A charging management method for electric vehicle

Abstract

The present application discloses a method for managing electric vehicle charging, including: obtaining the multi-feature fusion vectors of each new energy power generation branch connected to the electric vehicle charging station and the current new energy power generation branch; inputting the multi-feature fusion vectors of each new energy power generation branch into the pre-trained stable supply prediction model corresponding to each new energy power generation branch, and outputting the stability index of each current new energy power generation branch; obtaining the power demand power of each charging pile to be worked in the charging station, the rated power generation power of each new energy power generation branch and its stability index, and determining the power supply object of the new energy power generation branch according to the preset power supply association algorithm. Thus, based on the multi-feature fusion vector, the stable supply prediction model of each new energy power generation branch is differentiated, and the power supply efficiency and the stability of charging management are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of electric vehicle charging management, and in particular to an electric vehicle charging management method. Background Art

[0002] As a new energy vehicle, electric vehicles are gaining more and more attention. Due to cost constraints, electric vehicles cannot be charged through power banks like mobile phones, but need to be charged through dedicated charging piles. However, fast charging of electric vehicles requires instantaneous strong power, which cannot be met by charging piles relying on conventional power grids. It is necessary to rely on a dedicated charging network and build charging piles at dedicated electric vehicle charging stations. Therefore, electric vehicle charging is a bottleneck restricting the development of electric vehicles. The number of existing dedicated electric vehicle charging stations is relatively small, and the construction of existing dedicated electric vehicle charging stations still has an impact on conventional power grids, affecting household electricity consumption.

[0003] The Chinese invention patent with patent application number 202110644277.8 discloses an electric vehicle charging management method, which uses renewable energy power generation as a supplement to the conventional power grid to charge electric vehicles, diverts the impact of electric vehicle charging on the conventional power grid, reduces the impact of electric vehicle charging on household electricity consumption, promotes the development of electric vehicles, and improves the utilization rate of new energy.

[0004] However, in actual scenarios, due to the complexity and variability of the power grid and the volatility of renewable energy power generation itself, the dynamic changes in the power generation power of the power generation source that supplies power to the charging piles is an issue that cannot be ignored, which may lead to power supply instability. In addition, the above-mentioned patent monitors and adjusts the allocation of power supply branches to charging piles in real time. When encountering large fluctuations, the allocation strategy of the power supply branches may even be frequently switched. This frequent switching may bring charging risks due to the unknown stability of the power generation power of the power supply source, thereby affecting the charging management efficiency. Summary of the invention

[0005] The present application provides an electric vehicle charging management method that can differentiate the stable supply prediction models of each new energy power generation branch based on a multi-feature fusion vector, thereby improving the power supply efficiency and the stability of charging management.

[0006] The present application provides an electric vehicle charging management method, comprising:

[0007] S101, obtaining the multi-feature fusion vectors of each new energy power generation branch connected to the electric vehicle charging station and each current new energy power generation branch; the multi-feature fusion vector of each current new energy power generation branch is expressed as: [a, b, c], where a is the power supply time feature, b is the power supply environment feature, and c is the adjacent power generation-charging covariance feature;

[0008] S102, inputting the multivariate feature fusion vector of each renewable energy power generation branch into the pre-trained stable supply prediction model corresponding to each renewable energy power generation branch, and outputting the stability index of each renewable energy power generation branch at present;

[0009] S103, obtaining the power demand of each charging pile to be operated in the charging station, the rated power generation power of each new energy power generation branch and its stability index, and determining the power supply object and power supply mode of the new energy power generation branch according to a preset power supply association algorithm.

[0010] Preferably, the process of acquiring the multivariate feature fusion vector of the new energy power generation branch is:

[0011] A1. Based on the first allocation rule, extract the power supply time characteristics corresponding to the current time node, where the power supply time characteristics include peak period, mid-peak period, and low-peak period;

[0012] A2. Based on the second allocation rule, extract the power supply environment feature corresponding to the current time node, and set the power supply environment feature to the environment severity level value, and the range interval of the environment severity level value is [0,3];

[0013] A3, obtaining the power generation sequence of the new energy power generation branch in the power supply window adjacent to the new energy power generation branch and the charging power sequence of the corresponding charging pile group, and obtaining the adjacent power generation-charging covariance characteristics of the current new energy power generation branch according to the first preset algorithm;

[0014] The power generation sequence is a power generation sequence formed by the power generation sampled sequentially in the order of time nodes in the adjacent power supply window; the charging power sequence is a charging power sequence formed by the charging power sampled sequentially in the order of time nodes in the adjacent power supply window; the adjacent power supply window is the time period when the new energy power generation branch last generated electricity to the charging station;

[0015] A4. Generate a multivariate feature fusion vector of each current renewable energy power generation branch based on power supply time characteristics, power supply environment characteristics, and adjacent power generation-charging covariance characteristics.

[0016] Preferably, the first allocation rule is set to: respectively obtain the change curve of the daily electric vehicle flow data of each working day and rest day in the history of the charging station, obtain the charging peak time period, mid-peak time period and low-peak time period corresponding to the working day and rest day according to the change curve of the daily electric vehicle flow data, determine the time period hit by the current time node, and obtain the power supply time characteristics of the current time node;

[0017] The second allocation rule is set based on the environmental parameters: according to a preset severity level classification mechanism, the level values ​​corresponding to the current environmental parameters are obtained respectively, and the level values ​​corresponding to the environmental parameters are weighted summed to obtain the environmental severity level value.

[0018] Preferably, the first preset algorithm is specifically:

[0019]

[0020] Among them, c is the adjacent generation-charging covariance characteristic of the new energy generation branch, n is the total number of time nodes sampled in the adjacent power supply window, is the power generation power of the new energy power generation branch at the i-th time node, is the mean value of the power generation in the power generation sequence, is the charging power of the corresponding charging pile group at the i-th time node, is the mean value of the charging power in the charging power sequence, is the variance value of the power generation in the power generation sequence, is the variance value of the charging power in the charging power sequence.

[0021] Preferably, the method for obtaining the pre-trained stable supply prediction model corresponding to each renewable energy power generation branch includes:

[0022] B1. Collect a large number of multivariate feature fusion vectors of the new energy power generation branch in the history of the charging station;

[0023] B2. Label a large number of multi-feature fusion vectors in history, and set the label content as stability index to form a training data set;

[0024] B3. Use the training data set to train the predetermined neural network structure, optimize the model parameters, and obtain the final stable supply prediction model corresponding to the new energy power generation branch.

[0025] Preferably, B2 further includes:

[0026] C1. Obtain the power generation sequence of the new energy power generation branch in the historical power supply window where each training data is located [ ], the charging power sequence of the corresponding charging pile group[ ];

[0027] C2. Calculate the stability index of each training data based on the power characteristic data of the power generation and receiving ends within the historical power supply window where the training data is located.

[0028] Preferably, before C2, the method further comprises:

[0029] D1, based on each training data, according to the power generation sequence in its historical power supply window [ ], and calculate the power generation fluctuation amplitude value at each time node: , is the power generation fluctuation amplitude value at the i-th time node, is the power generation power at the i-th time node in the power generation power sequence, is the mean value of the power generation in the power generation sequence;

[0030] D2. Count the number of time nodes T in the historical power supply window where the power generation fluctuation amplitude is greater than the preset amplitude threshold, and the power generation and charging power corresponding to T time nodes to form a new power generation sequence [ ], the charging power sequence of the corresponding charging pile group[ ], replacing the power generation sequence and charging power sequence in step C1.

[0031] Preferably, in C2, the stability index of each training data is calculated, specifically:

[0032]

[0033] in, is the stability index of the rth training data, is the stable correlation factor of the rth training data, T is the total number of time nodes in the corresponding historical power supply window where the power generation fluctuation amplitude is greater than the preset amplitude threshold, t is the number of the time node, and its value range is [1, T]. is the power generation at the tth time node in the new power generation sequence, is the mean value of the power generation in the new power generation sequence, is the charging power at the tth time node in the new charging power, is the mean value of the charging power in the new charging power sequence, is the variance value of the power generation in the new power generation sequence, is the variance value of the charging power in the new charging power sequence, is the adjustment coefficient of the stable correlation factor.

[0034] Preferably, in S103, the preset power supply association algorithm specifically includes:

[0035] S201, obtaining the relationship attributes between all the charging piles to be operated currently in the charging station and each of the charging piles to be operated, the relationship attributes including parallel relationship and non-parallel relationship, and obtaining at least one charging pile group according to the attribute relationship between each of the charging piles to be operated, each charging pile group including at least one charging pile to be operated;

[0036] S202, for each charging pile group, calculating its total power demand, and obtaining the rated power generation power and stability index of each current new energy power generation branch;

[0037] S203, sorting the new energy power generation branches in descending order according to the stability index to generate a new energy power generation branch list, and sorting the charging pile groups in descending order according to the total power demand power to generate a charging pile group list;

[0038] S204, traverse the list of charging pile groups, for each charging pile group, traverse the list of new energy power generation branches, and search for new energy power generation branches that meet the power adaptation conditions. If they exist, select the branch with the highest stability index from among the new energy power generation branches that meet the power adaptation conditions as the power supply source for the charging pile group. If they do not exist, combine several new energy power generation branches with the highest stability indexes in parallel to meet the power adaptation conditions as the power supply source for the charging pile group, or select the branch with the highest stability index from among the new energy power generation branches that still have surplus power generation to connect in parallel with the main power grid as the power supply source for the charging pile group.

[0039] S205, update the remaining power generation of all new energy power generation branches, and remove the branches that have been allocated power from the list of new energy power generation branches, repeat steps S204 to S205 until all charging pile groups are traversed and an allocation result is generated, which includes the power supply source of each charging pile group.

[0040] Preferably, after S103, the method further includes:

[0041] S104, based on a preset time interval, periodically execute steps S101 to S103 at each time interval, update the multivariate feature fusion vector of each current renewable energy power generation branch, and update the adjacent power supply windows of each current renewable energy power generation branch according to the power supply windows of each renewable energy power generation branch in the previous time interval.

[0042] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0043] In the current multi-feature fusion vector of the new energy power generation branch, the adjacent power generation-charging covariance feature is introduced. By monitoring the power generation power sequence in the previous power supply window and the charging power sequence of the corresponding receiving charging pile group, the power supply fluctuation characteristics between the power generation and charging of the new energy power generation branch closest to the current can be quantified. This historical dynamic relationship is crucial for predicting the power supply stability in the current and future power supply windows; the introduction of the historical covariance feature enables the model to more accurately identify the historical fluctuation characteristics, thereby making more accurate and targeted predictions based on more similar historical situations; by introducing the multi-feature fusion vector, the training of the stable supply prediction model of each new energy power generation branch is differentiated, and by comprehensively considering the power supply time, power supply environment and adjacent power generation-charging covariance multiple dimensions, the current stability of the new energy power generation branch for charging the charging pile power supply is more comprehensively evaluated. Combined with the stability index and power generation power of the power supply end and the power demand power of the receiving section, a more accurate and reliable information support basis is provided for the subsequent power supply allocation decision, which can more reasonably allocate the power supply tasks of the new energy power generation branch and the trunk power grid, and improve the power supply efficiency and the stability of charging management;

[0044] By intelligently labeling the stability index of each training data, the process of manual labeling or expert experience labeling is replaced, which reduces manual intervention and improves the efficiency and accuracy of data processing. The calculation of the stability index comprehensively considers the characteristics of the actual power generation power at the power supply end and the actual charging power at the power receiving end as well as the fluctuation correlation between them, which can more comprehensively reflect the power supply stability of the renewable energy power generation branch. The targeted stability index labeling provides high-quality training data for the stable supply prediction model, which helps the model better learn the characteristics of the power supply stability of different renewable energy power generation branches, thereby improving the accuracy of the prediction.

[0045] By screening out the time nodes with significant fluctuations in power generation for each historical power supply window of training data, further refining the power data features of the associated receiving end, updating the sequence data corresponding to the time nodes in the historical power supply window, and calculating the stability index based on these nodes, the stability of the power supply end can be more accurately evaluated, and the power data features of the power generation end and the power receiving end can be effectively associated accurately; further providing more accurate and targeted training data for the stable power supply prediction model, which helps to improve the prediction accuracy of the model;

[0046] Based on the current stability index and rated power generation of each renewable energy power generation branch, the preset power supply association algorithm is used, that is, by dividing the charging pile groups, sorting the renewable energy power generation branches and the charging pile groups, and traversing the priority allocation conditions, to ensure the optimization of the power supply allocation plan; by periodically executing the power supply association algorithm, updating the relevant data parameters in real time, and monitoring the charging progress of the charging pile group in real time, the cycle time interval is dynamically adjusted to ensure that the power supply allocation plan can adapt to changes in actual conditions at any time, and adapt to the dynamic changes in renewable energy power generation and charging needs. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 The figure is a flow chart of an electric vehicle charging management method according to an embodiment of the present invention. DETAILED DESCRIPTION

[0048] To facilitate the understanding of the present invention, the present application will be described more comprehensively below with reference to the relevant drawings; the drawings show preferred embodiments of the present invention, but the present invention can be implemented in many different forms and is not limited to the embodiments described herein; on the contrary, the purpose of providing these embodiments is to enable a more thorough and comprehensive understanding of the disclosed content of the present invention.

[0049] It should be noted that the terms “vertical”, “horizontal”, “up”, “down”, “left”, “right” and similar expressions used in this document are only for illustrative purposes and do not represent the only implementation method.

[0050] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by technicians in the technical field to which the present invention belongs; the terms used in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention; the term "and / or" used herein includes any and all combinations of one or more related listed items.

[0051] Embodiment 1: Figure 1 It is a flow chart of an electric vehicle charging management method according to an embodiment of the present invention.

[0052] like Figure 1 As shown, a method for managing charging of an electric vehicle comprises the following steps:

[0053] S101, obtaining each new energy power generation branch connected to the electric vehicle charging station and a multi-feature fusion vector of each current new energy power generation branch.

[0054] It should be noted that the power supply end of the electric vehicle charging station is provided with several new energy power generation branches, which are mounted to the original main power grid, and the power supply mode of the new energy power generation branch is set to any one of a single power supply and a multi-parallel power supply; specifically, a single power supply means that a new energy power generation branch supplies power to a charging pile group alone, and the charging pile group includes at least one charging pile to be worked, and the multiple charging piles to be worked are in a parallel relationship; multi-parallel power supply means that multiple power supply sources supply power to a charging pile group, and the multiple power supply sources can be multiple parallel new energy power generation branches and a new energy power generation branch auxiliary main power grid. The specific power supply distribution is set according to the actual situation, and the power generation power of the power supply end meets the sum of the power demand of the charging pile group, and the present invention does not elaborate or limit this.

[0055] In some embodiments, the multivariate feature fusion vector of each current renewable energy power generation branch is expressed as: [a, b, c], where a is the power supply time feature, b is the power supply environment feature, and c is the adjacent power generation-charging covariance feature. The process of obtaining the multivariate feature fusion vector of the renewable energy power generation branch is:

[0056] A1. Based on the first allocation rule, extract the power supply time characteristics corresponding to the current time node, the power supply time characteristics include peak period, mid-peak period, and low-peak period, that is, the distribution of electric vehicles at the charging station at the current time node;

[0057] Among them, the first allocation rule is set as follows: respectively obtain the change curve of the daily electric vehicle flow data of each working day and rest day in the history of the charging station (the horizontal axis is set to the time node, and the vertical axis is set to the electric vehicle flow corresponding to the time node), and obtain the charging peak period, medium peak period, and low peak period corresponding to working days and rest days according to the change curve of the daily electric vehicle flow data (the electric vehicle flow range corresponding to different time periods can be set according to expert experience, and the present invention does not repeat and limit this), judge the time period hit by the current time node, and obtain the power supply time characteristics of the current time node. For example, the corresponding low peak period of working days is 9:00-11:00, the current time node is 10:00 am on Monday, and the hit period is the low peak period of working days, so the corresponding power supply time characteristics are low peak period.

[0058] A2. Based on the second allocation rule, extract the power supply environment characteristics corresponding to the current time node. The power supply environment characteristics are set as the environmental severity level value. The range of the environmental severity level value is [0,3]. 0 represents the best environment and 3 represents the worst environment. The second allocation rule is mainly based on environmental parameters (temperature, humidity, wind speed, rainfall) to set the severe situation.

[0059] The second allocation rule is set as follows: according to a preset severity level classification mechanism, the level values ​​corresponding to the current environmental parameters are obtained respectively; the level values ​​corresponding to the environmental parameters are weighted summed to obtain the environmental severity level value as the power supply environment feature at the current time node:

[0060]

[0061] is the environmental severity level, n is the total number of environmental parameters, is the level value corresponding to the i-th environmental parameter, is the weight value of the influence of the i-th environmental parameter on the environmental severity level value, which can be set according to expert experience or historical data, and the present invention will not elaborate on this;

[0062] Specifically, the pre-set bad grade classification mechanism is:

[0063] Level 0: Environmental parameters are within the appropriate range;

[0064] Level 1: Environmental parameters slightly deviate from the appropriate range;

[0065] Level 2: Environmental parameters significantly deviate from the appropriate range;

[0066] Level 3: Environmental parameters deviate extremely from the appropriate range;

[0067] Among them, the suitable range of environmental parameters and the definition of slight, significant and extreme deviations are set according to actual conditions and expert experience, and the present invention does not elaborate on this. Taking temperature as an example, level 0: temperature is within the suitable range (15-25°C), level 1: temperature slightly deviates from the suitable range (10-15°C or 25-30°C), level 2: temperature significantly deviates from the suitable range (5-10°C or 30-35°C), level 3: temperature extremely deviates from the suitable range (<5°C or >35°C);

[0068] A3. Obtain the power generation sequence of the new energy power generation branch in the adjacent power supply window (the previous power supply window, the time period of the most recent power generation to the charging station) of the new energy power generation branch (the power generation sequence formed by the power generation power sampled sequentially in the power supply window according to the time node sequence), and the charging power sequence of the corresponding charging pile group (the charging power sequence formed by the charging power sampled sequentially in the power supply window according to the time node sequence), and obtain the adjacent power generation-charging covariance feature of the current new energy power generation branch according to the first preset algorithm;

[0069] The first preset algorithm is specifically:

[0070]

[0071] c is the adjacent generation-charging covariance characteristic of the new energy generation branch, n is the total number of time nodes sampled in the adjacent power supply window, is the power generation power of the new energy power generation branch at the i-th time node, is the mean value of the power generation in the power generation sequence, is the charging power of the corresponding charging pile group at the i-th time node, is the mean value of the charging power in the charging power sequence, is the variance value of the power generation in the power generation sequence, is the variance value of the charging power in the charging power sequence.

[0072] A4. Generate a multivariate feature fusion vector of each current renewable energy power generation branch based on power supply time characteristics, power supply environment characteristics, and adjacent power generation-charging covariance characteristics.

[0073] S102, inputting the multivariate feature fusion vector of each renewable energy power generation branch into the pre-trained stable supply prediction model corresponding to each renewable energy power generation branch, and outputting the stability index of each current renewable energy power generation branch.

[0074] In some embodiments, the method for obtaining the pre-trained stable supply prediction model corresponding to each renewable energy power generation branch includes:

[0075] B1. Collect a large number of multi-feature fusion vectors of the new energy power generation branch in the charging station history. The generation method of the multi-feature fusion vectors in history refers to steps A1 to A3, which will not be elaborated in the present invention.

[0076] B2. Label a large number of multi-feature fusion vectors in history, and set the label content as the stability index to form a training data set.

[0077] B3. Use the training data set to train the predetermined neural network structure, optimize the model parameters, and obtain the final stable supply prediction model corresponding to the new energy power generation branch.

[0078] In some embodiments, in the stable supply prediction model, a comprehensive evaluation output module is also connected before the output end, which is specifically used to: based on the input multi-feature fusion vector, select training data whose similarity with the input multi-feature fusion vector is greater than a preset similarity threshold as a hit object, perform weighted summation of the stability indexes of all hit objects to obtain a comprehensive stability index, transmit the comprehensive stability index to the output end of the stable supply prediction model, and output it as the stability index of the current new energy power generation branch.

[0079] The stability indexes of all hit objects are weighted and summed, and the weight value of the stability index of each hit object is determined as follows:

[0080]

[0081] is the weight value of the stability index of the kth hit object, is the similarity between the kth hit object and the input multi-feature fusion vector, which can be calculated according to the Euclidean distance algorithm, which is not described in detail in the present invention, m is the total number of all hit objects selected, is the similarity between the jth hit object and the input multi-feature fusion vector.

[0082] S103, obtaining the power demand of each charging pile to be operated in the charging station, the rated power generation power of each new energy power generation branch and its stability index, and determining the power supply object and power supply mode of the new energy power generation branch according to a preset power supply association algorithm.

[0083] The technical solutions in the above embodiments of the present application have at least the following technical effects or advantages:

[0084] The adjacent power generation-charging covariance feature is introduced into the current multi-feature fusion vector of the new energy power generation branch. By monitoring the power generation power sequence in the previous power supply window and the charging power sequence of the corresponding charging pile group, the power supply fluctuation characteristics between the power generation and charging of the new energy power generation branch closest to the current can be quantified. This historical dynamic relationship is crucial for predicting the power supply stability in the current and future power supply windows. With the introduction of this historical covariance feature, the model can more accurately identify the historical fluctuation characteristics, thereby making more accurate and targeted predictions based on more similar historical situations.

[0085] By introducing multi-feature fusion vectors, the training of stable supply prediction models of various renewable energy power generation branches is differentiated. By comprehensively considering the power supply time, power supply environment and adjacent power generation-charging covariance (multi-angle monitoring stability relationship between the power supply end and the power receiving end), the stability of the new energy power generation branch for charging pile power supply is more comprehensively evaluated. Combined with the stability index and power generation power of the power supply end and the power demand power of the power receiving section, a more accurate and reliable information support basis is provided for subsequent power supply allocation decisions, which can more reasonably allocate the power supply tasks of the renewable energy power generation branch and the main power grid, and improve the power supply efficiency and the stability of charging management.

[0086] By using the historical power supply data of each renewable energy power generation branch of the charging station to train a stable power supply prediction model, it is possible to predict the stability coefficient of each current renewable energy power generation branch, thereby evaluating the power supply stability before actual power supply, reducing charging anomalies caused by unstable power supply or subsequent real-time monitoring of frequent switching of power supply branches, which affects the charging efficiency and safety of electric vehicles at the charging station.

[0087] Embodiment 2: In embodiment 1, the training data set of the stable supply prediction model corresponding to each renewable energy power generation branch is traditionally labeled manually or based on the information of the multi-feature fusion vector, or by expert experience. Moreover, the labeling process may be affected by subjective factors, resulting in inaccurate labeling results. The preprocessing process of the training data set is too cumbersome, which increases the difficulty of model training preparation and lacks intelligent processing of charging station data.

[0088] Therefore, the embodiments of the present application are optimized to a certain extent based on the above embodiments.

[0089] In some embodiments, step B2 further includes:

[0090] C1. Obtain the power generation sequence of the new energy power generation branch in the historical power supply window where each training data is located [ ], the charging power sequence of the corresponding charging pile group[ ];

[0091] Therefore, the actual power generation and charging data in history are used to reflect the situation of power generation and charging in the historical power supply window, providing data support for the subsequent calculation of the stability index of the new energy power generation branch as a power supply;

[0092] C2. Based on the power characteristic data of the power generation and receiving ends in the historical power supply window where the training data is located, the power characteristic data includes the power generation power and its mean value, variance value and the charging power and its mean value, variance value, calculate the stability index of each training data:

[0093]

[0094] in, is the stability index of the rth training data, is the stable correlation factor of the r-th training data, n is the total number of time nodes sampled in the historical power supply window where the r-th training data is located, is the power generation power at the i-th time node in the power generation power sequence, is the mean value of the power generation in the power generation sequence, is the charging power at the i-th time node in the charging power sequence, is the mean value of the charging power in the charging power sequence, is the variance value of the power generation in the power generation sequence, is the variance value of the charging power in the charging power sequence, In order to stabilize the adjustment coefficient of the correlation factor, it is set according to the actual situation to control the steepness of the exponential function. Better distributed between 0 and 1.

[0095] By calculating the deviation of the power generation and charging power at each time node in the historical power supply window where each training data is located from the mean, and dividing it by the corresponding variance for standardization, the stability contribution value of each time node is obtained, and the stability contribution values ​​of all time nodes in the historical power supply window are averaged to obtain the stability index of the entire power supply window. In this way, the fluctuation correlation between power generation and charging in the historical power supply window is quantified, and accurate stability index annotation is provided for the training data set, thereby improving the training effect of the stable supply prediction model.

[0096] The technical solutions in the above embodiments of the present application have at least the following technical effects or advantages:

[0097] By intelligently labeling the stability index of each training data, the process of manual labeling or expert experience labeling is replaced, which reduces human intervention and improves the efficiency and accuracy of data processing. The calculation of the stability index comprehensively considers the characteristics of the actual power generation power at the power supply end and the actual charging power at the power receiving end as well as the fluctuation correlation between them, which can more comprehensively reflect the power supply stability of the renewable energy power generation branch. Targeted stability index labeling provides high-quality training data for the stable supply prediction model, which helps the model better learn the characteristics of the power supply stability of different renewable energy power generation branches, thereby improving the accuracy of the prediction.

[0098] Embodiment 3: In embodiment 2, the label annotation of the training data of the model is set to the actual stability index, and the power characteristic fluctuation correlation at each time node of the power generation end and the power receiving end in the historical power supply window is introduced. Although it can more comprehensively reflect the stability of the power generation branch, in actual situations, what is considered is the stability of the power supply of the new energy power generation branch during the charging process, so that scientific allocation is made according to the stability of each new energy power generation branch before power supply. In fact, the working power fluctuation characteristics of the power receiving end (charging pile) at different time nodes are not necessarily attributed to the power supply end, and may be caused by other factors. The fluctuation characteristics, then the quantification of the fluctuation correlation characteristics of the two ends may have errors and deviations for the causal reflection of the power generation end on the power receiving end.

[0099] Therefore, the embodiments of the present application are optimized to a certain extent based on the above embodiments.

[0100] In some embodiments, before step C2, the method further comprises:

[0101] D1, based on each training data, according to the power generation sequence in its historical power supply window [ ], and calculate the power generation fluctuation amplitude value at each time node: , is the power generation fluctuation amplitude value at the i-th time node, is the power generation power at the i-th time node in the power generation power sequence, is the mean value of the power generation in the power generation sequence;

[0102] D2. Count the number of time nodes T in the historical power supply window where the power generation fluctuation amplitude is greater than the preset amplitude threshold, and the power generation and charging power corresponding to T time nodes to form a new power generation sequence [ ], the charging power sequence of the corresponding charging pile group[ ], replace the power generation sequence and charging power sequence in step C1;

[0103] The preset amplitude threshold is set according to the actual situation and expert experience to define the fluctuation of the generated power. The purpose is to filter out the fluctuation data that is truly related to the stability of the power supply end and exclude the fluctuation results caused by other factors at the power receiving end.

[0104] It should be noted that the new power generation sequence [ ], the charging power sequence of the corresponding charging pile group[ ], corresponding to the calculation formula in step C2, replace n with T, and replace other parameters with the corresponding new sequence, specifically:

[0105]

[0106] in, is the stable correlation factor of the rth training data, T is the total number of time nodes in the corresponding historical power supply window where the power generation fluctuation amplitude is greater than the preset amplitude threshold, t is the number of the time node, and its value range is [1, T]. is the power generation at the tth time node in the new power generation sequence, is the mean value of the power generation in the new power generation sequence, is the charging power at the tth time node in the new charging power, is the mean value of the charging power in the new charging power sequence, is the variance value of the power generation in the new power generation sequence, is the variance value of the charging power in the new charging power sequence.

[0107] The technical solutions in the above embodiments of the present application have at least the following technical effects or advantages:

[0108] By screening out the time nodes with significant fluctuations in power generation in the historical power supply window of each training data, further refining the associated power data features of the receiving end, updating the sequence data corresponding to the time nodes in the historical power supply window, and calculating the stability index based on these nodes, the stability of the power supply end can be more accurately evaluated, and the power data features of the power generation end and the receiving end can be effectively associated accurately; further providing more accurate and targeted training data for the stable supply prediction model, which helps to improve the prediction accuracy of the model.

[0109] Embodiment 4: Further limit the power supply association algorithm in Embodiment 1.

[0110] Therefore, the embodiments of the present application are optimized to a certain extent based on the above embodiments.

[0111] In some embodiments, in step S103, the preset power supply association algorithm specifically includes:

[0112] S201, obtaining the relationship attributes between all the charging piles to be operated currently in the charging station and each of the charging piles to be operated, the relationship attributes including parallel relationship and non-parallel relationship, and obtaining at least one charging pile group according to the attribute relationship between each of the charging piles to be operated, each charging pile group including at least one charging pile to be operated in a parallel relationship.

[0113] Specifically, the charging piles with a parallel relationship attribute are divided into one charging pile group, and the charging piles to be worked in a non-parallel relationship are respectively formed into a separate charging pile group, forming several charging pile groups.

[0114] S202, for each charging pile group, calculate its total power demand (the sum of the power demand of all charging piles to be operated in the charging pile group), and obtain the rated power generation power and stability index of each current new energy power generation branch.

[0115] S203, sorting the new energy power generation branches in descending order according to the stability index to generate a new energy power generation branch list, and sorting the charging pile groups in descending order according to the total power demand power to generate a charging pile group list.

[0116] S204, traverse the list of charging pile groups, for each charging pile group, traverse the list of new energy power generation branches, and search for a new energy power generation branch that meets the power adaptation condition (that is, the rated power generation power is greater than the total power demand power of the charging pile group and exceeds the preset power margin). If it exists, then among the new energy power generation branches that meet the power adaptation, select the one with the highest stability index as the power supply source for the charging pile group. If it does not exist, combine several new energy power generation branches with the highest stability index in parallel to meet the power adaptation condition as the power supply source for the charging pile group, or select the one with the highest stability index among the new energy power generation branches that still have surplus power and connect it in parallel with the main power grid as the power supply source for the charging pile group.

[0117] Among them, under the premise of meeting the power adaptation requirements, priority is given to new energy power generation branches with higher stability indexes for power supply. The power adaptation is set as follows: the rated power generation power of the selected new energy power generation branch can meet the total power demand of the charging pile group, and there is a certain power margin to cope with possible power fluctuations. The power margin is set according to actual conditions. For charging pile groups that include fast charging mode charging piles, priority is given to new energy power generation branches with high power supply stability and larger rated power generation power to meet the dual requirements of fast charging for power and stability.

[0118] S205, update the remaining power generation power of all new energy power generation branches (i.e., the rated power generation power minus the power demand power allocated to the charging pile group, replacing the initial rated power generation power), and remove the branches that have been allocated power from the list of new energy power generation branches, repeat steps S204 to S205 until all charging pile groups are traversed and an allocation result is generated, which includes the power supply source for each charging pile group.

[0119] As an example, a charging station currently has three new energy power generation branches (A, B, C), with rated power generation of 100kW, 80kW, and 60kW respectively, and the current output stability index is 0.95, 0.85, and 0.75 respectively. It is monitored that there are currently two charging pile groups (G1 and G2), of which G1 contains two conventional charging piles with a total power demand of 50kW; G2 contains one fast charging pile with a power demand of 60kW; the preset power margin is set to 5kW:

[0120] The order of new energy power generation branches is A > B > C; the order of charging pile groups is G2 > G1;

[0121] For charging pile group G2 (60kW), select branch A (100kW, stability 0.95) for power supply;

[0122] Update the remaining power generation capacity of the new energy power generation branch: branch A: 40kW, branch B: 80kW, branch C: 60kW;

[0123] For charging pile group G1 (50kW), branch B is selected for power supply among the remaining A (40kW, stability index 0.95), B (80kW, stability index 0.85), and C (60kW, stability index 0.75), because the remaining power generation power of branch B meets the power adaptation and has higher stability;

[0124] Generate allocation results: power supply source-A branch of charging pile group G2, and branch-B of charging pile group G1.

[0125] In some embodiments, after step S103, the method further includes:

[0126] S104, based on a preset time interval, periodically execute steps S101 to S103 at each time interval, update the multivariate feature fusion vector of each current renewable energy power generation branch, and update the adjacent power supply windows of each current renewable energy power generation branch according to the power supply windows of each renewable energy power generation branch in the previous time interval to adapt to the dynamic changes in renewable energy power generation and charging needs.

[0127] It should be noted that the charging progress of each charging pile in each charging pile group in the current cycle is monitored in real time. If any charging pile has finished charging, the current cycle is ended, and a new cycle is entered with the duration of the current cycle updated at the same time interval, and steps S101 to S103 are executed.

[0128] The technical solutions in the above embodiments of the present application have at least the following technical effects or advantages:

[0129] Based on the current stability index and rated power generation of each renewable energy power generation branch, the preset power supply association algorithm is used to ensure the optimization of the power supply allocation scheme by dividing the charging pile groups, sorting the renewable energy power generation branches and the charging pile groups, and traversing the priority allocation conditions. Under the premise of meeting the power adaptation, the renewable energy power generation branches with high stability index are given priority for power supply, which meets the high-standard power supply requirements such as fast charging and improves the reliability of charging services.

[0130] By periodically executing the power supply association algorithm, updating relevant data parameters in real time, and monitoring the charging progress of the charging pile group in real time, and dynamically adjusting the cycle time interval, it is ensured that the power supply distribution plan can adapt to changes in actual conditions at any time and adapt to the dynamic changes in new energy power generation and charging needs.

[0131] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for managing electric vehicle charging, characterized in that: include: S101, obtaining the multi-feature fusion vectors of each new energy power generation branch connected to the electric vehicle charging station and each current new energy power generation branch; the multi-feature fusion vector of each current new energy power generation branch is expressed as: [a, b, c], where a is the power supply time feature, b is the power supply environment feature, and c is the adjacent power generation-charging covariance feature; S102, inputting the multivariate feature fusion vector of each renewable energy power generation branch into the pre-trained stable supply prediction model corresponding to each renewable energy power generation branch, and outputting the stability index of each renewable energy power generation branch at present; S103, obtaining the power demand of each charging pile to be operated at the charging station, the rated power generation power of each new energy power generation branch and its stability index, and determining the power supply object of the new energy power generation branch according to a preset power supply association algorithm; The method for obtaining the pre-trained stable supply prediction model corresponding to each renewable energy power generation branch includes: B1. Collect a large number of multivariate feature fusion vectors of the new energy power generation branch in the history of the charging station; B2. Label a large number of multi-feature fusion vectors in history, and set the label content as stability index to form a training data set; B3. Use the training data set to train the predetermined neural network structure, optimize the model parameters, and obtain the final stable supply prediction model corresponding to the new energy power generation branch; The B2 also includes: C1. Obtain the power generation sequence of the new energy power generation branch in the historical power supply window where each training data is located [ ], the charging power sequence of the corresponding charging pile group[ ]; C2. Calculate the stability index of each training data according to the power characteristic data of the power generation and receiving ends in the historical power supply window where the training data is located; In C2, the stability index of each training data is calculated, specifically: in, is the stability index of the rth training data, is the stable correlation factor of the rth training data, T is the total number of time nodes in the corresponding historical power supply window where the power generation fluctuation amplitude is greater than the preset amplitude threshold, is the power generation at the tth time node in the new power generation sequence, is the mean value of the power generation in the new power generation sequence, is the charging power at the tth time node in the new charging power, is the mean value of the charging power in the new charging power sequence, is the variance value of the power generation in the new power generation sequence, is the variance value of the charging power in the new charging power sequence, is the adjustment coefficient of the stable correlation factor.

2. The electric vehicle charging management method according to claim 1, characterized in that: The process of obtaining the multi-feature fusion vector of the new energy power generation branch is as follows: A1. Based on the first allocation rule, extract the power supply time characteristics corresponding to the current time node, where the power supply time characteristics include peak period, mid-peak period, and low-peak period; A2. Based on the second allocation rule, extract the power supply environment feature corresponding to the current time node, and set the power supply environment feature to the environment severity level value, and the range interval of the environment severity level value is [0,3]; A3, obtaining the power generation sequence of the new energy power generation branch in the power supply window adjacent to the new energy power generation branch and the charging power sequence of the corresponding charging pile group, and obtaining the adjacent power generation-charging covariance characteristics of the current new energy power generation branch according to the first preset algorithm; The power generation sequence is a power generation sequence formed by the power generation sampled sequentially in the order of time nodes in the adjacent power supply window; the charging power sequence is a charging power sequence formed by the charging power sampled sequentially in the order of time nodes in the adjacent power supply window; the adjacent power supply window is the time period when the new energy power generation branch last generated electricity to the charging station; A4. Generate a multivariate feature fusion vector of each current renewable energy power generation branch based on power supply time characteristics, power supply environment characteristics, and adjacent power generation-charging covariance characteristics.

3. The electric vehicle charging management method according to claim 2, characterized in that: The first allocation rule is set as follows: respectively obtain the change curves of the daily electric vehicle flow data of each working day and rest day in the history of the charging station, obtain the charging peak time period, mid-peak time period, and low-peak time period corresponding to working days and rest days according to the change curves of the daily electric vehicle flow data, determine the time period hit by the current time node, and obtain the power supply time characteristics of the current time node; The second allocation rule is set based on the environmental parameters: according to a preset severity level classification mechanism, the level values ​​corresponding to the current environmental parameters are obtained respectively, and the level values ​​corresponding to the environmental parameters are weighted summed to obtain the environmental severity level value.

4. The electric vehicle charging management method according to claim 2, characterized in that: The first preset algorithm is specifically: Among them, c is the adjacent generation-charging covariance characteristic of the new energy generation branch, n is the total number of time nodes sampled in the adjacent power supply window, is the power generation power of the new energy power generation branch at the i-th time node, is the mean value of the power generation in the power generation sequence, is the charging power of the corresponding charging pile group at the i-th time node, is the mean value of the charging power in the charging power sequence, is the variance value of the power generation in the power generation sequence, is the variance value of the charging power in the charging power sequence.

5. The electric vehicle charging management method according to claim 1, characterized in that: Before C2, the method further includes: D1, based on each training data, according to the power generation sequence in its historical power supply window [ ], and calculate the power generation fluctuation amplitude value at each time node: , is the power generation fluctuation amplitude value at the i-th time node, is the power generation power at the i-th time node in the power generation power sequence, is the mean value of the power generation in the power generation sequence; D2. Count the number of time nodes T in the historical power supply window where the power generation fluctuation amplitude is greater than the preset amplitude threshold, and the power generation and charging power corresponding to T time nodes to form a new power generation sequence [ ], the charging power sequence of the corresponding charging pile group[ ], replacing the power generation sequence and charging power sequence in step C1.

6. The electric vehicle charging management method according to claim 1, characterized in that: In S103, the preset power supply association algorithm specifically includes: S201, obtaining the relationship attributes between all the charging piles to be operated currently in the charging station and each of the charging piles to be operated, the relationship attributes including parallel relationship and non-parallel relationship, and obtaining at least one charging pile group according to the attribute relationship between each of the charging piles to be operated, each charging pile group including at least one charging pile to be operated; S202, for each charging pile group, calculating its total power demand, and obtaining the rated power generation power and stability index of each current new energy power generation branch; S203, sorting the new energy power generation branches in descending order according to the stability index to generate a new energy power generation branch list, and sorting the charging pile groups in descending order according to the total power demand power to generate a charging pile group list; S204, traverse the list of charging pile groups, for each charging pile group, traverse the list of new energy power generation branches, and search for new energy power generation branches that meet the power adaptation conditions. If they exist, select the branch with the highest stability index from among the new energy power generation branches that meet the power adaptation conditions as the power supply source for the charging pile group. If they do not exist, combine several new energy power generation branches with the highest stability indexes in parallel to meet the power adaptation conditions as the power supply source for the charging pile group, or select the branch with the highest stability index from among the new energy power generation branches that still have surplus power generation to connect in parallel with the main power grid as the power supply source for the charging pile group. S205, update the remaining power generation of all new energy power generation branches, and remove the branches that have been allocated power from the list of new energy power generation branches, repeat steps S204 to S205 until all charging pile groups are traversed and an allocation result is generated, which includes the power supply source of each charging pile group.

7. The electric vehicle charging management method according to claim 1, characterized in that: After S103, the method further includes: S104, based on a preset time interval, periodically execute steps S101 to S103 at each time interval, update the multivariate feature fusion vector of each current renewable energy power generation branch, and update the adjacent power supply windows of each current renewable energy power generation branch according to the power supply windows of each renewable energy power generation branch in the previous time interval.

Images