A method and system for predicting charging load of electric vehicle

Abstract

The present invention relates to a method and system for predicting the charging load of an electric vehicle. The method comprises: obtaining the driving law of an electric vehicle, and fitting the driving law probability distribution characteristic parameters according to the driving law of the electric vehicle; fitting the charging characteristics of the electric vehicle according to the driving law of multiple electric vehicles; determining the electric vehicle parameters according to the actual number of electric vehicles in the area; determining random driving data according to the electric vehicle parameters, and converting the random driving data into a multidimensional normal distribution function to generate driving law random numbers containing coupling characteristics; according to the driving law of the electric vehicle, the charging characteristics and the driving law random numbers containing coupling characteristics, using a variable bandwidth kernel density function to establish a simulation model of variable bandwidth kernel density estimation, and determine the optimal charging load simulation curve. The present invention improves the prediction accuracy of the charging load of the electric vehicle.

Description

Technical Field

[0001] The present invention relates to the field of electric vehicle charging load prediction, and in particular to a method and system for predicting electric vehicle charging load. Background Art

[0002] Nowadays, the widespread application of electric vehicles needs to be coordinated with the planning, construction and operation of the power grid in many aspects. The universal charging of electric vehicles will have a certain impact on the planning and operation of the power grid. Due to the subjectivity of the charging behavior of electric vehicle owners, the charging load of electric vehicles will have strong randomness and uncertainty in time and space, which will bring certain challenges to the operation of the power grid, such as load rate, increased network loss, introduction of harmonic pollution, and increased difficulty in optimizing the dispatch of the power grid. Therefore, by reasonably predicting the charging load of electric vehicles and guiding the charging load reasonably, the operation efficiency of the power grid will be improved, which can solve people’s concerns about insufficient charging, and at the same time, it can lay a solid foundation for the realization of the goal of carbon neutrality as scheduled.

[0003] Existing patents and literature are basically based on the deterministic probability distribution model built by the Monte Carlo method, and some use kernel density estimation. Although it is more random than the Monte Carlo method, the accuracy of the final result will still be affected by the bandwidth problem of kernel density. Summary of the invention

[0004] The purpose of the present invention is to provide a method and system for predicting the charging load of an electric vehicle, so as to solve the problem of low prediction accuracy of the charging load of an electric vehicle.

[0005] To achieve the above object, the present invention provides the following solutions:

[0006] A method for predicting charging load of an electric vehicle, comprising:

[0007] Obtaining the driving pattern of the electric vehicle, and fitting the driving pattern probability distribution characteristic parameters according to the driving pattern of the electric vehicle; the driving pattern probability distribution characteristic parameters include the starting driving time, the ending driving time, the probability distribution of the driving mileage, the charging start time and the charging end time of each day;

[0008] Fitting charging characteristics of electric vehicles according to the driving rules of the electric vehicles of the plurality of electric vehicles; the charging characteristics including a charging start time and a charging end time;

[0009] Determine electric vehicle parameters according to the actual number of electric vehicles in the area; the electric vehicle parameters include charging power, power consumption characteristics and charging mode;

[0010] Determine random driving data according to the electric vehicle parameters, and convert the random driving data into a multidimensional normal distribution function to generate a driving law random number containing coupling characteristics; the random driving data includes a charging start time, a charging end time, and a driving mileage;

[0011] According to the driving law of the electric vehicle, the charging characteristics and the driving law random number containing coupling characteristics, a simulation model of variable bandwidth kernel density estimation is established using a variable bandwidth kernel density function to determine an optimal charging load simulation curve; the optimal charging load simulation curve is used to predict the charging charge of the electric vehicle.

[0012] Optionally, the variable bandwidth kernel density function is:

[0013]

[0014] Among them, h j is the bandwidth of the jth sample point required; is the product of k=1 to k=M fixed kernel density functions; M is the number of sample points; f(X k ) is the kernel density function of the selected fixed k-th sample point; X k is the variable value of the kth sample point; k is the sequence number of the kth sample point; f(X j ) is the kernel density function of the jth sample point; X j is the variable value of the j-th sample point; j is the sequence number of the j-th sample point; α is the sensitivity factor.

[0015] Optionally, determining random driving data according to the electric vehicle parameters, and converting the random driving data into a multidimensional normal distribution function to generate driving law random numbers with coupling characteristics specifically includes:

[0016] Using the formula Calculate the marginal distribution function of each of the random driving data; wherein, is the marginal distribution function; x is the random variable of random driving data; is the random variable density function;

[0017] According to the marginal distribution function, generating a uniform random number sequence from the processed random driving data;

[0018] Convert the homogenized random number sequence into a standard normal distribution random number sequence;

[0019] estimating the correlation of normal random variables and generating a multidimensional standard normal distribution function based on the correlation;

[0020] Generate a multivariate random number pair whose correlation factor is the correlation degree by using a multidimensional normal distribution function;

[0021] Convert the multi - variable random number pairs into original - type data; the original - type data is a random number of driving rules with coupling characteristics.

[0022] Optionally, according to the driving rules of the electric vehicle, the charging characteristics, and the random number of driving rules with coupling characteristics, a simulation model of variable - bandwidth kernel density estimation is established using a variable - bandwidth kernel density function to determine the optimal charging load simulation curve. After that, it further includes:

[0023] Model the uncertainty of the vehicle owner's charging behavior according to the optimal charging load simulation curve; the vehicle owner's charging behavior includes charging behavior and non - charging behavior; the vehicle owner's charging behavior is an uncertain behavior.

[0024] Optionally, modeling the uncertainty of the vehicle owner's charging behavior according to the optimal charging load simulation curve specifically includes:

[0025] Cluster the vehicle owner's charging behaviors approximately following a binomial distribution B(m, p) with parameters m and p; where m is the total number of samples and p is the charging probability.

[0026] Set the cluster charging probability as p, p ∈ [0, 1]. For the i - th electric vehicle, randomly generate a random number rand following a U(0, 1) uniform distribution; U(0, 1) is a uniform distribution function over the interval [0, 1].

[0027] When rand < p, select the charging behavior;

[0028] When rand ≥ p, select the non - charging behavior.

[0029] A prediction system for the charging load of an electric vehicle, including:

[0030] A driving - rule probability - distribution characteristic parameter fitting module, configured to obtain the driving rules of the electric vehicle and fit the driving - rule probability - distribution characteristic parameters according to the driving rules of the electric vehicle; the driving - rule probability - distribution characteristic parameters include the starting driving time, ending driving time, probability distribution of the driving mileage, charging start time, and charging end time of each day;

[0031] A charging - characteristic fitting module, configured to fit the charging characteristics of the electric vehicle according to the driving rules of multiple electric vehicles; the charging characteristics include the starting charging time and the ending charging time;

[0032] An electric - vehicle parameter determination module, configured to determine the electric - vehicle parameters according to the actual number of electric vehicles in the region; the electric - vehicle parameters include the charging power, power - consumption characteristics, and charging mode;

[0033] A driving pattern random number generation module is used to determine random driving data according to the electric vehicle parameters, and convert the random driving data into a multidimensional normal distribution function to generate a driving pattern random number with coupling characteristics; the random driving data includes a charging start time, a charging end time and a driving mileage;

[0034] The module for determining the optimal charging load simulation curve is used to establish a simulation model of variable bandwidth kernel density estimation using a variable bandwidth kernel density function according to the driving law of the electric vehicle, the charging characteristics and the driving law random number containing coupling characteristics, so as to determine the optimal charging load simulation curve; the optimal charging load simulation curve is used to predict the charging charge of the electric vehicle.

[0035] Optionally, the variable bandwidth kernel density function is:

[0036]

[0037] Among them, h j is the bandwidth of the jth sample point required; is the product of k=1 to k=M fixed kernel density functions; M is the number of sample points; f(X k ) is the kernel density function of the selected fixed k-th sample point; X k is the variable value of the kth sample point; k is the sequence number of the kth sample point; f(X j ) is the kernel density function of the jth sample point; X j is the variable value of the j-th sample point; j is the sequence number of the j-th sample point; α is the sensitivity factor.

[0038] Optionally, the driving pattern random number generation module specifically includes:

[0039] Marginal distribution function calculation unit, used to use the formula Calculate the marginal distribution function of each of the random driving data; wherein, is the marginal distribution function; x is the random variable of random driving data; is the random variable density function;

[0040] A random number sequence generating unit, used for generating a uniform random number sequence from the processed random driving data according to the marginal distribution function;

[0041] A conversion unit, used for converting the homogenized random number sequence into a standard normal distribution random number sequence;

[0042] A multidimensional standard normal distribution function generating unit, used for estimating the correlation of normal random variables and generating a multidimensional standard normal distribution function according to the correlation;

[0043] A multi - variable random number pair generation unit, which is used to generate multi - variable random number pairs with a correlation factor of the said correlation degree by using a multi - dimensional normal distribution function;

[0044] A driving pattern random number generation unit, which is used to convert the said multi - variable random number pairs into original type data; the original type data is driving pattern random numbers with coupling characteristics.

[0045] Optionally, it further includes:

[0046] A modeling module, which is used to model the uncertainty of the charging behavior of vehicle owners according to the said optimal charging load simulation curve; the charging behavior of vehicle owners includes charging behavior and non - charging behavior; the charging behavior of vehicle owners is an uncertain behavior.

[0047] Optionally, the said modeling module specifically includes:

[0048] A clustering unit, which is used to cluster that the charging behavior of vehicle owners approximately follows a binomial distribution B(m, p) with parameters m and p; where m is the total number of samples and p is the charging probability.

[0049] A random number generation unit, which is used to set the clustering charging probability as p, p ∈ [0, 1], and for the i - th electric vehicle, randomly generate a random number rand that follows a uniform distribution U(0, 1); U(0, 1) is a uniform distribution function that follows the interval [0, 1].

[0050] A charging behavior selection unit, which is used to select the charging behavior when rand < p;

[0051] A non - charging behavior selection unit, which is used to select the non - charging behavior when rand ≥ p.

[0052] According to the specific embodiments provided by the present invention, the following technical effects are disclosed: The present invention provides a method and system for predicting the charging load of electric vehicles. After a large number of electric vehicles are simulated for a certain number of days, the variable - bandwidth kernel density is used to accurately fit the charging load during this period, and finally the optimal charging load simulation curve is fitted, so as to reasonably and effectively predict the charging load of electric vehicles in the future, improve the prediction accuracy of the charging load of electric vehicles, and provide help for the power supply quality of the power grid. Description of the Drawings

[0053] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required to be used in the embodiments. Obviously, the following - described drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0054] Figure 1This is a flow chart of the method for predicting the charging load of an electric vehicle provided by the present invention;

[0055] Figure 2 A calculation flow chart for generating random numbers of driving rules with coupling characteristics according to the present invention;

[0056] Figure 3 The charging load calculation flow chart provided by the present invention;

[0057] Figure 4 This is a structural diagram of the electric vehicle charging load prediction system provided by the present invention. DETAILED DESCRIPTION

[0058] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only 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] The purpose of the present invention is to provide a method and system for predicting the charging load of an electric vehicle, thereby improving the prediction accuracy of the charging load of the electric vehicle.

[0060] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0061] Figure 1 The flowchart of the method for predicting the charging load of an electric vehicle provided by the present invention is as follows: Figure 1 As shown, a method for predicting charging load of an electric vehicle includes:

[0062] Step 101: Acquire the driving pattern of the electric vehicle, and fit the driving pattern probability distribution characteristic parameters according to the driving pattern of the electric vehicle; the driving pattern probability distribution characteristic parameters include the starting driving time, the ending driving time, the probability distribution of the driving mileage, the charging start time and the charging end time of each day.

[0063] In practical applications, the driving pattern probability distribution characteristic parameters are determined according to the driving pattern of the vehicle owner.

[0064] For example, the charging simulation curve is constructed by fitting the initial charging time and the end charging time, and then the mileage simulation curve is constructed according to the mileage to better describe the changes in driving patterns. From this, we can get the peak charging time of the day, or the mileage range of most electric vehicles in a day.

[0065] Step 102: Fitting charging characteristics of electric vehicles according to the driving rules of multiple electric vehicles; the charging characteristics include a charging start time and a charging end time.

[0066] In practical applications, the charging characteristics are fitted according to the charging characteristics of the car owner and the probability distribution characteristic parameters of the driving pattern, and the time changes of the charging load can be intuitively seen.

[0067] Step 103: Determine electric vehicle parameters according to the actual number of electric vehicles in the area; the electric vehicle parameters include charging power, power consumption characteristics and charging mode.

[0068] In practical applications, the number N of electric vehicles is determined according to the actual number of electric vehicles in the area, and the electric vehicle parameters are determined; by determining the characteristics of the electric vehicles, the charging loads of different electric vehicles can be fitted, and the power loss of different electric vehicles can also be obtained.

[0069] Step 104: Determine random driving data according to the electric vehicle parameters, and convert the random driving data into a multidimensional normal distribution function to generate driving law random numbers containing coupling characteristics; the random driving data includes the start time of charging, the end time of charging and the driving mileage.

[0070] The step 104 specifically includes: using the formula Calculate the marginal distribution function of each random driving data; generate a uniform random number sequence from the processed random driving data according to the marginal distribution function; convert the uniform random number sequence into a standard normal distribution random number sequence; estimate the correlation of normal random variables, and generate a multidimensional standard normal distribution function according to the correlation; use the multidimensional normal distribution function to generate a multivariate random number pair whose correlation factor is the correlation; convert the multivariate random number pair into original type data; the original type data is a driving law random number containing coupling characteristics; wherein, is the marginal distribution function; x is the random variable of random driving data; is the random variable density function.

[0071] Figure 2 This is a calculation flow chart of the random number generation of driving rules with coupling characteristics of the present invention. The calculation steps of the random number generation of driving rules with coupling characteristics are as follows:

[0072] (1) According to Sklar's theorem, a system with cumulative marginal distribution F 1 , F 2 ,…,F N The N-dimensional joint distribution function F, for x i ∈(-∞,+∞); x i∈(-∞,+∞) is the i-th variable; there exists a Copula function such that: F(x 1 ,x 2 ,...,x N )=C[F 1 (x 1 ),F(x 2 ),...,F N (x N )]

[0073] (2) Adoption The marginal distribution function of each random variable is calculated, and the starting time (T 1 )End time (T 2 ) The statistical sample of mileage (D) is homogenized, that is, converted into a random number sequence in the interval (0, 1): U T1 =F T1 (T 1 ), U T2 =F T2 (T 2 ), U D =F D (D).

[0074] (3) The start time of the uniform random number sequence (U T1 )、End Time(U T2 ), mileage (U D ) is converted to a standard normal distribution random number sequence: Y T1 =Ф -1 (U T1 ), Y T2 =Ф -1 (U T2 ), Y D =Ф -1 (U D ).

[0075] (4) Estimate Y T1 ,Y T2 ,Y D The correlation degree ρ of normal random variables forms a multidimensional standard normal distribution function with a correlation degree ρ; the correlation expression between two random variables is as follows:

[0076]

[0077] in, represent the sample means of random variables X and Y respectively.

[0078] (5) Use the multidimensional normal distribution function to generate multivariate random number pairs with correlation factor ρ, start time, end time, and mileage (y t1 ,yt2 ,y d );

[0079] (6) The generated (y t1 ,y t2 ,y d ) is converted to the original type of data t 1 =F -1 T1 (y t1 ), t 2 =F -1 T2 (y t2 ), d = F -1 D (y d ).

[0080] Step 105: According to the driving law of the electric vehicle, the charging characteristics and the driving law random number containing coupling characteristics, a simulation model of variable bandwidth kernel density estimation is established using a variable bandwidth kernel density function to determine an optimal charging load simulation curve; the optimal charging load simulation curve is used to predict the charging charge of the electric vehicle.

[0081] Figure 3 The charging load calculation flow chart provided by the present invention, the charging load calculation flow chart has the following steps:

[0082] (1) Using the formula Calculate variable bandwidth; h j is the bandwidth of the jth sample point required; is the product of k=1 to k=M fixed kernel density functions; M is the number of sample points; f(X k ) is the kernel density function of the selected fixed k-th sample point; X k is the variable value of the kth sample point; k is the sequence number of the kth sample point; f(X j ) is the kernel density function of the jth sample point; X j is the variable value of the jth sample point; j is the serial number of the jth sample point; α is the sensitivity factor. (2) Generally, the value of α is 0.5, and the appropriate value can be adjusted within 0<α<1 according to the situation, and it is substituted into (1) to calculate h; h represents the calculated variable bandwidth.

[0083] (3) Use the variable bandwidth kernel density function to fit the car driving pattern and obtain the mileage range of different cars.

[0084] (4) Fit the characteristics of the charging start time, end time, etc. using a variable bandwidth kernel density function to obtain the charging load for different time periods. Among them, deterministic charging behavior refers to the situation where the current remaining battery power is no longer sufficient to meet the remaining driving mileage of the vehicle owner to the destination, and it has a normal driving pattern. For example, taxis charging one or two times a day is deterministic charging, etc.

[0085] (5) Set the number of simulation days, and use conditional statements to perform iterative calculations on the charging loads of the first i vehicles to finally obtain the total charging load.

[0086] In practical applications, for kernel density estimation with variable bandwidth, a reasonable bandwidth h needs to be calculated, and the calculated h is substituted into the kernel density formula for application in charging load prediction to obtain the charging load curve.

[0087] After the step 105, it further includes: modeling the uncertainty of the vehicle owner's charging behavior according to the optimal charging load simulation curve; the vehicle owner's charging behavior includes charging behavior and non-charging behavior; the vehicle owner's charging behavior is an uncertain behavior.

[0088] The modeling of the uncertainty of the vehicle owner's charging behavior according to the optimal charging load simulation curve specifically includes: clustering the vehicle owner's charging behavior approximately follows a binomial distribution B(m, p) with parameters m and p; where m is the total number of samples and p is the charging probability. Set the clustering charging probability as p, p ∈ [0, 1]. For the i-th electric vehicle, a random number rand that follows a U(0, 1) uniform distribution is randomly generated; U(0, 1) is a uniform distribution function that follows the interval [0, 1]. When rand < p, select the charging behavior; when rand ≥ p, select the non-charging behavior.

[0089] In practical applications, the present invention includes the following steps to solve the technical problems of the present application.

[0090] (1) Determine the characteristic parameters of the driving pattern probability distribution according to the driving pattern of private electric vehicles.

[0091] (2) Fit its charging characteristics according to the charging patterns of a large number of vehicle owners, including the charging start time, charging end time, etc.

[0092] (3) Determine the number N of electric vehicles (determined according to the actual number of electric vehicles in the area) and the parameters of electric vehicles, including the power consumption characteristics and battery parameters of electric vehicles, and perform separate calculations for different power consumption characteristics and battery parameters.

[0093] (4) According to the obtained random driving data of electric vehicles, transform these random data into a multi-dimensional normal distribution function through a function.

[0094] (5) Use the multidimensional normal distribution function to generate a multivariate random number pair (yt1, yt2, yd) with a correlation factor of ρ, including the start time, end time, mileage, etc.

[0095] (6) Convert the generated multivariate random number pair (yt1, yt2, yd) into the original type of data t1 = F-1T1(yt1), t2 = F-1T2(yt2), d = F-1D(yd).

[0096] (7) According to the charging characteristics and driving patterns of an electric vehicle, a variable bandwidth kernel density function is used to fit its charging load at different time periods of the day.

[0097] (8) Based on the charging load of an electric vehicle, considering that N electric vehicles are connected, the variable bandwidth kernel density estimation method is used to fit the charging load of N electric vehicles.

[0098] (9) After a certain number of days of simulation for a large number of electric vehicles, the variable bandwidth kernel density is used to accurately fit the charging load during this period, thereby making a reasonable and effective prediction of the future charging load of electric vehicles.

[0099] Figure 4 This is a structural diagram of the prediction system for the electric vehicle charging load provided by the present invention, such as Figure 4 As shown, a prediction system for electric vehicle charging load includes:

[0100] The driving pattern probability distribution characteristic parameter fitting module 401 is used to obtain the driving pattern of the electric vehicle and fit the driving pattern probability distribution characteristic parameters according to the driving pattern of the electric vehicle; the driving pattern probability distribution characteristic parameters include the starting driving time, the ending driving time, the probability distribution of the driving mileage, the charging start time and the charging end time of each day.

[0101] The charging characteristic fitting module 402 is used to fit the charging characteristics of electric vehicles according to the driving rules of the electric vehicles of multiple electric vehicles; the charging characteristics include the start time of charging and the end time of charging.

[0102] The electric vehicle parameter determination module 403 is used to determine the electric vehicle parameters according to the actual number of electric vehicles in the area; the electric vehicle parameters include charging power, power consumption characteristics and charging mode.

[0103] The driving pattern random number generation module 404 is used to determine random driving data according to the electric vehicle parameters, and convert the random driving data into a multidimensional normal distribution function to generate a driving pattern random number containing coupling characteristics; the random driving data includes the start charging time, the end charging time and the driving mileage.

[0104] The driving pattern random number generation module 404 specifically includes: a marginal distribution function calculation unit for using the formula Calculate the marginal distribution function of each random driving data; a random number sequence generating unit, used to generate a uniform random number sequence from the processed random driving data according to the marginal distribution function; a conversion unit, used to convert the uniform random number sequence into a standard normal distribution random number sequence; a multidimensional standard normal distribution function generating unit, used to estimate the correlation of normal random variables, and generate a multidimensional standard normal distribution function according to the correlation; a multivariate random number pair generating unit, used to generate a multivariate random number pair whose correlation factor is the correlation using a multidimensional normal distribution function; a driving pattern random number generating unit, used to convert the multivariate random number pair into original type data; the original type data is a driving pattern random number containing coupling characteristics; wherein, is the marginal distribution function; x is the random variable of random driving data; is the random variable density function.

[0105] The optimal charging load simulation curve determination module 405 is used to establish a simulation model of variable bandwidth kernel density estimation using a variable bandwidth kernel density function according to the driving law of the electric vehicle, the charging characteristics and the driving law random number containing coupling characteristics, and determine the optimal charging load simulation curve; the optimal charging load simulation curve is used to predict the charging charge of the electric vehicle.

[0106] The variable bandwidth kernel density function is:

[0107]

[0108] Among them, h j is the bandwidth of the jth sample point required; is the product of k=1 to k=M fixed kernel density functions; M is the number of sample points; f(X k ) is the kernel density function of the selected fixed k-th sample point; X k is the variable value of the kth sample point; k is the sequence number of the kth sample point; f(X j ) is the kernel density function of the jth sample point; X j is the variable value of the j-th sample point; j is the sequence number of the j-th sample point; α is the sensitivity factor.

[0109] The present invention also includes: a modeling module, which is used to model the uncertainty of the car owner's charging behavior according to the optimal charging load simulation curve; the car owner's charging behavior includes charging behavior and non-charging behavior; the car owner's charging behavior is uncertain behavior.

[0110] The modeling module specifically includes: a clustering unit for clustering that the charging behavior of the vehicle owners approximately follows a binomial distribution B(m, p) with parameters m and p; where m is the total number of samples and p is the charging probability; a random number generation unit for setting the clustering charging probability as p, p ∈ [0, 1], and for the i-th electric vehicle, randomly generating a random number rand that follows a uniform distribution U(0, 1); U(0, 1) is a uniform distribution function that follows the interval [0, 1]. A charging behavior selection unit for selecting a charging behavior when rand < p; a non-charging behavior selection unit for selecting a non-charging behavior when rand ≥ p.

[0111] The present invention uses a kernel density function with a variable bandwidth to fit the charging load curve, which can be helpful for the power supply quality of the power grid.

[0112] In this specification, each embodiment is described in a progressive manner. The key point of each embodiment is to illustrate the differences from other embodiments. For the same or similar parts among the embodiments, reference can be made to each other. For the system disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple, and reference can be made to the description in the method part for related parts.

[0113] In this article, specific examples are used to elaborate on the principle and implementation manner of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea; at the same time, for those of ordinary skill in the art, according to the idea of the present invention, there will be changes in the specific implementation manner and application scope. In summary, the content of this specification should not be construed as a limitation to the present invention.

Claims

1. A method for predicting the charging load of electric vehicles, characterized in that, it includes: Obtain the driving rules of electric vehicles, and fit the characteristic parameters of the driving rule probability distribution according to the driving rules of the electric vehicles; the characteristic parameters of the driving rule probability distribution include the starting driving time, ending driving time, probability distribution of driving mileage, charging start time, and charging end time of each day; Fit the charging characteristics of electric vehicles according to the driving rules of multiple electric vehicles; the charging characteristics include the start charging time and the end charging time; Determine the electric vehicle parameters according to the actual number of electric vehicles in the area; the electric vehicle parameters include charging power, power consumption characteristics, and charging mode; Determine the random driving data according to the electric vehicle parameters, and convert the random driving data into a multi-dimensional normal distribution function to generate a driving rule random number with coupling characteristics; The random driving data includes the start charging time, end charging time, and driving mileage; According to the driving rules of the electric vehicles, the charging characteristics, and the driving rule random number with coupling characteristics, establish a simulation model of variable bandwidth kernel density estimation using a variable bandwidth kernel density function, and determine the optimal charging load simulation curve; the optimal charging load simulation curve is used to predict the charging charge of electric vehicles; Model the uncertainty of the charging behavior of vehicle owners according to the optimal charging load simulation curve, specifically including: Cluster the charging behaviors of vehicle owners approximately follow a binomial distribution B(m, p) with parameters m and p; where m is the total number of samples and p is the charging probability; Set the cluster charging probability to p, p ∈ [0, 1]. For the i-th electric vehicle, randomly generate a random number rand that follows a U(0, 1) uniform distribution; U(0, 1) is a uniform distribution function that follows the interval [0 1]; When rand < p, select the charging behavior; When rand ≥ p, select the non-charging behavior; the charging behaviors of vehicle owners include charging behaviors and non-charging behaviors; the charging behaviors of vehicle owners are uncertain behaviors.

2. The method for predicting the charging load of electric vehicles according to claim 1, characterized in that, The variable bandwidth kernel density function is: Among them, h j is the bandwidth of the jth sample point required; is the product of k=1 to k=M fixed kernel density functions; M is the number of sample points; f(X k ) is the kernel density function of the selected fixed k-th sample point; X k is the variable value of the kth sample point; k is the sequence number of the kth sample point; f(X j ) is the kernel density function of the jth sample point; X j is the variable value of the j-th sample point; j is the sequence number of the j-th sample point; α is the sensitivity factor.

3. The method for predicting the charging load of electric vehicles according to claim 1, characterized in that, The determining the random driving data according to the electric vehicle parameters, and converting the random driving data into a multi-dimensional normal distribution function to generate a driving rule random number with coupling characteristics specifically includes: Using the formula Calculate the marginal distribution function of each of the random driving data; wherein, is the marginal distribution function; x is the random variable of random driving data; is the random variable density function; Generate a homogenized random number sequence according to the marginal distribution function for the processed random driving data; Convert the homogenized random number sequence into a standard normal distribution random number sequence; Estimate the correlation degree of the normal random variable, and generate a multi-dimensional standard normal distribution function according to the correlation degree; Generate a multivariate random number pair with the correlation factor being the correlation degree using the multi-dimensional normal distribution function; Convert the multivariate random number pair into the original type data; the original type data is a driving rule random number with coupling characteristics.

4. A system for predicting the charging load of electric vehicles, characterized in that, it includes: The driving pattern probability distribution characteristic parameter fitting module is used to obtain the driving pattern of the electric vehicle and fit the driving pattern probability distribution characteristic parameters according to the driving pattern of the electric vehicle; the driving pattern probability distribution characteristic parameters include the starting driving time, ending driving time, probability distribution of driving mileage, charging start time, and charging end time of each day; The charging characteristic fitting module is used to fit the charging characteristics of the electric vehicle according to the driving pattern of multiple electric vehicles; the charging characteristics include the charging start time and the charging end time; The electric vehicle parameter determination module is used to determine the electric vehicle parameters according to the actual number of electric vehicles in the area; the electric vehicle parameters include charging power, power consumption characteristics, and charging mode; The driving pattern random number generation module is used to determine random driving data according to the electric vehicle parameters, and convert the random driving data into a multi-dimensional normal distribution function to generate driving pattern random numbers with coupling characteristics; The random driving data includes the charging start time, the charging end time, and the driving mileage; The optimal charging load simulation curve determination module is used to establish a simulation model of variable bandwidth kernel density estimation using a variable bandwidth kernel density function according to the driving pattern of the electric vehicle, the charging characteristics, and the driving pattern random numbers with coupling characteristics, and determine the optimal charging load simulation curve; the optimal charging load simulation curve is used to predict the charging charge of the electric vehicle; The modeling module is used to model the uncertainty of the owner's charging behavior according to the optimal charging load simulation curve; The owner's charging behavior includes charging behavior and non-charging behavior; The owner's charging behavior is an uncertain behavior; The modeling module specifically includes: The clustering unit is used to cluster that the owner's charging behavior approximately follows a binomial distribution B(m, p) with parameters m and p; where m is the total number of samples and p is the charging probability; The random number generation unit is used to set the clustering charging probability as p, p ∈ [0, 1], and for the i-th electric vehicle, randomly generate a random number rand that follows a U(0, 1) uniform distribution; U(0, 1) is a uniform distribution function that follows the interval [0 1]; The charging behavior selection unit is used to select the charging behavior when rand < p; The non-charging behavior selection unit is used to select the non-charging behavior when rand ≥ p.

5. The prediction system for the charging load of an electric vehicle according to claim 4, wherein, The variable bandwidth kernel density function is: Among them, h j is the bandwidth of the jth sample point required; is the product of k=1 to k=M fixed kernel density functions; M is the number of sample points; f(X k ) is the kernel density function of the selected fixed k-th sample point; X k is the variable value of the kth sample point; k is the sequence number of the kth sample point; f(X j ) is the kernel density function of the jth sample point; X j is the variable value of the j-th sample point; j is the sequence number of the j-th sample point; α is the sensitivity factor.

6. The prediction system for the charging load of an electric vehicle according to claim 4, wherein, The driving pattern random number generation module specifically includes: Marginal distribution function calculation unit, used to use the formula Calculate the marginal distribution function of each of the random driving data; wherein, is the marginal distribution function; x is the random variable of random driving data; is the random variable density function; The random number sequence generation unit is used to generate a homogenized random number sequence according to the marginal distribution function for the processed random driving data; The conversion unit is used to convert the homogenized random number sequence into a standard normal distribution random number sequence; The multi-dimensional standard normal distribution function generation unit is used to estimate the correlation degree of the normal random variable and generate a multi-dimensional standard normal distribution function according to the correlation degree; A multivariate random number pair generating unit, used for generating a multivariate random number pair whose correlation factor is the correlation degree by using a multidimensional normal distribution function; The driving pattern random number generating unit is used to convert the multivariate random number pair into original type data; the original type data is a driving pattern random number containing coupling characteristics.

Images