Charging pile adaptive charging planning method based on data mining

Abstract

The invention discloses a charging pile self-adaptive charging planning method based on data mining, and relates to the technical field of self-adaptive charging, and the method comprises the steps: carrying out the real-time data collection and load prediction, dynamically adjusting a charging strategy, intelligently distributing charging resources in a peak load period, limiting high-power charging, and avoiding the overload of a power grid; in the idle load period, the idle capacity of the power grid is fully utilized, the power grid load is balanced, and the stability and safety of the power grid are guaranteed; through behavior data analysis and user portrait construction, users are divided into a high-priority type and a low-priority type, the charging rate and resource allocation are flexibly adjusted according to the requirements of different users, the high-priority users can obtain faster charging services, the low-priority users can enjoy more economical charging options, resource waste is avoided, and the user experience is improved. And meanwhile, the overall satisfaction degree of the user is improved.

Description

Technical Field

[0001] The present invention relates to the field of adaptive charging technology, and in particular to a method for adaptive charging planning of charging piles based on data mining. Background Art

[0002] With the popularization of electric vehicles, the number and frequency of use of charging piles continue to increase, and the demand in urban areas is even greater. This concentrated growth in demand has also brought significant pressure to the power system, especially when the peak usage period of charging piles overlaps with the peak period of residential electricity consumption. The power grid will face serious load imbalance problems, which may cause local power grid overload.

[0003] Most of the current power grid load management methods use static load control strategies to deal with the problem of overlapping power consumption during peak periods. The control strategies based on preset rules or historical data cannot respond to the dynamic changes of charging demand and power grid load in real time, resulting in the difficulty of effective diversion of power grids during peak load periods, aggravating the pressure on local power grids. Some of them also adopt peak-shifting power consumption strategies, which seriously affect user experience. In addition, with the increase in the number of electric vehicles, the use scenarios of charging piles have become diversified, and users' charging needs have become more and more diverse. Some users may need fast charging, while some users may need cheap electricity. The existing charging pile charging methods are generally queuing strategies, which cannot be optimized and scheduled according to the specific needs of users, such as urgent charging or low-power long-term charging. There is a lack of priority management and it is impossible to effectively cope with the peak charging period. The charging piles used by users who are in urgent need of charging may be occupied by users who are charging at low power for a long time, resulting in waste of resources. Therefore, there is an urgent need for an adaptive charging planning method for charging piles based on data mining to solve such problems. Summary of the invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a charging pile adaptive charging planning method based on data mining to solve the problem that the control strategy based on preset rules or historical data for scheduling cannot respond to the dynamic changes of charging demand and grid load in real time; the charging pile strategy cannot be adjusted according to the specific needs of users, or the charging efficiency is low, or it causes waste of resources.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] The present invention provides a charging pile adaptive charging planning method based on data mining, which comprises:

[0008] Step S1, charging pile and power grid data collection, based on the charging pile and power grid data, conduct behavior data analysis on each user who uses the charging pile;

[0009] Step S2, load forecasting, forecasting the peak load period and idle load period of the power grid based on the historical real-time load of the power grid;

[0010] Step S3: targeted resource allocation, intelligently allocate charging resources based on the results of load forecasting.

[0011] Furthermore, the charging pile and power grid data collected in step S1 include: the current use status of the charging pile, output power, information of the vehicle being charged, and real-time load of the power grid.

[0012] Furthermore, the behavior data analysis in step S1 includes: the user's historical charging behavior, priority requirements, and travel habits, thereby constructing a user portrait and dividing the users into high-priority users and low-priority users.

[0013] Furthermore, in step S1, the user division method is:

[0014] Assume the charging frequency of user i is f i The average charging time is The average charging power is The charging time period preference is τ i and price sensitivity is γ i ;

[0015] Calculate the comprehensive priority score S for user i i :

[0016] Among them, w1, w2, w3, w4, w5 represent the weights of the corresponding features;

[0017] Based on the comprehensive priority score S i , use Kmeans clustering algorithm to divide users into high priority and low priority categories, clustering objective function:

[0018] Among them, i represents the i-th cluster, x j is the user's feature vector, i.e. cluster C i The jth data point in i is the cluster center, C i is the i-th cluster, k is the number of clusters, here it is 2, where,

[0019] According to the clustering results, users are divided into high-priority users and low-priority users.

[0020] Furthermore, in the user partitioning and clustering objective function of step S1:

[0021] Charging time period preference τ iIndicates the charging behavior tendency of user i at different time periods of the day:

[0022] Divide a day into six time periods: T1:00:00-04:00, T2:04:00-08:00, T3:08:00-12:00, T4:12:00-16:00, T5:16:00-20:00, T6:20:00-00:00, and calculate the number of times n that user i charges in each time period. i,k , where k represents the time period (16);

[0023] Normalize the number of charges in each time period: Among them, p i,k Indicates that user i is in time period T k The charging preference weight of

[0024] but Among them, T k is the center time of the corresponding time period (e.g. the center time of T1 is 02:00, which is expressed as 2 hours).

[0025] Furthermore, in the user partitioning and clustering objective function of step S1:

[0026] Price Sensitivity i Indicates the user's sensitivity to charging prices:

[0027] Get the price C of each charge for user i i,j And the corresponding charging power P i,j , where j represents the jth charge, and the unit power cost of each charge is calculated as:

[0028] Normalize the unit power cost of each user: Among them, min(u i ) and max(u i ) are the minimum and maximum values ​​of the unit power cost of user i, is the normalized unit power cost;

[0029] but Where N is the number of times user i charges.

[0030] Furthermore, the load prediction method in step S2 is:

[0031] The real-time load data L(t) of the past T time points are extracted from the power grid, where t represents the time point. The data set is {L(1), L(2), …, L(T)}, denoised, and standardized: in,

[0032] represents the normalized load, L(t) represents the original grid load at time t, μ L is the mean of the historical load data L(t), indicating the average level of grid load, σ L It represents the standard deviation of the historical load data L(t), which is used to measure the amplitude of load fluctuation.

[0033] Furthermore, the load prediction method in step S2 also includes:

[0034] Use the ARIMA model to model the normalized load time:

[0035] in, represents the normalized load at time t, i.e., the prediction target, φ1, φ2, …, φ p Represents the parameters of the autoregressive AR model, that is, the impact of the load values ​​at the past p time points on the current load, represents the normalized load at time ti, i.e., the historical load, θ1,θ2,…,θ q represents the parameters of the moving average MA model, that is, the impact of the error term at the past q time points on the current load, ∈(ti) represents the random error term at time ti, and ∈(t) represents the error term at time t;

[0037] Estimate the parameter φ of the ARIMA model by maximum likelihood estimation MLE i and θ j :

[0038] in, represents the load value predicted by the model after given parameters φ and θ, represents the model parameter estimate obtained by minimizing the sum of squared forecast errors, and T represents the length of the historical time series;

[0039] Load forecasting using the estimated ARIMA model:

[0040] in, represents the load forecast at time t+h, represents the normalized load forecast, μ L represents the mean of historical load data, σ L represents the standard deviation of historical load data, h represents the prediction step length, and after denormalization,

[0041] Furthermore, the load prediction method in step S2 also includes:

[0042] For the predicted load sequence To perform smoothing:

[0043] in, represents the smoothed predicted load value, N represents the smoothing window size, and i represents the index of the time step;

[0044] Set the load peak and idle thresholds, and divide the load peak and idle based on the predicted load.

[0045] Furthermore, the method of intelligently allocating charging resources in step S3 is:

[0046] Add two additional options, high-power charging rate and low-power charging rate, to the existing fast charging and regular charging options of charging piles, for a total of four options. The low-power charging rate option is enabled in any scenario.

[0047] If the load forecast result is a peak load period, conventional charging is allocated to all users. On this basis, fast charging and high-power charging rate options are enabled according to user priority levels, and the allocation of charging resources does not exceed the grid load threshold;

[0048] The existing charging piles have fast charging and regular charging options. The high-priority user allocation here provides a more expensive and faster charging rate based on the original fast charging option in the existing solution. The low-priority user allocation provides a lower-priced low-power charging rate based on the original regular charging option in the existing solution. The high and low-power charging rates here are relative and are set according to the actual user charging speed, charging price, and grid load.

[0049] If the load forecast result is an idle load period, conventional charging will be allocated to all users, and all options will be enabled on this basis; regardless of user priority, the allocated charging resources will not exceed the grid load threshold.

[0050] The beneficial effects of the present invention are:

[0051] The present invention dynamically adjusts the charging strategy through real-time data collection and load prediction, intelligently allocates charging resources during peak load periods, limits high-power charging, and avoids grid overload; during idle load periods, fully utilizes the idle capacity of the grid, balances the grid load, and ensures the stability and security of the grid.

[0052] The present invention divides users into high-priority and low-priority categories through behavioral data analysis and user portrait construction, and flexibly adjusts the charging rate and resource allocation according to the needs of different users. High-priority users can obtain faster charging services, while low-priority users can enjoy more economical charging options, avoiding resource waste and improving the overall user satisfaction.

[0053] The present invention introduces high-power and low-power charging rate options and provides four different charging modes to meet the needs of different users in different scenarios. During peak load periods of the power grid, the system limits high-power charging, while all options are open during idle periods, which not only ensures the safety of the power grid, but also improves the economy and flexibility of the charging process.

[0054] The present invention distributes conventional charging during peak load periods of the power grid through accurate load forecasting and intelligent scheduling to avoid overloading of the power grid. At the same time, it encourages users to charge during idle load periods to reduce the pressure on the power grid during peak periods. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] 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.

[0056] Figure 1 It is a schematic diagram of the process flow of the charging pile adaptive charging planning method of the present invention. DETAILED DESCRIPTION

[0057] In order to make the above-mentioned objects, 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 accompanying drawings.

[0058] 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.

[0059] 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 does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0060] Example 1, reference Figure 1 This embodiment provides a charging pile adaptive charging planning method based on data mining, comprising the following steps:

[0061] Step S1, charging pile and power grid data collection, based on the charging pile and power grid data, conduct behavioral data analysis on each user who uses the charging pile; the charging pile and power grid data include: the current use status of the charging pile, output power, information of the vehicle being charged, and real-time load of the power grid; the behavioral data analysis content includes: the user's historical charging behavior, priority requirements, travel habits, and thus construct a user profile, and divide the users into high-priority users and low-priority users;

[0062] The user division method is:

[0063] Assume the charging frequency of user i is f i The average charging time is The average charging power is The charging time period preference is τ i and price sensitivity is γ i ;

[0064] Calculate the comprehensive priority score S for user i i :

[0065] Among them, w1, w2, w3, w4, w5 represent the weights of the corresponding features;

[0066] Based on the comprehensive priority score S i , use Kmeans clustering algorithm to divide users into high priority and low priority categories, clustering objective function:

[0067] Among them, i represents the i-th cluster, x j is the user's feature vector, i.e. cluster C i The jth data point in i is the cluster center, C i is the i-th cluster, k is the number of clusters, here it is 2, where,

[0068] According to the clustering results, users are divided into high-priority users and low-priority users;

[0069] In the user partitioning and clustering objective function of step S1:

[0070] Charging time period preference τ i Indicates the charging behavior tendency of user i at different time periods of the day:

[0071] Divide a day into six time periods: T1:00:00-04:00, T2:04:00-08:00, T3:08:00-12:00, T4:12:00-16:00, T5:16:00-20:00, T6:20:00-00:00, and calculate the number of times n that user i charges in each time period. i,k , where k represents the time period (16);

[0072] Normalize the number of charges in each time period: Among them, p i,k Indicates that user i is in time period T k The charging preference weight of

[0073] but Among them, T k is the central time of the corresponding time period (e.g. the central time of T1 is 02:00, which is expressed as 2 hours);

[0074] In the user partitioning and clustering objective function of step S1:

[0075] Price Sensitivity i Indicates the user's sensitivity to charging prices:

[0076] Get the price C of each charge for user i i,j And the corresponding charging power P i,j , where j represents the jth charge, and the unit power cost of each charge is calculated as:

[0077] Normalize the unit power cost of each user: Among them, min(u i ) and max(u i ) are the minimum and maximum values ​​of the unit power cost of user i, is the normalized unit power cost;

[0078] but Where N is the number of times user i charges;

[0079] Specifically, through detailed analysis of user behavior, high-priority and low-priority users are identified, and charging resources are allocated according to users' charging needs and behavioral habits. High-priority users can obtain higher charging power and faster charging speeds when needed, while low-priority users can choose lower-priced, lower-power charging options. Personalized resource allocation not only improves charging efficiency, but also greatly improves users' charging experience, so that users who urgently need fast charging will not be blocked by users with low power requirements.

[0080] Step S2, load forecasting, forecasting the peak load period and idle load period of the power grid based on the historical real-time load of the power grid. The load forecasting method is:

[0081] The real-time load data L(t) of the past T time points are extracted from the power grid, where t represents the time point. The data set is {L(1), L(2), …, L(T)}, denoised, and standardized: in,

[0082] represents the normalized load, L(t) represents the original grid load at time t, μ L is the mean of the historical load data L(t), indicating the average level of grid load, σ L Represents the standard deviation of historical load data L(t), which is used to measure the amplitude of load fluctuation;

[0083] The load prediction method in step S2 also includes:

[0084] Use the ARIMA model to model the normalized load time:

[0085] in, represents the normalized load at time t, i.e., the prediction target, φ1, φ2, …, φ p Represents the parameters of the autoregressive AR model, that is, the impact of the load values ​​at the past p time points on the current load, represents the normalized load at time ti, i.e., the historical load, θ1,θ2,…,θ q represents the parameters of the moving average MA model, that is, the impact of the error term at the past q time points on the current load, ∈(ti) represents the random error term at time ti, and ∈(t) represents the error term at time t;

[0087] Estimate the parameter φ of the ARIMA model by maximum likelihood estimation MLE i and θ j :

[0088] in, represents the load value predicted by the model after given parameters φ and θ, represents the model parameter estimate obtained by minimizing the sum of squared forecast errors, and T represents the length of the historical time series;

[0089] Load forecasting using the estimated ARIMA model:

[0090] in, represents the load forecast at time t+h, represents the normalized load forecast, μ L represents the mean of historical load data, σ L represents the standard deviation of historical load data, h represents the prediction step length, and after denormalization,

[0091] The load prediction method in step S2 also includes:

[0092] For the predicted load sequence To perform smoothing:

[0093] in, represents the smoothed predicted load value, N represents the smoothing window size, and i represents the index of the time step;

[0094] Set the load peak and idle thresholds, and divide the load peak and idle based on the predicted load;

[0095] Specifically, through the analysis and prediction of historical load data, the peak load periods and idle load periods of the power grid can be accurately identified. Based on the prediction results, the use of high-power charging can be limited during peak load periods to prevent power grid overload. At the same time, priority is given to meeting the needs of users who urgently need fast charging. During idle periods when the power grid load is lower, all charging options are open to encourage users to charge during these periods, thereby achieving balance and optimization of the power grid load.

[0096] Step S3, targeted resource allocation, intelligently allocating charging resources based on the load forecast results;

[0097] Specifically, this solution adds high-power and low-power charging rate options based on existing charging piles, provides four charging modes, and greatly improves the flexibility of charging piles.

[0098] The way to intelligently allocate charging resources is:

[0099] Add two additional options, high-power charging rate and low-power charging rate, to the existing fast charging and regular charging options of charging piles, for a total of four options. The low-power charging rate option is enabled in any scenario.

[0100] If the load forecast result is a peak load period, conventional charging is allocated to all users. On this basis, fast charging and high-power charging rate options are enabled according to user priority levels, and the allocation of charging resources does not exceed the grid load threshold;

[0101] The existing charging piles have fast charging and regular charging options. The high-priority user allocation here provides a more expensive and faster charging rate based on the original fast charging option in the existing solution. The low-priority user allocation provides a lower-priced low-power charging rate based on the original regular charging option in the existing solution. The high and low-power charging rates here are relative and are set according to the actual user charging speed, charging price, and grid load.

[0102] If the load forecast result is an idle load period, all users are assigned regular charging, and all options are enabled on this basis; regardless of user priority, the charging resources allocated do not exceed the grid load threshold;

[0103] Specifically, through dynamic prediction of grid load and intelligent adjustment of charging strategy, the risk of grid overload during peak load periods can be effectively avoided, the safety and stability of the grid can be enhanced, and during peak load periods, the charging power can be reasonably restricted and the allocation of charging resources can be optimized to ensure the stable operation of the grid. During idle load periods, a comprehensive open strategy can be adopted to make full use of the idle capacity of the grid and maximize the utilization of resources.

[0104] 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 can be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should be included in the scope of the claims of the present invention.

Claims

1. A charging pile adaptive charging planning method based on data mining, characterized by: include, Step S1, collecting charging pile and power grid data, and analyzing the behavior data of each user using the charging pile based on the charging pile and power grid data; Step S2, load forecasting, forecasting the peak load period and idle load period of the power grid based on the historical real-time load of the power grid; Step S3: targeted resource allocation, intelligently allocate charging resources based on the results of load forecasting.

2. A charging pile adaptive charging planning method based on data mining as claimed in claim 1, characterized in that: The charging pile and power grid data collected in step S1 include: the current use status of the charging pile, output power, information of the vehicle being charged, and real-time load of the power grid.

3. A charging pile adaptive charging planning method based on data mining as claimed in claim 2, characterized in that: The behavior data analysis in step S1 includes: the user's historical charging behavior, priority requirements, travel habits, and dividing the users into high-priority users and low-priority users.

4. A charging pile adaptive charging planning method based on data mining as claimed in claim 3, characterized in that: In step S1, the user division method is: Assume the charging frequency of user i is f i The average charging time is The average charging power is The charging time period preference is τ i and price sensitivity is γ i ; Calculate the comprehensive priority score S for user i i : Among them, w1, w2, w3, w4, w5 represent the weights of the corresponding features; Based on the comprehensive priority score S i , use Kmeans clustering algorithm to divide users into high priority and low priority categories, clustering objective function: Among them, i represents the i-th cluster, x j is the user's feature vector, i.e. cluster C i The jth data point in i is the cluster center, C i is the i-th cluster, k is the number of clusters, here it is 2, where, According to the clustering results, users are divided into high-priority users and low-priority users.

5. A charging pile adaptive charging planning method based on data mining as claimed in claim 4, characterized in that: In the user partitioning and clustering objective function of step S1: Charging time period preference τ i Indicates the charging behavior tendency of user i at different time periods of the day: Divide a day into six time periods: T1:00:00-04:00, T2:04:00-08:00, T3:08:00-12:00, T4:12:00-16:00, T5:16:00-20:00, T6:20:00-00:00, and calculate the number of times n that user i charges in each time period. i,k , where k represents the time period (16); Normalize the number of charges in each time period: Among them, p i,k Indicates that user i is in time period T k The charging preference weight of but Among them, T k is the center time of the corresponding time period.

6. A charging pile adaptive charging planning method based on data mining as claimed in claim 5, characterized in that: In the user partitioning and clustering objective function of step S1: Price Sensitivity i Indicates the user's sensitivity to charging prices: Get the price C of each charge for user i i,j And the corresponding charging power P i,j , where j represents the jth charge, and the unit power cost of each charge is calculated as: Normalize the unit power cost of each user: Among them, min(u i ) and max(u i ) are the minimum and maximum values ​​of the unit power cost of user i, is the normalized unit power cost; but Where N is the number of times user i charges.

7. A charging pile adaptive charging planning method based on data mining as claimed in claim 6, characterized in that: The load prediction method in step S2 is: The real-time load data L(t) of the past T time points are extracted from the power grid, where t represents the time point. The data set is {L(1), L(2), …, L(T)}, denoised, and standardized: in, represents the normalized load, L(t) represents the original grid load at time t, μ L is the mean of the historical load data L(t), indicating the average level of grid load, σ L It represents the standard deviation of the historical load data L(t), which is used to measure the amplitude of load fluctuation.

8. A charging pile adaptive charging planning method based on data mining as claimed in claim 7, characterized in that: The load prediction method in step S2 also includes: Use the ARIMA model to model the normalized load time: in, represents the normalized load at time t, i.e., the prediction target, φ1, φ2, …, φ p Represents the parameters of the autoregressive AR model, that is, the impact of the load values ​​at the past p time points on the current load, represents the normalized load at time ti, i.e., the historical load, θ1,θ2,…,θ q represents the parameters of the moving average MA model, that is, the impact of the error term at the past q time points on the current load, ∈(ti) represents the random error term at time ti, and ∈(t) represents the error term at time t; Estimate the parameter φ of the ARIMA model by maximum likelihood estimation MLE i and θ j : in, represents the load value predicted by the model after given parameters φ and θ, represents the model parameter estimate obtained by minimizing the sum of squared forecast errors, and T represents the length of the historical time series; Load forecasting using the estimated ARIMA model: in, represents the load forecast at time t+h, represents the normalized load forecast, μ L represents the mean of historical load data, σ L represents the standard deviation of historical load data, h represents the prediction step length, and after denormalization, 9. A charging pile adaptive charging planning method based on data mining as claimed in claim 8, characterized in that: The load prediction method in step S2 also includes: For the predicted load sequence To perform smoothing: in, represents the smoothed predicted load value, N represents the smoothing window size, and i represents the index of the time step; Set the load peak and idle thresholds, and divide the load peak and idle based on the predicted load.

10. A charging pile adaptive charging planning method based on data mining as claimed in claim 9, characterized in that: The method of intelligently allocating charging resources in step S3 is: Add high-power charging rate and low-power charging rate options to the basic fast charging and regular charging options. The low-power charging rate option is enabled in all scenarios. If the load forecast result is a peak load period, conventional charging is allocated to all users. On this basis, fast charging and high-power charging rate options are enabled according to user priority levels, and the allocated charging resources do not exceed the grid load threshold; If the load forecast result is an idle load period, conventional charging will be allocated to all users, and all options will be enabled on this basis; regardless of user priority, the allocated charging resources will not exceed the grid load threshold.

Images