electric vehicle cluster charging control, and in particular to a method and device for distributed charging control of electric vehicle clusters based on mixed Gaussian. Background Art

[0002] With the mass production and use of electric vehicles, we are facing more severe challenges. For example, after 6 o'clock in the evening, it is the peak period of power supply, and electric car owners need to charge their electric cars after get off work. Therefore, a large number of electric cars are randomly connected to the power grid as loads, which is bound to cause the phenomenon of "peak on peak". However, because the battery characteristics, usage methods, charging types and charging habits of electric car owners are different, these electric cars are usually randomly connected to the power grid for disorderly charging. This will inevitably increase the voltage deviation of the power grid, affect the power quality of the power grid, and then affect the economy of the power grid and the charging costs of users.

[0003] Disordered charging of large-scale electric vehicles will have an impact on the maximum load and the peak-to-valley difference of load. Therefore, orderly charging management of electric vehicles is of great significance to reducing the negative impact of disordered charging of electric vehicles and improving the stable operation of the distribution network. However, in the prior art, the characteristics of charging behavior of electric vehicles are usually ignored. For example, the conventional load is predicted based on the historical load of a typical day in a residential area. Since the historical load of a typical day in a residential area is uncontrollable, the prediction result is unreliable and inaccurate, which in turn leads to the peak load still being uncontrollable when electric vehicles are connected to the residential area for charging, and it cannot ensure that the probability of transformer overload is effectively reduced.

[0004] Chinese patent, publication number: CN118163658A, publication date: June 11, 2024, discloses a method for controlling the orderly charging of electric vehicles in residential areas. According to the historical load of the residential area on a typical day, the conventional load of the residential area on that day is predicted; according to the predicted conventional load of the residential area and the power supply capacity of the community distribution transformer, the total power of the working charging piles in each time period is calculated; the charging demand information of electric vehicle users is collected by the electric vehicle charging management system; according to the predicted conventional load data of the residential area, the total power of the working charging piles, and the charging demand of the user, the optimal charging power of each electric vehicle in each time period is solved based on the pre-set control strategy, so as to obtain the optimal charging plan for the electric vehicle; according to the solved charging plan, the output power of the charging piles in each time period is controlled, so as to control the orderly charging of the vehicle. This invention sacrifices the charging time of some electric vehicle users and the continuity of electric vehicle charging, but according to the historical load of the typical day of the residential area, the method has uncertainty in actual application, resulting in unreliable prediction results, which further makes the applicability of the control method low and has certain limitations. Summary of the invention

[0005] The purpose of the present invention is to address the problem that the prior art lacks comprehensive consideration of the baseline of an electric vehicle group and the probability of calculating the time when charging can start in each time period, resulting in uncontrollable transformer overload when the electric vehicle is connected to a residential area for charging; a distributed charging control method and device for an electric vehicle cluster based on a mixed Gaussian is proposed, by calculating the cluster baseline according to the charging behavior of the electric vehicle group, and calculating the capacity margin; the charging demand of the electric vehicle is collected through the capacity margin, the probability of the first start time that the electric vehicle obtains the right to start charging in each time period is calculated, sampling is performed to determine the charging start time, and a start charging instruction is generated to make a charging decision for the electric vehicle; the problem that the prior art does not comprehensively consider the baseline of the electric vehicle group and calculate the probability of the time when charging can start in each time period, resulting in uncontrollable and difficult to predict transformer overload when the electric vehicle is connected to a residential area for charging is overcome, the distributed charging control efficiency of the electric vehicle cluster is significantly improved, and the probability of transformer overload is effectively reduced.

[0006] To solve the above technical problems, according to a first aspect of an embodiment of the present invention, a method for controlling distributed charging of electric vehicle clusters based on mixed Gaussian is provided, comprising the following steps: S1. Calculate the cluster baseline based on the charging behavior of the electric vehicle group and calculate the capacity margin; S2. collecting charging demand of electric vehicles in response to the capacity margin; S3, calculating the first start time probability of the electric vehicle obtaining the charging start permission in each time period according to the electric vehicle charging demand and the capacity margin; S4, sampling the first start time probability, determining the charging start time of the electric vehicle, and generating a charging start instruction; S5. Execute the electric vehicle charging decision based on the start charging instruction to complete the electric vehicle cluster charging control.

[0007] In this scheme, by calculating the cluster baseline according to the charging behavior of the electric vehicle group, the uncontrollable electric vehicle load can be accurately and reasonably modeled and estimated, providing basic data support for the effective scheduling of controllable electric vehicles; and it is helpful to calculate the electric vehicle load that can be connected to the transformer without exceeding the transformer capacity when the electric vehicle is connected to the transformer for charging, so as to obtain the conservative charging margin of the charging vehicle and ensure the safe and stable operation of the transformer during peak hours; at the same time, by calculating the capacity margin, it provides a basis for calculating the probability that the electric vehicle can start charging in each time period, that is, calculating the probability of the first start time that the electric vehicle obtains the right to start charging in each time period, and can obtain accurate and reliable start charging time, so as to realize orderly charging control of electric vehicles and balance the power grid load, thereby avoiding transformer overload, improving charging efficiency, and improving the overall operation of the power grid, effectively solving the problem of distributed charging control of electric vehicle clusters in residential areas, and significantly reducing the probability of charging out of control of electric vehicle groups.

[0008] Preferably, before executing S1, the step includes: Construct a charging behavior record model for electric vehicles and obtain the optimal charging data component in the model; The specific steps include: Obtain the historical charging data set of electric vehicles, simulate the joint distribution of the first charging behavior of the electric vehicle group based on the Gaussian mixture model, and establish a charging behavior record model; The components in the charging behavior record model are evaluated to determine the optimal component group in the model, and to obtain the optimal charging data components of the electric vehicle cluster.

[0009] In this scheme, the Gaussian mixture model can be used to simulate the joint distribution of uncontrollable electric vehicle charging behavior to accurately quantify multiple random variables in the charging behavior of charging vehicles, including charging time, charging power, etc. It can effectively represent the correlation and mutual influence between different variables, and build a more comprehensive and accurate charging behavior recording model.

[0010] Preferably, in said S1, calculating the cluster baseline based on the charging behavior of the electric vehicle group includes the following sub-steps: generating a second charging behavior of the electric vehicle through the charging behavior record model, and obtaining an optimal charging data component of the second charging behavior; Calculating the charging power of any electric vehicle in the electric vehicle cluster based on the optimal charging data component; The cluster baseline is calculated based on the charging power.

[0011] Preferably, in S1, calculating the capacity margin includes the following sub-steps: Based on the cluster baseline combined with the predicted load of the user's power load, the load baseline of the electric vehicle connected to the transformer for charging is predicted; and the capacity margin is calculated based on the load baseline.

[0012] Preferably, before executing S3, the step includes: Determine a load feasible region representing an optional charging start time for the electric vehicle based on the charging power of the electric vehicle in combination with the capacity margin; According to the electric vehicle charging scenario, constraint conditions of the load feasible area are added, the load feasible area is optimized, and the continuous charging time of the electric vehicle is obtained.

[0013] Preferably, the constraint conditions include at least: user charging demand constraint and distributed scenario constraint.

[0014] Preferably, S3 includes the following sub-steps: Calculating a second charging probability that the electric vehicle is currently being charged based on the demand information for charging the electric vehicle and combining the capacity margin; Extracting the charging duration of the electric vehicle based on the load feasible area and constructing the electric vehicle charging coefficient matrix; The first start time probability of the electric vehicle starting to charge in each time period is calculated based on the second charging probability and the electric vehicle charging coefficient matrix.

[0015] Preferably, S4 includes the following sub-steps: Normalizing and binning the first start time probability to obtain a first start time distribution; Using a sampling method to sample the first start time distribution to obtain a first start time series; Randomly sample from the first start time sequence and determine the charging start time; A charge start instruction is generated based on the determined charge start time, and the charge start instruction is sent to the electric vehicle.

[0016] According to another aspect of an embodiment of the present invention, there is provided a hybrid Gaussian-based distributed charging control device for electric vehicle clusters, comprising: a central controller, a data concentrator, and a charging actuator; The central controller is installed in the power distribution area where each transformer is located, and is used to regularly read the data from the data concentrator terminal, calculate the cluster baseline and capacity margin according to the data from the data concentrator terminal, calculate the first start time probability of the electric vehicle obtaining the right to start charging in each time period, generate a start charging instruction, and send the start charging instruction to the charging execution mechanism; The data concentrator is used to collect a first historical charging data set, send it to the central controller, and broadcast the capacity margin to electric vehicle users; The charging execution mechanism is used to execute the charging decision according to the start charging instruction; The charging execution mechanism is built with an information interaction module for performing data interaction between the charging execution mechanism and the electric vehicle, including: receiving the charging demand set by the user on the electric vehicle.

[0017] Preferably, it also includes an edge calculator for calculating the start time of charging of the electric vehicle according to the charging demand of the electric vehicle and making charging decisions; The edge calculator is installed in the data concentrator or the charging actuator.

[0018] In this solution, the central controller reads the historical charging data of electric vehicles and obtains the charging demand of charging vehicles, so as to calculate the cluster baseline and capacity margin of charging vehicles, and provide basic data support for the effective scheduling of controllable electric vehicles; the data concentrator collects the historical charging data of electric vehicles, which helps to obtain the second charging behavior of electric vehicles (i.e. the charging behavior of electric vehicles on the second day), and then obtains the charging power of the electric vehicles to calculate the cluster baseline, so as to achieve accurate and reasonable modeling and estimation of uncontrollable electric vehicle loads; by building an edge calculator in the charging actuator or data concentrator, the charging start time of electric vehicles can be calculated according to the charging demand of electric vehicles, and charging decisions can be made to achieve orderly control of the charging of distributed clusters of electric vehicles, effectively solving the problem of distributed charging control of electric vehicle clusters in residential areas, and significantly reducing the probability of out-of-control charging of electric vehicle clusters.

[0019] Beneficial effects of the present invention: 1. By calculating the cluster baseline based on the charging behavior of the electric vehicle group, accurate and reasonable modeling and estimation of the uncontrollable electric vehicle load are achieved, which provides a reference for subsequent charging decisions of subsequent users and provides basic data support for the effective scheduling of controllable electric vehicles; 2. The probability that electric vehicles can start charging in each time period is calculated based on the charging demand and capacity margin of electric vehicles, the charging start time is determined by sampling, and the charging decision is executed based on the start charging instruction, which effectively solves the distributed charging control problem of electric vehicle clusters in residential areas and significantly reduces the probability of charging out of control of electric vehicle groups. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Other features, objects and advantages of the present invention will become more apparent by reading the detailed description of non-limiting embodiments made with reference to the following drawings. The drawings are only for the purpose of illustrating preferred embodiments and are not to be considered as limiting the present invention. Also, the same reference symbols are used throughout the drawings to represent the same parts.

[0021] Figure 1 The present invention is a flow chart of a method for controlling distributed charging of electric vehicle clusters based on mixed Gaussian charging.

[0022] Figure 2 The present invention is a schematic diagram of the structure of a distributed charging control device for electric vehicle clusters based on mixed Gaussian according to an embodiment of the present invention. DETAILED DESCRIPTION

[0023] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific implementation method described herein is only an optimal embodiment of the present invention, which is only used to explain the present invention and does not limit the scope of protection of the present invention. All other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0024] Example 1: Figure 1 As shown, a distributed charging control method for electric vehicle clusters based on mixed Gaussian includes steps S1-S5, wherein: S1. Calculate the cluster baseline based on the charging behavior of the electric vehicle group and calculate the capacity margin.

[0025] Specifically, before executing S1, it includes: Construct a charging behavior record model for electric vehicles and obtain the optimal charging data component in the model.

[0026] The specific steps include: Obtain the historical charging data set of electric vehicles, simulate the joint distribution of the first charging behavior of the electric vehicle group based on the Gaussian mixture model, and establish a charging behavior record model; The components in the charging behavior record model are evaluated, the optimal component group in the model is determined, and the optimal charging data components of the electric vehicle cluster are obtained.

[0027] Specifically, in S1, the cluster baseline is calculated based on the charging behavior of the electric vehicle group, including the following sub-steps: Generate a second charging behavior of the electric vehicle through a charging behavior record model, and obtain an optimal charging data component of the second charging behavior; Calculate the charging power of any electric vehicle in the electric vehicle cluster based on the optimal charging data component; The cluster baseline is calculated based on the charging power.

[0028] Specifically, in S1, calculating the capacity margin includes the following sub-steps: Based on the cluster baseline and the predicted load of the user's electricity load, the load baseline for electric vehicles connected to the transformer for charging is predicted; The capacity margin is calculated based on the load baseline.

[0029] As an implementation method, the start time, duration, and target charging energy that can be used to fully describe the charging behavior record are selected to simulate the joint distribution of the charging behavior, as follows: Consider that there are N1 behavior records in the historical charging data set Φ, and the i-th charging behavior record φ i It is expressed as: Among them, α i represents the charging start time in the i-th record, β i represents the charging duration in the i-th record, e i represents the target charging energy (kWh) in the i-th record, H represents the number of all time periods of the behavior record, and R is a real number set; according to formula (1), the charging behavior record φ i The probability p(φ i ) The charging behavior record model is simulated by using the ternary GMMψ (Gaussian mixture model), and the charging behavior record model ψ is expressed as follows: Where K represents the number of groups of the charging behavior recording model ψ, k represents the Gaussian component in the group, μ k / Σ k represents the mean and covariance matrix of Gaussian component k, N(·) represents Gaussian distribution, π k Represents the mixing ratio of Gaussian component k; The variables in formula (2) are estimated to determine the optimal number of components in the charging behavior record model, that is, the optimal charging data component; After the charging behavior record model is constructed, the electric vehicle group baseline (here the cluster baseline) is calculated based on the average load of the X highest consumption days among Y similar days in the history of electric vehicle charging, that is, by generating The data records are used to generate the electric vehicle charging behavior of the next day (the second charging behavior). The optimal charging data component is obtained from the data set composed of pieces of data to calculate the charging power of the jth electric vehicle Charging power It is expressed as follows: Among them, α j represents the charging start time of the jth electric vehicle, β j represents the charging duration of the jth electric vehicle, e j represents the target charging energy (kWh) of the jth electric vehicle, and h represents the index of all equal time periods in a day; Based on the cluster baseline of the electric vehicle population it is possible to calculate: Furthermore, the baseline load T of the electric vehicle group one day ago can be estimated according to equations (3) and (4): EV , and then according to T EV Estimate the charging load baseline of the electric vehicle group one day ago to predict the load baseline T of the electric vehicle connected to the transformer for charging load , thereby estimating the capacity margin Z.

[0030] Furthermore, the capacity margin Z is expressed as follows: Among them, h represents the day-ahead capacity margin in period h, and C1 represents the maximum capacity of the transformer.

[0031] As an optional implementation, when determining the optimal component group in the model and obtaining the optimal charging data component of the electric vehicle cluster, the maximum expectation value method can be used to estimate the parameters of the charging behavior record model; in addition, the information criterion (a standard for the goodness of statistical model fit), the Bayesian information criterion or the negative log-likelihood loss function can be used to quickly calculate the charging behavior record model and accurately determine the optimal charging data component.

[0032] S2. Collecting electric vehicle charging demand in response to capacity margin.

[0033] Specifically, before executing S3, the following steps are included: Determine a load feasible region representing an optional charging start time for the electric vehicle based on the charging power of the electric vehicle combined with the capacity margin; According to the electric vehicle charging scenario, constraints on the load feasible area are added, the load feasible area is optimized, and the continuous charging time of the electric vehicle is obtained.

[0034] Specifically, the constraints include at least: user charging demand constraints and distributed scenario constraints.

[0035] It can be understood that the feasible area of ​​EV charging control consists of charging start time, duration and charging power, where y is used to represent the charging scheduling result of the feasible area, and y is expressed as: Among them, α′ represents the charging start time of the charging scheduling result, β′ represents the charging duration, It represents the charging power of a controllable electric vehicle in the h period.

[0036] Furthermore, each time the user finishes using the electric vehicle and arrives at the parking lot, he or she needs to input the demand information for the next use of the electric vehicle, including the energy demand E d and departure time T d , therefore, the user charging demand constraint is: α'≥T arr (7) β'+α'-1≤T d (8) Among them, T arr represents the arrival time of the electric vehicle, E d represents the energy demand of the electric vehicle after arrival, T d represents the departure time of the electric vehicle, Q cha represents the charging power set of controllable electric vehicles in all planned time periods; In order to ensure high charging efficiency of electric vehicles and reduce the complexity of decision-making in non-communication distributed charging scenarios, the distributed scenario constraints are as follows: Among them, Q r represents the fixed charging power of the electric vehicle; Based on equations (9) and (11), the charging duration β′ of the electric vehicle after arrival can be calculated: According to the above constraints, the feasible region for controlling the charging of electric vehicles can be simplified so that the feasible region only includes the charging start time.

[0037] In this embodiment, the continuity of the electric vehicle charging power is ensured through distributed scenario constraints, the complexity of the feasible area is reduced, and the high efficiency of the electric vehicle charging process is ensured.

[0038] S3. Calculate the probability of the first start time for the electric vehicle to obtain the charging start permission in each time period according to the charging demand and capacity margin of the electric vehicle.

[0039] Specifically, S3 includes the following sub-steps: According to the demand information of the electric vehicle charging, the second charging probability that the electric vehicle is currently charging is calculated in combination with the capacity margin; Extract the charging duration of electric vehicles based on the feasible load area and construct the electric vehicle charging coefficient matrix; The first starting time probability of the electric vehicle starting to charge in each time period is calculated based on the second charging probability and the electric vehicle charging coefficient matrix.

[0040] S4. Sampling the first start time probability, determining the charging start time of the electric vehicle, and generating a charging start instruction.

[0041] Specifically, S4 includes the following sub-steps: Normalizing and intervalizing the first start time probability to obtain the first start time distribution; The first start time distribution is sampled by a sampling method to obtain a first start time series; Randomly sampling from the first start time sequence and determining a charging start time; A charge start instruction is generated based on the determined charge start time, and the charge start instruction is sent to the electric vehicle.

[0042] S5. Execute the electric vehicle charging decision based on the start charging instruction to complete the electric vehicle cluster charging control.

[0043] As an implementation method, by determining the day-ahead load and capacity margin forecast results when the electric vehicle arrives, the feasible area, that is, the optional charging start time of the electric vehicle, can be determined. Since it is difficult for a single electric vehicle to obtain global demand information and transformer load, probability distribution sampling can be used instead of online optimization to determine the control command. The probability P of the electric vehicle being charged cha (here is the second charging probability) can be expressed and estimated by the capacity margin Z, which is expressed as follows: Among them, P cha represents the probability set that the electric vehicle is charging, P h cha represents the probability that the electric vehicle is charging in the h period; based on the second charging probability combined with the coefficient matrix Γ1 of whether the electric vehicle is charging (here is the electric vehicle charging coefficient matrix), the probability of starting charging in each period can be obtained: Γ1P st =P cha (15) Among them, P st represents the probability set of the electric vehicle starting charging (the first start time probability), P h st represents the probability that the electric vehicle starts charging during period h, γ a,b Represents the element located in row a and column b in the coefficient matrix Γ1.

[0044] Furthermore, for each electric vehicle, the probability distribution P of starting charging in each period is obtained. st After that, the start time that meets the distribution can be randomly sampled (the first start time distribution). The specific steps are as follows: For the probability distribution P st After normalization and interval processing, we get the distribution P st,nar (First start time distribution), expressed as follows: Among them, P h st,nar Indicates P in h period st,nar The probability of A sampling method, for example, acceptance-rejection sampling method is used to obtain a sequence A′ (first starting time series) that obeys the distribution; wherein, for the variable group that meets formula (19) if Then randomly sample N s records, and obtain the s A sequence of data items A′; Among them, y α″ represents the probability of random sampling, α″ represents the time period of random sampling, U(·) represents uniform distribution, Indicates the charging start time α′ i The probability in .

[0045] As an optional implementation, equations (15)-(17) can be solved using the least squares method with constraints, and the objective function to be solved is min||P cha -Γ1P st || 2 (20) Example 2: Figure 2 As shown, a distributed charging control device for electric vehicle clusters based on mixed Gaussian includes: a central controller, a data concentrator and a charging actuator; The central controller is installed in the distribution area where each transformer is located, and is used to regularly read the data concentrator terminal data, calculate the cluster baseline and capacity margin based on the data concentrator terminal data, calculate the first start time probability of electric vehicles obtaining the right to start charging in each time period, generate a start charging instruction, and send the start charging instruction to the charging execution mechanism; The data concentrator is used to collect a first historical charging data set, send it to the central controller, and broadcast the capacity margin to electric vehicle users; The charging execution mechanism is used to execute the charging decision according to the charge start instruction; The charging actuator is equipped with an information interaction module for data interaction between the charging actuator and the electric vehicle, including receiving the charging requirements set by the user on the electric vehicle.

[0046] Specifically, it also includes an edge calculator for calculating the start time of electric vehicle charging according to the electric vehicle charging demand and making charging decisions; The edge computer is installed in the data concentrator or charging actuator.

[0047] As an implementation method, the central controller can be a cloud computing machine, the data concentrator can be integrated into a smart meter, and the charging actuator can be integrated into a smart charging pile; the central controller serves as the connection point between the smart meter terminal and the power grid, and its basic functions include regular reading of terminal data, cloud system command transmission, data communication, etc.; it is mainly responsible for storing the historical charging behavior of each electric vehicle, modeling the charging behavior of the electric vehicle, and then estimating the benchmark load of the electric vehicle group. In addition, the cloud computing machine is also responsible for predicting the resident load, and estimating the total benchmark load in combination with the benchmark load of the electric vehicle group, and determining the charging time and power of the electric vehicle; the smart meter function includes broadcasting the capacity margin of the residential transformer to users.

[0048] It is understandable that the function of traditional charging piles is to provide power to electric vehicles. The electric vehicles plugged in them are called uncontrollable electric vehicles. Compared with traditional charging piles, smart charging piles can be used to interact with electric vehicles, receive demand information set by users on electric vehicles, such as target charging energy and duration, and transmit it to the edge calculator to make charging decisions.

[0049] In this embodiment, the central controller reads the historical charging data of the electric vehicle and obtains the charging demand of the charging vehicle, so as to calculate the cluster baseline and capacity margin of the charging vehicle, and provide basic data support for the effective scheduling of controllable electric vehicles; the historical charging data of the electric vehicle is collected by the data concentrator, which is helpful to obtain the second charging behavior of the electric vehicle (i.e., the charging behavior of the electric vehicle on the second day), and then obtain the charging power of the electric vehicle to calculate the cluster baseline, so as to achieve accurate and reasonable modeling and estimation of the uncontrollable electric vehicle load; by building an edge calculator in the charging actuator or the data concentrator, the charging start time of the electric vehicle can be calculated according to the charging demand of the electric vehicle, and charging decisions can be made, so as to achieve orderly control of the charging of distributed clusters of electric vehicles, effectively solve the problem of distributed charging control of electric vehicle clusters in residential areas, and significantly reduce the probability of out-of-control charging of electric vehicle clusters.

[0050] The above specific embodiments are preferred embodiments of the present invention, and are not intended to limit the specific implementation scope of the present invention. The scope of the present invention includes but is not limited to the present specific embodiments. All equivalent changes made in accordance with the shape, structure, and method of the present invention are within the protection scope of the present invention.

Claims

1. A distributed charging control method for electric vehicle clusters based on mixed Gaussian, characterized in that: The steps include: S1. Calculate the cluster baseline based on the charging behavior of the electric vehicle group and calculate the capacity margin; S2. collecting charging demand of electric vehicles in response to the capacity margin; S3, calculating the first start time probability of the electric vehicle obtaining the charging start permission in each time period according to the electric vehicle charging demand and the capacity margin; S4, sampling the first start time probability, determining the charging start time of the electric vehicle, and generating a charging start instruction; S5. Execute the electric vehicle charging decision based on the start charging instruction to complete the electric vehicle cluster charging control.

2. The electric vehicle cluster distributed charging control method based on mixed Gaussian according to claim 1 is characterized by: Before executing S1, the steps include: Construct a charging behavior record model for electric vehicles and obtain the optimal charging data component in the model; The specific steps include: Obtain the historical charging data set of electric vehicles, simulate the joint distribution of the first charging behavior of the electric vehicle group based on the Gaussian mixture model, and establish a charging behavior record model; The components in the charging behavior record model are evaluated to determine the optimal component group in the model, and to obtain the optimal charging data components of the electric vehicle cluster.

3. The electric vehicle cluster distributed charging control method based on mixed Gaussian according to claim 2 is characterized by: In S1, the cluster baseline is calculated based on the charging behavior of the electric vehicle group, including the following sub-steps: Generate a second charging behavior of the electric vehicle through the charging behavior record model, and obtain an optimal charging data component of the second charging behavior; Calculating the charging power of any electric vehicle in the electric vehicle cluster based on the optimal charging data component; The cluster baseline is calculated based on the charging power.

4. The electric vehicle cluster distributed charging control method based on mixed Gaussian according to claim 1 is characterized by: In S1, calculating the capacity margin includes the following sub-steps: Based on the cluster baseline combined with the predicted load of the user's power load, the load baseline of the electric vehicle connected to the transformer for charging is predicted; and the capacity margin is calculated based on the load baseline.

5. The electric vehicle cluster distributed charging control method based on mixed Gaussian according to any one of claim 1, characterized in that: Before executing S3, include: Determine a load feasible region representing an optional charging start time for the electric vehicle based on the charging power of the electric vehicle in combination with the capacity margin; According to the electric vehicle charging scenario, constraint conditions of the load feasible area are added, the load feasible area is optimized, and the continuous charging time of the electric vehicle is obtained.

6. The electric vehicle cluster distributed charging control method based on mixed Gaussian according to claim 5 is characterized by: The constraint conditions include at least: user charging demand constraint and distributed scenario constraint.

7. The electric vehicle cluster distributed charging control method based on mixed Gaussian according to claim 5 is characterized by: The S3 comprises the following sub-steps: Calculating a second charging probability that the electric vehicle is currently being charged based on the demand information for charging the electric vehicle and combining the capacity margin; Extracting the charging duration of the electric vehicle based on the load feasible area and constructing the electric vehicle charging coefficient matrix; The first start time probability of the electric vehicle starting to charge in each time period is calculated based on the second charging probability and the electric vehicle charging coefficient matrix.

8. The electric vehicle cluster distributed charging control method based on mixed Gaussian according to claim 1 is characterized by: The S4 comprises the following sub-steps: Normalizing and binning the first start time probability to obtain a first start time distribution; Using a sampling method to sample the first start time distribution to obtain a first start time series; Randomly sample from the first start time sequence and determine the charging start time; A charge start instruction is generated based on the determined charge start time, and the charge start instruction is sent to the electric vehicle.

9. A distributed charging control device for electric vehicle clusters based on mixed Gaussian, applicable to a distributed charging control method for electric vehicle clusters based on mixed Gaussian as claimed in any one of claims 1 to 8, characterized in that: include: Central controller, data concentrator and charging actuator; The central controller is installed in the power distribution area where each transformer is located, and is used to regularly read the data from the data concentrator terminal, calculate the cluster baseline and capacity margin according to the data from the data concentrator terminal, calculate the first start time probability of the electric vehicle obtaining the right to start charging in each time period, generate a start charging instruction, and send the start charging instruction to the charging execution mechanism; The data concentrator is used to collect a first historical charging data set, send it to the central controller, and broadcast the capacity margin to electric vehicle users; The charging execution mechanism is used to execute the charging decision according to the start charging instruction; The charging execution mechanism is built with an information interaction module for performing data interaction between the charging execution mechanism and the electric vehicle, including: receiving the charging demand set by the user on the electric vehicle.

10. The electric vehicle cluster distributed charging control device based on mixed Gaussian according to claim 9, characterized in that: Also included is an edge calculator for calculating the start time of electric vehicle charging based on the electric vehicle charging demand and making charging decisions; The edge calculator is installed in the data concentrator or the charging actuator.