Method for predicting charging load of electric vehicles

Abstract

The present invention provides a method for predicting the charging load of an electric vehicle, and the method mainly includes: step S1, dividing the travel destinations of electric vehicles into four categories, counting the daily travel data of electric vehicles to obtain the travel characteristics of electric vehicles, and establishing a travel model of electric vehicles; step S2, establishing a traffic network topology model, and filtering the data set whose travel time falls within the congestion time period of the city from NHTS 2017; step S3, establishing a multiple linear regression model of travel time and other variables based on the filtered data; step S4, predicting the charging load of electric vehicles in a certain area based on the travel model and multiple linear regression model established above. The method for predicting the charging load of electric vehicles of the present invention makes the electric vehicle load prediction method closer to reality and has very good applicability.

Description

Technical Field

[0001] The invention relates to a method for predicting charging load of an electric vehicle, and belongs to the technical field of prediction of charging load of an electric vehicle. Background Art

[0002] As an important member of the new energy industry structure, electric vehicles have played a significant role in improving the global climate and alleviating energy problems. my country attaches great importance to the electric vehicle industry. However, with the popularization of electric vehicles, a large number of electric vehicles are connected to the power grid. Their disorder, instantaneity and imbalance will cause voltage quality degradation and three-phase voltage imbalance on the power distribution network. Accurate electric vehicle load prediction curves can give the charging load conditions in various time periods of the day. Combining this result with relevant research can solve the above problems caused by electric vehicles connecting to the power grid in a targeted manner.

[0003] In recent years, domestic and foreign scholars have conducted a lot of research on electric vehicle load forecasting. The time when electric vehicles start charging is an important basis for load forecasting. Therefore, some previous studies mainly focus on the time when electric vehicles start charging. By calculating the time of departure and mileage, the time when electric vehicles arrive at the destination and the remaining battery power are calculated to determine whether the electric vehicle needs to be charged and obtain the time when charging starts. However, considering that the charging of electric vehicles is a behavior based on the power grid-transportation network and is easily affected by user choices, some scholars have proposed combining the road network with the power grid, and introducing factors such as road section impedance and node impedance to simulate electric vehicle travel, and thus predict the charging load of electric vehicles. In addition, related research focuses on the travel demand of electric vehicles, and describes the charging behavior of electric vehicles by predicting the travel behavior of car owners. However, there are many defects in the current research methods, such as ignoring the impact of travel date type on electric vehicle travel behavior, ignoring the impact of traffic peak time on the travel time of car owners, etc.

[0004] In view of this, it is necessary to propose a method for predicting electric vehicle charging load to solve the above problems. Summary of the invention

[0005] The purpose of the present invention is to provide a method for predicting the charging load of an electric vehicle, so that the method for predicting the charging load of an electric vehicle is closer to reality and has very good applicability.

[0006] To achieve the above object, the present invention provides a method for predicting the charging load of an electric vehicle, the method mainly comprising:

[0007] Step S1, classifying the travel destinations of electric vehicles into four categories, counting the daily travel data of electric vehicles to obtain the travel characteristics of electric vehicles, and establishing a travel model of electric vehicles;

[0008] Step S2, building a traffic network topology model and selecting a data set from NHTS 2017 whose travel time falls within the congestion period of the city;

[0009] Step S3, establishing a multiple linear regression model of travel time and other variables based on the screened data;

[0010] Step S4, based on the above established electric vehicle travel model and multivariate linear regression model, the charging load of electric vehicles in a certain area is predicted.

[0011] As a further improvement of the present invention, step S1 specifically includes:

[0012] Step S11, dividing the destination into a living area, a work area, a commercial area and an entertainment area;

[0013] Step S12, fitting the travel characteristics of electric vehicles on weekends and weekdays from the NHTS 2017 data set;

[0014] Step S13, based on the fitting result of the previous step and the basic model of electric vehicle travel, a travel model of electric vehicles on weekdays and weekends is established.

[0015] As a further improvement of the present invention, in step S12, the travel characteristics are the starting point, destination, and stay time at the destination of the electric vehicle.

[0016] As a further improvement of the present invention, the first travel time of the electric vehicle on weekends and weekdays and the residence time of the electric vehicle in different areas under different date types in step S12 satisfy the normal distribution:

[0017]

[0018] Among them, μ 0、 σ0 is obtained by fitting the dataset in NHTS 2017.

[0019] As a further improvement of the present invention, the basic model of electric vehicle travel includes travel time, the time when the i-th trip arrives at the destination and the start time of the (i+1)-th trip, the remaining power of the electric vehicle, and the charging probability model of the electric vehicle;

[0020] The travel time is as follows:

[0021]

[0022] Among them, L ij represents the actual distance between destination i and destination j, and V represents the average driving speed of the electric vehicle;

[0023] The time t at which the i-th trip arrives at the destination is di and the starting time t of the i+1th trip o(i+1) as follows:

[0024] t di =t oi +T

[0025] t o(i+1) =t oi +t pi +T

[0026] Among them, T is the driving time, t pi is the parking time of the i-th trip;

[0027] The remaining battery power during charging needs to be calculated based on the last mileage. The calculation formula is as follows:

[0028]

[0029] Among them, L is the last mileage, SOC end The remaining battery power after the last mileage, SOC before The remaining battery power before the last journey. max It is the maximum driving range of an electric vehicle when fully charged.

[0030] As a further improvement of the present invention, step S2 specifically includes:

[0031] Step S21, building a road network model based on graph theory and making assumptions about the peak traffic time period of urban roads;

[0032] Step S22, filter out the data set in the urban road traffic peak time period from the NHTS 2017 data set, and divide it into weekday morning peak, weekday evening peak, weekend morning peak, and weekend evening peak.

[0033] As a further improvement of the present invention, step S3 specifically includes:

[0034] Step S31, preprocessing the data set;

[0035] Step S32, describing the processed data, observing the characteristics of the data set, and determining whether to use this data set for multivariate linear regression;

[0036] Step S33, establish a multivariate linear regression model, perform least squares estimation of parameters, obtain a linear regression equation of travel time and other variables, and perform a significance test.

[0037] As a further improvement of the present invention, in step S31, the obtained data set needs to be further processed to delete data segments that are useless for establishing the multivariate linear regression model.

[0038] As a further improvement of the present invention, in step S32, a regression model is used to perform data analysis. If the error term does not satisfy the normality assumption, a parameter λ value is given and a Box-Cox transformation is performed on the explained variable:

[0039]

[0040] Among them, the parameter λ is determined by maximum likelihood estimation.

[0041] As a further improvement of the present invention, in step S33,

[0042]

[0043] Among them, Y is the travel time of electric vehicles, X is the independent variable that has a greater impact on the travel time of electric vehicles, β is the parameter of the model, and ε is the error term, which reflects other factors that affect the variation of the value of the dependent variable Y in addition to X;

[0044] Then the multiple linear regression model can be expressed as:

[0045] Y=Xβ+ε

[0046] The parameters of the established multiple linear regression model are β0, β1, …β k is unknown, and a sample estimator is needed Estimation is performed and the estimated multiple regression equation is obtained:

[0047]

[0048] in represents the estimated driving time of the electric vehicle, It means that when the independent variables X2, X3, ..., X k When X1 changes by one unit, the average change in the driving time of electric vehicles is units, It means that when the independent variables X1, X3, ..., X k When X2 changes by one unit, the average change in the driving time of electric vehicles is units, ..., It means that when the independent variables X1, X3, ..., X k-1 When X k For every unit change, the average change in electric vehicle driving time units;

[0049] Estimate the parameters in the multiple linear regression model using the least squares method:

[0050]

[0051] The following equations are used to find the minimum Q

[0052]

[0053] In the multiple linear regression model, the estimator of the variance of the random error term ε is

[0054]

[0055] The beneficial effects of the present invention are as follows: the method for predicting the charging load of an electric vehicle of the present invention establishes an electric vehicle travel model, fully considering the impact of date factors on the travel behavior of electric vehicles and the impact of peak traffic time periods on the travel duration of electric vehicles, making the electric vehicle load prediction method closer to reality and having very good applicability. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 The present invention is a flow chart of the method for predicting the charging load of an electric vehicle.

[0057] Figure 2 It is a charging load curve of electric vehicles in a certain area on weekdays taking into account the peak period of urban road traffic.

[0058] Figure 3 This is a curve of electric vehicle charging load in a certain area on a weekend day taking into account the peak traffic period on urban roads.

[0059] Figure 4 The basic topology model in the present invention establishes a road network model. DETAILED DESCRIPTION

[0060] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention is described in detail below with reference to the accompanying drawings and specific embodiments.

[0061] like Figure 1 As shown, the present invention discloses a method for predicting the charging load of an electric vehicle, and the method for predicting the charging load of an electric vehicle mainly includes:

[0062] Step S1, classifying the travel destinations of electric vehicles into four categories, counting the daily travel data of electric vehicles to obtain the travel characteristics of electric vehicles, and establishing a travel model of electric vehicles;

[0063] Step S2, building a traffic network topology model and selecting a data set from NHTS 2017 whose travel time falls within the congestion period of the city;

[0064] Step S3, establishing a multiple linear regression model of travel time and other variables based on the screened data;

[0065] Step S4, based on the above established electric vehicle travel model and multivariate linear regression model, the charging load of electric vehicles in a certain area is predicted.

[0066] The following will describe steps S1-S3 in detail:

[0067] Step S1 specifically includes the following steps:

[0068] Step S11: According to the travel purposes of travelers in the NHTS 2017 dataset, the area is divided into four functional areas: living area, working area, commercial area and entertainment area.

[0069]

[0070]

[0071] Step S12: Using the NHTS 2017 data set, different characteristics of electric vehicles traveling on weekends and weekdays are fitted. The first travel time of electric vehicles on weekends and weekdays and the residence time of electric vehicles in different areas on different date types satisfy the normal distribution:

[0072]

[0073] The μ0 and σ0 are obtained by fitting the dataset in NHTS 2017.

[0074] The basic model of electric vehicle travel includes driving time, the time when the i-th trip arrives at the destination and the starting time of the i+1-th trip, the remaining power of the electric vehicle, and the electric vehicle charging probability model.

[0075] The travel time is as follows:

[0076]

[0077] Among them, L ij represents the actual distance between destination i and destination j, and V represents the average driving speed of the electric vehicle.

[0078] The time t at which the i-th trip arrives at the destination is di and the starting time t of the i+1th trip o(i+1) as follows:

[0079] t di =t oi +T

[0080] t o(i+1) =t oi +t pi +T

[0081] Among them, T is the driving time, t pi is the parking time of the i-th trip.

[0082] The remaining battery power during charging needs to be calculated based on the previous mileage, and the calculation formula is as follows.

[0083]

[0084] Among them, L is the last mileage, SOC end The remaining battery power after the last mileage, SOC before The remaining battery power before the last journey. max It is the maximum driving range of an electric vehicle when fully charged.

[0085] The charging probability model of electric vehicles is as follows:

[0086]

[0087] Among them, w is the power consumption of electric vehicles per kilometer, L next For the next trip.

[0088] Step S13: Based on the fitting result of the previous step and the basic model of electric vehicle travel, a travel model for electric vehicles on weekdays and weekends is established, wherein the number of travel times of electric vehicles per day is set to 3 times, and charging is performed by default after the last trip.

[0089] Step S2 specifically includes the following steps:

[0090] Step S21: Model the road network model based on graph theory and make assumptions about the peak traffic time period of urban roads; that is, based on Figure 4 The basic topology model establishes the road network model.

[0091] in, Figure 4 The one-way arrow in the middle indicates a one-way street, and the two-way arrow indicates a two-way street. The mathematical model can be expressed as:

[0092]

[0093] Where: G is the traffic road network; D is the set of all nodes in graph G; E is the set of all road sections in graph G; W is the set of road section lengths.

[0094] Combination Figure 2 and Figure 3As shown, step S22, a data set in the peak time period of urban road traffic is filtered out from the NHTS 2017 data set, which is divided into morning peak time on weekdays, evening peak time on weekdays, morning peak time on weekends, and evening peak time on weekends. Assuming the peak time period of urban road traffic, the present invention assumes that the morning peak time period on weekdays is 7:00-9:00, the evening peak time period is 17:00-19:00, and the morning peak time period on weekends is 10:00-11:00, and the evening peak time period is 18:00-20:00. According to this time period, the data set with a travel range of the city in NHTS 2017 is filtered.

[0095] Step S3 specifically includes the following steps:

[0096] Step S31, during data preprocessing, the obtained data set is further processed to delete data segments that are useless for establishing a multivariate linear regression model. For example, if there is a variable in the data set that is the travel destination, if the value corresponding to this variable is -8, it means that the respondent has forgotten the destination of the trip at that time, and such useless data is deleted.

[0097] Step S32, using the regression model to perform data analysis, if the error term does not meet the normality assumption, a parameter λ value is given and a Box-Cox transformation is performed on the explained variable:

[0098]

[0099] Among them, the parameter λ is determined by maximum likelihood estimation.

[0100] Step S33, establish a multiple linear regression model, let

[0101]

[0102] Among them, Y is the driving time of electric vehicles, X is the independent variable that has a greater impact on the driving time of electric vehicles, β is the parameter of the model, and ε is the error term, which reflects other factors that affect the variation of the value of the dependent variable Y except X.

[0103] Then the multiple linear regression model can be expressed as:

[0104] Y=Xβ+ε

[0105] The parameters of the established multiple linear regression model are β0, β1, …β k is unknown, and a sample estimator is needed Estimation is performed and the estimated multiple regression equation is obtained:

[0106]

[0107] in represents the estimated driving time of the electric vehicle, It means that when the independent variables X2, X3, ..., X k When X1 changes by one unit, the average change in the driving time of electric vehicles is units, It means that when the independent variables X1, X3, ..., X k When X2 changes by one unit, the average change in the driving time of electric vehicles is units, ..., It means that when the independent variables X1, X3, ..., X k-1 When X k For every unit change, the average change in electric vehicle driving time units.

[0108] Estimate the parameters in the multiple linear regression model using the least squares method:

[0109]

[0110] The following equations are used to find the minimum Q

[0111]

[0112] In the multiple linear regression model, the estimator of the variance of the random error term ε is

[0113]

[0114] Step S4: According to the above steps, the number of electric vehicles in the area is set, and each electric vehicle is selected in turn. The model established according to the above steps is used to simulate the travel process of a day and predict the spatiotemporal distribution of electric vehicle charging in the area.

[0115] In summary, the method for predicting the charging load of an electric vehicle of the present invention establishes an electric vehicle travel model, fully considering the impact of date factors on the travel behavior of electric vehicles and the path selection behavior of electric vehicles during driving, making the electric vehicle load prediction method closer to reality and has very good applicability.

[0116] The above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention.

Claims

1. A method for predicting electric vehicle charging load, characterized in that: The method for predicting the charging load of an electric vehicle mainly includes: Step S1, classifying the travel destinations of electric vehicles into four categories, counting the daily travel data of electric vehicles to obtain the travel characteristics of electric vehicles, and establishing a travel model of electric vehicles; Step S2, building a traffic network topology model and selecting a data set from NHTS2017 whose travel time falls within the congestion period of the city; Step S3, establishing a multiple linear regression model of travel time and other variables based on the screened data; Step S4, predicting the charging load of electric vehicles in a preset area based on the travel model and multivariate linear regression model of electric vehicles established above; Step S1 specifically includes: Step S11, dividing the destination into a living area, a work area, a commercial area and an entertainment area; Step S12, fitting the travel characteristics of electric vehicles on weekends and weekdays from the NHTS2017 data set; Step S13, based on the fitting result of the previous step and the basic model of electric vehicle travel, a travel model of electric vehicles on weekdays and weekends is established; The first travel time of the electric vehicle on weekends and weekdays and the residence time of the electric vehicle in different areas under different date types in step S12 satisfy the normal distribution: The μ0 and σ0 are obtained by fitting the data set in NHTS2017; Step S3 specifically includes: Step S31, preprocessing the data set; Step S32, describing the processed data, observing the characteristics of the data set, and determining whether to use this data set for multivariate linear regression; Step S33, establishing a multivariate linear regression model, and performing least squares estimation of parameters to obtain a linear regression equation of travel time and other variables, and performing a significance test; In step S32, the regression model is used to analyze the data. If the error term does not meet the normality assumption, a parameter λ value is given and a Box-Cox transformation is performed on the explained variable: Among them, the parameter λ is determined by maximum likelihood estimation; In step S33, Among them, Y is the travel time of electric vehicles, X is the independent variable that has a greater impact on the travel time of electric vehicles, β is the parameter of the model, and ε is the error term, which reflects other factors that affect the variation of the value of the dependent variable Y in addition to X; The multiple linear regression model is expressed as: Y=Xβ+ε, The parameters of the established multiple linear regression model are β0, β1, …, β k is unknown, and a sample estimator is needed Estimation is performed and the estimated multiple regression equation is obtained: in, represents the estimated driving time of the electric vehicle, It means that when the independent variables X2, X3, …, X k When X1 changes by one unit, the average change in the driving time of electric vehicles is units, It means that when the independent variables X1, X3, …, X k When X2 changes by one unit, the average change in the driving time of electric vehicles is units, ..., It means that when the independent variables X1, X3, …, X k-1 When X k For every unit change, the average change in electric vehicle driving time units; Estimate the parameters in the multiple linear regression model using the least squares method: The following equations are used to find the minimum value of Q. In the multiple linear regression model, the estimator of the variance of the random error term ε is 2. The method for predicting electric vehicle charging load according to claim 1, characterized in that: In step S12, the travel characteristics are the starting point, destination, and stay time at the destination of the electric vehicle.

3. The method for predicting electric vehicle charging load according to claim 1, characterized in that: The basic model of electric vehicle travel includes travel time, the time when the i-th trip arrives at the destination and the start time of the (i+1)-th trip, the remaining power of the electric vehicle, and the charging probability model of the electric vehicle; The travel time is as follows: Among them, L ij represents the actual distance between destination i and destination j, and V represents the average driving speed of the electric vehicle; The time t at which the i-th trip reaches the destination is di and the starting time t of the i+1th trip o(i+1) as follows: t di =t oi +T, t o(i+1) =t oi +t pi +T, Among them, T is the driving time, t pi is the parking time of the i-th trip; The remaining battery power during charging needs to be calculated based on the last mileage. The calculation formula is as follows: Among them, L is the last mileage, SOC end SOC is the remaining battery power after the last mileage. before The remaining battery power before the last journey. max It is the maximum driving range of an electric vehicle when fully charged.

4. The method for predicting electric vehicle charging load according to claim 1, characterized in that: Step S2 specifically includes: Step S21, building a road network model based on graph theory and making assumptions about the peak traffic time period of urban roads; Step S22, filter out the data set in the urban road traffic peak time period from the NHTS2017 data set, and divide it into weekday morning peak, weekday evening peak, weekend morning peak, and weekend evening peak.

5. The method for predicting electric vehicle charging load according to claim 1, characterized in that: In step S31, the obtained data set needs to be further processed to delete data segments that are not useful for establishing a multivariate linear regression model.

Images