A method and device for predicting charging behavior of an electric vehicle user, and an electronic device

Abstract

The application belongs to the field of automobiles, and particularly relates to a charging behavior prediction method and device for electric vehicle users and electronic equipment; the method comprises the following steps: obtaining historical behavior data and travel environment data of electric vehicle users; calculating a battery state of charge energy consumption value through a physical model, optimizing the battery state of charge energy consumption value through a rain and snow weather influence coefficient, combining driving behavior data to determine a user driving style, updating the battery state of charge energy consumption value by using the driving style, so as to calculate a more reasonable battery state of charge energy consumption value; and finally performing binary classification processing on the battery state of charge end value, so that the charging behavior of the electric vehicle user can be predicted, and the layout planning of charging facilities can be reasonably selected.

Description

Technical Field

[0001] This invention belongs to the automotive field, specifically relating to a method, device, and electronic device for predicting the charging behavior of electric vehicle users. Background Technology

[0002] Today, countries around the world are facing increasingly scarce energy resources and severe environmental pollution. To address climate change and promote green energy development, and with the maturation of technology, the market share of electric vehicles is rapidly increasing. While electric vehicles offer significant advantages, the disorderly charging of large numbers of electric vehicles not only affects the safe and stable operation of the power grid but also impacts road traffic efficiency and ultimately harms the user experience. Therefore, it is necessary to predict the charging intentions of electric vehicle users. This allows for the design of reasonable charging scheduling strategies to improve user experience while preventing large-scale electric vehicle grid connection from causing grid shocks.

[0003] Currently, researchers divide the task of predicting electric vehicle users' charging intentions into three processes. First, they identify the factors that affect the energy consumption of electric vehicles. Second, they use appropriate analytical methods to process these factors and analyze the energy consumption of electric vehicles. Finally, based on the analysis results, they predict the charging intentions of electric vehicle users.

[0004] Currently, the main factors affecting the energy consumption of pure electric vehicles can be divided into three categories: vehicle structure, external environment during driving, and driving style. Studies have shown that optimizing the vehicle structure of electric vehicles (vehicle weight, frontal area, transmission system efficiency, etc.) can effectively reduce overall vehicle energy consumption. However, current analyses of the impact of vehicle structure on energy consumption are mainly applied to the vehicle research and development stage and are only suitable for predicting electric vehicles in the laboratory, not for predicting energy consumption and charging intentions of private car users. Existing research on the external environment mainly focuses on driving conditions, road conditions, and travel temperature. Current technologies require a large amount of real-vehicle data collection, equipping the vehicle with numerous data collection devices, which may affect the user experience. Furthermore, laboratory conditions are somewhat limited compared to real road conditions. Existing research on driving style is also based on a large amount of real-vehicle data. Driving style refers to the comprehensive behavioral characteristics exhibited by a driver while driving. Different drivers drive differently, leading to significant differences in vehicle energy consumption.

[0005] Existing technologies mainly acquire experimental data to analyze the energy consumption of pure electric vehicles from three aspects: first, acquiring experimental data in a test chamber under given preset operating conditions; second, establishing a physical model and simulating the energy consumption of the entire vehicle; and third, using a data acquisition system to acquire real-vehicle operating data. Each of these three methods has its own drawbacks. The first two methods differ significantly from the actual energy consumption of the vehicle during operation. The real-vehicle data acquired by the third method suffers from defects such as parameter asynchrony and noise interference. Furthermore, collecting too much data can reduce the user experience.

[0006] In predicting electric vehicle (EV) charging demand, the distribution patterns of charging demand vary significantly due to the distinct charging behavior characteristics of different user types. Existing research often focuses on predicting charging loads for a specific type, leading to somewhat limited accuracy. Secondly, in stochastic simulations of EV travel, current studies frequently neglect the impact of temperature and terrain on energy consumption, and many simulation scenarios fail to accurately reflect reality. Factors such as environmental conditions and road network conditions can cause discrepancies between simulation results and actual conditions. Finally, some studies using existing statistical patterns as the basis for calculating EV charging load fail to consider the differences in charging choices and habits among EV users, resulting in predictions that may not accurately reflect actual charging demand.

[0007] In summary, how to process the selected factors affecting the energy consumption of electric vehicles through reasonable analysis methods to more closely reflect the actual energy consumption of electric vehicles when driving on real roads, and thus obtain more accurate prediction results of electric vehicle charging demand, is a technical problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0008] Addressing the problems of existing technologies, the purpose of this invention is to provide a method, device, and electronic device for predicting the charging behavior of electric vehicle users. This method calculates the battery's state-of-charge (SOC) energy consumption value using a physical model, optimizes the SOC energy consumption value using the influence coefficient of rain and snow weather, and then combines driving behavior data to determine the user's driving style. The driving style is then used to update the battery SOC energy consumption value, resulting in a more reasonable SOC energy consumption calculation. Finally, by performing binary classification on the battery SOC end value, the charging behavior of electric vehicle users can be predicted, enabling the rational selection of charging facility layout planning. This invention employs a method combining big data technology and physical models. It acquires user driving data without interfering with the user's experience, and simultaneously considers environmental conditions and road slope to establish a physical model for energy consumption research. Furthermore, it incorporates driver characteristics, using the t-SNE algorithm combined with the K-Means clustering algorithm to make the classification of driving styles more accurate. The energy consumption model incorporating driver characteristics has better adaptability to different drivers, resulting in more accurate energy consumption prediction. In terms of predicting users' charging intentions, based on travel energy consumption and pre-trip SOC prediction values, a relatively accurate SOC value is obtained after the trip ends. At the same time, by learning from users' historical charging behavior and considering users' habits and personalities, the GBDT algorithm is used to make a more accurate prediction of whether users will charge.

[0009] In a first aspect, the present invention provides a method for predicting the charging behavior of electric vehicle users, the method comprising:

[0010] Acquire historical behavioral data and travel environment data of electric vehicle users; the historical behavioral data includes travel behavior data, driving behavior data, and parking and charging behavior data;

[0011] The travel behavior data is processed to predict the initial value of the battery state of charge;

[0012] The vehicle driving behavior data in the driving behavior data is segmented, and the energy consumption of the electric vehicle in the driving state in each segment is physically modeled to calculate the first energy consumption value of each segment.

[0013] The travel environment data is processed, and a physical model of the electric vehicle energy consumption is performed based on the temperature effect to calculate the second energy consumption value.

[0014] The first and second energy consumption values ​​of each segment are superimposed, and the energy consumption value of the battery state of charge is calculated by combining the influence coefficient of rain and snow weather.

[0015] The end value of the battery's state of charge at the end of the trip is calculated based on the difference between the initial value of the battery's state of charge and the energy consumption value of the battery's state of charge.

[0016] The battery state of charge end value is subjected to binary classification processing, and the charging behavior of electric vehicle users is predicted based on parking and charging behavior data.

[0017] In a second aspect, the present invention also provides a device for predicting the charging behavior of electric vehicle users, the device comprising:

[0018] The acquisition unit is used to acquire historical behavior data and travel environment data of electric vehicle users; the historical behavior data includes travel behavior data, driving behavior data, and parking and charging behavior data.

[0019] The first calculation unit is used to process the travel behavior data and predict the initial value of the battery state of charge.

[0020] The second calculation unit is used to segment the vehicle driving behavior data in the driving behavior data, physically model the energy consumption of the electric vehicle in each segment under driving conditions, and calculate the first energy consumption value of each segment.

[0021] The third calculation unit is used to process the travel environment data, perform physical modeling of the electric vehicle energy consumption based on the temperature effect, and calculate the second energy consumption value.

[0022] The fourth calculation unit is used to superimpose the first and second energy consumption values ​​of each segment, and calculate the battery state-of-charge energy consumption value by combining the rain and snow weather influence coefficient.

[0023] The fifth calculation unit is used to calculate the end value of the battery state of charge at the end of the trip based on the difference between the initial value of the battery state of charge and the energy consumption value of the battery state of charge.

[0024] The prediction unit performs binary classification on the battery state of charge end value and predicts the charging behavior of electric vehicle users based on parking and charging behavior data.

[0025] In a third aspect, the present invention also provides an electronic device, including a memory and a processor, wherein the memory stores a computer program executable on the processor, and the processor executes the computer program to implement the steps of the method described in the first aspect of the present invention.

[0026] The beneficial effects of this invention are:

[0027] This invention establishes a physical model of electric vehicles driving in different environments and terrains. It also considers the impact of rain and snow on energy consumption, as well as the additional impact of temperature on energy consumption. Furthermore, based on the driver's driving behavior during the driving process, it classifies the driver's driving style and establishes an energy consumption model for electric vehicles in motion. Compared with traditional models, this energy consumption model combines the advantages of big data and physical models, making the energy consumption prediction for users during driving more realistic and authentic.

[0028] The research on electric vehicle charging demand prediction in this invention revolves around the actual historical behavioral data of electric vehicle users. By statistically analyzing the behavioral patterns of existing users' historical data as the basis for simulating charging demand, the invention uses a combination of physical models and big data technology in energy consumption prediction. It takes into account the travel habits of different drivers, making the energy consumption prediction results more personalized. At the same time, it learns the charging habits of each individual based on their historical charging data, and finally obtains the prediction results of electric vehicle charging demand. The prediction results are more accurate than traditional charging demand prediction. Attached Figure Description

[0029] Figure 1 This is a flowchart of a method for predicting the charging behavior of electric vehicle users according to an embodiment of the present invention;

[0030] Figure 2 This is a physical model architecture diagram of the energy consumption of an electric vehicle in a driving state according to an embodiment of the present invention;

[0031] Figure 3 This is an architectural diagram of a charging behavior prediction device for electric vehicle users according to an embodiment of the present invention;

[0032] Figure 4 This is a schematic diagram of an electronic device structure according to an embodiment of the present invention. Detailed Implementation

[0033] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0034] Figure 1 This is a flowchart of a method for predicting the charging behavior of electric vehicle users according to an embodiment of the present invention, as follows: Figure 1 As shown, the method includes:

[0035] 101. Obtain historical behavioral data and travel environment data of electric vehicle users; the historical behavioral data includes travel behavior data, driving behavior data, and parking and charging behavior data;

[0036] In this embodiment of the invention, historical behavior data of vehicles can be collected at a certain frequency. For example, the Internet of Vehicles can collect historical travel behavior data and historical parking behavior data of vehicles every 10 seconds.

[0037] In this embodiment of the invention, the travel behavior data is the initial state of charge (SOC) of the battery for each travel cycle of the electric vehicle user; the driving behavior data includes vehicle driving behavior data and driver behavior data. The vehicle driving behavior data includes vehicle driving parameters, which include, but are not limited to, data such as gradient, speed, acceleration, vehicle weight, and vehicle frontal area during driving. The driver behavior data includes, but is not limited to, parameters that reflect the driver's control behaviors such as starting, accelerating, decelerating, and braking. For example, it may include three driver speed parameters, three driver gear shift parameters, three driver pedal operation parameters, and three driver time ratio parameters. These vehicle driving parameters can be obtained through various sensors mounted on the vehicle, such as the vehicle's horizontal gyroscope recording the road gradient at different times during driving; the parking and charging behavior data includes the SOC value of the electric vehicle user after returning home from a previous trip and whether charging is being performed during the current travel cycle.

[0038] In this embodiment of the invention, the travel environment data of the electric vehicle user includes travel environment information such as temperature, humidity, and weather during the user's historical travel cycle. The weather status during the user's travel cycle can be obtained from the network. The temperature and weather during the travel cycle can be obtained from the network. The weather is mainly divided into rainy / snowy weather and ordinary weather.

[0039] 102. Process the travel behavior data to predict the initial value of the battery state of charge;

[0040] In this embodiment of the invention, the SOC value of a user before their first trip each day can be statistically analyzed, and the SOC values ​​before each trip over the past three months can be statistically analyzed. A probability distribution chart can be generated to determine which interval the SOC value appears most frequently; that is, the user's initial SOC value is considered to follow a random distribution within that interval. These values ​​will be used as empirical values ​​to predict the initial SOC value for subsequent user trips.

[0041] Specifically, a user's initial SOC for a trip can be set to follow a random distribution of [u1, u2], where u1 and u2 are the intervals in which the user's historical initial SOC values ​​have a relatively high probability. Since each user has different habits, the distribution parameters will vary slightly. For example, if a user's historical initial SOC value has a relatively high probability of being between 80-100%, then the user's initial SOC for a trip is considered to follow a random distribution of [80, 100].

[0042] In some embodiments, this embodiment records the SOC value of electric vehicle users before their first trip each day and calculates the SOC value before each trip over the past three months. Since the SOC value before each trip is time series data and does not have obvious seasonal or trend characteristics, an exponential smoothing model is used to predict the initial value of future SOC.

[0043] The advantages of exponential smoothing are: (1) It does not require collecting a lot of historical travel behavior data, but takes into account the importance of travel behavior data in each period, and uses all historical travel data. It is an improvement and development of the moving average method and is widely used; (2) It has the advantages of simple calculation, small sample size requirement, strong adaptability, and stable results; (3) It can be used not only for short-term prediction, but also for medium- and long-term prediction. In this embodiment, the user's initial SOC value before travel in the past 3 months is used to predict this behavior, which belongs to medium-term prediction. Therefore, the exponential smoothing method can achieve a better prediction effect in this embodiment.

[0044] In some embodiments of the present invention, the initial values ​​of the user's historical travel SOC over the past t travel cycles are first obtained as y1, y2, ..., y t The formula for the first exponential smoothing method used is:

[0045]

[0046] Among them, s t The first exponential smoothing value is given by s, where 'a' is the weighting coefficient with a value of 0 < a < 1. t-1 This is the exponentially smoothed value of the previous travel cycle, which is y. t With s t-1 The weighted average is used to predict the initial SOC value (SOC0) of a user in the (t+1)th travel cycle.

[0047] In a preferred embodiment of the present invention, this embodiment combines the optimal constant regression method and the exponential smoothing method to predict the initial value of the battery state of charge. The prediction result of the exponential smoothing method is used to make reference corrections to the optimal constant regression method. The initial value of the battery state of charge of the most recent historical travel cycle is used to find the regression prediction value of the minimum loss function to fit the initial value of the battery state of charge that is closest to the actual situation; thus realizing a reliable assessment of the initial value of the battery state of charge for electric vehicle users.

[0048] Specifically, in a preferred embodiment of the present invention, the initial state of charge (SOC) of the battery for each travel cycle of an electric vehicle user is statistically analyzed, and the historical SOC values ​​of the M travel cycles closest to the travel cycle to be predicted are selected as the rule validation set. The initial SOC value of each day in the M travel cycles is used as the initial regression value for the corresponding travel cycle. The initial regression value of the M travel cycles is then used to predict the intermediate regression result for the travel cycle to be predicted. By minimizing the loss function between the initial regression value and the intermediate regression result, the allocation coefficient corresponding to the travel cycle is determined. Under this allocation coefficient, the regression value under the minimized loss function is determined by changing the initial regression value. The product of the allocation coefficient for each travel cycle and the regression value under the minimized loss function for the corresponding travel cycle is used as the initial SOC value for the travel cycle to be predicted.

[0049] 103. The vehicle driving behavior data in the driving behavior data is segmented, and the energy consumption of the electric vehicle in the driving state in each segment is physically modeled to calculate the first energy consumption value of each segment.

[0050] In this embodiment of the invention, weather information, slope data, and mileage data obtained from vehicle driving behavior data can be used to perform force analysis and physical modeling of a moving vehicle. By summing up the mileage of each segment, the total mileage of the user in the current travel cycle can be obtained. The user's total energy consumption in the current travel cycle is directly proportional to the total mileage. To more accurately estimate the user's energy consumption in the current travel cycle, and to differentiate between slope, weather, and other factors under different driving conditions, this embodiment of the invention calculates the energy consumption for each segment of the current travel cycle, and then sums up the energy consumption of each segment to obtain the total energy consumption of the electric vehicle user in the current travel cycle, providing a basis for predicting the user's charging needs in the future.

[0051] For each segment of the electric vehicle in motion, a physical model is performed. Taking an uphill slope as an example, the state and force analysis of the electric vehicle during motion are as follows: Figure 2 As shown. Figure 2 For an electric vehicle in motion, the main resistance factors include rolling resistance, uphill resistance, air resistance, and acceleration resistance. The traction force of a pure electric vehicle overcomes these resistances, propelling the vehicle forward. The force analysis during the driving process of a pure electric vehicle is shown in the figure. When traveling at a constant speed on a level road, there is no slope resistance or acceleration resistance. Uphill, slope resistance is positive; downhill, slope resistance is negative.

[0052] In this embodiment of the invention, based on the acquired vehicle driving parameters, the energy consumption of the electric vehicle during driving can take into account the effects of gradient, speed, acceleration, and resistance. The maximum longitudinal slope of roads in mountainous cities is greater than that in general cities, and those skilled in the art can determine the corresponding maximum longitudinal slope ratio based on the actual situation.

[0053]

[0054]

[0055] In the formula, F t For the driving wheel traction, F f For rolling resistance, F i For slope resistance, F j To represent acceleration resistance, M is the curb weight of the pure electric vehicle, g is the acceleration due to gravity, v is the vehicle speed; δ is the rotational mass conversion factor, f is the tire rolling resistance coefficient, θ is the road gradient, and F... w For air resistance, C D Let A be the drag coefficient, A be the frontal area, and ρ be the air density; a represent the acceleration of the electric vehicle. The work done by the electric vehicle against drag during travel is:

[0056] E m =ηF t S

[0057]

[0058] Where S is the driving distance and η is the conversion efficiency of the electric vehicle's operating system (battery, transmission system, motor, etc.).

[0059] The driving state of an electric vehicle is divided into multiple short-duration kinematic segments. Motion parameters for each kinematic skewness are extracted, including average vehicle speed, average acceleration, and average gradient. Based on these parameters, an energy consumption prediction model is constructed. The driving state is divided into segments of equal length, with a segmentation period of 150 seconds. The first energy consumption value for each driving segment can then be expressed as:

[0060]

[0061] Where v, in a travel segment where Δt is sufficiently small, can be represented by the average speed of that travel segment.

[0062] 104. Process the travel environment data, perform physical modeling of electric vehicle energy consumption based on temperature effect, and calculate the second energy consumption value;

[0063] In this embodiment of the invention, the impact of temperature on the energy consumption of electric vehicles is mainly reflected in:

[0064] It affects battery heat dissipation. As temperature changes, the battery's internal resistance also changes accordingly. Increased internal resistance leads to increased heat dissipation from the power battery; this affects the energy consumption of the air conditioning and heating systems. This energy consumption is mainly affected by the ambient temperature during travel; high-temperature cooling conditions and low-temperature heating conditions will significantly increase the proportion of energy consumption in the air conditioning system.

[0065] Therefore, the energy consumption of accessories and the energy consumption of battery heat dissipation are uniformly identified as the effect of temperature on the energy consumption of the whole vehicle. According to the fitted relationship between travel temperature and the energy consumption of the whole vehicle, it is found that as the temperature increases, the energy consumption of the whole vehicle first decreases and then increases, which is more in line with a cubic function. Therefore, the effect of temperature on energy consumption is a cubic function of temperature, which conforms to the relationship of cubic regression.

[0066]

[0067] In the formula, E t This represents the second energy consumption value, which is the energy consumed by the temperature effect, T. i The temperature is the travel temperature, β0 is a constant term, and λ0, λ1, and λ2 are the first-order, second-order, and third-order parameters of the effect of temperature on the vehicle's energy consumption, respectively.

[0068] 105. The first and second energy consumption values ​​of each segment are superimposed, and the energy consumption value of the battery state of charge is calculated by combining the influence coefficient of rain and snow weather.

[0069] In this embodiment of the invention, the first energy consumption value of each segment after superimposing the rain and snow weather impact coefficient can be superimposed with the second energy consumption value of the travel cycle to calculate the median value E of the battery state of charge energy consumption value. mid =(E d +E t Because rain and snow increase the friction between electric vehicles and the ground, the coefficient of friction increases, the climbing resistance increases, and the energy consumption of electric vehicles also increases accordingly; therefore, in this embodiment of the invention, the influence coefficient f of rain and snow is considered. rain The median value of energy consumption E for battery state of charge mid After processing, the optimized battery state-of-charge energy consumption value E = f is obtained. rain E mid .

[0070] This embodiment classifies weather into rainy / snowy weather (winter) and ordinary weather. Of course, in other embodiments, weather can also be classified into various different types, such as high temperature weather and low temperature weather. Using different types of weather classification can accurately depict the impact of different environmental factors on the energy consumption of electric vehicles, thereby improving the accuracy of energy consumption simulation results.

[0071] Taking the rainy / snowy weather and normal weather conditions given in this embodiment as examples, the total energy consumption of the car is E, expressed as follows:

[0072] E = f rain (E d +E t )

[0073]

[0074]

[0075] Among them, f rain E represents the impact coefficient of rain and snow weather on vehicle energy consumption. d E is the energy consumed by electric vehicles while driving. t E represents the electricity consumed due to temperature. E represents the total energy consumption, which includes factors such as temperature, weather, slope, and vehicle driving conditions.

[0076] In a preferred embodiment of the present invention, driving style is further incorporated into the established power consumption model, making the model more targeted. The driver's control over starting, acceleration, deceleration, and braking during the driving process reflects the intensity of their driving and allows for the determination of their driving style. The driving energy consumption model incorporates driver characteristics and mountainous / urban characteristics, considering the day's travel environment, weather, and road gradient. A comprehensive classification of driver behavior characteristics is used to determine energy consumption during driving. Table 1 provides some driving behavior characteristic parameters from embodiments of the present invention.

[0077] Table 1 Driving Behavior Characteristic Parameters

[0078]

[0079] It is understood that those skilled in the art may select some or all of the above-mentioned driving behavior characteristic parameters, or may select other common driving behavior characteristic parameters in the art according to the actual situation, as long as these driving behavior characteristic parameters can reflect the behaviors that affect driving style during the driving process.

[0080] In this embodiment of the invention, the classification of driving behavior is relatively objective; however, these data variables have a large dimensionality, and the high-dimensional feature variables increase the difficulty of driving style classification, making it difficult for ordinary clustering methods to meet the requirements. Therefore, this invention adopts a deep learning algorithm, namely the t-SNE algorithm, which is more suitable for dimensionality reduction of high-dimensional data, and combines it with the K-Means clustering algorithm to classify driving styles.

[0081] First, t-SNE is used to reduce the dimensionality of driving behavior feature parameters, and then the dimensionality-reduced data is clustered to obtain the classification results of driving style.

[0082] t-SNE works by first calculating the similarity probability of points in the high-dimensional space, and then calculating the similarity probability of points in the corresponding low-dimensional space. The similarity of points is calculated as conditional probabilities; if a point A chooses its neighbors proportionally according to its probability density within a Gaussian (normal) distribution centered at A, then point A will choose point B as its neighbor. It then attempts to minimize the difference between these conditional probabilities (or similarities) in the high-dimensional and low-dimensional spaces to perfectly represent the data points in the low-dimensional space.

[0083] The K-Means algorithm, also known as the k-means algorithm, uses the k in K-means to represent the number of clusters and the mean to represent the centroid of each cluster, which is the mean of the data values ​​in each cluster.

[0084] The algorithm is roughly as follows: First, randomly select k samples from the sample set as cluster centers, and calculate the distance between all samples and these k cluster centers. For each sample, assign it to the cluster containing the nearest cluster center. For the new cluster, calculate the new cluster center for each cluster.

[0085] The four main points for implementing the k-means algorithm:

[0086] (1) The choice of the number of clusters k.

[0087] (2) The distance from each sample point to the "cluster center" can be Euclidean distance or Manhattan distance.

[0088] (3) Update the "cluster center" based on the newly divided clusters.

[0089] (4) Repeat steps 2 and 3 above until the "cluster center" no longer moves.

[0090] In some embodiments of the present invention, the K-Means algorithm is used to cluster the dimensionality-reduced driving behavior features, which includes selecting k driving behavior features as initial cluster centers; assigning all driving behavior features to clusters represented by the nearest cluster centers according to the nearest proximity principle; calculating the mean of all driving behavior features in each cluster as a new cluster center and calculating the cluster radius; finding the nearest neighbor clusters of each cluster based on the distance relationship between the cluster radius and the cluster centers; calculating the distance between each driving behavior feature and the cluster center of its nearest neighbor cluster, and assigning it to the nearest cluster according to the proximity principle; until the cluster centers no longer change, the clustering results are output as normal driving style, cautious driving style, and aggressive driving style.

[0091] In some embodiments of the present invention, selecting k driving behavior features as initial cluster centers from the dimensionality-reduced driving behavior features may include: randomly selecting a driving behavior feature from the dimensionality-reduced driving behavior feature set as the first initial cluster center; using the Markov Model Kallo method to extract a Markov chain of length 3k from the driving behavior feature set, and using the 3k data points on the Markov chain as candidate initial cluster centers; for these 3k candidate initial cluster centers in the driving behavior feature set, using the Prim minimum spanning tree method, repeatedly merging the two nearest initial cluster centers into a new initial cluster center, until only k data points remain as initial cluster centers.

[0092] In other embodiments, selecting k driving behavior features as initial cluster centers from the dimensionality-reduced driving behavior features may further include: randomly selecting a driving behavior feature from the set of driving behavior features using reservoir sampling as the first initial cluster center; wherein the first k driving behavior features in the set of driving behavior features are all placed into the reservoir, and for the m-th element thereafter, a certain driving behavior feature set in the reservoir is replaced with a probability of k / m, and the final selected k driving behavior feature sets are used as the initial cluster centers.

[0093] Different driving styles can be weighted differently on the battery state-of-charge (SOC) energy consumption value before the update in the energy consumption update. The clustering result shows three driving styles: normal, cautious, and aggressive. Studies have shown that drivers with different driving styles have different energy consumption per 100 kilometers. The driving style coefficient represents the impact of different driving styles on energy consumption. i :i = 1, 2, 3, where 1 represents a normal driving style, 2 represents a cautious driving style, and 3 represents an aggressive driving style.

[0094] Therefore, the power consumption model considering driver behavior is as follows:

[0095]

[0096] Using SOC E The energy consumption for all travel segments during the trip is shown in the following formula:

[0097]

[0098] Where Num is the number of all motion segments, T is the period of each segment (e.g., 150s), and SOC is the number of segments. E It is used to represent the electricity consumption of an electric vehicle for all its trips in a day.

[0099] 106. Calculate the battery state of charge ending value at the end of the trip based on the difference between the initial value of the battery state of charge and the energy consumption value of the battery state of charge.

[0100] By combining the predicted SOC at the start of the trip (SOC0) and the predicted energy consumption during the trip with a joint probabilistic prediction, the user's SOC upon returning home after several trips can be obtained. end .

[0101] SOC end =SOC0-SOC E

[0102] 107. Perform binary classification on the battery state of charge end value, and predict the charging behavior of electric vehicle users based on parking and charging behavior data.

[0103] In this embodiment of the invention, it is necessary to record the SOC value of electric vehicle users after returning home from their historical trips and whether they charged the battery. Let x represent the SOC value of the user after returning home from their historical trips on that day. end Let y represent whether the user charged the device on that day, where y = 0 indicates no charging and y = 1 indicates charging.

[0104] The GBDT (Gradient Boosting Decision Tree) classification algorithm is used to predict a user's willingness to charge, with the result being either charge or not charge.

[0105] The prediction function for logistic regression is:

[0106]

[0107] function h θ The value (x) has a special meaning; it represents the probability that the result is 1. Therefore, the probabilities of classifying input x as category 1 and category 0 are respectively:

[0108] P(Y=1|x;θ)=h θ (x)

[0109] P(Y=0|x;θ)=1-h θ (x)

[0110] The process of the GBDT binary classification algorithm is as follows:

[0111] (1) Initialize the first weak learner:

[0112]

[0113] Where P(Y=1|x) is the proportion of y=1 in the training samples, and the learner is initialized using prior information.

[0114] (2) Construct M classification regression trees, m = 1, 2, ..., M

[0115] a) For i = 1, 2, ..., N, calculate the response value of the m-th tree (the negative gradient of the loss function, i.e., the pseudo residual):

[0116]

[0117] b) For i = 1, 2, ..., N, fit the data (x) using a CART regression tree. i ,r m,i This yields the m-th regression tree, whose corresponding leaf node region is R. m,j Where j = 1, 2, ..., J m And J m Let be the number of leaf nodes of the m-th regression tree.

[0118] c) For J m Leaf node regions j = 1, 2, ..., J m The best-fit value was calculated:

[0119]

[0120] d) Update the strong learner F m (x)

[0121]

[0122] (3) Obtain the final strong learner F M The expression for (x)

[0123]

[0124] Therefore, the classification model can be expressed as the probability of whether a user will charge their phone after completing the current cycle of their trip:

[0125] The probability that a user will charge their phone after completing the current cycle of their trip is:

[0126] The probability that a user will not charge after the current cycle ends is: P(Y=0|x)=1-P(Y=1|x).

[0127] like Figure 3 As shown in the figure, this application provides an electric vehicle user charging behavior prediction device 500, such as... Figure 3 As shown, it includes:

[0128] The acquisition unit is used to acquire historical behavior data and travel environment data of electric vehicle users; the historical behavior data includes travel behavior data, driving behavior data, and parking and charging behavior data.

[0129] The first calculation unit is used to process the travel behavior data and predict the initial value of the battery state of charge.

[0130] The second calculation unit is used to segment the vehicle driving behavior data in the driving behavior data, physically model the energy consumption of the electric vehicle in each segment under driving conditions, and calculate the first energy consumption value of each segment.

[0131] The third calculation unit is used to process the travel environment data, perform physical modeling of the electric vehicle energy consumption based on the temperature effect, and calculate the second energy consumption value.

[0132] The fourth calculation unit is used to superimpose the first and second energy consumption values ​​of each segment, and calculate the battery state-of-charge energy consumption value by combining the rain and snow weather influence coefficient.

[0133] The fifth calculation unit is used to calculate the end value of the battery state of charge at the end of the trip based on the difference between the initial value of the battery state of charge and the energy consumption value of the battery state of charge.

[0134] The prediction unit performs binary classification on the battery state of charge end value and predicts the charging behavior of electric vehicle users based on parking and charging behavior data.

[0135] like Figure 4 As shown in the embodiment of this application, an electronic device 600 includes a processor 601, a memory 602, and a bus 603. The memory 602 stores machine-readable instructions executable by the processor 601. When the electronic device is running, the processor 601 communicates with the memory 602 via the bus, and the processor 601 executes the machine-readable instructions to perform the steps of the electric vehicle charging behavior prediction method described above.

[0136] Specifically, the memory 602 and processor 601 mentioned above can be general-purpose memory and processor, without any specific limitations. When the processor 601 runs the computer program stored in the memory 602, it can execute the above-mentioned method for predicting the charging behavior of electric vehicle users.

[0137] Processor 601 may be an integrated circuit chip with signal processing capabilities. In implementation, each step of the above method can be completed by the integrated logic circuitry in the hardware of processor 601 or by instructions in software form. Processor 601 can be a general-purpose processor, including a Central Processing Unit (CPU), a Network Processor (NP), etc.; it can also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field-Programmable Gate Array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this application can be directly manifested as execution by a hardware decoding processor, or execution by a combination of hardware and software modules in the decoding processor. The software module can reside in a mature storage medium in the art, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, or registers. This storage medium is located in memory 602, and processor 601 reads the information from memory 602 and, in conjunction with its hardware, completes the steps of the above method.

[0138] Corresponding to the above-described method for predicting the charging behavior of electric vehicle users, this application also provides a computer-readable storage medium storing machine-executable instructions. When these machine-executable instructions are invoked and executed by a processor, they cause the processor to perform the steps of the above-described method for predicting the charging behavior of electric vehicle users.

[0139] The electric vehicle user charging behavior prediction device provided in this application embodiment can be specific hardware on the device or software or firmware installed on the device. The implementation principle and technical effects of the device provided in this application embodiment are the same as those in the foregoing method embodiments. For the sake of brevity, any parts not mentioned in the device embodiment can be referred to the corresponding content in the foregoing method embodiments. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can all be referred to the corresponding processes in the above method embodiments, and will not be repeated here.

[0140] In the embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the displayed or discussed mutual couplings, direct couplings, or communication connections may be through some communication interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms.

[0141] For example, the flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of apparatus, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram and / or flowchart, and combinations of blocks in block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0142] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0143] In addition, the functional units in the embodiments provided in this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0144] If the aforementioned function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause an electronic device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the electric vehicle charging behavior prediction method described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0145] Understandably, this invention employs a method combining physical modeling with real-world user travel history data for energy consumption prediction. This approach complements the shortcomings of both methods, resulting in a more accurate prediction of electric vehicle energy consumption on actual roads, without significantly interfering with users. Furthermore, the research on electric vehicle charging demand prediction in this invention revolves around real-world historical behavioral data (including travel behavior data, driving behavior data, and parking behavior data). By statistically analyzing the behavioral patterns of existing users' historical data as the basis for charging demand simulation, a physical model is used for energy consumption prediction, supplemented by environmental factors and driver characteristics. This considers the travel habits of different drivers, making the energy consumption prediction results more personalized. Simultaneously, by learning each individual's charging habits based on their historical charging data, the invention ultimately obtains a more accurate prediction of electric vehicle charging demand compared to traditional charging demand prediction methods.

[0146] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. In addition, the terms "first", "second", "third", etc. are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0147] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for predicting the charging behavior of electric vehicle users, characterized in that, The method includes: Acquire historical behavioral data and travel environment data of electric vehicle users; the historical behavioral data includes travel behavior data, driving behavior data, and parking and charging behavior data; The travel behavior data is processed to predict the initial value of the battery state of charge; The process of processing the travel behavior data to predict the initial battery state of charge (SBC) value includes: statistically analyzing the initial SBC value of electric vehicle users for each travel cycle; selecting the historical initial SBC values ​​of the M travel cycles closest to the travel cycle to be predicted as the rule validation set; using the initial SBC value of each day in the M travel cycles as the initial regression value for the corresponding travel cycle; employing exponential smoothing to predict the initial regression values ​​of the M travel cycles to obtain intermediate regression results for the travel cycle to be predicted; determining the allocation coefficient corresponding to the travel cycle by minimizing the loss function between the initial regression value and the intermediate regression results; and determining the regression value under the minimized loss function by changing the initial regression value under this allocation coefficient. The product of the allocation coefficient for each travel cycle and the regression value under the minimized loss function for the corresponding travel cycle is used as the initial SBC value for the travel cycle to be predicted. The vehicle driving behavior data in the driving behavior data is segmented, and the energy consumption of the electric vehicle in the driving state in each segment is physically modeled to calculate the first energy consumption value of each segment. The energy consumption model of an electric vehicle in motion is expressed as follows: In the formula, This indicates the first energy consumption value. M is the conversion efficiency of the electric vehicle's working system; g is the acceleration due to gravity; v is the speed of the electric vehicle. In sufficiently small segments of travel, the average speed of the segment is used to represent the speed. This is the rotational mass conversion factor. This is the tire rolling resistance coefficient. For road slope, Where A is the drag coefficient and A is the frontal area. air density; Indicates the acceleration of an electric vehicle; The travel environment data is processed, and a physical model of the electric vehicle energy consumption is performed based on the temperature effect to calculate the second energy consumption value. The first and second energy consumption values ​​of each segment are superimposed, and the energy consumption value of the battery state of charge is calculated by combining the influence coefficient of rain and snow weather. The battery state of charge energy consumption value also includes determining the driving style of the electric vehicle user based on the driver behavior data in the driving behavior data, and updating the battery state of charge energy consumption value of the corresponding electric vehicle user as an influencing factor using the driving style of the electric vehicle user. The driving style includes extracting driving behavior features from driving behavior data, using a nonlinear dimensionality reduction algorithm t-SNE to reduce the dimensionality of the driving behavior features, and using the K-Means algorithm to cluster the dimensionality-reduced driving behavior features into normal driving style, cautious driving style and aggressive driving style, and assigning different numerical values ​​to the three driving styles. The end value of the battery's state of charge at the end of the trip is calculated based on the difference between the initial value of the battery's state of charge and the energy consumption value of the battery's state of charge. The battery state of charge end value is subjected to binary classification processing, and the charging behavior of electric vehicle users is predicted based on parking and charging behavior data.

2. The method for predicting the charging behavior of electric vehicle users according to claim 1, characterized in that, The K-Means algorithm is used to cluster the dimensionality-reduced driving behavior features, which includes selecting k driving behavior features as initial cluster centers; and assigning all driving behavior features to the cluster spheres represented by the nearest cluster center according to the nearest proximity principle. Calculate the mean of all driving behavior features in each cluster as the new cluster center, and calculate the cluster radius; find the nearest neighbor clusters of each cluster based on the distance relationship between the cluster radius and the cluster center; calculate the distance between each driving behavior feature and the cluster center of its nearest neighbor cluster, and assign it to the nearest cluster according to the proximity principle; continue until the cluster center no longer changes, and output the clustering results, which are the normal driving style, the cautious driving style, and the aggressive driving style.

3. The method for predicting the charging behavior of electric vehicle users according to claim 1, characterized in that, The process of performing binary classification on the battery state of charge end value to predict the charging behavior of electric vehicle users includes using a gradient boosting decision tree to perform binary classification on the battery state of charge end value to predict the charging probability of electric vehicle users after the trip ends.

4. A device for predicting the charging behavior of electric vehicle users, characterized in that, The method for predicting the charging behavior of electric vehicle users as described in any one of claims 1-3 includes: The acquisition unit is used to acquire historical behavior data and travel environment data of electric vehicle users; the historical behavior data includes travel behavior data, driving behavior data, and parking and charging behavior data. The first calculation unit is used to process the travel behavior data and predict the initial value of the battery state of charge. The second calculation unit is used to segment the vehicle driving behavior data in the driving behavior data, physically model the energy consumption of the electric vehicle in each segment under driving conditions, and calculate the first energy consumption value of each segment. The third calculation unit is used to process the travel environment data, perform physical modeling of the electric vehicle energy consumption based on the temperature effect, and calculate the second energy consumption value. The fourth calculation unit is used to superimpose the first and second energy consumption values ​​of each segment, and calculate the battery state-of-charge energy consumption value by combining the rain and snow weather influence coefficient. The fifth calculation unit is used to calculate the end value of the battery state of charge at the end of the trip based on the difference between the initial value of the battery state of charge and the energy consumption value of the battery state of charge. The prediction unit performs binary classification on the battery state of charge end value and predicts the charging behavior of electric vehicle users based on parking and charging behavior data.

5. An electronic device, characterized in that, The method includes a memory and a processor, wherein the memory stores a computer program that can run on the processor, and the processor executes the computer program to implement the steps of the method described in any one of claims 1 to 3.

Images