Electric vehicle energy consumption prediction method and device and nonvolatile storage medium

Abstract

The invention discloses an electric vehicle energy consumption prediction method and device and a nonvolatile storage medium. The method comprises the steps that electric vehicle energy consumption data time sequences and predictive factors corresponding to a plurality of sub-areas in a target area are acquired, and the predictive factors represent factors influencing electric vehicle energy consumption; on the basis of the prediction factor and the electric vehicle energy consumption data time series corresponding to the multiple sub-regions, predicting future energy consumption conditions corresponding to the multiple sub-regions; acquiring historical electric vehicle energy consumption data corresponding to the target area; and based on the historical electric vehicle energy consumption data and the future energy consumption conditions corresponding to the plurality of sub-regions, predicting the future energy consumption condition corresponding to the target region. The technical problem that the prediction efficiency and precision are low due to the fact that an existing electric vehicle energy consumption prediction method cannot effectively process periodicity and volatility of large-scale electric vehicles is solved.

Description

Technical Field

[0001] The present invention relates to the field of electric power technology, and in particular to a method and device for predicting energy consumption of an electric vehicle and a non-volatile storage medium. Background Art

[0002] The use pattern of electric vehicles is irregular, and it is difficult to capture fixed load patterns; weather and temperature changes in different seasons will significantly affect the use frequency and charging demand of electric vehicles: short-term weather changes, such as rainfall or extreme temperatures, may cause sudden fluctuations in electricity demand; at the same time, holidays, special events or social activities will change people's travel and car habits, which in turn affects charging demand; individual users' charging decisions are highly random and difficult to accurately predict, and charging behavior may fluctuate significantly in a short period of time, increasing the difficulty of prediction. These factors interact with each other to form a dynamic and unpredictable system. Traditional probabilistic prediction methods cannot effectively handle the periodicity and volatility of large-scale electric vehicles, resulting in low prediction efficiency and accuracy in predicting the energy consumption of electric vehicles, affecting subsequent research on optimization operations.

[0003] To address the above-mentioned problems, no effective solution has been proposed yet. Summary of the invention

[0004] The embodiments of the present invention provide an electric vehicle energy consumption prediction method, device and non-volatile storage medium to at least solve the technical problem that the current electric vehicle energy consumption prediction method cannot effectively handle the periodicity and volatility of large-scale electric vehicles, resulting in low prediction efficiency and accuracy.

[0005] According to one aspect of an embodiment of the present invention, a method for predicting energy consumption of an electric vehicle is provided, comprising: obtaining time series of energy consumption data of electric vehicles corresponding to a plurality of sub-regions in a target region and prediction factors, wherein the prediction factors represent factors affecting the energy consumption of electric vehicles; based on the prediction factors and the time series of energy consumption data of electric vehicles corresponding to the plurality of sub-regions, predicting the future energy consumption conditions corresponding to the plurality of sub-regions; obtaining historical energy consumption data of electric vehicles corresponding to the target region; and based on the historical energy consumption data of electric vehicles and the future energy consumption conditions corresponding to the plurality of sub-regions, predicting the future energy consumption conditions corresponding to the target region.

[0006] Optionally, obtaining a time series of energy consumption data of electric vehicles corresponding to each of the multiple sub-areas in the target area includes: obtaining the energy consumption data of electric vehicles corresponding to each of the multiple sub-areas; selecting energy consumption data within a preset time period from the energy consumption data corresponding to each of the multiple sub-areas and arranging them in chronological order to obtain a time series of energy consumption data corresponding to each of the multiple sub-areas.

[0007] Optionally, the prediction factor includes at least one of the following: a weather prediction factor, a calendar prediction factor, and a super user prediction factor, wherein the weather prediction factor represents the impact factor of weather on the electric vehicle energy consumption data in the target area, the calendar prediction factor represents the impact factor of different time periods on the electric vehicle energy consumption data in the target area, and the super user prediction factor represents the impact factor of electric vehicles whose energy consumption is higher than the average energy consumption of electric vehicles in the target area on the electric vehicle energy consumption data in the target area.

[0008] Optionally, based on the prediction factors and the time series of electric vehicle energy consumption data corresponding to each of the multiple sub-regions, the future energy consumption conditions corresponding to each of the multiple sub-regions are predicted, including: inputting the prediction factors into a preset target regression tree model, and outputting the future energy consumption conditions corresponding to each of the multiple sub-regions, wherein the target regression tree model is obtained by training an initial regression tree model using the time series of electric vehicle energy consumption data corresponding to each of the multiple sub-regions.

[0009] Optionally, before the prediction factor is input into a preset target regression tree model, the method includes: constructing an initial regression tree model and setting a loss function, wherein the loss function characterizes the dependency between the electric vehicle energy consumption data and the prediction factor; based on the initial regression tree model, using the electric vehicle energy consumption data time series corresponding to each of the multiple sub-regions for training, and calculating the negative gradient of the loss function; using the negative gradient as the target variable, iteratively updating the initial regression tree model to obtain the target regression tree model.

[0010] Optionally, based on historical electric vehicle energy consumption data and the future energy consumption conditions corresponding to each of the multiple sub-regions, the future energy consumption conditions corresponding to the target area are predicted, including: inputting the future energy consumption conditions corresponding to each of the multiple sub-regions, prediction factors and historical electric vehicle energy consumption data into a quantile linear regression model to obtain the future energy consumption conditions corresponding to the target area, wherein the quantile linear regression model is obtained by training an initial linear regression model using historical electric vehicle energy consumption data.

[0011] Optionally, the quantile linear regression model is obtained by training an initial linear regression model using historical electric vehicle energy consumption data, including: setting quantiles, wherein the quantiles represent the probability distribution of the predicted values; setting an initial linear regression model according to the quantiles; and training the initial linear regression model using historical electric vehicle energy consumption data to obtain a quantile linear regression model.

[0012] According to another aspect of an embodiment of the present invention, there is also provided an electric vehicle energy consumption prediction device, comprising: a first acquisition module, used to obtain a time series of electric vehicle energy consumption data and a prediction factor corresponding to each of a plurality of sub-regions in a target region, wherein the prediction factor represents a factor affecting the energy consumption of an electric vehicle; a first prediction module, used to predict the future energy consumption conditions corresponding to each of the plurality of sub-regions based on the prediction factor and the time series of electric vehicle energy consumption data corresponding to each of the plurality of sub-regions; a second acquisition module, used to obtain historical electric vehicle energy consumption data corresponding to the target region; and a second prediction module, used to predict the future energy consumption conditions corresponding to the target region based on the historical electric vehicle energy consumption data and the future energy consumption conditions corresponding to each of the plurality of sub-regions.

[0013] According to another aspect of an embodiment of the present invention, a non-volatile storage medium is provided, the non-volatile storage medium including a stored program, wherein when the program is running, the device where the non-volatile storage medium is located is controlled to execute any one of the above-mentioned electric vehicle energy consumption prediction methods.

[0014] According to another aspect of the embodiments of the present invention, a computer device is provided. The computer device includes a processor, and the processor is used to run a program. When the program is run, any one of the above-mentioned electric vehicle energy consumption prediction methods is executed.

[0015] According to another aspect of the embodiments of the present invention, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, any one of the above-mentioned electric vehicle energy consumption prediction methods is implemented.

[0016] In an embodiment of the present invention, an electric vehicle energy consumption prediction method is adopted, by obtaining electric vehicle energy consumption data time series and prediction factors corresponding to multiple sub-areas in a target area, wherein the prediction factors represent factors affecting the energy consumption of electric vehicles; based on the prediction factors and the electric vehicle energy consumption data time series corresponding to the multiple sub-areas, predicting the future energy consumption corresponding to the multiple sub-areas; obtaining historical electric vehicle energy consumption data corresponding to the target area; based on the historical electric vehicle energy consumption data and the future energy consumption corresponding to the multiple sub-areas, predicting the future energy consumption corresponding to the target area, the purpose of performing hierarchical prediction of electric vehicle energy consumption taking into account the periodicity and volatility of electric vehicles is achieved, thereby realizing the technical effect of improving the accuracy of energy consumption prediction, and further solving the technical problem that the current electric vehicle energy consumption prediction method cannot effectively deal with the periodicity and volatility of large-scale electric vehicles, resulting in low prediction efficiency and accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of this application. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0018] Figure 1 A hardware structure block diagram of a computer terminal for implementing an electric vehicle energy consumption prediction method is shown;

[0019] Figure 2 is a flow chart of an electric vehicle energy consumption prediction method provided according to an embodiment of the present invention;

[0020] Figure 3 is a schematic diagram of the architecture of an electric vehicle energy consumption prediction method provided according to an optional embodiment of the present invention;

[0021] Figure 4 It is a structural block diagram of an electric vehicle energy consumption prediction device provided according to an embodiment of the present invention. DETAILED DESCRIPTION

[0022] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings 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 should fall within the scope of protection of the present invention.

[0023] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0024] According to an embodiment of the present invention, a method embodiment of an electric vehicle energy consumption prediction method is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0025] The method embodiment provided in the first embodiment of the present application can be executed in a mobile terminal, a computer terminal or a similar computing device. Figure 1The hardware structure block diagram of a computer terminal for implementing an electric vehicle energy consumption prediction method is shown. Figure 1 As shown, the computer terminal 10 may include one or more (102a, 102b, ..., 102n are used to illustrate) processors (the processor may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA), a memory 104 for storing data. In addition, it may also include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of the BUS bus), a network interface, a power supply and / or a camera. It can be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the above electronic device. Figure 1 More or fewer components as shown, or with Figure 1 Different configurations are shown.

[0026] It should be noted that the one or more processors and / or other data processing circuits described above may generally be referred to herein as "data processing circuits". The data processing circuits may be embodied in whole or in part as software, hardware, firmware, or any other combination thereof. In addition, the data processing circuit may be a single independent processing module, or may be incorporated in whole or in part into any of the other components in the computer terminal 10. As described in the embodiments of the present application, the data processing circuit acts as a processor control (e.g., selection of a variable resistor terminal path connected to an interface).

[0027] The memory 104 can be used to store software programs and modules of application software, such as the program instructions / data storage device corresponding to the electric vehicle energy consumption prediction method in the embodiment of the present invention. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory 104, that is, the electric vehicle energy consumption prediction method of the above-mentioned application is realized. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include a memory remotely arranged relative to the processor, and these remote memories may be connected to the computer terminal 10 via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0028] The display may be, for example, a touch screen liquid crystal display (LCD) that enables a user to interact with a user interface of the computer terminal 10 .

[0029] Figure 2FIG. 1 is a flow chart of a method for predicting energy consumption of an electric vehicle according to an embodiment of the present invention. Figure 2 As shown, the method comprises the following steps:

[0030] Step S202, obtaining electric vehicle energy consumption data time series and prediction factors corresponding to each of a plurality of sub-areas in the target area, wherein the prediction factors represent factors affecting the energy consumption of the electric vehicle.

[0031] In this step, there will be multiple charging stations in the target area. Affected by geographical factors, the activity level in different charging stations, that is, the number of electric vehicles charged and the charging frequency will be different. Therefore, the charging stations that are more active in the target period can be screened out. Similarly, the uncertainty of a single charging station is large, so multiple charging stations can be spatially aggregated as a sub-area based on geographical factors, and multiple sub-areas in the target area can be obtained. That is, multiple sub-areas may not completely cover the target area. The electric vehicle energy consumption data corresponding to each sub-area can be determined based on the energy provided by the charging station in the sub-area to charge the electric vehicle. For example, the past seven days can be used as the target period, and each hour can be used as a sub-period. Then, the energy consumption of the sub-area in different sub-periods is obtained, and arranged in chronological order to form a time series of electric vehicle energy consumption data corresponding to different sub-areas.

[0032] Specifically, for time period selection, considering that the dataset covers a long time range, the number and management of charging stations have changed over time. In order to reduce the impact of these changes on data analysis, relatively stable time periods can be selected, and only charging stations that have been active throughout the selected time period can be considered. For geographic area selection, that is, the division of sub-regions, it is very challenging and difficult to generalize because the energy consumption of a single charging station is affected by the randomness of the driver's charging behavior and the battery status of the car. Therefore, the energy consumption data of multiple charging stations are spatially aggregated to provide more useful predictions. For energy consumption time series, a time series is a set of observations, each of which is defined at a specific time point, and the time intervals between all observations are equal. The energy consumption data for each hour can be determined by the meter readings of the last 15 minutes. If the 15-minute reading falls entirely within an hour, the entire reading is counted into the energy value of that hour. If the 15-minute reading spans two hours, the energy value is distributed to the two hours in proportion to the overlapping time.

[0033] The prediction factors obtained are factors that can affect the energy consumption of electric vehicles, such as time factors. The energy consumption of electric vehicles during the day is usually greater than that at night, and the energy consumption of electric vehicles during holidays and weekdays is also different. It can also include weather factors, which will also affect the use of electric vehicles and thus affect the energy consumption of electric vehicles.

[0034] Step S204: predicting the future energy consumption of each of the multiple sub-regions based on the prediction factors and the time series of the electric vehicle energy consumption data corresponding to each of the multiple sub-regions.

[0035] In this step, considering the existence of the prediction factors, the future energy consumption of each of the multiple sub-regions can be predicted based on the acquired electric vehicle energy consumption data time series and the influence of the prediction factors on the electric vehicle energy consumption. A prediction model can be used to predict the future energy consumption of the sub-region, for example, a long short-term memory network model, a support vector regression model, a deep neural network model, a random forest, etc., all of which can predict future energy consumption.

[0036] Step S206, obtaining historical electric vehicle energy consumption data corresponding to the target area.

[0037] In this step, the historical electric vehicle energy consumption data in the target area is obtained, that is, the energy provided by all charging stations in the target area for charging electric vehicles. The historical electric vehicle energy consumption data should cover the electric vehicle energy consumption data corresponding to multiple sub-areas. The prediction results of future energy consumption in multiple sub-areas can only reflect the data of more active charging stations. Using the historical electric vehicle energy consumption data corresponding to the target area can make the predicted data more stable.

[0038] Step S208: predicting the future energy consumption of the target area based on the historical electric vehicle energy consumption data and the future energy consumption of each of the multiple sub-areas.

[0039] In this step, a hierarchical probability prediction model can be used to predict the future energy consumption of the target area by combining the future energy consumption of electric vehicles in multiple sub-areas. A weighted average method can also be used, and the weighted average based on the prediction results of small areas can be used to predict the energy consumption of large areas. The weights can be the number of electric vehicles in the small area, the area of ​​the area, the historical energy consumption ratio, etc., so as to ensure that the prediction results better reflect the actual energy consumption distribution of the large area. A hybrid model can also be used to merge the prediction results of the small area with the results of other prediction models (such as time series analysis, deep learning models, etc.) in the large area. This hybrid model may include linear regression, neural networks, or support vector machines, etc., which can improve the accuracy and stability of the prediction by learning the relationship between the prediction of the small area and the actual energy consumption of the large area.

[0040] Through the above steps, the purpose of hierarchical prediction of electric vehicle energy consumption taking into account the periodicity and volatility of electric vehicles is achieved, thereby achieving the technical effect of improving the accuracy of energy consumption prediction, and further solving the technical problem that the current electric vehicle energy consumption prediction method cannot effectively deal with the periodicity and volatility of large-scale electric vehicles, resulting in low prediction efficiency and accuracy.

[0041] As an optional embodiment, obtaining time series of energy consumption data of electric vehicles corresponding to multiple sub-areas in the target area includes: obtaining energy consumption data of electric vehicles corresponding to multiple sub-areas; selecting energy consumption data within a preset time period from the energy consumption data corresponding to the multiple sub-areas and arranging them in chronological order to obtain time series of energy consumption data corresponding to the multiple sub-areas.

[0042] Optionally, the energy consumption data of electric vehicles corresponding to each of the multiple sub-areas in the historical period can be obtained, and the energy consumption data within the preset period can be curled down and arranged to obtain a time series of energy consumption data. Because a time series is a set of observations, each observation is defined at a specific time point, and the time intervals between all observations are equal. For example, the energy consumption data per hour can be determined by the meter readings of the last 15 minutes, that is, the preset period is the last 15 minutes. If the 15-minute reading falls completely within one hour, the entire reading is counted into the energy value of that hour. If the 15-minute reading spans two hours, the energy value is distributed to the two hours in proportion to the overlapping time.

[0043] Before determining the time series, the acquired energy consumption data can also be preprocessed, including data cleaning, time period selection, and geographic area selection. Among them, for data cleaning, in order to ensure the validity and consistency of the analyzed data, check whether there are missing values ​​(i.e., missing data) and outliers (such as abnormally high electric vehicle energy consumption values ​​or negative values, which do not conform to the actual situation) in the data, and the detected outliers can be deleted to avoid affecting the subsequent prediction results. For time period selection, considering that the data set covers a long time range, the number and management of charging stations have changed over time. In order to reduce the impact of these changes on data analysis, relatively stable time periods are selected, and only charging stations that have been active throughout the selected time period are considered. For geographic area selection, since the energy consumption of a single charging station is affected by the randomness of the driver's charging behavior and the battery status of the car, it is very challenging and difficult to generalize. Therefore, the energy consumption data of multiple charging stations are spatially aggregated to provide more useful predictions.

[0044] As an optional embodiment, the prediction factor includes at least one of the following: a weather prediction factor, a calendar prediction factor, and a super user prediction factor, wherein the weather prediction factor represents the impact factor of weather on the energy consumption data of electric vehicles in the target area, the calendar prediction factor represents the impact factor of different time periods on the energy consumption data of electric vehicles in the target area, and the super user prediction factor represents the impact factor of electric vehicles whose energy consumption is higher than the average energy consumption of electric vehicles in the target area on the energy consumption data of electric vehicles in the target area.

[0045] Optionally, the predictors may include weather predictors, calendar predictors, and super user predictors. For weather predictors, considering that charging stations are unevenly distributed and stations are more dense in cities, the impact of weather forecasts in different regions is uneven. However, considering that some weather predictors may have a smaller impact on electric vehicles, the proportion of weather predictors can be smaller in the subsequent main influencing factor analysis. For calendar predictors, calendar predictors are typical input variables in load forecasting. They reflect the social behavior of electric vehicle users, such as holidays, weekdays, or different time periods of the day, including weekdays / non-working days and hours of the day, where the observed value corresponds to a weekday, the value is 0; otherwise (non-working days) the value is 1; the number of hours is an integer variable, indicating the hour when the observed value occurs (0 to 23). For super user predictors, super users are a relatively small group of electric vehicle users whose energy consumption is higher than the average level. Considering that the charging pattern of super users is less random, it means that their charging behavior is more regular than that of general users, and their charging behavior is relatively stable, so they can be used as useful predictors to help predict future energy consumption.

[0046] Since the training process of the prediction model is very time-consuming and very sensitive to the number of predictors, the input dimension can be reduced. That is, by capturing the greatest variability in the data, the number of predictors can be reduced, thereby improving model performance and reducing calculation time. Therefore, the predictors with the greatest impact can be selected according to the actual situation.

[0047] As an optional embodiment, based on the prediction factors and the time series of electric vehicle energy consumption data corresponding to the multiple sub-regions, the future energy consumption conditions corresponding to the multiple sub-regions are predicted, including: inputting the prediction factors into a preset target regression tree model, and outputting the future energy consumption conditions corresponding to the multiple sub-regions, wherein the target regression tree model is obtained by training the initial regression tree model using the time series of electric vehicle energy consumption data corresponding to the multiple sub-regions.

[0048] Optionally, a gradient boosting regression tree can be used to predict the future energy consumption of each of the multiple sub-regions. The time series of electric vehicle energy consumption data corresponding to each of the multiple sub-regions and the prediction factor are input into the gradient boosting regression tree to obtain the future energy consumption. The target regression tree model is obtained by training the initial regression tree model using the time series of electric vehicle energy consumption data corresponding to each of the multiple sub-regions.

[0049] For example, the regression tree model can be trained by the following method:

[0050] Initialization: First, the prediction function is initialized to a constant value, which can be determined by minimizing the loss function. For example, for the square loss function, the initialization value can be the average of the energy consumption values ​​of all training samples.

[0051] Loss function selection: Choose an appropriate loss function to evaluate the prediction performance of the model. In electric vehicle energy consumption prediction, a common choice is Pinball Score (PS) or square loss function. Pinball Score is particularly important for generating probabilistic predictions because it can evaluate the performance of the model at different quantiles.

[0052] Iteration process: a. Calculate negative gradient: In each round of iteration, calculate the negative gradient of the loss function with respect to the current prediction. The negative gradient indicates the direction in which the loss function decreases, helping to update the model to better fit the data. For different loss functions, the calculation method of negative gradient will be different. b. Fit regression tree: Use negative gradient as the target variable to train a regression tree model. This regression tree will split the data into multiple leaf nodes according to the different values ​​of the prediction factor. Each leaf node will have a corresponding output value, indicating the expected energy consumption value under the action of the prediction factor in the area. c. Update the predicted value: Update the prediction function based on the output of the regression tree. Usually, the update of the predicted value is determined by the output of the regression tree and a reduction coefficient. The reduction coefficient is used to control the magnitude of each update to prevent overfitting. d. Evaluate model performance: After each round of iteration, evaluate whether the performance of the model has reached the expected threshold or whether the predetermined number of iterations has been reached. If the model performance meets the requirements, the iteration ends; otherwise, continue to train more regression trees until the termination condition is met.

[0053] Prediction: Once the model is trained, it can be used to predict new predictor data. At this point, the prediction function will give a final prediction value by accumulating the output of each tree and the effect of the reduction coefficient. Probabilistic prediction generation: Through multiple predictions, probabilistic predictions can be generated at different quantiles. For example, the 5th, 50th, and 95th percentile predictions can be generated to provide distribution information on electric vehicle energy consumption.

[0054] Specifically, the gradient boosted regression tree evaluates the response variable P by minimizing the specified loss function Ψ i (Energy consumption of electric vehicles in the i-th sub-region) and the corresponding prediction factor x i The unknown functional dependence between i (·):

[0055] P i =f i (x i )

[0056]

[0057] Then the dependent function is solved, including two cases: initial iteration (m=0) and subsequent iteration (m≥1). In the initial iteration (m=0), the function Initialize to a constant value This constant value can be determined by minimizing a loss function, where the loss function is related to the energy consumption values ​​of electric vehicles in the time series of the sub-region.

[0058] In subsequent iteration steps (m≥1), the training process is based on the gradient descent method. In each general mth step, the estimated value provided by the "regression tree model" is compared with the actual result of the target variable to calculate the negative gradient of the loss function. The negative gradient indicates the direction in which the model parameters need to be changed to reduce the error in the next iteration. The regression tree is fitted using non-repeating subsamples randomly selected from the original data, that is, the electric vehicle energy consumption data corresponding to each of the multiple sub-regions, and the negative gradient is used as the response variable to obtain the output result. The output result of the regression tree is added to the previous estimate to obtain an improved estimate. That is, in the iterative process of the gradient enhanced regression tree, the gradient descent step size Used to update the function f i (x i ) to obtain an improved estimate. In the mth iteration, the gradient descent step size can be determined by minimizing the loss function. In the mth iteration of the gradient enhanced regression tree model, the updated regression tree formula is as follows:

[0059]

[0060] Among them, κ represents the reduction coefficient, which controls the weight of each regression tree; is the output of the current regression tree. During the training process of the gradient boosted regression tree model, when the step size When the estimation algorithm ends, the prediction factor x for the given target period h is i,h , the predicted value of electric vehicle energy consumption in the i-th region Can be based on Sure.

[0061] As an optional embodiment, before the prediction factor is input into the preset target regression tree model, it includes: constructing an initial regression tree model and setting a loss function, wherein the loss function characterizes the dependency relationship between the electric vehicle energy consumption data and the prediction factor; based on the initial regression tree model, using the electric vehicle energy consumption data time series corresponding to each of the multiple sub-regions for training, and calculating the negative gradient of the loss function; using the negative gradient as the target variable, iteratively updating the initial regression tree model to obtain the target regression tree model.

[0062] Optionally, a loss function needs to be set in the training regression tree model, wherein the loss function represents the dependency between the electric vehicle energy consumption data and the prediction factor. Based on the negative gradient of the loss function, the initial regression tree model can be iteratively trained to obtain the final target regression tree model. Among them, a constant value can be initialized first, and this constant value is determined by minimizing the loss function Ψ.

[0063]

[0064] Among them, P i,n is the nth historical electric vehicle energy consumption value in the ith sub-region.

[0065] In subsequent iterations (m ≥ 1), the training process is based on the gradient descent method. At each generic m-th step, the estimated value provided by the "regression tree model" is compared with the actual result of the target variable to calculate the negative gradient of the loss function. The negative gradient indicates the direction in which the model parameters need to be changed to reduce the error in the next iteration. The negative gradient is relative to the historical observation value P i,n Compute the negative partial derivative of the loss function:

[0066]

[0067] Among them, x i,n is the prediction factor vector associated with the nth EV energy consumption observation in the ith sub-region.

[0068] Then, the regression tree is fitted using a non-repeated subsample randomly drawn from the original data, using the negative gradient as the response variable and the predictor as the input variable. The specific prediction value calculation formula is as follows:

[0069]

[0070] in, Represents a given predictor x i,n The predicted value of the nth negative gradient of ; S is the total number of leaves in the regression tree; represents the average value of negative gradients in the sth leaf of the fitted tree, It is an indicator function. If the predictor x i,n The rectangular subspace belonging to the s-th leaf Then the function takes the value of 1, otherwise it takes the value of 0. The output of the regression tree is added to the previous estimate to get the improved estimate. That is, in the iterative process of the gradient boosting regression tree, the gradient descent step size is Used to update the function f i (x i) to obtain an improved estimate. In the mth iteration, the gradient descent step size can be determined by minimizing the following loss function:

[0071]

[0072] in, is the estimated value of the previous iteration. Considering In by x i,n The terminal leaf determined is a constant, that is, That is, the average value of the negative gradient in the sth leaf. Therefore, the problem in formula (6) can be solved separately for each sth leaf subspace, and the simplified expression is as follows:

[0073]

[0074] The gradient boosted regression tree model can select Pinball Score (PS) as the loss function, so as to generate a probability prediction of the energy consumption of electric vehicles corresponding to multiple sub-regions, that is, a set of N prediction quantiles PS loss function is used to predict the actual energy consumption of electric vehicles P i,n and the predicted α n Quantile The calculation formula is as follows:

[0075]

[0076] in, is a binary function that indicates whether the actual load, i.e., the energy consumption of the charging station, is less than or equal to the predicted quantile. If yes, the function takes the value 1, otherwise it takes the value 0.

[0077] As an optional embodiment, based on historical electric vehicle energy consumption data and the future energy consumption conditions corresponding to multiple sub-areas, the future energy consumption conditions corresponding to the target area are predicted, including: inputting the future energy consumption conditions corresponding to the multiple sub-areas, prediction factors and historical electric vehicle energy consumption data into a quantile linear regression model to obtain the future energy consumption conditions corresponding to the target area, wherein the quantile linear regression model is obtained by training an initial linear regression model using historical electric vehicle energy consumption data.

[0078] Optionally, the future energy consumption corresponding to the target area can be predicted based on the quantile linear regression model. Large-scale electric vehicle energy consumption prediction based on hierarchical probability prediction: The prediction of the small area and the historical load data of the large area are combined through a hierarchical quantile linear regression model to generate a prediction for the large area.

[0079] Based on the hierarchical probability prediction, the large-scale electric vehicle energy consumption prediction can be achieved by combining the small-area prediction with the large-area historical data using a hierarchical quantile linear regression model. The process of achieving large-scale prediction is as follows: the historical electric vehicle energy consumption data of the target area can be collected first, and then the quantiles can be defined. A series of quantiles (for example, 0.1, 0.5, 0.9, etc.) can be selected, which will be used to describe the probability distribution of electric vehicle energy consumption in the target area. Each quantile will correspond to a regression model. Let's build a quantile linear regression model and train the quantile linear regression model using historical data. This usually involves minimizing the Pinball loss (or absolute quantile loss) between the predicted value and the actual value. The loss function is designed according to the definition of quantiles and can reflect the accuracy of the model prediction on the quantile. Then, the small-area prediction can be combined, that is, the future energy consumption corresponding to each of the multiple sub-areas. For large-scale prediction, the prediction factors may include but are not limited to the prediction results of the small area, weather prediction factors, super user prediction factors, calendar prediction factors, and any other factors that may affect the energy consumption of electric vehicles. These factors need to be combined with the historical load data of the large area to train the quantile regression model. The predicted values ​​for multiple sub-regions and the associated predictors can be input into the quantile linear regression model together with the historical EV energy consumption data for the target region. The model will learn how to fuse the predictions for small regions with the historical data for large regions to predict the EV energy consumption for the large region at a specific quantile. For each defined quantile, the trained quantile linear regression model is used to predict the EV energy consumption for a large area. This will produce a set of predicted values, one for each quantile. The quantile prediction results can provide a multi-faceted probabilistic prediction of EV energy consumption for a large region. For example, the 0.5 quantile prediction represents the median prediction, while the 0.1 and 0.9 quantile predictions can be used to determine the confidence interval of the prediction, helping to assess the uncertainty of the prediction.

[0080] Specifically, the target period h returned by the hierarchical probability prediction model is the predicted energy consumption of electric vehicles in the target area in the future period α n The analytical formula for the quantile prediction value can be:

[0081]

[0082] in, α is the energy consumption of electric vehicles in the target area n Quantile forecast value, i.e. future energy consumption; is the predictor vector for the target period h; is the model in α n Vector of regression coefficients at quantiles.

[0083] As an optional embodiment, the quantile linear regression model is obtained by training the initial linear regression model using historical electric vehicle energy consumption data, including: setting the quantile, wherein the quantile represents the probability distribution of the predicted value; setting the initial linear regression model according to the quantile; and training the initial linear regression model using historical electric vehicle energy consumption data to obtain the quantile linear regression model.

[0084] Optionally, quantiles can be set, where quantiles can be used to describe the probability distribution of electric vehicle energy consumption in the target area. For example, the 0.5 quantile prediction represents the median prediction, while the 0.1 and 0.9 quantile predictions can be used to determine the confidence interval of the prediction, helping to assess the uncertainty of the prediction.

[0085] Specifically, in the hierarchical probability prediction model of electric vehicle energy consumption in a large area, that is, the quantile linear regression model, for α n Coefficient vector of quantiles The estimation of can be done by solving the following minimization problem with L1 penalty. For the estimation of the coefficient vector of quantiles, by solving the minimization problem with L1 penalty, it is essentially performing a regularized regression analysis to determine the weights of each predictor in the quantile regression model. This regularization method, especially L1 regularization (also known as Lasso regression), helps to control the complexity of the model, avoid overfitting, and achieve feature selection by compressing the coefficients of some unimportant predictors to zero.

[0086] It should be noted that, for the above-mentioned method embodiments, for the sake of simplicity, they are all described as a series of action combinations, but those skilled in the art should know that the present invention is not limited by the described action sequence, because according to the present invention, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present invention.

[0087] Through the description of the above implementation methods, those skilled in the art can clearly understand that the electric vehicle energy consumption prediction method according to the above embodiment can be implemented by means of software plus a necessary general hardware platform, and of course it can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for a terminal device (which can be a mobile phone, computer, server, or network device, etc.) to execute the methods described in each embodiment of the present invention.

[0088] Figure 3 is a schematic diagram of an electric vehicle energy consumption prediction method according to an optional embodiment of the present invention. Figure 4 As shown, it includes a data preprocessing module, a small area gradient enhanced regression tree energy consumption prediction module and a large area hierarchical probability prediction module. Among them, the data preprocessing module is used for data preprocessing, specifically including the time series of energy consumption and the extraction of prediction related factors. The small area gradient enhanced regression tree energy consumption prediction module is used to predict the electric vehicle load in a small area based on the gradient enhanced regression tree, wherein the small area is a plurality of sub-areas, and the large area hierarchical probability prediction module is used to predict the energy consumption of large-scale electric vehicles based on hierarchical probability prediction, and the prediction of the small area and the historical load data of the large area are combined through the hierarchical quantile linear regression model to generate the prediction of the large area, wherein the large area is the target area, and the purpose of hierarchical prediction of electric vehicle energy consumption is achieved by taking into account the periodicity and volatility of electric vehicles, thereby achieving the technical effect of improving the accuracy of energy consumption prediction, and then solving the technical problem that the current prediction method of electric vehicle energy consumption cannot effectively deal with the periodicity and volatility of large-scale electric vehicles, resulting in low prediction efficiency and accuracy.

[0089] According to an embodiment of the present invention, there is also provided an electric vehicle energy consumption prediction device for implementing the above electric vehicle energy consumption prediction method. Figure 4 is a structural block diagram of an electric vehicle energy consumption prediction device provided according to an embodiment of the present invention. Figure 4 As shown, the electric vehicle energy consumption prediction device includes: a first acquisition module 42, a first prediction module 44, a second acquisition module 46 and a second prediction module 48. The electric vehicle energy consumption prediction device is described below.

[0090] The first acquisition module 42 is used to acquire the electric vehicle energy consumption data time series and prediction factors corresponding to each of the plurality of sub-areas in the target area, wherein the prediction factors represent factors affecting the energy consumption of the electric vehicle.

[0091] The first prediction module 44 is connected to the first acquisition module 42, and is used to predict the future energy consumption of each of the multiple sub-regions based on the prediction factor and the electric vehicle energy consumption data time series corresponding to each of the multiple sub-regions.

[0092] The second acquisition module 46 is connected to the first prediction module 44 and is used to acquire the historical electric vehicle energy consumption data corresponding to the target area.

[0093] The second prediction module 48 is connected to the second acquisition module 46 and is used to predict the future energy consumption of the target area based on the historical electric vehicle energy consumption data and the future energy consumption of each of the multiple sub-areas.

[0094] It should be noted that the first acquisition module 42, the first prediction module 44, the second acquisition module 46 and the second prediction module 48 correspond to steps S202 to S208 in the embodiment, and the examples and application scenarios implemented by the multiple modules and the corresponding steps are the same, but are not limited to the contents disclosed in the above embodiment. It should be noted that the above modules, as part of the device, can be run in the computer terminal 10 provided in the embodiment.

[0095] An embodiment of the present invention may provide a computer device. Optionally, in this embodiment, the computer device may be located in at least one network device among multiple network devices of a computer network. The computer device includes a memory and a processor.

[0096] Among them, the memory can be used to store software programs and modules, such as the program instructions / modules corresponding to the electric vehicle energy consumption prediction method device in the embodiment of the present invention. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory, that is, the above-mentioned electric vehicle energy consumption prediction method is realized. The memory may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory may further include a memory remotely arranged relative to the processor, and these remote memories may be connected to the computer terminal via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0097] The processor can call the information and application programs stored in the memory through the transmission device to execute the following steps: obtain the electric vehicle energy consumption data time series and prediction factors corresponding to each of the multiple sub-areas in the target area, wherein the prediction factors represent the factors affecting the energy consumption of the electric vehicle; based on the prediction factors and the electric vehicle energy consumption data time series corresponding to each of the multiple sub-areas, predict the future energy consumption conditions corresponding to each of the multiple sub-areas; obtain the historical electric vehicle energy consumption data corresponding to the target area; based on the historical electric vehicle energy consumption data and the future energy consumption conditions corresponding to each of the multiple sub-areas, predict the future energy consumption conditions corresponding to the target area.

[0098] Optionally, the processor may also execute program code for the following steps: obtaining a time series of energy consumption data of electric vehicles corresponding to each of the multiple sub-areas in the target area, including: obtaining energy consumption data of electric vehicles corresponding to each of the multiple sub-areas; selecting energy consumption data within a preset time period from the energy consumption data corresponding to each of the multiple sub-areas and arranging them in chronological order to obtain a time series of energy consumption data corresponding to each of the multiple sub-areas.

[0099] Optionally, the processor may also execute the program code of the following steps: the prediction factor includes at least one of the following: a weather prediction factor, a calendar prediction factor, and a super user prediction factor, wherein the weather prediction factor represents the influence factor of weather on the energy consumption data of electric vehicles in the target area, the calendar prediction factor represents the influence factor of different time periods on the energy consumption data of electric vehicles in the target area, and the super user prediction factor represents the influence factor of electric vehicles whose energy consumption is higher than the average energy consumption of electric vehicles in the target area on the energy consumption data of electric vehicles in the target area.

[0100] Optionally, the processor may also execute the program code of the following steps: based on the prediction factors and the time series of the electric vehicle energy consumption data corresponding to each of the multiple sub-regions, predicting the future energy consumption conditions corresponding to each of the multiple sub-regions, including: inputting the prediction factors into a preset target regression tree model, and outputting the future energy consumption conditions corresponding to each of the multiple sub-regions, wherein the target regression tree model is obtained by training the initial regression tree model using the time series of the electric vehicle energy consumption data corresponding to each of the multiple sub-regions.

[0101] Optionally, the processor may also execute the following program code: before inputting the prediction factor into the preset target regression tree model, the steps include: constructing an initial regression tree model and setting a loss function, wherein the loss function characterizes the dependency between the electric vehicle energy consumption data and the prediction factor; based on the initial regression tree model, using the electric vehicle energy consumption data time series corresponding to each of the multiple sub-regions for training, and calculating the negative gradient of the loss function; using the negative gradient as the target variable, iteratively updating the initial regression tree model to obtain the target regression tree model.

[0102] Optionally, the processor may also execute the following program code: based on historical electric vehicle energy consumption data and the future energy consumption conditions corresponding to each of the multiple sub-regions, predict the future energy consumption conditions corresponding to the target area, including: inputting the future energy consumption conditions corresponding to each of the multiple sub-regions, prediction factors and historical electric vehicle energy consumption data into a quantile linear regression model to obtain the future energy consumption conditions corresponding to the target area, wherein the quantile linear regression model is obtained by training an initial linear regression model using historical electric vehicle energy consumption data.

[0103] Optionally, the processor may also execute program code for the following steps: a quantile linear regression model is obtained by training an initial linear regression model using historical electric vehicle energy consumption data, including: setting quantiles, wherein the quantiles represent the probability distribution of predicted values; setting an initial linear regression model according to the quantiles; and training the initial linear regression model using historical electric vehicle energy consumption data to obtain a quantile linear regression model.

[0104] By adopting the embodiment of the present invention, a method for predicting energy consumption of electric vehicles is provided, which obtains time series of electric vehicle energy consumption data and prediction factors corresponding to multiple sub-regions in a target area, wherein the prediction factors represent factors affecting the energy consumption of electric vehicles; based on the prediction factors and the time series of electric vehicle energy consumption data corresponding to the multiple sub-regions, predicts the future energy consumption conditions corresponding to the multiple sub-regions; obtains historical electric vehicle energy consumption data corresponding to the target area; based on the historical electric vehicle energy consumption data and the future energy consumption conditions corresponding to the multiple sub-regions, predicts the future energy consumption conditions corresponding to the target area, thereby achieving the purpose of hierarchical prediction of electric vehicle energy consumption taking into account the periodicity and volatility of electric vehicles, thereby realizing the technical effect of improving the accuracy of energy consumption prediction, and further solving the technical problem that the current method for predicting the energy consumption of electric vehicles cannot effectively deal with the periodicity and volatility of large-scale electric vehicles, resulting in low prediction efficiency and accuracy.

[0105] A person of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing the hardware related to the terminal device through a program, and the program can be stored in a non-volatile storage medium, and the storage medium may include: a flash drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0106] The embodiment of the present invention further provides a non-volatile storage medium. Optionally, in this embodiment, the non-volatile storage medium can be used to store the program code executed by the electric vehicle energy consumption prediction method provided in the above embodiment.

[0107] Optionally, in this embodiment, the non-volatile storage medium may be located in any computer terminal in a computer terminal group in a computer network, or in any mobile terminal in a mobile terminal group.

[0108] Optionally, in this embodiment, the non-volatile storage medium is configured to store program codes for executing the following steps: obtaining time series of electric vehicle energy consumption data and prediction factors corresponding to each of the multiple sub-areas in the target area, wherein the prediction factors represent factors affecting the energy consumption of electric vehicles; predicting future energy consumption conditions corresponding to each of the multiple sub-areas based on the prediction factors and the time series of electric vehicle energy consumption data corresponding to each of the multiple sub-areas; obtaining historical electric vehicle energy consumption data corresponding to the target area; predicting future energy consumption conditions corresponding to the target area based on the historical electric vehicle energy consumption data and future energy consumption conditions corresponding to each of the multiple sub-areas.

[0109] Optionally, in this embodiment, the non-volatile storage medium is configured to store program codes for executing the following steps: obtaining a time series of electric vehicle energy consumption data corresponding to each of the multiple sub-areas in the target area, including: obtaining the electric vehicle energy consumption data corresponding to each of the multiple sub-areas; selecting energy consumption data within a preset time period from the energy consumption data corresponding to each of the multiple sub-areas and arranging them in chronological order to obtain a time series of energy consumption data corresponding to each of the multiple sub-areas.

[0110] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for executing the following steps: the prediction factor includes at least one of the following: a weather prediction factor, a calendar prediction factor, and a super user prediction factor, wherein the weather prediction factor represents the impact factor of weather on the energy consumption data of electric vehicles in the target area, the calendar prediction factor represents the impact factor of different time periods on the energy consumption data of electric vehicles in the target area, and the super user prediction factor represents the impact factor of electric vehicles with energy consumption higher than the average energy consumption of electric vehicles in the target area on the energy consumption data of electric vehicles in the target area.

[0111] Optionally, in this embodiment, the non-volatile storage medium is configured to store program codes for executing the following steps: based on the prediction factors and the time series of electric vehicle energy consumption data corresponding to each of the multiple sub-regions, predicting the future energy consumption conditions corresponding to each of the multiple sub-regions, including: inputting the prediction factors into a preset target regression tree model, and outputting the future energy consumption conditions corresponding to each of the multiple sub-regions, wherein the target regression tree model is obtained by training an initial regression tree model using the time series of electric vehicle energy consumption data corresponding to each of the multiple sub-regions.

[0112] Optionally, in this embodiment, the non-volatile storage medium is configured to store program codes for executing the following steps: before the prediction factor is input into a preset target regression tree model, it includes: constructing an initial regression tree model and setting a loss function, wherein the loss function characterizes the dependency relationship between the electric vehicle energy consumption data and the prediction factor; based on the initial regression tree model, using the electric vehicle energy consumption data time series corresponding to each of the multiple sub-regions for training, and calculating the negative gradient of the loss function; using the negative gradient as the target variable, iteratively updating the initial regression tree model to obtain the target regression tree model.

[0113] Optionally, in this embodiment, the non-volatile storage medium is configured to store program codes for executing the following steps: based on historical electric vehicle energy consumption data and future energy consumption conditions corresponding to multiple sub-areas, predicting future energy consumption conditions corresponding to the target area, including: inputting the future energy consumption conditions corresponding to each of the multiple sub-areas, prediction factors and historical electric vehicle energy consumption data into a quantile linear regression model to obtain future energy consumption conditions corresponding to the target area, wherein the quantile linear regression model is obtained by training an initial linear regression model using historical electric vehicle energy consumption data.

[0114] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for executing the following steps: the quantile linear regression model is obtained by training the initial linear regression model using historical electric vehicle energy consumption data, including: setting the quantile, wherein the quantile represents the probability distribution of the predicted value; setting the initial linear regression model according to the quantile; training the initial linear regression model using historical electric vehicle energy consumption data to obtain the quantile linear regression model.

[0115] An embodiment of the present invention also provides a computer program product, including a computer program. Optionally, in this embodiment, when the computer program is executed by a processor, it can achieve: obtaining electric vehicle energy consumption data time series and prediction factors corresponding to multiple sub-areas in the target area, wherein the prediction factors represent factors affecting the energy consumption of electric vehicles; based on the prediction factors and the electric vehicle energy consumption data time series corresponding to the multiple sub-areas, predicting the future energy consumption conditions corresponding to the multiple sub-areas; obtaining historical electric vehicle energy consumption data corresponding to the target area; based on the historical electric vehicle energy consumption data and the future energy consumption conditions corresponding to the multiple sub-areas, predicting the future energy consumption conditions corresponding to the target area.

[0116] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0117] In the above embodiments of the present invention, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0118] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the units can be a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0119] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0120] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0121] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a non-volatile storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, server or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, magnetic disk or optical disk and other media that can store program codes.

[0122] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for predicting energy consumption of electric vehicles, characterized in that: include: Obtaining electric vehicle energy consumption data time series and prediction factors corresponding to each of a plurality of sub-regions in a target region, wherein the prediction factors represent factors affecting the energy consumption of electric vehicles; Based on the prediction factor and the time series of electric vehicle energy consumption data corresponding to each of the multiple sub-areas, predict the future energy consumption corresponding to each of the multiple sub-areas; Acquire historical electric vehicle energy consumption data corresponding to the target area; Based on the historical electric vehicle energy consumption data and the future energy consumption conditions corresponding to each of the multiple sub-areas, the future energy consumption conditions corresponding to the target area are predicted.

2. The method according to claim 1, characterized in that The step of obtaining the electric vehicle energy consumption data time series corresponding to each of the multiple sub-areas in the target area includes: Acquire the electric vehicle energy consumption data corresponding to each of the plurality of sub-areas; Energy consumption data within a preset time period from the energy consumption data corresponding to each of the multiple sub-areas are selected and arranged in chronological order to obtain a time series of energy consumption data corresponding to each of the multiple sub-areas.

3. The method according to claim 1, characterized in that The prediction factor includes at least one of the following: a weather prediction factor, a calendar prediction factor, and a super user prediction factor, wherein the weather prediction factor represents the influence factor of weather on the electric vehicle energy consumption data in the target area, the calendar prediction factor represents the influence factor of different time periods on the electric vehicle energy consumption data in the target area, and the super user prediction factor represents the influence factor of electric vehicles with energy consumption higher than the average energy consumption of electric vehicles in the target area on the electric vehicle energy consumption data in the target area.

4. The method according to claim 1, characterized in that The predicting of the future energy consumption conditions corresponding to each of the plurality of sub-regions based on the prediction factors and the time series of the electric vehicle energy consumption data corresponding to each of the plurality of sub-regions includes: The prediction factor is input into a preset target regression tree model, and the future energy consumption corresponding to each of the multiple sub-areas is output, wherein the target regression tree model is obtained by training an initial regression tree model using the electric vehicle energy consumption data time series corresponding to each of the multiple sub-areas.

5. The method according to claim 4, characterized in that Before inputting the prediction factors into the preset target regression tree model, the method includes: Constructing the initial regression tree model and setting a loss function, wherein the loss function characterizes the dependency relationship between the electric vehicle energy consumption data and the prediction factor; Based on the initial regression tree model, the electric vehicle energy consumption data time series corresponding to each of the multiple sub-regions are used for training to calculate the negative gradient of the loss function; The negative gradient is used as the target variable to iteratively update the initial regression tree model to obtain the target regression tree model.

6. The method according to claim 1, characterized in that The predicting of the future energy consumption situation corresponding to the target area based on the historical electric vehicle energy consumption data and the future energy consumption situation corresponding to each of the multiple sub-areas includes: The future energy consumption conditions corresponding to each of the multiple sub-areas, the prediction factors and the historical electric vehicle energy consumption data are input into a quantile linear regression model to obtain the future energy consumption conditions corresponding to the target area, wherein the quantile linear regression model is obtained by training an initial linear regression model using the historical electric vehicle energy consumption data.

7. The method according to claim 6, characterized in that The quantile linear regression model is obtained by training an initial linear regression model using the historical electric vehicle energy consumption data, and includes: Setting a quantile, wherein the quantile represents a probability distribution of a predicted value; According to the quantiles, an initial linear regression model is set; The historical electric vehicle energy consumption data is used to train the initial linear regression model to obtain the quantile linear regression model.

8. An electric vehicle energy consumption prediction device, characterized in that: include: A first acquisition module is used to acquire electric vehicle energy consumption data time series and prediction factors corresponding to each of a plurality of sub-areas in a target area, wherein the prediction factors represent factors affecting the energy consumption of electric vehicles; A first prediction module is used to predict the future energy consumption of each of the multiple sub-areas based on the prediction factor and the electric vehicle energy consumption data time series corresponding to each of the multiple sub-areas; A second acquisition module is used to acquire the historical electric vehicle energy consumption data corresponding to the target area; The second prediction module is used to predict the future energy consumption situation corresponding to the target area based on the historical electric vehicle energy consumption data and the future energy consumption situation corresponding to each of the multiple sub-areas.

9. A non-volatile storage medium, characterized in that: The non-volatile storage medium includes a stored program, wherein when the program is running, the device where the non-volatile storage medium is located is controlled to execute the electric vehicle energy consumption prediction method according to any one of claims 1 to 7.

10. A computer device, characterized in that: include: Memory and processor, The memory stores a computer program; The processor is used to execute the computer program stored in the memory, and when the computer program is running, the processor executes the electric vehicle energy consumption prediction method according to any one of claims 1 to 7.

11. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the electric vehicle energy consumption prediction method described in any one of claims 1 to 7 is implemented.

Images