Method and device for predicting adjustable potential of electric vehicle
By establishing an electric vehicle charging and discharging load model and compressing its cluster into generalized energy storage equipment, combined with the long and short-term memory time series method, the grid problem caused by electric vehicle charging is solved, and the accurate prediction of the adjustable potential of electric vehicles is achieved, and the power grid is achieved. Adjustment provides a basis.
Patent Information
- Application Number
- CN202411184742.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-27
- Publication Date
- 2024-11-26
AI Technical Summary
Random disordered charging of large-scale electric vehicles leads to problems such as low voltage in the station area, three-phase imbalance, and distribution network blockage, and brings huge challenges to grid balance adjustment.
By determining the influencing factors of charge and discharge load of electric vehicles, a charging and discharge load model was established, and the electric vehicle cluster was compressed into generalized energy storage equipment by using Minkovsky's summation, and a dispatchable capacity prediction model was established based on the long and short-term memory time series method to evaluate the model Performance to achieve accurate prediction of the adjustable potential of electric vehicles.
Accurate prediction of the adjustable potential of electric vehicles is achieved, providing a basis for it to participate in demand response transactions, scheduling, etc. as adjustable resources, and helping the power grid to balance and adjust.
Smart Images

Figure CN119026865A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the field of electrical technology, and in particular to a method and a device for predicting the adjustable potential of an electric vehicle. Background Art
[0002] In the field of transportation, the development and construction trend of new energy vehicles and supporting infrastructure has swept the world. By the end of 2021, the number of new energy vehicles in China reached 7.84 million, accounting for 2.60% of the total number of vehicles. Excluding the number of scrapped and deregistered vehicles, there were 2.92 million more vehicles than in 2020, an increase of 59.25%. Among them, the number of pure electric vehicles was 6.4 million, accounting for 81.63% of the total number of new energy vehicles, and the number of charging piles was 2.617 million. According to the China Society of Automotive Engineers, the number of new energy vehicles in my country will reach 300 million in 2040. Therefore, China will continue to be the main force in the world's new energy vehicle market in the future.
[0003] Random and disordered charging of large-scale electric vehicles will cause many problems such as low voltage in the substation, three-phase imbalance, and distribution network congestion; and with the advancement of the construction of new power systems, the high proportion of new energy and large-scale electric vehicle access have made the "double high" characteristics of the power system increasingly prominent, which will bring huge challenges to the balanced regulation of the power grid; at the same time, electric vehicles have the characteristics of flexible adjustment and mobile energy storage, and also have huge regulatory value under the new power system. Summary of the invention
[0004] The purpose of the present invention is to provide a method and device for predicting the adjustable potential of electric vehicles, which is used to solve the problems in the prior art that random and disordered charging of large-scale electric vehicles may cause low voltage in the substation, three-phase imbalance, distribution network congestion, and bring huge challenges to the balance and regulation of the power grid. The method can accurately predict the adjustable potential of electric vehicles, which is reasonable and can provide a basis for electric vehicles to participate in demand response transactions and scheduling as adjustable resources.
[0005] In order to achieve the above-mentioned object, in a first aspect, the present invention provides a method for predicting adjustable potential of an electric vehicle, comprising:
[0006] Step 1: Determine the influencing factors of the electric vehicle charging and discharging load and the relationship between the influencing factors, and establish an electric vehicle charging and discharging load model based on the charging and discharging behavior of the electric vehicle;
[0007] Step 2: Based on the electric vehicle charging and discharging load model, the electric vehicle cluster is compressed into a generalized energy storage device through Minkowski summation, and the parameters of the generalized energy storage device are used to characterize the adjustable potential of the electric vehicle;
[0008] Step 3: Use the long short-term memory time series method to identify the parameters of the generalized energy storage equipment, establish a dispatchable capacity prediction model, and introduce evaluation indicators to evaluate the model performance of the dispatchable capacity prediction model.
[0009] According to a method for predicting adjustable potential of an electric vehicle provided by the present invention, factors affecting the charge and discharge load of the electric vehicle include the starting charge and discharge time, daily mileage, the starting charge state of charge and the charge and discharge duration.
[0010] According to a method for predicting adjustable potential of an electric vehicle provided by the present invention, the probability density function of the starting charging and discharging time obeys a normal distribution, and the probability density function of the starting state of charge of charging and discharging obeys a normal distribution;
[0011] The probability density function of daily mileage is expressed as:
[0012]
[0013] Where, d is the daily mileage; μ d , σ d are the expected value and standard deviation of daily mileage respectively;
[0014] The calculation formula for charging and discharging time is:
[0015]
[0016] Where, T c is the charge and discharge time; SOC e SOC is the state of charge of the electric vehicle after charging and discharging. start is the state of charge of the electric vehicle at the beginning of charging and discharging; η is the charging and discharging transmission efficiency; P c is the charging and discharging power; E is the battery capacity of the electric vehicle.
[0017] According to a method for predicting adjustable potential of electric vehicles provided by the present invention, the charging and discharging load model of the electric vehicle is:
[0018]
[0019] In the formula, are the charging and discharging power of electric vehicle n at time t; T str,n is the starting charging and discharging time of electric vehicle n; SOC e,n SOC is the state of charge of the electric vehicle after charging and discharging. start,n E is the state of charge of the electric vehicle n at the beginning of charging and discharging; n is the battery capacity of electric vehicle n.
[0020] According to a method for predicting adjustable potential of electric vehicles provided by the present invention, step 2 specifically includes:
[0021] The boundary constraints of electric vehicle charging and discharging power are:
[0022]
[0023] In the formula, are the charging power and discharging power of electric vehicle n at time t respectively; are the upper limit of charging power and the lower limit of discharging power of electric vehicle n at time t respectively; is the time set of electric vehicle n accessing the power grid;
[0024] The boundary constraint of the electric vehicle battery power is:
[0025]
[0026] In the formula, and are the battery power of electric vehicle n at time t and time t-1 respectively; η ch and η dis are the charging efficiency and discharging efficiency of electric vehicles respectively; Δt represents the unit dispatching time; η ref Indicates the discharge compensation coefficient; and They represent the lower and upper boundaries of the battery power of electric vehicle n at time t respectively;
[0027] Expanding the time set of electric vehicles connected to the grid to the full dispatch time, we get:
[0028]
[0029] in,
[0030] Where N EV represents the electric car collection; X n,t represents the grid-connected state of electric vehicle n at time t, X n,t =1 means that electric vehicle n is connected to the grid at time t, X n,t =0 means that electric vehicle n is off-grid at time t; Indicates the time when the electric vehicle n starts charging and discharging. Indicates the moment when the charging and discharging of electric vehicle n is completed;
[0031] Transforming formula (13), we get:
[0032]
[0033] In the formula, s n,arrival represents the initial charge of electric vehicle n;
[0034]
[0035] In the formula, s n,leaveIndicates the amount of electricity after the electric vehicle n is charged and discharged;
[0036] Integrating equations (15), (16) and (17) we can obtain:
[0037]
[0038] Through Minkowski addition, we get:
[0039]
[0040] This compresses the electric vehicle cluster into a generalized energy storage device;
[0041] Define the parameters of generalized energy storage equipment to characterize the dispatchable capacity of electric vehicle clusters. The parameters of generalized energy storage equipment are:
[0042]
[0043] In the formula, and They represent the maximum charging and discharging power of the generalized energy storage device at time t respectively; and They represent the lower and upper boundaries of the amount of electricity in the generalized energy storage device at time t; ΔS t Represents the change in the amount of electricity generated by a generalized energy storage device at time t due to external factors.
[0044] According to a method for predicting adjustable potential of electric vehicles provided by the present invention, step 3 specifically includes:
[0045] Collect historical data of electric vehicle fleets;
[0046] The historical dispatchable capacity of the electric vehicle cluster is calculated based on formula (22);
[0047] A dispatchable capacity prediction model is established based on the long short-term memory time series method, and the data set consisting of the historical dispatchable capacity of the electric vehicle cluster is divided into a training set and a test set;
[0048] The training set is normalized and then input into the dispatchable capacity prediction model for training. The test set is input into the trained dispatchable capacity prediction model for prediction to obtain the prediction results. The prediction results are inversely normalized and the evaluation index is calculated to evaluate the model performance of the dispatchable capacity prediction model.
[0049] According to a method for predicting adjustable potential of electric vehicles provided by the present invention, the evaluation index is a root mean square error.
[0050] In a second aspect, the present invention further provides an electric vehicle adjustable potential prediction device, comprising:
[0051] Establish a unit for determining the influencing factors of the electric vehicle charging and discharging load and the relationship between the influencing factors, and establish an electric vehicle charging and discharging load model in combination with the charging and discharging behavior of the electric vehicle;
[0052] A processing unit, for compressing the electric vehicle cluster into a generalized energy storage device through Minkowski summation based on the electric vehicle charging and discharging load model, and using the parameters of the generalized energy storage device to characterize the adjustable potential of the electric vehicle;
[0053] The identification and evaluation unit is used to identify the parameters of the generalized energy storage equipment using the long short-term memory time series method, establish a dispatchable capacity prediction model, and introduce evaluation indicators to evaluate the model performance of the dispatchable capacity prediction model.
[0054] The present invention provides a method and device for predicting the adjustable potential of electric vehicles, the method comprising: step 1, determining the influencing factors of the charging and discharging load of electric vehicles and the relationship between the influencing factors, and establishing an electric vehicle charging and discharging load model in combination with the charging and discharging behavior of electric vehicles; step 2, based on the electric vehicle charging and discharging load model, compressing the electric vehicle cluster into a generalized energy storage device through Minkowski summation, and using the parameters of the generalized energy storage device to characterize the adjustable potential of the electric vehicle; step 3, using the long short-term memory time series method to identify the parameters of the generalized energy storage device, establishing a dispatchable capacity prediction model, and introducing evaluation indicators to evaluate the model performance of the dispatchable capacity prediction model. The present invention establishes an electric vehicle charging and discharging load model, which can summarize the changing law of the electric vehicle cluster load from the two dimensions of time and space; aggregates electric vehicles as generalized energy storage devices, quantifies their dispatchable potential, and combines deep learning algorithms to form a prediction method for the adjustable potential of electric vehicle charging and discharging resources, which can accurately predict the adjustable potential of electric vehicles and provide a basis for electric vehicles to participate in demand response transactions and scheduling as adjustable resources. BRIEF DESCRIPTION OF THE DRAWINGS
[0055] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0056] In the attached picture:
[0057] Figure 1 The present invention provides a flow chart of the method for predicting the adjustable potential of electric vehicles. DETAILED DESCRIPTION
[0058] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0059] Some embodiments of the present invention will be described in detail below in conjunction with the accompanying drawings. In the absence of conflict, the following embodiments and features in the embodiments can be combined with each other.
[0060] See also Figure 1 The present invention provides a method for predicting adjustable potential of electric vehicles, comprising:
[0061] Step 1: Determine the influencing factors of the electric vehicle charging and discharging load and the relationship between the influencing factors, and establish an electric vehicle charging and discharging load model based on the charging and discharging behavior of the electric vehicle;
[0062] Specifically, unlike conventional electricity loads, electric vehicle loads are highly flexible and random, and will be affected by factors such as the start and end time of charging and discharging, power demand, and battery characteristics.
[0063] Different types of electric vehicles have slightly different charging characteristics, charging time and frequency according to their different uses. The main factors affecting the charging and discharging load include: the starting charging and discharging time, daily mileage, the starting state of charge of charging and discharging, and the charging and discharging time.
[0064] (1) Starting charging and discharging time of electric vehicles
[0065] Generally speaking, when a car owner finishes a journey, he or she may engage in charging and discharging behavior. According to statistics, the probability density function of the starting charging and discharging time obeys the normal distribution, and the distribution pattern is similar to the peak load period of the power grid.
[0066] The probability density function of the initial charging and discharging time of a private car user is as follows:
[0067]
[0068] In the formula, the starting charge and discharge time is The expected value of μ is s , with standard deviation σ s .
[0069] (2) Daily mileage of electric vehicles
[0070] The daily mileage will directly affect the use of electric vehicles and the user's charging and discharging choices. The larger the value, the higher the power consumption of the electric vehicle. If the charging power remains unchanged, the charging and discharging time and the mileage will show a mapping relationship. At the same time, due to the differences in behavioral habits between users, the daily mileage will also be different. The probability density function of the daily mileage of electric vehicles is as follows:
[0071]
[0072] Where, d is the daily mileage, unit: km; μ d ,σ d are the expected value and standard deviation of daily mileage respectively.
[0073] (3) Initial charge state of electric vehicle charging and discharging
[0074] The state of charge (SOC) of an electric vehicle refers to the ratio of the remaining capacity of the battery in its current state to the capacity in a fully charged state, and the SOC satisfies the range of 0 to 1. If SOC=0, it means that the battery is exhausted, and SOC=1 means that the battery is fully charged.
[0075] By using statistics to count the state of charge (SOC at the start of charge and discharge) of electric vehicles when they are charged and discharged for the first time, it can be obtained that the probability density function of the SOC at the start of charge and discharge of electric vehicles obeys the normal distribution, and its probability density function is shown in formula (3):
[0076]
[0077] In the formula, SOC is the state of charge at the beginning of charge and discharge, and its expected value is μ soc , with standard deviation σ soc .
[0078] (4) Charging and discharging time of electric vehicles
[0079] The charging and discharging time of an electric vehicle is related to factors such as the user's expected charge, charging power, and the initial charge state of charge. The calculation of the charging and discharging time is shown in formula (5).
[0080]
[0081] Where, T c is the charge and discharge time, SOC e SOC is the state of charge of the electric vehicle after charging and discharging; start is the state of charge of the electric vehicle at the beginning of charging and discharging; η is the charging and discharging transmission efficiency; P c Indicates the charge and discharge power in kW. There are two states: P c_slow Indicates slow charging power, P c_quickIt represents fast charging power. The charging and discharging power is selected according to the fast and slow charging standards of electric vehicles. E is the battery capacity of the electric vehicle.
[0082] (5) Electric vehicle charging and discharging load model
[0083] From the above analysis, it can be concluded that the charging and discharging behavior of electric vehicles is related to a variety of factors. Based on the various factors that affect the charging and discharging load of electric vehicles proposed previously, a charging and discharging load model that is generally applicable to electric vehicles is obtained as shown in formula (6):
[0084]
[0085] In the formula, are the charging and discharging power of electric vehicle n at time t; P c Select fast charge or slow charge according to national standards and charging mode; η is the charge and discharge transmission efficiency; T str,n is the starting time of charging and discharging of electric vehicle n; SOC e,n SOC is the state of charge of electric vehicle n after charging and discharging (leaving the charging station); start,n E is the state of charge of electric vehicle n at the beginning of charging and discharging (arriving at the charging station); n is the battery capacity of electric vehicle n.
[0086] Step 2: Based on the electric vehicle charging and discharging load model, the electric vehicle cluster is compressed into a generalized energy storage device through Minkowski summation, and the parameters of the generalized energy storage device are used to characterize the adjustable potential of the electric vehicle;
[0087] Specifically, when the owner of an electric vehicle selects the DC fast charging mode, the electric vehicle will be charged at the maximum power, and its charging power cannot be controlled. When the AC slow charging mode is selected, the electric vehicle has the ability to load shift and reverse power supply. Therefore, the boundary constraints of the electric vehicle charging and discharging power are shown in Equations (7) and (8).
[0088]
[0089] In the formula, are the charging power and discharging power of electric vehicle n at time t respectively; are the upper limit of charging power and the lower limit of discharging power of electric vehicle n at time t respectively; is the time set when electric vehicle n is connected to the grid.
[0090] The boundary constraint of electric vehicle battery power is shown in formula (9).
[0091]
[0092] In the formula, and are the battery power of electric vehicle n at time t and time t-1 respectively; η ch and η dis are the charging efficiency and discharging efficiency of electric vehicles respectively; Δt represents the unit dispatching time; η ref Represents the discharge compensation coefficient, which is used to solve the complex variables introduced by the battery loss of electric vehicles; and They represent the lower and upper boundaries of the battery power of electric vehicle n at time t respectively.
[0093] Since the power grid contains a large number of electric vehicles, if a single electric vehicle is modeled, on the one hand, individual parameters are difficult to predict, and on the other hand, the dimension of the model is too high. Considering that electric vehicles have similar mathematical models, and in the market transaction process, electric vehicles are usually managed by operators and participate in power grid dispatch as a whole. Therefore, the present invention proposes an aggregation method for electric vehicle clusters, compressing electric vehicle clusters into generalized energy storage devices through Minkowski summation, that is, establishing an electric vehicle aggregation model.
[0094] The premise of Minkowski summation is that the two variable spaces have the same domain. However, due to the difference in the grid connection time of individual electric vehicles, their domains are There is heterogeneity, so it needs to be extended to the same full scheduling time T. In order to realize the extension of the definition domain of individual electric vehicles, the state variable X is first introduced n,t Used to characterize the grid-connected status of electric vehicles, X n,t It can be directly calculated from the parking time of the electric vehicle, as shown in formula (11).
[0095]
[0096] In the formula, X m,t represents the grid-connected state of electric vehicle n at time t, X m,t =1 means that electric vehicle n is connected to the grid at time t, X m,t =0 means that electric vehicle n is off-grid at time t; represents the time when electric vehicle n arrives at the charging station, represents the time when electric vehicle n leaves the charging station.
[0097] Since electric vehicles do not charge and discharge at the same time, equations (7) to (9) can be further transformed from the definition domain Expanded to the full scheduling time T, as shown in equations (12) to (14):
[0098]
[0099] Where N EC Represents a collection of electric vehicles.
[0100] Among them, equations (12) and (14) have Minkowski additivity, while equation (13) requires some transformation. Therefore, further discussion divides it into three cases, as shown in equations (15) to (17). Among them, equation (15) corresponds to the time when the electric vehicle enters the grid, and its characteristic is that the initial power of the electric vehicle s needs to be considered. n,arrival ; Formula (16) corresponds to the normal grid-connected time of electric vehicles, also known as the general time, which degenerates into Formula (9); Formula (17) corresponds to the time when electric vehicles are off-grid, and its characteristic is that the off-grid power s of electric vehicles needs to be considered m,leave , s in formula (17) n,t The value is 0, but s n,t-1 The value is not 0. Since the power of an electric vehicle will not disappear out of thin air, the physical meaning of formula (17) is that when an electric vehicle leaves the charging station, its corresponding power will return to zero.
[0101]
[0102] In the formula, s n,arrival Represents the initial charge of electric vehicle n.
[0103]
[0104] In the formula, s n,leave It represents the power of electric vehicle n when it leaves the charging station, which is also its expected power.
[0105] Note that equation (15) implies s n,t-1 = 0, which is determined by equation (14); similarly, equation (17) implies and These two conditions are determined by formula (12). Equivalent to satisfying X n,t (X n,t -X n,t-1 )=1, Equivalent to satisfying X n,t-1 (X n,t-1 -X n,t )=1, so equations (15) to (17) can be further integrated into equation (18), which has Minkowski additivity and can be compressed into a generalized energy storage device.
[0106]
[0107] By using Minkowski addition to process equations (12), (14) and (18), we can obtain the envelope space corresponding to equations (19) to (21).
[0108]
[0109]
[0110] Further define variables As a variable of generalized energy storage equipment, and in, and They represent the charging and discharging dispatching power of the generalized energy storage device at time t, S t Represents the power of the generalized energy storage device at time t.
[0111] And define the parameters As the parameter of the generalized energy storage device, it is calculated by formula (22).
[0112]
[0113] In the formula, and They represent the maximum charging and discharging power of the generalized energy storage device at time t respectively; and They represent the lower and upper boundaries of the amount of electricity in the generalized energy storage device at time t; ΔS t Represents the change in the amount of electricity generated by a generalized energy storage device at time t due to external factors.
[0114] By replacing equations (19) to (21) with equation (22), the variable space of the electric vehicle cluster can be compressed into the variable space of the generalized energy storage device, as shown in equation (23).
[0115]
[0116] The essence of formula (23) is to project the variable space of individual electric vehicles into a hypercube space while retaining the constraint relationship between variables. EV The electric vehicles are compressed into a generalized energy storage device, which greatly reduces the dimension of the model. The hypercube space contains all feasible charging and discharging decisions of the electric vehicle cluster, and the parameters of the generalized energy storage device are The potential of electric vehicle clusters as flexible load storage resources is determined, so the present invention sets the parameters Considered as the dispatchable capacity of the electric vehicle cluster.
[0117] Step 3: Use the long short-term memory (LSTM) time series method to identify the parameters of the generalized energy storage equipment, establish a dispatchable capacity prediction model, and introduce evaluation indicators to evaluate the model performance of the dispatchable capacity prediction model.
[0118] Specifically, formula (23) mathematically establishes an aggregation model of flexible resources, but the parameters in the model have not yet been determined. Therefore, the present invention proposes a parameter identification method for the electric vehicle aggregation model, that is, a dispatchable capacity prediction method for an electric vehicle cluster.
[0119] First, EV operators can predict dispatchable capacity based on historical data sets and first define their historical operation data as shown in Equation (24).
[0120]
[0121] In the formula, Ω v is the electric vehicle data in the historical dataset v; is the number of electric vehicles in the historical data set v. In the present invention, the unit of the historical data set v is 1 day.
[0122] Secondly, the electric vehicle operator will record the data of the electric vehicles it serves every day, including the arrival time, departure time, battery power at arrival, battery power at departure, and model of the electric vehicle (which includes the battery power boundary and maximum charge and discharge power), as shown in formula (25).
[0123]
[0124] In the formula, EV n is the historical record data of electric vehicle n.
[0125] Therefore, step 3 may specifically include: firstly collecting historical record data of the electric vehicle cluster;
[0126] Based on formula (22), the parameters of the electric vehicle cluster after being compressed into a generalized energy storage device are further calculated and used as the historical dispatchable capacity of the electric vehicle cluster obtained a posteriori, as shown in formula (26).
[0127]
[0128] At this point, the historical data of the electric vehicle cluster is compressed into historical dispatchable capacity, which can be used as training data, and further establish a dispatchable capacity prediction model based on the long short-term memory time series method. Compared with the method of directly using individual parameters of electric vehicles as training data, on the one hand, it reduces the tensor of the data set, and on the other hand, it takes into account the correlation between variables, which can improve the accuracy of the prediction model.
[0129] Load forecasting is essentially a point-to-point, sequence-to-point or sequence-to-sequence input-output model. It combines the image factors of various loads and historical load data as the input matrix, and uses the model to capture the changing patterns of historical load data, thereby outputting forecast data for the next time point or multiple time points.
[0130] Load forecasting models can be divided into single-variable models and multi-variable models. Some traditional load forecasting methods must establish a mathematical relationship between historical load data and characteristic variables to make effective forecasts, because these methods cannot capture the changing patterns from a single load variable, while the powerful learning ability of deep learning models can well capture subtle changing patterns from historical load data and make effective forecasts.
[0131] In order to improve the prediction accuracy, it is necessary to consider the impact of more factors on the electric vehicle load, which will increase the dimension of the characteristic variables and increase the calculation cost. The present invention takes into account the three factors affecting the electric vehicle load: day type, electricity price, and temperature. A single-step prediction method is used to predict the electric vehicle load value at 96 time points on the next day based on the known load value at 96 time points on a certain day. A dispatchable capacity prediction model is established based on the long short-term memory time series method. The data set composed of the historical dispatchable capacity of the electric vehicle cluster is divided into a training set and a test set with a ratio of 8:2. The training set is normalized and input into the dispatchable capacity prediction model for training. The test set is input into the trained dispatchable capacity prediction model for prediction to obtain the prediction result. The prediction result is inversely normalized and the evaluation index is calculated to evaluate the model performance of the dispatchable capacity prediction model.
[0132] It should be noted that in the time series forecasting task, there are three commonly used evaluation indicators for evaluating model performance, namely root mean square error (RMSE), mean percentage error (MAPE), and mean absolute error (MAE). The calculation formulas are as follows:
[0133] Assume that the predicted value is The true value is y={y1,y2,…,y n},but
[0134]
[0135] Where N is the number of samples. In the present invention, the charging load of electric vehicles is different from the traditional power load. Due to the durability of the battery and the behavior of the driver, in certain time periods, all charging piles in the charging station are not used. This leads to many data points with a value of 0 in the data. However, this is not abnormal data, but reflects the authenticity of the data. Most of the previous studies on the charging load of electric vehicles are based on simulation data, and there is often no 0 value, which seriously affects the credibility of the prediction results. However, it is precisely because of the existence of the 0 value that the present invention cannot use MAPE as an evaluation indicator for the prediction results, because it can be seen from formula (28) that y iWhen it exists as a denominator, it cannot be 0. Therefore, the evaluation index of the present invention selects RMSE as a metric, and adds a loss iteration change curve to measure the effect of model training.
[0136] Based on the same inventive concept, the present invention provides an electric vehicle adjustable potential prediction device, which corresponds to the method of the above embodiment, and includes:
[0137] Establish a unit for determining the influencing factors of the electric vehicle charging and discharging load and the relationship between the influencing factors, and establish an electric vehicle charging and discharging load model in combination with the charging and discharging behavior of the electric vehicle;
[0138] A processing unit, for compressing the electric vehicle cluster into a generalized energy storage device through Minkowski summation based on the electric vehicle charging and discharging load model, and using the parameters of the generalized energy storage device to characterize the adjustable potential of the electric vehicle;
[0139] The identification and evaluation unit is used to identify the parameters of the generalized energy storage equipment using the long short-term memory time series method, establish a dispatchable capacity prediction model, and introduce evaluation indicators to evaluate the model performance of the dispatchable capacity prediction model.
[0140] In summary, the electric vehicle adjustable potential prediction method and device provided by the present invention establish an electric vehicle charging and discharging load model, which can summarize the changing law of the electric vehicle cluster load from the two dimensions of time and space; electric vehicles are aggregated and quantified as generalized energy storage devices, and a deep learning algorithm is combined to form a prediction method for the adjustable potential of electric vehicle charging and discharging resources, which can accurately predict the adjustable potential of electric vehicles and provide a basis for electric vehicles to participate in demand response transactions and scheduling as adjustable resources.
[0141] Those skilled in the art will readily appreciate other embodiments of the present invention after considering the specification and practicing the embodiments disclosed herein. The present invention is intended to cover any variations, uses or adaptations of the present invention that follow the general principles of the present invention and include common knowledge or customary techniques in the art that are not disclosed by the present invention. It should be understood that the present invention is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from the scope thereof. The scope of the present invention is limited only by the appended claims.
Claims
1. A method for predicting the adjustable potential of electric vehicles, It is characterized in that include: Step 1: Determine the influencing factors of the electric vehicle charging and discharging load and the relationship between the influencing factors, and establish an electric vehicle charging and discharging load model based on the charging and discharging behavior of the electric vehicle; Step 2: Based on the electric vehicle charging and discharging load model, the electric vehicle cluster is compressed into a generalized energy storage device through Minkowski summation, and the parameters of the generalized energy storage device are used to characterize the adjustable potential of the electric vehicle; Step 3: Use the long short-term memory time series method to identify the parameters of the generalized energy storage equipment, establish a dispatchable capacity prediction model, and introduce evaluation indicators to evaluate the model performance of the dispatchable capacity prediction model.
2. The electric vehicle adjustable potential prediction method according to claim 1, It is characterized in that The influencing factors of the electric vehicle charging and discharging load include the starting charging and discharging time, daily mileage, the starting state of charge of charging and discharging, and the charging and discharging time.
3. The electric vehicle adjustable potential prediction method according to claim 2, It is characterized in that The probability density function of the starting charge and discharge time obeys a normal distribution, and the probability density function of the charge state at the start of charge and discharge obeys a normal distribution; The expression of the probability density function of the daily mileage is: Where, d is the daily mileage; μ d , σ d are the expected value and standard deviation of daily mileage respectively; The calculation formula for the charge and discharge time is: Where, T c is the charge and discharge time; SOC e SOC is the state of charge of the electric vehicle after charging and discharging. start is the state of charge of the electric vehicle at the beginning of charging and discharging; η is the charging and discharging transmission efficiency; P c is the charging and discharging power; E is the battery capacity of the electric vehicle.
4. The electric vehicle adjustable potential prediction method according to claim 3, It is characterized in that The electric vehicle charging and discharging load model is: In the formula, are the charging and discharging power of electric vehicle n at time t; T str,n is the starting charging and discharging time of electric vehicle n; SOC e,n SOC is the state of charge of the electric vehicle after charging and discharging. start,n E is the state of charge of the electric vehicle n at the beginning of charging and discharging; n is the battery capacity of electric vehicle n.
5. The method for predicting the adjustable potential of electric vehicles according to claim 4, It is characterized in that The step 2 specifically includes: The boundary constraints of electric vehicle charging and discharging power are: In the formula, are the charging power and discharging power of electric vehicle n at time t respectively; are the upper limit of charging power and the lower limit of discharging power of electric vehicle n at time t respectively; is the time set of electric vehicle n connecting to the power grid; The boundary constraint of the electric vehicle battery power is: In the formula, and are the battery power of electric vehicle n at time t and time t-1 respectively; η ch and η dis are the charging efficiency and discharging efficiency of electric vehicles respectively; Δt represents the unit dispatching time; η ref Indicates the discharge compensation coefficient; and They represent the lower and upper boundaries of the battery power of electric vehicle n at time t respectively; Expanding the time set of electric vehicles connected to the grid to the full dispatch time, we get: in, Where N EV represents the electric car collection; X n,t represents the grid-connected state of electric vehicle n at time t, X n,t =1 means that electric vehicle n is connected to the grid at time t, X n,t =0 means that electric vehicle n is off-grid at time t; Indicates the time when the electric vehicle n starts charging and discharging. Indicates the moment when the charging and discharging of electric vehicle n is completed; Transforming formula (13), we get: In the formula, s n,arrival represents the initial charge of electric vehicle n; In the formula, s n,leave Indicates the amount of electricity after the electric vehicle n is charged and discharged; Integrating equations (15), (16) and (17) we can obtain: Through Minkowski addition, we get: This compresses the electric vehicle cluster into a generalized energy storage device; The parameters of the generalized energy storage device are defined to characterize the dispatchable capacity of the electric vehicle cluster. The parameters of the generalized energy storage device are: In the formula, and They represent the maximum charging and discharging power of the generalized energy storage device at time t respectively; and They represent the lower and upper boundaries of the amount of electricity in the generalized energy storage device at time t; ΔS t Represents the change in the amount of electricity generated by a generalized energy storage device at time t due to external factors.
6. The electric vehicle adjustable potential prediction method according to claim 5, It is characterized in that The step 3 specifically includes: Collect historical data of electric vehicle fleets; The historical dispatchable capacity of the electric vehicle cluster is calculated based on formula (22); A dispatchable capacity prediction model is established based on the long short-term memory time series method, and the data set consisting of the historical dispatchable capacity of the electric vehicle cluster is divided into a training set and a test set; The training set is normalized and then input into the dispatchable capacity prediction model for training, the test set is input into the trained dispatchable capacity prediction model for prediction to obtain prediction results, the prediction results are inversely normalized and then the evaluation index is calculated to evaluate the model performance of the dispatchable capacity prediction model.
7. The method for predicting the adjustable potential of electric vehicles according to claim 6, It is characterized in that The evaluation index is the root mean square error.
8. An adjustable potential prediction device for electric vehicles, It is characterized in that include: Establish a unit for determining the influencing factors of the electric vehicle charging and discharging load and the relationship between the influencing factors, and establish an electric vehicle charging and discharging load model in combination with the charging and discharging behavior of the electric vehicle; A processing unit, configured to compress the electric vehicle cluster into a generalized energy storage device through Minkowski summation based on the electric vehicle charging and discharging load model, and characterize the adjustable potential of the electric vehicle using the parameters of the generalized energy storage device; The identification and evaluation unit is used to identify the parameters of the generalized energy storage device by using the long short-term memory time series method, establish a dispatchable capacity prediction model, and introduce evaluation indicators to evaluate the model performance of the dispatchable capacity prediction model.
Citation Information
Cited By
Electric vehicle generalized energy storage space-time prediction method based on BERT
CN119849693A
A Spatiotemporal Prediction Method for Generalized Energy Storage of Electric Vehicles Based on BERT
CN119849693B
