CN-121092869-B - Electric vehicle state of charge data completion method based on improved SVR

CN121092869BCN 121092869 BCN121092869 BCN 121092869BCN-121092869-B

Abstract

The invention discloses an electric vehicle State of Charge data completing method based on an improved SVR model, which comprises the steps of 1, cleaning existing data, deleting abnormal data and the like, dividing an overall data sample into a fast Charge data sample and a slow Charge data sample, 2, predicting the State of Charge (SOC) of a battery before charging of the slow Charge data sample by using the improved SVR model based on the fast Charge data sample, 3, correcting the predicted value of the SOC before charging of the slow Charge data sample by using physical constraint, and calculating to obtain the SOC after slow Charge based on the corrected SOC before slow Charge, so as to complete the State of Charge data completing. Aiming at the problems that the charging state data of the electric automobile is missing and is difficult to effectively apply, the invention completes data complementation by combining SVR with physical correction, and provides support for improving the quality of the charging data of the electric automobile.

Inventors

HE YE
TIAN ZHENG
WU HONGBIN
JI BIN
BI RUI
HAN PINGPING
WANG LEI
MA YINGHAO

Assignees

合肥工业大学

Dates

Publication Date: 20260512
Application Date: 20250912

Claims (5)

1. The electric vehicle state of charge data completing method based on the improved SVR model is characterized by comprising the following steps: step one, if Y charging pile charging characteristic sequences of ith electric automobile user Current feature set of (2) Then it means Is characterized by alternating current and will Divided into slow charge samples, if Current feature set of (2) Then it means Is characterized by direct current and will Dividing into quick charge samples so as to construct a quick charge characteristic matrix of an electric automobile user with dimension of AxY Slow-charging characteristic matrix of electric automobile user with dimension of D multiplied by Y , wherein, Representing the quick charge characteristics of the y-th charging pile of the a-th electric automobile user, and taking the quick charge characteristics of the y-th charging pile of the a-th electric automobile user as a-th quick charge sample , Representing the slow charging characteristic of the Y charging pile of the d electric automobile user, and taking the slow charging characteristic sequence of the Y charging pile of the d electric automobile user as a d slow charging sample , , , A represents the number of quick charge samples of an electric automobile user, D represents the number of slow charge samples of the electric automobile user, and Y represents the characteristic number of the charge samples of the electric automobile user; construction of a dimension 1 XA Pre-fast Charge SOC feature set for an electric automobile user Electric automobile user's SOC feature set after quick charge of dimension 1 xA , wherein, Representing the pre-fast charge SOC characteristics of the a-th electric vehicle user, Representing the SOC characteristics of a user of the a-th electric automobile after quick charge; step two, using improved SVR model pair Prediction is performed to obtain Corresponding prediction slow charge pre-charge SOC feature set , wherein, Representing the predicted pre-slow charge SOC characteristics of the d-th electric vehicle user; Step 2.1, normalizing AX to obtain a standardized fast charge characteristic matrix of the electric automobile user , wherein, Representing the rapid charging characteristics of the Y-th charging pile standardized by the a-th electric automobile user, and taking the rapid charging characteristics sequence of the Y-th charging pile standardized by the a-th electric automobile user in XA as a-th rapid charging sample ; Based on AX, obtaining the standardized y-th charging pile slow charging characteristic of the d-th electric automobile user by using the formula (1) Thereby obtaining the standardized slow-charging characteristic matrix of the electric automobile user ; (1) In the formula (1): the average value of the y-th quick charge characteristic sequences after the standardization of all electric automobile users is shown, Representing standard deviation of a y-th quick charging characteristic sequence after all electric automobile users are standardized; Step 2.2, calculating the absolute value of the Pearson correlation coefficient between the y-th column quick charge characteristic sequence and FAST in XA Thereby obtaining the average value of all the columns of the quick charge characteristic sequences ; Step 2.3, calculating an a-th fast charge sample in XA The sum of squares of the euclidean norms of (c), thereby obtaining all the quick charge samples The sum of squares of the euclidean norms and the inverse of its mean value as the global scale ; Step 2.4, calculating the scale of the y-th column quick charge characteristic sequence in XA by using the formula (2) : (2) Step 2.5, calculating to obtain an a-th quick charge sample in XA by using the formula (3) And the first Sample of quick charge The values of the kernel matrix between Thereby obtaining the nuclear matrix value of XA : (3) In the formula (3), the amino acid sequence of the compound, Represent the first The y-th charging pile after the standardization of the electric automobile users is provided with a quick charging characteristic; Calculating to obtain the (d) slow charge sample by using the (4) And the a-th fast charge sample The values of the kernel matrix between : (4) Step 2.6, calculating the a-th quick charge sample by using the method (5) Tolerance of (2) : (5) In formula (5): Representation of Is used for the control of the fluctuation coefficient of (c), Representation of Is used for the sparse coefficients of (1), Indicating that the reference tolerance is to be given, A lower tolerance limit is indicated and is indicated, An upper tolerance limit is indicated and is indicated, Representing the fusion weight; step 2.7, constructing an objective function by using the method (6) And solving to obtain And is recorded as the value of ; (6) In formula (6): And Representing 2 dual variables of dimension 1 xA, an , , Representation of Is selected from the group consisting of the a-th element, Representation of A-th element of (a); Representing a transpose; Step 2.8, calculating the slow charge pre-SOC characteristics of the d-th electric vehicle user by using the step (7) : (7) In the formula (7): indicating a baseline offset; If it is Order in principle , Order in principle Thereby obtaining Corresponding prediction slow charge pre-charge SOC feature set ; Step three, based on the capacity upper limit constraint, the lowest SOC constraint under full charge and the acceleration constraint, respectively for Physical correction is carried out, so that a SOC (state of charge) characteristic set before slow charge after the completion of the electric automobile user is obtained Wherein, the method comprises the steps of, Representing the SOC characteristics before slow charge after the user correction of the d-th electric automobile; Based on Calculating SOC (state of charge) feature set after slow charge after completion of electric automobile user Wherein, the method comprises the steps of, And representing the SOC characteristics after slow charge after the user of the d-th electric automobile is complemented.
2. The method for supplementing state of charge data of an electric vehicle based on an improved SVR model according to claim 1, wherein the first step is performed as follows: Step 1.1, all are obtained respectively The method comprises the steps of preprocessing a charging electric quantity set and a current set of each electric automobile user, and correspondingly obtaining a preprocessed charging electric quantity characteristic set And a preprocessed current feature set , wherein, Representing the characteristics of the charge quantity of the i-th electric automobile after the pretreatment, Indicating the current characteristics of the ith electric vehicle user after pretreatment, wherein Z indicates the total number of the electric vehicle users, and Z=A+D; Acquiring all The other Y-2 charging pile charging data sets of the electric automobile users are preprocessed, and the preprocessed charging pile charging feature sets are obtained and then are combined with the other Y-2 charging pile charging data sets of the electric automobile users And Charging pile feature matrix for jointly forming Z rows and Y columns And taking the Y charging pile charging characteristic sequences of the ith electric automobile user as the ith charging sample Wherein, the method comprises the steps of, Representing the charging characteristics of the y-th charging pile of the i-th electric automobile user, ; Step 1.2, will The corresponding charge quantity feature set is recorded as a fast charge quantity feature set , wherein, Representing the characteristic of the quick charge electric quantity of the ith electric automobile user; Will be The corresponding charge quantity feature set is recorded as a slow charge quantity feature set , wherein, Representing the slow charge electric quantity characteristics of the d-th electric automobile user; Will be In (a) The corresponding characteristic set of the slow charge duration is recorded as , wherein, Indicating the slow charge time length characteristic of the d-th electric automobile user; Step 1.3, obtaining Corresponding pre-charge SOC characteristics and post-charge SOC characteristics, thereby constructing a pre-charge SOC characteristic set of an electric vehicle user with dimension of 1 xA Electric automobile user's SOC feature set after quick charge of dimension 1 xA , wherein, Representing the pre-fast charge SOC characteristics of the a-th electric vehicle user, Representing the SOC characteristics after the fast charge of the a-th electric vehicle user.
3. The method for supplementing state of charge data of an electric vehicle based on an improved SVR model according to claim 2, wherein the third step is performed as follows: Step 3.1, calculating an a-th quick charge sample by using the method (8) Is a battery capacity of (a) Thereby obtaining a battery capacity set And deriving therefrom a minimum battery capacity Maximum battery capacity Average battery capacity : (8) Step 3.2, using the pair of (9) Performing first-layer correction to obtain the slow charge pre-charge SOC characteristics of the d electric vehicle user after the first-layer correction : (9) Step 3.3, utilizing the pair of (10) Performing second-layer correction to obtain the slow charge pre-charge SOC characteristics of the d electric vehicle user after the second-layer correction : (10) Step 3.4, pairing using (11) Performing third-layer correction to obtain the SOC characteristics before slow charge after the user of the d-th electric automobile is complemented : (11) Step 3.5, calculating and obtaining the SOC characteristics after slow charge after the user completion of the d-th electric automobile by using the step (12) ; (12)。
4. An electronic device comprising a memory and a processor, wherein the memory is configured to store a program that supports the processor to perform the electric vehicle state of charge data population method of any one of claims 1-3, the processor being configured to execute the program stored in the memory.
5. A computer readable storage medium having a computer program stored thereon, characterized in that the computer program when executed by a processor performs the steps of the electric vehicle state of charge data complementing method of any one of claims 1 to 3.

Description

Electric vehicle state of charge data completion method based on improved SVR Technical Field The invention relates to the field of application of charging data of electric vehicles, in particular to an electric vehicle state of charge data supplementing method based on improved SVR. Background The method is influenced by commute rules, time-of-use electricity prices and the like, the charging requirement of the electric automobile is obviously concentrated in time and space at night and holidays and is easy to overlap with resident electricity, and meanwhile, the distributed renewable energy source output fluctuation, price signal time variation and user travel uncertainty coexist, so that the source-network-load-storage cooperative scheduling and load prediction are more dependent on high-quality charging operation data. In an actual scene, the slow charging pile generally lacks the State of Charge (SOC) of a battery before and after charging or is inconsistent, on one hand, the slow charging pile has low reporting granularity and various measuring channels with a vehicle end, and on the other hand, the slow charging process of a user at home or the slow charging pile has long time, the power change is slow and is influenced by external factors, so that SOC records are missed, misrecorded and noise are caused. The existing completion method often depends on a fixed threshold value or simple linear interpolation, and is difficult to describe a nonlinear relation, and the existing completion method also directly uses an end-to-end model, but ignores a physical feasible domain, and is easy to generate reverse physical predictions such as overcharge, overspeed and the like. The above problems limit the accuracy and reliability of load forecasting, settlement reconciliation, demand response, and other applications. Disclosure of Invention In order to overcome the defects in the prior art, the invention provides an electric vehicle state of charge data supplementing method based on an improved SVR, so that the nonlinear mapping of an improved SVR model learned on a fast charge sample can be migrated to the slow charge sample under the condition that the slow charge SOC is absent or unreliable, and the SOC before and after charging of the slow charge sample is completely and reliably supplemented by combining with necessary physical consistency constraint, thereby providing support for improving the quality of electric vehicle charging data. In order to achieve the above purpose, the technical scheme adopted by the invention is as follows: the invention discloses an electric vehicle state of charge data complement method based on an improved SVR model, which is characterized by comprising the following steps: step one, constructing a quick charge characteristic matrix of an electric automobile user with dimension of AxY Slow-charging characteristic matrix of electric automobile user with dimension of D multiplied by Y, wherein,Representing the quick charge characteristics of the y-th charging pile of the a-th electric automobile user, and taking the quick charge characteristics of the y-th charging pile of the a-th electric automobile user as a-th quick charge sample,Representing the slow charging characteristic of the Y charging pile of the d electric automobile user, and taking the slow charging characteristic sequence of the Y charging pile of the d electric automobile user as a d slow charging sample,,,A represents the number of quick charge samples of an electric automobile user, D represents the number of slow charge samples of the electric automobile user, and Y represents the characteristic number of the charge samples of the electric automobile user; construction of a dimension 1 XA Pre-fast Charge SOC feature set for an electric automobile user Electric automobile user's SOC feature set after quick charge of dimension 1 xA, wherein,Representing the pre-fast charge SOC characteristics of the a-th electric vehicle user,Representing the SOC characteristics of a user of the a-th electric automobile after quick charge; step two, using improved SVR model pair Prediction is performed to obtainCorresponding prediction slow charge pre-charge SOC feature set, wherein,Representing the predicted pre-slow charge SOC characteristics of the d-th electric vehicle user; step three, based on the capacity upper limit constraint, the lowest SOC constraint under full charge and the acceleration constraint, respectively for Physical correction is carried out, so that a SOC (state of charge) characteristic set before slow charge after the completion of the electric automobile user is obtainedWherein, the method comprises the steps of,Representing the SOC characteristics before slow charge after the user correction of the d-th electric automobile; Based on Calculating SOC (state of charge) feature set after slow charge after completion of electric automobile userWherein, the method comprises the steps of,And representing the