CN116278903B - Dynamic charging control method and system for electric automobile based on federal learning - Google Patents
Dynamic charging control method and system for electric automobile based on federal learning Download PDFInfo
- Publication number
- CN116278903B CN116278903B CN202310589331.2A CN202310589331A CN116278903B CN 116278903 B CN116278903 B CN 116278903B CN 202310589331 A CN202310589331 A CN 202310589331A CN 116278903 B CN116278903 B CN 116278903B
- Authority
- CN
- China
- Prior art keywords
- charging
- electric
- electric automobile
- data
- global
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 44
- 238000005457 optimization Methods 0.000 claims abstract description 86
- 238000012549 training Methods 0.000 claims abstract description 43
- 230000004931 aggregating effect Effects 0.000 claims abstract description 13
- 239000004973 liquid crystal related substance Substances 0.000 claims description 14
- 230000008569 process Effects 0.000 claims description 10
- 230000006870 function Effects 0.000 claims description 8
- 230000002776 aggregation Effects 0.000 claims description 6
- 238000004220 aggregation Methods 0.000 claims description 6
- 238000007781 pre-processing Methods 0.000 claims description 3
- 230000005611 electricity Effects 0.000 description 4
- 238000004891 communication Methods 0.000 description 3
- 238000012351 Integrated analysis Methods 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 238000004140 cleaning Methods 0.000 description 2
- 238000013135 deep learning Methods 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 238000004134 energy conservation Methods 0.000 description 1
- 238000005265 energy consumption Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000002986 genetic algorithm method Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000012417 linear regression Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 239000002245 particle Substances 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60L—PROPULSION OF ELECTRICALLY-PROPELLED VEHICLES; SUPPLYING ELECTRIC POWER FOR AUXILIARY EQUIPMENT OF ELECTRICALLY-PROPELLED VEHICLES; ELECTRODYNAMIC BRAKE SYSTEMS FOR VEHICLES IN GENERAL; MAGNETIC SUSPENSION OR LEVITATION FOR VEHICLES; MONITORING OPERATING VARIABLES OF ELECTRICALLY-PROPELLED VEHICLES; ELECTRIC SAFETY DEVICES FOR ELECTRICALLY-PROPELLED VEHICLES
- B60L53/00—Methods of charging batteries, specially adapted for electric vehicles; Charging stations or on-board charging equipment therefor; Exchange of energy storage elements in electric vehicles
- B60L53/60—Monitoring or controlling charging stations
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60L—PROPULSION OF ELECTRICALLY-PROPELLED VEHICLES; SUPPLYING ELECTRIC POWER FOR AUXILIARY EQUIPMENT OF ELECTRICALLY-PROPELLED VEHICLES; ELECTRODYNAMIC BRAKE SYSTEMS FOR VEHICLES IN GENERAL; MAGNETIC SUSPENSION OR LEVITATION FOR VEHICLES; MONITORING OPERATING VARIABLES OF ELECTRICALLY-PROPELLED VEHICLES; ELECTRIC SAFETY DEVICES FOR ELECTRICALLY-PROPELLED VEHICLES
- B60L53/00—Methods of charging batteries, specially adapted for electric vehicles; Charging stations or on-board charging equipment therefor; Exchange of energy storage elements in electric vehicles
- B60L53/60—Monitoring or controlling charging stations
- B60L53/62—Monitoring or controlling charging stations in response to charging parameters, e.g. current, voltage or electrical charge
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60L—PROPULSION OF ELECTRICALLY-PROPELLED VEHICLES; SUPPLYING ELECTRIC POWER FOR AUXILIARY EQUIPMENT OF ELECTRICALLY-PROPELLED VEHICLES; ELECTRODYNAMIC BRAKE SYSTEMS FOR VEHICLES IN GENERAL; MAGNETIC SUSPENSION OR LEVITATION FOR VEHICLES; MONITORING OPERATING VARIABLES OF ELECTRICALLY-PROPELLED VEHICLES; ELECTRIC SAFETY DEVICES FOR ELECTRICALLY-PROPELLED VEHICLES
- B60L53/00—Methods of charging batteries, specially adapted for electric vehicles; Charging stations or on-board charging equipment therefor; Exchange of energy storage elements in electric vehicles
- B60L53/60—Monitoring or controlling charging stations
- B60L53/63—Monitoring or controlling charging stations in response to network capacity
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60L—PROPULSION OF ELECTRICALLY-PROPELLED VEHICLES; SUPPLYING ELECTRIC POWER FOR AUXILIARY EQUIPMENT OF ELECTRICALLY-PROPELLED VEHICLES; ELECTRODYNAMIC BRAKE SYSTEMS FOR VEHICLES IN GENERAL; MAGNETIC SUSPENSION OR LEVITATION FOR VEHICLES; MONITORING OPERATING VARIABLES OF ELECTRICALLY-PROPELLED VEHICLES; ELECTRIC SAFETY DEVICES FOR ELECTRICALLY-PROPELLED VEHICLES
- B60L53/00—Methods of charging batteries, specially adapted for electric vehicles; Charging stations or on-board charging equipment therefor; Exchange of energy storage elements in electric vehicles
- B60L53/60—Monitoring or controlling charging stations
- B60L53/65—Monitoring or controlling charging stations involving identification of vehicles or their battery types
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/60—Other road transportation technologies with climate change mitigation effect
- Y02T10/70—Energy storage systems for electromobility, e.g. batteries
Landscapes
- Engineering & Computer Science (AREA)
- Power Engineering (AREA)
- Transportation (AREA)
- Mechanical Engineering (AREA)
- Charge And Discharge Circuits For Batteries Or The Like (AREA)
Abstract
The invention discloses an electric automobile dynamic charging control method and system based on federal learning, wherein the method comprises the following steps: collecting charging data of a plurality of electric vehicles in a designated area respectively for training based on federal learning, aggregating training results of the edge computing terminals, and obtaining global charging optimization models corresponding to different vehicle types according to the duty ratio of data of different vehicle types in the edge computing terminals; when the electric automobile is connected to a power grid through the intelligent charging pile, acquiring the automobile type and state information of the electric automobile; calling a corresponding global charging optimization model according to the vehicle type and state information of the electric vehicle to be connected; and controlling the charging of each electric automobile according to the state and the charging requirement of each electric automobile to be connected in the platform area and the corresponding global charging optimization model. The method has the advantages of simple realization method, low cost, high charging efficiency and precision, strong flexibility and expandability and the like, and can realize data privacy protection.
Description
Technical Field
The invention relates to the technical field of electric vehicle charging control, in particular to an electric vehicle dynamic charging control method and system based on federal learning.
Background
At present, an electric automobile is usually charged by directly adopting a rough step power charging mode, namely, a charging pile is used for charging with high power or power correction according to the battery electric quantity state of the electric automobile, for example, the electric automobile is charged with high power when the battery electric quantity is smaller, and the electric automobile is charged with lower power when the battery electric quantity is larger. With the continuous increase of the holding capacity of electric vehicles, the charging service demands are rapidly increased, the types of electric vehicles are various, the optimal charging curves of different types of electric vehicles are greatly different due to different configurations of power batteries, and even the characteristics of the same vehicle type in different battery states are different. The traditional direct adoption of the rough type step power charging mode can not be matched with the charging characteristic curves of different vehicle types to realize optimal charging, so that the charging efficiency can be reduced, the energy consumption cost can be increased, the safety of a power battery system can be influenced, and the current requirements on the charging efficiency of an electric vehicle, the energy conservation and emission reduction and the like can not be met.
In order to solve the above-mentioned problem, there is a practitioner who puts forward to train a charge prediction model of an electric vehicle by monitoring the charge state of the electric vehicle and using a deep learning method to early warn the fault state of the electric vehicle in the charging process. However, by adopting a deep learning charging prediction model construction mode, a large amount of charging information of users needs to be uploaded for centralized training, the privacy problem of user data can be related, and the cost for acquiring a large amount of user data for learning is high.
Disclosure of Invention
The technical problem to be solved by the invention is as follows: aiming at the technical problems existing in the prior art, the invention provides the dynamic charging control method and system for the electric automobile based on federal learning, which have the advantages of simple implementation method, low cost, high charging efficiency and precision, and strong flexibility and expandability, and can realize data privacy protection.
In order to solve the technical problems, the technical scheme provided by the invention is as follows:
a dynamic charging control method of an electric automobile based on federal learning comprises the following steps:
collecting charging data of a plurality of electric vehicles in a designated area respectively for training based on federal learning, aggregating training results of the edge computing terminals, and obtaining global charging optimization models corresponding to different vehicle types according to the duty ratio of data of different vehicle types in the edge computing terminals;
when the electric automobile is connected to a power grid through the intelligent charging pile, acquiring the automobile type and state information of the electric automobile;
calling the corresponding global charging optimization model according to the vehicle type and state information of each electric vehicle to be connected;
and controlling the charging of each electric automobile according to the state of each electric automobile to be connected in the platform area, the charging requirement and the corresponding global charging optimization model, wherein the charging requirement comprises charging time and charging mode.
Further, the charging data comprises any one or more of a vehicle type, a charging start time, a charging end time, a total charging amount and a battery state of charge, and further comprises any one or more of a charging power curve, a temperature and a humidity, wherein the temperature curve, the temperature curve and the humidity curve are changed with time, and the state information comprises any one or more of a power battery state of charge, a battery temperature, a current temperature, the humidity and a charging start time.
Further, the step of collecting charging data of a plurality of electric vehicles for training by each edge computing terminal in a designated area based on federal learning, aggregating training results of each edge computing terminal, and obtaining global charging optimization models corresponding to different vehicle types according to the duty ratio of data of different vehicle types in each edge computing terminal includes:
data preprocessing is carried out on the collected charging data at each edge computing terminal, and classification is carried out according to vehicle types;
training the classified charging data to obtain charging optimization models corresponding to different vehicle types;
aggregating the charge optimization models trained by the edge computing terminals to obtain a global unified optimization model;
and obtaining the global charging optimization model corresponding to different vehicle types according to the global unified optimization model and the duty ratio of the data of different vehicle types in each edge computing terminal.
Further, the expression of the global charging optimization model corresponding to different vehicle types is as follows:
wherein, the liquid crystal display device comprises a liquid crystal display device,representation ofMElectric automobile of motorcycle type is intA global charge optimization model of the moment in time,D n (M) Represented by the numbernIs involved in model training in edge computing terminalsMNumber of vehicle model data sets->At all of the participation in the trainingMNumber of vehicle modelsThe ratio of (a)>，/>Representing the number of edge computing terminals,W N representing weight vector, ++>、…、/>Respectively represent the weights corresponding to the respective influencing factors,Grepresenting a charging power influencing factor vector,/->、…、/>Each of the influence factors is represented by a respective one,Iindicating the number of influencing factors.
Further, controlling the charging of each electric automobile according to the state and the charging requirement of each electric automobile to be connected in the platform area and the corresponding global charging optimization model includes:
calculating according to the state of each electric automobile to be connected and the corresponding global charging optimization model to obtain the initial value of the charging power of each electric automobile to be connectedPz；
According to the pressure coefficient of transformer in the transformer areaαThe initial value of the charging power to be connected into each electric automobilePz, adjusting the charging power of each electric automobile to be connected in, the charging mode of each electric automobile to be connected in, and the required charging time length, wherein the transformer pressure coefficient of the transformer areaαFor the current power of the transformerPdAnd rated powerPeIs a ratio of (2).
Further, if the charging modes of the electric vehicles to be connected in the transformer area are all charging efficiency priority, and the charging modes are%Pz+Pd)/Pe exceeds a first preset ratio, and according to the charged time of each electric automobile to be connectedtyCharged electric quantityQyAnd an estimated charge amountQAdjusting charging power; if the charging mode of the electric vehicles to be connected in the station area is the charging efficiency priority and the charging mode of the electric vehicles to be connected in the station area is the load adjustment priority, the charging mode in the station area is adjusted to be the charging power of each electric vehicle with the load adjustment priority.
Further, the adjusting the charging mode in the platform area to adjust the charging power of each electric automobile with priority for load adjustment includes: the reduced charging power of each electric automobile isPz/m 1 ，m 1 Adjusting the number of electric vehicles with priority for charging mode as load if anyIn the time-course of which the first and second contact surfaces,Py(n) Represent the firstnThe original charging power of the electric automobile before adjustment,a 0 indicating a second preset ratio, and reducing the charging power of the electric automobilea 0 Py(n) Each of which is prioritized in charging efficiency by the charging system in excess of the charging rateAnd adjusting the electric automobile.
Furthermore, the charging modes of the electric vehicles to be connected in the transformer area are all charging efficiency priority, and the charging modes are as followsPz+Pd)/Pe exceeds a first preset ratio, and the objective function of the charging power is adjusted as follows:
wherein, the liquid crystal display device comprises a liquid crystal display device,ris the serial number of the electric automobile,represent the firstrReduced charging power of electric vehicle, constraint condition is +.><a 0 Py(r)，m 0 For the number of electric vehicles whose charging mode is priority in charging efficiency,Py(r) Represent the firstrThe original charging power of the electric automobile before adjustment,a 0 representing a second preset ratio;
if the charging mode of the electric automobile to be connected in the station area is charging efficiency priority, and the charging mode of the electric automobile to be connected in part is load adjustment priority, the objective function of adjusting the charging power is as follows:
。
further, the method further comprises the step of continuously judging deviation between charging data and actual charging data in the charging process of each electric automobile, and transmitting the charging data to an edge computing terminal to perform charging optimization model training if the deviation is in a preset range, so that iterative optimization of an electric automobile charging scheme based on federal learning is realized.
An electric vehicle dynamic charge control system based on federal learning, comprising:
the edge computing terminals are respectively used for respectively collecting charging data of a plurality of electric vehicles in a designated area for training based on federal learning;
the global model aggregation module is used for aggregating training results of the edge computing terminals and obtaining global charging optimization models corresponding to different vehicle types according to the duty ratio of the data of the different vehicle types in the edge computing terminals;
the information acquisition module is used for acquiring the vehicle type and state information of the electric vehicle to be connected when the electric vehicle is connected into the power grid through the intelligent charging pile;
the model calling module is used for calling the corresponding global charging optimization model according to the vehicle type and the state information of each electric vehicle to be connected;
the charging control module is used for controlling the charging of each electric automobile according to the state, the charging requirement and the corresponding global charging optimization model of each electric automobile to be connected in the platform area, and the charging requirement comprises the charging duration and the charging mode.
Compared with the prior art, the invention has the advantages that:
1. according to the invention, the global charging mode of the electric vehicle is established by utilizing federal learning, the charging data of a plurality of electric vehicles are collected by each edge computing terminal to carry out local training, the charging characteristics of different types of electric vehicles are fully learned, the training results obtained by each edge computing terminal are aggregated by the federal learning model, so that the charging demand data of each electric vehicle is subjected to integrated analysis, the global charging optimization model capable of reflecting the overall charging demand is obtained, meanwhile, according to the occupation ratio condition of the data of different vehicle types in each edge computing terminal, the global charging optimization model corresponding to different vehicle types can be obtained, the characteristics of the electric vehicles of different vehicle types during charging can be rapidly and accurately predicted by the model, therefore, the charging modes of various electric vehicles can be optimized by fully utilizing the federal learning mode, the accurate and intelligent charging control of various types of electric vehicles can be realized on the premise of ensuring the privacy protection of user data, the charging efficiency and the charging accuracy of different types of electric vehicles are effectively improved, meanwhile, the charging time and the cost are reduced, and the overlarge burden on a power grid can be avoided.
2. According to the invention, the corresponding global optimal charging model is called from the edge computing terminal according to the state of the electric automobile, the charging power initial value is determined based on the global optimal charging model, and meanwhile, the charging power is adjusted by combining the transformer pressure coefficient of the station area, the access condition of the charging station and the charging requirement of a user, so that the electric automobiles in different automobile types and different states can be optimally charged.
Drawings
Fig. 1 is a schematic implementation flow chart of an electric vehicle dynamic charging control method based on federal learning in this embodiment.
Fig. 2 is a schematic structural diagram of an electric vehicle dynamic charging control system based on federal learning in this embodiment.
Detailed Description
The invention is further described below in connection with the drawings and the specific preferred embodiments, but the scope of protection of the invention is not limited thereby.
As used in this disclosure, the terms "a," "an," "the," and/or "the" are not intended to be limiting, but rather are to be construed as covering the singular and the plural, unless the context clearly dictates otherwise. The terms "first," "second," and the like, as used in this disclosure, do not denote any order, quantity, or importance, but rather are used to distinguish one element from another. Likewise, the word "comprising" or "comprises", and the like, means that elements or items preceding the word are included in the element or item listed after the word and equivalents thereof, but does not exclude other elements or items.
Federal learning is a machine learning method that can be used for distributed model training without exposing user privacy. In federal learning, multiple participants use their own local data for model training, and aggregate the training results to obtain a global model. According to the invention, the global charging mode of the electric vehicle is established by utilizing federal learning, the charging data of a plurality of electric vehicles are collected by each edge computing terminal to carry out local training, the charging characteristics of different types of electric vehicles are fully learned, the training results obtained by each edge computing terminal are aggregated by the federal learning model, so that the charging demand data of each electric vehicle is subjected to integrated analysis, the global charging optimization model capable of reflecting the overall charging demand is obtained, meanwhile, according to the occupation ratio condition of the data of different vehicle types in each edge computing terminal, the global charging optimization model corresponding to different vehicle types can be obtained, the characteristics of the electric vehicles of different vehicle types during charging can be rapidly and accurately predicted by the model, therefore, the charging modes of various electric vehicles can be optimized by fully utilizing the federal learning mode, the accurate and intelligent charging control of various types of electric vehicles can be realized on the premise of ensuring the privacy protection of user data, the charging efficiency and the charging accuracy of different types of electric vehicles are effectively improved, meanwhile, the charging time and the cost are reduced, and the overlarge burden on a power grid can be avoided.
As shown in fig. 1, the steps of the electric vehicle dynamic charging control method based on federal learning in this embodiment include:
step S01, building a federal learning model: and respectively collecting charging data of a plurality of electric vehicles in the designated area based on federal learning by each edge computing terminal, training, aggregating training results of each edge computing terminal, and obtaining global charging optimization models corresponding to different vehicle types according to the duty ratio of data of different vehicle types in each edge computing terminal.
And S101, collecting charging data of a plurality of electric vehicles by each edge computing terminal, preprocessing the collected charging data, and classifying according to vehicle types.
The charging data specifically includes a vehicle model, a charging start time, a charging end time, a total charge amount, a battery state of charge, a charging rate, a charging power curve over time, a temperature over time, a humidity over time, and the like. The charging data can be specifically collected through intelligent charging piles or arranged sensors, and the collected data is uploaded to corresponding edge computing terminals for local training. The edge computing terminal can be various devices such as a server, a processor, a PC and the like, and can also be realized by adopting a charging management system which is configured in the charging station and can realize charging management.
In a specific application embodiment, first, charging data of a plurality of electric vehicles are collected through intelligent charging piles or other sensors and transmitted to edge computing terminals corresponding to charging stations, and a data set of each electric vehicle can be expressed asX={M，V s ，V e ，P(t)，Q，C L (t)，C T (t)，T(t)，H(t) }, whereinMThe model of the vehicle is indicated,V s the charging start time is indicated as the time at which charging is started,V e the time at which the charging is completed is indicated,P(t) A curve representing the change of the charging power with time,Qindicating the total charge amount,C L (t) AndC T (t) Curves representing the state of charge of the battery and the temperature over time respectively,T(t) Indicating the temperature of the air over time,H(t) Indicating humidity over time. Then, data cleaning and processing are carried out on the collected charging data at each edge computing terminal, and the cleaned data are carried out according to the vehicle typeMThe classification process is performed, and the data format can be expressed as:Y M ={M，P(t)，Q，C L (t)，C T (t)，T(t)，H(t)}。
and S102, training the classified charging data to obtain charging optimization models corresponding to different vehicle types.
In the embodiment, the constant power model is based on chargingP S (t) Establishing a charging model based on federal learning, and charging a constant power modelP S (t) Namely, the model is charged according to constant power, and the classified charging data is used for training, so that the charging optimization models of different vehicle types can be obtainedP M (t). Charging optimization modelP M (t) The expression of (c) may specifically be:
（1）
in the method, in the process of the invention,Wthe weight vector is represented by a weight vector,、…、/>respectively represent the weights corresponding to the respective influencing factors,Grepresenting a charging power influencing factor vector,/->、…、/>Each of the influence factors is represented by a respective one,Iindicating the number of influencing factors.
And S103, aggregating the charge optimization models trained by the edge computing terminals to obtain a global unified optimization model.
In this embodiment, each edge computing terminal is trained to complete the charge optimization modelP M (t) Uploading to a cloud server, and aggregating the training results of all edge computing terminals by the cloud server to form a global unified optimization modelP U (t). Only the models obtained by the edge computing terminals are aggregated in the cloud server, and local charging data of the electric automobile are not required to be exchanged, so that data privacy protection can be ensured. It can be understood that the cloud server may be a remote server, a controller, a control terminal, or other types of devices.
The global unified optimization modelP U (t) Namely, the charge optimization is performed for the situation that the model of the vehicle is unknown, and the calculation mode can be expressed as follows:
（2）
wherein, the liquid crystal display device comprises a liquid crystal display device,representing the number of edge computing terminals.
And S104, obtaining global charging optimization models corresponding to different vehicle types according to the global unified optimization model and the duty ratio of the data of the different vehicle types in each edge computing terminal.
Global unified optimization modelP U (t) In order to accurately obtain global charge optimization models for different vehicle types for charge optimization under the condition of unknown vehicle models, the embodiment further combines the duty ratio condition of each vehicle type data participating in model training in the edge computing terminal to obtain the global charge optimization models corresponding to different vehicle typesP S (t) The method is global charging optimization for the known vehicle type. Global charging optimization model corresponding to different vehicle typesP S (t) The computational expression of (2) can be expressed as:
（3）
wherein, the liquid crystal display device comprises a liquid crystal display device,representation ofMElectric automobile of motorcycle type is intA global charge optimization model of the moment in time,D n (M) Represented by the numbernIs involved in model training in edge computing terminalsMNumber of vehicle model data sets->At all of the participation in the trainingMNumber of vehicle modelsThe ratio of (a), namely:
（4）
the cloud server then optimizes the global charging scheme obtained aboveP M (t) Issuing to each edge computing terminal to establish a charging optimization scheme of different types of electric vehicles based on federal learning through integration, wherein the global charging optimization schemeP M (t) Two types of global models are included: global unified optimization modelP U (t) Global charging optimization model corresponding to different vehicle typesP S (t) By globally unifying optimization modelsP U (t) Global charge optimization model for charge optimization under unknown vehicle modelP S (t) Global charge optimization corresponding to known vehicle models of various types.
Step S02, information acquisition: when the electric automobile is connected into the power grid through the intelligent charging pile, the automobile type and state information of the electric automobile are obtained.
In this embodiment, the state information is specific to the state of charge of the power batteryC L Battery temperatureC T Current air temperatureTAnd humidity ofHCharge start timeV s Etc. When the electric automobile is connected into the power grid through the intelligent charging pile, the specific automobile type of the electric automobile can be identified through the intelligent charging pile or other sensorsMAnd reads the state information of the electric automobile.
S03, model calling: and calling a corresponding global charging optimization model according to the current electric automobile type and the state information.
Through the global charging optimization model corresponding to different vehicle types obtained in the step S01P S (t) When the electric automobile needs to be connected to the power grid, a corresponding global charging optimization model can be matched according to the automobile type of the electric automobileP S (t) The comprehensive state information in the matching process can further improve the matching precision so as to accurately determine the global charging optimization model applicable to different vehicle types and different vehicle statesP S (t)。
Step S04, charging control: controlling the charging of each electric automobile according to the state of each electric automobile to be connected in the platform area, the charging requirement and the corresponding global charging optimization model, wherein the charging requirement comprises the charging durationt s Charging modeβ。
In the present embodiment, the charging modeβThe values of (1) are 0 and 1,0 represents the charging efficiency priority, and 1 represents the load adjustment priority.The charge demand may also include an estimated charge amountQAnd the like, and can be determined according to the selection of a charging user.
S401, calculating to obtain initial charging power values of the electric vehicles to be connected according to the states of the electric vehicles to be connected and the corresponding global charging optimization modelPz。
In the present embodiment, according to the state of charge of the power battery to be connected to the electric vehicleC L Battery temperatureC T Current air temperatureTAnd humidity ofHCharge start timeV s Global optimization model based on corresponding vehicle type of electric vehicle to be connectedP S (t) Calculating to obtain initial value of charging powerPz。
S402, reading the pressure coefficient of the transformer in the transformer areaαAccording to the pressure coefficient of transformer in the transformer areaαCharging power initial value to be connected into each electric automobilePAnd z, adjusting the charging power of each electric automobile to be connected in the charging mode of each electric automobile to be connected in.
In this embodiment, if the charging modes of the electric vehicles to be connected in the station area are all charging efficiency priority, andPz+Pd)/Pe exceeds a first preset ratio, and according to the charged time of each electric automobile to be connectedtyCharged electric quantityQyAnd an estimated charge amountQAdjusting the charging power, wherein an objective function of the adjusting charging power is as follows:
（5）
wherein, the liquid crystal display device comprises a liquid crystal display device,ris the serial number of the electric automobile,represent the firstrReduced charging power of electric vehicle, constraint condition is +.><a 0 Py(r)，m 0 For the number of electric vehicles whose charging mode is priority in charging efficiency,Py(r) Represent the firstrThe original charging power of the electric automobile before adjustment,a 0 representing a second preset ratio.
In this embodiment, if the charging mode of the electric vehicle to be connected in the bay is charging efficiency priority and the charging mode of the electric vehicle to be connected in part is load adjustment priority, the charging mode in the bay is adjusted to be charging power of each electric vehicle with load adjustment priority. The charging power adjustment method for adjusting the charging mode in the platform area for each electric automobile with priority for load adjustment specifically comprises the following steps: the reduced charging power of each electric automobile isPz/m 1 ，m 1 Adjusting the number of electric vehicles with priority for charging mode as load if anyIn the time-course of which the first and second contact surfaces,Py(r) Represent the firstrThe original charging power of the electric automobile of the vehicle,a 0 indicating a second preset ratio, and reducing the charging power of the electric automobilea 0 Py(r) More than part of the electric automobiles with priority for charging efficiency are adjusted by the charging mode, and the objective function for adjusting the charging power is specifically as follows:
（6）
according to the embodiment, the corresponding global optimal charging model is called from the edge computing terminal according to the state of the electric automobile, and the initial value of the charging power is determined based on the global optimal charging modelPz, simultaneously combining transformer pressure coefficients in the station areaαCharging power is adjusted according to the access condition of the charging station and the charging requirement of a user, and the electric vehicles in different vehicle types and different states can be optimally charged.
Electric automobile with charging station to be accessed is set as the firstMVehicles, i.e. with access beforeM-1 electric vehicle is charging, wherein the charging mode isβThe number of the values of 0 ism 0 Charging modeβThe number of the values of 1 ism 1 ，m 0 +m 1 =MThe following is the first embodimentOne preset ratio is 85%, the second preset ratio is 40% as an example, and the above detailed steps for realizing the electric automobile charging control are as follows:
if the charging mode of the electric automobile to be connected is 0, the charging mode is as followsPz+Pd)/Pe is less than or equal to 85 percent, the power is obtained according to calculationPz, charging;
when (a ]Pz+Pd)/Pe>At 85%, ifm 1 =0, i.e. the charging modes of the electric vehicles in the areas are all efficiency priority, according to the charged time of each electric vehicletyCharged electric quantityQyAnd an estimated charge amountQAdjusting the charging scheme according to the formula (5) as an objective function, wherein the constraint condition isf(ty(r),Qy(r),Q(r))<0.4Py(r)，Py(r) Represent the firstrThe original charging power of the electric vehicle, namely the reduced charging power cannot exceed 40% of the original charging power;
if it ism 1 >0, the charging mode in the adjustment area is 1m 1 Charging scheme of electric automobile, each electric automobile reduces charging power to bePz/m 1 The method comprises the steps of carrying out a first treatment on the surface of the If presentPz/m 1 >0.4Py(r) When the influence on the original electric charging power is large, the electric charging power of the electric automobile is reduced by 0.4Py(r) More than part of the charge is 0m 0 The electric car is adjusted according to the objective function of the formula (6). Through the steps, the optimal charging of various electric automobiles can be realized.
The dynamic adjustment of the charging scheme can be realized by adopting a linear regression method, a particle swarm method, a genetic algorithm method and other machine learning methods.
In this embodiment, the method further includes continuously judging deviation between charging data and actual charging data in the charging process of each electric automobile, and if the deviation is in a preset range, transmitting the charging data to an edge computing terminal to perform charging optimization model training, so as to realize iterative optimization of the charging scheme of the electric automobile based on federal learning.
In a specific application embodiment, in a charging process, a charging station keeps communication with an electric automobile at all times, a power battery state is continuously detected, a charging plan is dynamically adjusted, charging data are recorded, if abnormal charging data conditions occur, the charging data are marked, the charging data are sent to an edge computing terminal to train a charging model, and the retrained model is sent to a cloud server. The abnormal condition judgment standard of the charging data can be specifically a standard deviation method, different battery temperatures and charge states are corresponding to different charging powers, and the standard deviation is calculated through data cleaning and statistics; when the corresponding data deviation in the actual charging data is between 2 and 3 times of standard deviation, the charging model is directly sent to an edge computing terminal to train the charging model, and the charging scheme is iteratively optimized based on a federal learning method through a cloud server; when the corresponding data deviation in the actual charging data exceeds 3 times of standard deviation, whether the charging data are invalid data can be further judged, and when the charging data are valid data, the charging data are sent to an edge computing terminal to train a charging model and iteratively optimize a charging scheme based on a federal learning method through a cloud server; if the data is judged to be invalid, the data is directly discarded. It can be understood that the standard deviation judging range can be specifically configured according to actual requirements.
Furthermore, a reward mechanism can be added to increase the control flexibility, for example, the charging mode is set to be different electricity prices, and the set efficiency priority electricity price is higher than the load adjustment priority. The intelligent charging pile can realize automatic control through communication with a charging management system, when the step electricity price exists, the charging cost preference option is increased, and when the step electricity price does not exist, the intelligent charging pile is influenced by the integral load of the distribution transformer area.
Furthermore, various data sources can be comprehensively used in the federal learning model building process, and besides charging data, for example, the testing data of a battery manufacturer to a battery, the testing data of an electric automobile manufacturer to an electric automobile and the like can be adopted, so that the accuracy and individuation degree of a charging plan are further improved. The method can be used for dynamically optimizing the charging scheme by combining weather, real-time states of a power grid and a power battery in the actual charging process, so that a more accurate personalized charging scheme is further formed according to different vehicle type information, the charging cost is reduced, the charging efficiency is improved, and the influence of charging on the power grid is reduced.
Furthermore, the federal learning model can also adopt encryption technology to improve the privacy protection and the safety reliability of the user data.
As shown in fig. 2, the electric vehicle dynamic charging control system based on federal learning in this embodiment includes:
the edge computing terminals are respectively used for respectively collecting charging data of a plurality of electric vehicles in a designated area for training based on federal learning;
the global model aggregation module is used for aggregating training results of all the edge computing terminals and obtaining global charging optimization models corresponding to different vehicle types according to the duty ratio of the data of different vehicle types in all the edge computing terminals;
the information acquisition module is used for acquiring the vehicle type and state information of the electric vehicle to be connected when the electric vehicle is connected into the power grid through the intelligent charging pile;
the model calling module is used for calling a corresponding global charging optimization model according to the vehicle type and the state information of each electric vehicle to be connected;
the charging control module is used for controlling the charging of each electric automobile according to the state of each electric automobile to be connected into the platform area, the charging requirement and the corresponding global charging optimization model, wherein the charging requirement comprises the charging duration and the charging mode.
In a specific application embodiment, the edge computing terminal, the model calling module and the charging control module are realized through a charging management system, and the model aggregation module can be realized by a cloud server and the like, namely, the charging management system collects charging data of a plurality of electric vehicles and obtains charging optimization models of different vehicle types by utilizing a federal learning model; the model aggregation module aggregates and anonymizes the charging data of the electric vehicles, and distributes the anonymized data to a plurality of participants for model training so as to obtain global charging optimization models of different electric vehicles. The intelligent charging pile is used for monitoring and managing the electric automobile in real time, and when the intelligent charging pile acquires the automobile type and state information of the electric automobile to be connected, the automatic control of charging of the electric automobile is realized through communication with the charging management system.
The electric vehicle dynamic charging control system based on federal learning in this embodiment corresponds to the electric vehicle dynamic charging control method based on federal learning in this embodiment, and will not be described in detail here.
According to the invention, the federal learning model is built by collecting the charging data of a plurality of electric vehicles, the charging characteristics of different types of electric vehicles are learned, and the charging optimization model of each electric vehicle is subjected to aggregation analysis to obtain the global charging optimization model capable of reflecting the overall charging requirement.
The foregoing is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. While the invention has been described with reference to preferred embodiments, it is not intended to be limiting. Therefore, any simple modification, equivalent variation and modification of the above embodiments according to the technical substance of the present invention shall fall within the scope of the technical solution of the present invention.
Claims (10)
1. The dynamic charging control method for the electric automobile based on federal learning is characterized by comprising the following steps:
collecting charging data of a plurality of electric vehicles in a designated area respectively for training based on federal learning, aggregating training results of the edge computing terminals, and obtaining global charging optimization models corresponding to different vehicle types according to the duty ratio of data of different vehicle types in the edge computing terminals;
when the electric automobile is connected to a power grid through the intelligent charging pile, acquiring the type and state information of the electric automobile to be connected;
calling the corresponding global charging optimization model according to the vehicle type and state information of each electric vehicle to be connected;
and controlling the charging of each electric automobile according to the state of each electric automobile to be connected in the platform area, the charging requirement and the corresponding global charging optimization model, wherein the charging requirement comprises charging time and charging mode.
2. The federal learning-based electric vehicle dynamic charge control method according to claim 1, wherein the charge data includes any one or more of a vehicle model, a charge start time, a charge end time, a total charge amount, and a battery state of charge, and further includes any one or more of a time-varying curve of charge power, a time-varying curve of temperature, a time-varying air temperature, and a time-varying humidity, and the state information includes any one or more of a power battery state of charge, a battery temperature, a current air temperature and humidity, and a charge start time.
3. The method for controlling dynamic charging of electric vehicles based on federal learning according to claim 1, wherein the step of collecting charging data of a plurality of electric vehicles for training by each edge computing terminal in a designated area based on federal learning, aggregating training results of each edge computing terminal, and obtaining global charging optimization models corresponding to different vehicle types according to the duty ratio of data of different vehicle types in each edge computing terminal comprises:
data preprocessing is carried out on the collected charging data at each edge computing terminal, and classification is carried out according to vehicle types;
training the classified charging data to obtain charging optimization models corresponding to different vehicle types;
aggregating the charge optimization models trained by the edge computing terminals to obtain a global unified optimization model;
and obtaining the global charging optimization model corresponding to different vehicle types according to the global unified optimization model and the duty ratio of the data of different vehicle types in each edge computing terminal.
4. The federal learning-based dynamic charge control method for an electric vehicle according to claim 3, wherein the expression of the global charge optimization model corresponding to different vehicle types is:
wherein, the liquid crystal display device comprises a liquid crystal display device,representation ofMElectric automobile of motorcycle type is intA global charge optimization model of the moment in time,D n (M) Represented by the numbernIs involved in model training in edge computing terminalsMNumber of vehicle model data sets->At all of the participation in the trainingMNumber of vehicle modelsThe ratio of (a)>，/>Representing the number of edge computing terminals,W N representing weight vector, ++>、…、/>Respectively represent the weights corresponding to the respective influencing factors,Grepresenting a charging power influencing factor vector,/->、…、/>Each of the influence factors is represented by a respective one,Ithe superscript T indicates the number of influencing factors and the transpose operation.
5. The method for controlling dynamic charging of electric vehicles based on federal learning according to claim 1, wherein controlling charging of each electric vehicle according to the state of each electric vehicle to be connected in the platform area, the charging requirement, and the corresponding global charging optimization model comprises:
calculating according to the state of each electric automobile to be connected and the corresponding global charging optimization model to obtain the initial value of the charging power of each electric automobile to be connectedPz；
According to the pressure coefficient of transformer in the transformer areaαThe initial value of the charging power to be connected into each electric automobilePz, adjusting the charging power of each electric automobile to be connected in, the charging mode of each electric automobile to be connected in, and the required charging time length, wherein the transformer pressure coefficient of the transformer areaαFor the current power of the transformerPdAnd rated powerPeIs a ratio of (2).
6. The method for dynamically charging electric vehicles based on federal learning according to claim 5, wherein if charging modes of electric vehicles to be connected in the transformer area are charging efficiency priority, andPz+Pd)/Pe exceeds a first preset ratio, and according to the charged time of each electric automobile to be connectedtyCharged electric quantityQyAnd an estimated charge amountQAdjusting charging power; if the charging mode of the electric vehicles to be connected in the station area is the charging efficiency priority and the charging mode of the electric vehicles to be connected in the station area is the load adjustment priority, the charging mode in the station area is adjusted to be the charging power of each electric vehicle with the load adjustment priority.
7. The federal learning-based electric vehicle dynamic charge control method according to claim 6, wherein the adjustment of the intra-bay charging mode is load adjustment priorityThe charging power of each electric automobile comprises: the reduced charging power of each electric automobile isPz/m 1 ，m 1 Adjusting the number of electric vehicles with priority for charging mode as load if anyIn the time-course of which the first and second contact surfaces,Py(n) Represent the firstnThe original charging power of the electric automobile before adjustment,a 0 indicating a second preset ratio, and reducing the charging power of the electric automobilea 0 Py(n) The electric vehicles, the charging efficiency of which is prioritized by the charging system, are adjusted in excess of the charging rate.
8. The method for dynamically charging and controlling electric vehicles based on federal learning according to claim 6, wherein the charging modes of the electric vehicles to be connected in the service area are charging efficiency priority, andPz+Pd)/Pe exceeds a first preset ratio, and the objective function of the charging power is adjusted as follows:
wherein, the liquid crystal display device comprises a liquid crystal display device,ris the serial number of the electric automobile,represent the firstrReduced charging power of electric vehicle, constraint condition is +.><a 0 Py(r)，m 0 For the number of electric vehicles whose charging mode is priority in charging efficiency,Py(r) Represent the firstrThe original charging power of the electric automobile before adjustment,a 0 representing a second preset ratio;
if the charging mode of the electric automobile to be connected in the station area is charging efficiency priority, and the charging mode of the electric automobile to be connected in part is load adjustment priority, the objective function of adjusting the charging power is as follows:
。
9. the method for controlling dynamic charging of electric vehicles based on federal learning according to any one of claims 1 to 8, further comprising continuously judging deviation between charging data and actual charging data in each electric vehicle charging process, and if the deviation is within a preset range, transmitting the charging data to an edge computing terminal to perform charging optimization model training, so as to realize iterative optimization of an electric vehicle charging scheme based on federal learning.
10. Electric automobile dynamic charge control system based on federal study, characterized by comprising:
the edge computing terminals are respectively used for respectively collecting charging data of a plurality of electric vehicles in a designated area for training based on federal learning;
the global model aggregation module is used for aggregating training results of the edge computing terminals and obtaining global charging optimization models corresponding to different vehicle types according to the duty ratio of the data of the different vehicle types in the edge computing terminals;
the information acquisition module is used for acquiring the vehicle type and state information of the electric vehicle to be connected when the electric vehicle is connected into the power grid through the intelligent charging pile;
the model calling module is used for calling the corresponding global charging optimization model according to the vehicle type and the state information of each electric vehicle to be connected;
the charging control module is used for controlling the charging of each electric automobile according to the state, the charging requirement and the corresponding global charging optimization model of each electric automobile to be connected in the platform area, and the charging requirement comprises the charging duration and the charging mode.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310589331.2A CN116278903B (en) | 2023-05-24 | 2023-05-24 | Dynamic charging control method and system for electric automobile based on federal learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310589331.2A CN116278903B (en) | 2023-05-24 | 2023-05-24 | Dynamic charging control method and system for electric automobile based on federal learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116278903A CN116278903A (en) | 2023-06-23 |
| CN116278903B true CN116278903B (en) | 2023-07-18 |
Family
ID=86820744
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310589331.2A Active CN116278903B (en) | 2023-05-24 | 2023-05-24 | Dynamic charging control method and system for electric automobile based on federal learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116278903B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111162575A (en) * | 2019-07-17 | 2020-05-15 | 上海钧正网络科技有限公司 | Charging curve updating method and device, cloud server and battery |
| CN111532150A (en) * | 2020-05-15 | 2020-08-14 | 国网辽宁省电力有限公司电力科学研究院 | Self-learning-based electric vehicle charging control strategy optimization method and system |
| CN112819203A (en) * | 2021-01-12 | 2021-05-18 | 湖北追日电气股份有限公司 | Charging management system and method based on deep learning |
| CN113541272A (en) * | 2021-08-26 | 2021-10-22 | 山东浪潮科学研究院有限公司 | Energy storage battery balanced charging and discharging method and device based on deep learning model and medium |
| CN114977162A (en) * | 2022-06-02 | 2022-08-30 | 芯城时代(深圳)科技有限公司 | Charging monitoring method and system for electric automobile |
| CN115395544A (en) * | 2022-08-16 | 2022-11-25 | 王琼 | Electric vehicle charging and discharging rate control system and method |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20220161677A1 (en) * | 2020-11-24 | 2022-05-26 | Ecamion Inc. | Battery-enabled, direct current, electric vehicle charging station, method and controller therefor |
-
2023
- 2023-05-24 CN CN202310589331.2A patent/CN116278903B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111162575A (en) * | 2019-07-17 | 2020-05-15 | 上海钧正网络科技有限公司 | Charging curve updating method and device, cloud server and battery |
| CN111532150A (en) * | 2020-05-15 | 2020-08-14 | 国网辽宁省电力有限公司电力科学研究院 | Self-learning-based electric vehicle charging control strategy optimization method and system |
| CN112819203A (en) * | 2021-01-12 | 2021-05-18 | 湖北追日电气股份有限公司 | Charging management system and method based on deep learning |
| CN113541272A (en) * | 2021-08-26 | 2021-10-22 | 山东浪潮科学研究院有限公司 | Energy storage battery balanced charging and discharging method and device based on deep learning model and medium |
| CN114977162A (en) * | 2022-06-02 | 2022-08-30 | 芯城时代(深圳)科技有限公司 | Charging monitoring method and system for electric automobile |
| CN115395544A (en) * | 2022-08-16 | 2022-11-25 | 王琼 | Electric vehicle charging and discharging rate control system and method |
Non-Patent Citations (1)
| Title |
|---|
| 联邦学习在电力数据分析中的应用及隐私保护研究;戴理朋、杨鑫、徐茹枝;电力信息与通讯技术;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116278903A (en) | 2023-06-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP7251503B2 (en) | Power management device and power management method | |
| Ahmed et al. | The role of artificial intelligence in the mass adoption of electric vehicles | |
| JP7251459B2 (en) | Power management system and server | |
| CN110957752B (en) | Regional intelligent substation scheduling system and method based on edge internet of things | |
| CN113147482B (en) | Ordered charging optimization method and system for electric automobile | |
| US11698611B2 (en) | Server and power management system | |
| CN110549902A (en) | Cloud server, electric automobile and management system and method of power battery in electric automobile | |
| CN102539016A (en) | Onboard adaptive battery core temperature estimation | |
| US20210053459A1 (en) | Electric power system and vehicle | |
| US11955616B2 (en) | Electric energy storage system and vehicle | |
| CN111446711A (en) | Load-storage combined optimization operation method based on demand response | |
| CN113428048A (en) | Information presentation system, server, information presentation method, and information presentation device | |
| US20230013447A1 (en) | Power supply/demand adjusting method and power supply/demand management device | |
| CN116278903B (en) | Dynamic charging control method and system for electric automobile based on federal learning | |
| CN112508288A (en) | Ordered charging scheduling system and method based on temperature control load prediction | |
| TWI706366B (en) | Electric vehicle charging station power management method | |
| CN116739292A (en) | Energy optimization scheduling method, system and storage medium of data center | |
| CN116896086A (en) | Virtual power plant adjustable resource regulation and control system and method considering demand response | |
| CN114590098B (en) | Intelligent control system of automobile air conditioner | |
| CN112810555B (en) | Intelligent energy management system for a vehicle and corresponding method | |
| CN115071490A (en) | Control system and energy management method | |
| DE102022103551A1 (en) | ELECTRIC VEHICLE CHARGING AGGREGATION | |
| JP2023013634A (en) | Control system and power adjustment method | |
| CN111404168A (en) | Flexible air conditioner load-based dispatching system and method for stabilizing overload of transformer substation | |
| CN113327036B (en) | Load aggregator adjustable capacity assessment method under multi-time space scale |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |