A charging selection probability prediction method and system for highway ev users
By using a single-class support vector machine model and a density partitioning interval method, the limitations of data sources and model adaptability issues related to the charging selection behavior of electric vehicles on highways were addressed. This enabled accurate prediction of the charging selection probability of EV users, providing data support for intelligent services for electric vehicles.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- SHANDONG UNIV
- Filing Date
- 2025-12-23
- Publication Date
- 2026-06-02
Smart Images
Figure CN121365997B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of charging behavior prediction technology, and in particular to a method and system for predicting the probability of charging choices for EV users on highways. Background Technology
[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.
[0003] In recent years, the new energy vehicle industry has experienced explosive growth. As a vital transportation network connecting cities and supporting long-distance travel, the layout of highway charging infrastructure and the accurate identification of user charging choices directly affect the convenience and user experience of long-distance electric vehicle travel, thus influencing the further promotion of new energy vehicles. However, in the field of research on electric vehicle (EV) user charging choices, existing research still has some shortcomings, making it difficult to meet the needs of accurate and efficient probabilistic modeling of charging choices in highway scenarios.
[0004] Firstly, the data sources have certain limitations. Currently, many studies rely on online questionnaires, simulating charging or not charging scenarios to allow participants to choose and obtain state information about their charging decisions. However, this method has inherent drawbacks: firstly, data collection is difficult, and the obtained state information largely depends on respondents' charging recollections, generally collecting interval-type option data for various charging state parameters, rather than real-time and specific numerical state data of users; secondly, data collection is highly subjective, with participants' answers easily influenced by the survey context and personal subjective cognitive biases, making it difficult to accurately reflect actual charging decision-making behavior. On the other hand, some studies utilize electric vehicle trajectory datasets. This type of data is unsupervised learning samples, typically requiring manual labeling using clustering methods such as K-means and hierarchical clustering. These studies focus on data similarity and incorporate researchers' subjective judgments during labeling, making it difficult for models to accurately characterize the impact of individual differences on charging choices.
[0005] Secondly, the model selection suffers from insufficient adaptability. In the fields of behavior prediction and behavior probability prediction, a series of derivative models derived from Logit memory have been widely used. However, the dependent variable in Logit models is typically binary, only able to output a "charge or not charge" decision. Although Logistic models can output charging probabilities, when there is a strong correlation between independent variables (such as the distance to charging stations and driving time), it can lead to unstable parameter estimation. Furthermore, these models are generally used in combination with clustering methods such as K-means, further limiting their effectiveness in complex scenarios. Summary of the Invention
[0006] To address the shortcomings of existing technologies, the purpose of this invention is to provide a method and system for predicting the probability of charging choices for EV users on highways. This method and system can accurately predict the charging choice behavior of EV users based on their historical data and real-time status, providing strong data support for the intelligent services of electric vehicles.
[0007] To achieve the above objectives, the present invention is implemented through the following technical solution:
[0008] The first aspect of this invention provides a method for predicting the probability of charging choices for EV users on highways, comprising the following steps:
[0009] Acquire the individual attribute data and charging status data to be predicted, and perform preprocessing operations;
[0010] A charging behavior prediction model is constructed using a single-class support vector machine, and the model is trained using a multi-dimensional training sample set. Specifically, an objective function is designed based on the classification of user charging behavior, and the decision value representing the characteristics of user charging behavior is obtained by solving the objective function. The decision value is derived based on the density partitioning interval method to obtain the user's charging choice probability.
[0011] The trained charging behavior prediction model is used to predict the probability of charging choices based on personal attribute data and charging status data.
[0012] Furthermore, personal attribute data includes gender and travel purpose, while charging status data includes remaining range, battery level when entering the highway, battery level when starting to charge, and driving range.
[0013] Further, the preprocessing operation involves cleaning the personal attribute data and charging status data, removing null and outlier data, and performing one-hot encoding on the personal attribute data.
[0014] Furthermore, the specific steps for designing an objective function based on the classification of user charging behavior and solving the objective function to obtain the decision values characterizing the user charging behavior features are as follows:
[0015] Based on the classification of user charging behavior, a hyperplane-based objective function is constructed. The inside of the hyperplane represents normal samples, i.e., EV users charging, while the outside represents abnormal samples, i.e., EV users not charging.
[0016] The objective function is optimized by introducing dual transformation and kernel function;
[0017] The decision value is obtained by solving the optimized objective function.
[0018] Furthermore, the specific steps for deriving the decision value based on the density-based interval method are as follows:
[0019] Determine the range of decision values based on the decision value boundaries;
[0020] Based on the density binning principle, intervals are divided according to the range of decision values;
[0021] Assign a corresponding charging user sample probability to each interval.
[0022] Furthermore, based on the density binning principle, the specific steps for dividing the interval according to the decision value range are as follows:
[0023] Sort the decision values from smallest to largest;
[0024] The sorted decision values are divided into intervals with equal sample sizes based on the preset number of intervals.
[0025] A second aspect of the present invention provides a charging selection probability prediction system for highway EV users, comprising:
[0026] The data acquisition module is configured to acquire the personal attribute data and charging status data to be predicted, and perform preprocessing operations.
[0027] The model training module is configured to build a charging behavior prediction model using a single-class support vector machine and train the charging behavior prediction model using a multi-dimensional training sample set. Specifically, an objective function is designed based on the classification of user charging behavior, and the objective function is solved to obtain the decision value that characterizes the user's charging behavior features. The decision value is derived based on the density partitioning interval method to obtain the user's charging choice probability.
[0028] The probability prediction module is configured to use a trained charging behavior prediction model to predict the probability of charging choices based on personal attribute data and charging status data.
[0029] A third aspect of the present invention provides a computer-readable storage medium storing a computer program adapted to be loaded by a processor and to execute steps in the charging selection probability prediction method for highway EV users as described in the first aspect of the present invention.
[0030] A fourth aspect of the present invention provides a computer device comprising:
[0031] A processor, adapted to execute computer programs;
[0032] A computer-readable storage medium storing a computer program, which, when executed by the processor, implements the charging selection probability prediction method for highway EV users as described in the first aspect of the present invention.
[0033] A fifth aspect of the present invention provides a computer program product or computer program comprising computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the steps in the charging selection probability prediction method for highway EV users as described in the first aspect of the present invention.
[0034] The above one or more technical solutions have the following beneficial effects:
[0035] This invention discloses a method and system for predicting the charging choice probability of EV users on highways. It obtains a dataset of EV user charging behavior selection indicators by combining on-site surveys and records from highway service areas with real-time charging status data displayed by intelligent charging piles in the service areas. This dataset includes driver personal attributes and charging parameters. Furthermore, taking advantage of the characteristics of highway service area sample collection, it innovatively employs a one-class support vector machine (SVM) model to predict the charging choices of highway EV users. This model can effectively capture the core features of user behavior in charging scenarios, providing technical support for accurate modeling of the charging choice probability of highway EV users.
[0036] This invention constructs the user charging choice problem as a hyperplane and introduces dual problem transformation and kernel function to simplify the objective function, thereby reducing computational complexity. Furthermore, it introduces a density binning method based on decision values to transform the user charging decision values obtained from the One-class SVM model into charging probabilities, thereby achieving accurate prediction of the user charging choice probability.
[0037] Advantages of additional aspects of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description
[0038] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0039] Figure 1 This is a flowchart of the charging selection probability prediction method for highway EV users in Embodiment 1 of the present invention;
[0040] Figure 2This is a frequency distribution histogram of the charging state attribute data in Embodiment 1 of the present invention;
[0041] Figure 3 This is a schematic diagram of the decision scores of the One-class SVM model in Embodiment 1 of the present invention;
[0042] Figure 4 This is a schematic diagram of the probability distribution of the One-class SVM model in Embodiment 1 of the present invention;
[0043] Figure 5 This is a QQ graph of the decision scores in Embodiment 1 of the present invention. Detailed Implementation
[0044] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.
[0045] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.
[0046] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.
[0047] Example 1:
[0048] Embodiment 1 of the present invention provides a method for predicting the probability of charging selection for EV users on highways, such as... Figure 1 As shown, it includes the following steps:
[0049] S1: Obtain the personal attribute data and charging status data to be predicted, and perform preprocessing operations.
[0050] In one specific implementation, personal attribute data includes gender and travel purpose, while charging status data includes remaining mileage, battery level when entering the highway, battery level at the start of charging, and remaining driving range. Preprocessing involves cleaning the personal attribute data and charging status data, removing null and outlier values, and performing one-hot encoding on the personal attribute data to map categorical values to integer values. Each integer value is then represented as a binary vector.
[0051] S2: Construct a charging behavior prediction model using a single-class support vector machine, and train the charging behavior prediction model using a multi-dimensional training sample set.
[0052] S2.1: Obtain a multi-dimensional training sample set.
[0053] In one specific implementation, this embodiment conducted on-site surveys in multiple service areas to obtain a training sample set. EV users' personal attributes and charging status data were recorded in detail through oral questioning, as shown in Table 1. The on-site surveys revealed that most service areas are equipped with intelligent charging equipment that can display the initial charge level, real-time charging power, and final charge level. Therefore, to ensure the reliability of the model input data, this embodiment combines the data collected from the questionnaires with the real-time data acquired by the charging equipment, selecting six variables as inputs to the charging choice behavior model: gender, travel purpose, remaining mileage, charge level when entering the highway, initial charge level, and driving range. After data preprocessing, removing null values, outliers, and other erroneous data, a total of 446 sample data points were obtained.
[0054] Table 1. Sample Questionnaire
[0055]
[0056] It should be noted that both field survey data and real-time acquired data have been obtained with the user's consent and are used solely for the research purposes of this embodiment. Furthermore, to ensure the quantity and quality of the dataset, this embodiment will expand the training sample set using existing public datasets or data simulations.
[0057] As shown in Table 2, descriptive statistics were performed on the collected sample data. Indicators such as mean and standard deviation were used to present the central tendency, dispersion, and distribution characteristics of the data, intuitively revealing hidden underlying patterns (such as user charging habits and vehicle range distribution). Figure 2 As shown in Table 2. (Analysis of Table 2 and...) Figure 2It can be seen that when entering the highway, the average battery level of users is 90.98% and the median is 100%, indicating that most people will fully charge their batteries to ensure highway travel. Only a few users have batteries as low as 25% due to emergencies or inconvenience in charging. Moreover, the data distribution is relatively concentrated without extreme anomalies. Users generally charge when the remaining battery level is around 35%. At the same time, the range of 5%-90% and the relatively flat data distribution show that there are differences in charging habits among different users.
[0058] Table 2. Descriptive statistics of charging status attribute data
[0059]
[0060] S2.2: Construct a charging behavior prediction model using a single-class support vector machine.
[0061] In one specific implementation, considering the characteristics of sample collection from highway service areas, this embodiment found that the proportion of vehicles entering service areas without charging is significantly smaller than that of vehicles charging. Furthermore, most vehicles not charging enter service areas for rest, eating, or other activities, with very few stopping to answer questions. This results in the collected samples being single-class data, meaning they only contain real data from user charging scenarios. Traditional behavioral probability prediction models are limited by data structures and struggle to solve the prediction problem for single-class samples. Therefore, this embodiment innovatively employs a one-class support vector machine (One-class SVM) model to predict the charging choices of EV users on highways.
[0062] Single-class SVM is a variant of Support Vector Machine (SVM). Its core objective is anomaly detection or novelty detection in unsupervised scenarios. That is, it trains the model using only normal samples of a single class, learns the feature distribution boundary of normal samples, and then determines whether a new sample belongs to the normal class (if it exceeds the boundary, it is an anomaly).
[0063] S2.3: Train the charging behavior prediction model using a multi-dimensional training sample set.
[0064] S2.3.1: Design an objective function based on the classification of user charging behavior, and solve the objective function to obtain decision values that characterize the features of user charging behavior.
[0065] S2.3.1.1: Construct a hyperplane-based objective function based on user charging behavior classification. The inside of the hyperplane represents normal samples, i.e., charging EV users, while the outside represents abnormal samples, i.e., non-charging EV users.
[0066] In one specific implementation, a training sample set of EV user charging behavior on highways is provided. for:
[0067] .
[0068] in, In this embodiment, the number of EV user samples is... , For user sample indexing, For the first One user sample, For the set of real numbers, In this embodiment, the feature dimension is used as the feature dimension. The core idea of One-Class SVM in this embodiment can be evolved into learning a hyperplane that "encloses normal samples (charging user behavior)" in a six-dimensional feature space containing gender, travel purpose, remaining mileage, battery level when entering the highway, battery level when starting charging, and driving range. The area inside this hyperplane represents normal samples (charging EV users), while the area outside represents abnormal samples (non-charging EV users). To achieve this goal, the original problem is defined as follows:
[0069] (1).
[0070] in, The normal vector of the separating hyperplane in the high-dimensional highway EV user charging behavior feature space; This represents the distance from the hyperplane to the origin, thus determining the radius of the normal region; it also sets up a fault-tolerance mechanism. As slack variables, For the first The slack variables for each user sample are set in this embodiment by setting a fault tolerance mechanism to allow a small number of charging user samples to fall outside the hyperplane. This is a regularization parameter, representing the upper bound of the proportion of abnormal samples identified in the training set of charging EV users. For kernel function mapping, the original low-dimensional samples are... Mapping to a high-dimensional feature space solves the nonlinear separability problem.
[0071] S2.3.1.2: The objective function is optimized by introducing dual transformation and kernel function.
[0072] In one specific implementation, considering the extremely high computational complexity of the high-dimensional feature space of highway EV user charging behavior, One-Class SVM employs dual transformation and kernel functions to simplify the computation. Based on the Lagrange multiplier method, by constructing a Lagrange function for the primal problem and finding its extrema, the optimal solution can be proven. satisfy:
[0073] (2).
[0074] in, For dual variables, satisfying .
[0075] Substituting formula (2) into the original constraints of formula (1), all calculations can be performed using the kernel function. Implementation. The commonly used kernel function is the radial basis function (RBF):
[0076] (3).
[0077] in, The kernel parameter controls the similarity range after sample mapping. and These are sample users With feature center point Kernel function mapping.
[0078] S2.3.1.3: Solve the optimized objective function to obtain the decision value.
[0079] In one specific implementation, One-Class SVM determines whether an EV user sample is a charging EV user sample based on the decision value. Substituting formula (2) into the hyperplane equation, a certain EV user sample... The decision value formula is:
[0080] (4).
[0081] like : A sample of EV users If it falls inside the hyperplane, it is identified as a charging EV user sample;
[0082] like : A sample of EV users Those falling outside the hyperplane are identified as EV user samples that are not charging.
[0083] S2.3.2: The decision value is derived based on the density partitioning interval method to obtain the user's charging selection probability.
[0084] In one specific implementation, since One-Class SVM only outputs decision values, it needs to be mapped to a probability transformation. The probability of an interval reflects the confidence that a sample is a charging user sample. Density binning is a non-parametric method based on the density of decision values to divide intervals. Its core principle is that "the probability distribution must simulate the density distribution of the decision values in the training data." The derivation process is as follows:
[0085] Since One-Class SVMs have no real labels, density binning is based on two key assumptions:
[0086] 1) The decision value is positively correlated with the charging probability of EV users: Decision value The larger the value, the higher the probability that the sample is a normal sample, because The corresponding EV user sample is charged, and the larger the value, the further away from the abnormal (not charging) boundary;
[0087] 2) Probability Continuity Matching Density: The density of the decision value region can reflect the concentration of the sample distribution. In dense decision value regions, the probability change should be more gradual; in sparse decision value regions, the probability change can be steeper to avoid the excessive influence of outliers on the probability.
[0088] S2.3.2.1: Determine the range of decision values based on the decision value boundaries.
[0089] In one specific implementation, the decision values for all samples of EV charging users on highways are first calculated. Determine the minimum value of the decision. Maximum value And the decision boundary (no charging / charging threshold). .
[0090] S2.3.2.2: Divide the interval according to the decision value range based on the density binning principle.
[0091] In one specific implementation, this embodiment sorts the decision values from smallest to largest. The sorted decision values are then divided into intervals of equal sample size based on a preset number of intervals. Unlike the equidistant binning rule that divides intervals at fixed intervals, density binning divides intervals based on the density distribution of the decision values. Specifically, the decision values of highway EV charging users are sorted from smallest to largest: Based on the preset number of intervals The sorted decision values are divided into Each interval contains equal sample sizes, meaning each interval contains... or A sample; for example: if , Each interval contains 100 samples; the goal is to make the interval width smaller in dense decision value regions and larger in sparse regions, matching the density distribution. Determine the boundaries of each interval: The left boundary of each interval is The right boundary is ,in, The number of samples for each interval.
[0092] S2.3.2.3: Assign a corresponding charging user sample probability to each interval.
[0093] In one specific implementation, assigning a corresponding charging user sample probability to each interval must satisfy two constraints:
[0094] 1. Probability monotonicity: The larger the decision value of the interval, the higher the probability of the allocation (consistent with hypothesis 1);
[0095] 2. Boundary calibration: Decision boundary The probability of the interval in which it falls needs to be calibrated to 0.5 (because...). It is the threshold between normal and abnormal, with a confidence level of 50%.
[0096] The specific probability allocation formula is derived as follows:
[0097] 1. Define the decision boundary for EV users. Falling in Within each interval, the probability of that interval. ;
[0098] 2. Regarding The abnormal side intervals are assigned probabilities in a linearly decreasing order, and the abnormal side intervals total [number missing]. indivual:
[0099] (6).
[0100] Among them, the minimum interval The probability of it being close to 0 reflects a high degree of anomaly confidence.
[0101] 3. Regarding The normal lateral intervals are assigned probabilities in a linearly increasing manner, and the normal lateral intervals total [number missing]. indivual:
[0102] (7).
[0103] Among them, the maximum interval The probability is close to 1, reflecting a high degree of normal confidence.
[0104] S3: Use the trained charging behavior prediction model to predict the probability of charging selection based on personal attribute data and charging status data.
[0105] In one specific implementation, for any test EV user sample Calculate its decision value back:
[0106] judge Which interval does it fall into?
[0107] Directly use the probability of this interval as The probability of the sample of charging users being evaluated.
[0108] To better illustrate the superiority of the method in this embodiment, the following experiments were conducted for verification:
[0109] A One-Class SVM model was built using Python software. The model output is the probability that each EV user will choose to charge at a service area. The model's independent variables can be divided into two categories:
[0110] Numerical characteristics, namely charging status data: battery level when on the highway, battery level when charging, driving range, and remaining range.
[0111] Category features, i.e., personal attribute data: gender, travel purpose.
[0112] Categorical features such as gender and travel purpose cannot be directly input into the model and require one-hot encoding. This first requires mapping categorical values to integer values. Then, each integer value is represented as a binary vector.
[0113] The most important parameter in the training process of a One-Class SVM model is the regularization parameter mentioned in formula (1). This represents the upper bound of the proportion of samples in the training set that are judged as "atypical charging samples," that is, individual data points in the charging user dataset that do not exhibit typical charging behavior. This embodiment incorporates the results of field research. When set to 0.05, users meeting the criteria of "high battery level (80–100%) but extremely low remaining range (0–100 km)" can be classified as "atypical charging samples". The model learns the charging characteristics of these atypical samples to distinguish between charging and non-charging users.
[0114] This model uses a density-based binning method to convert the decision scores of highway EV users into probabilities, taking into account the sample size and the number of intervals. Set it to 6, and map each box to a uniform probability of [0.001, 0.999] (avoiding extreme values of 0 / 1).
[0115] This embodiment utilizes decision scores, probability distributions, and quantile-quantile plots (QQ plots) to analyze the model training results, such as... Figure 3 , Figure 4 and Figure 5 As shown. By Figure 3 and Figure 4 It can be seen that the decision score distribution of the samples is significantly right-skewed, with most samples concentrated in the 0.2–1.2 range, indicating that the model considers the vast majority of EV user samples to be "normal charging" behavior; there is a significant peak near 0 (the highest bar), indicating that many EV user samples are "just on the boundary", which may be normal charging samples on the edge or slightly abnormal. Figure 5The X-axis represents the theoretical quantiles of the standard normal distribution, and the Y-axis represents the actual quantiles of the sample decision scores. The red diagonal line indicates that, ideally, the sample data should fall entirely on this line (i.e., perfectly conform to a normal distribution), and the blue dots represent the quantile correspondences of the actual samples. Figure 5 As can be seen, most points are close to the red diagonal, indicating that the decision scores of the training set are generally close to a normal distribution. On the left side (the negative value region), the points deviate significantly from the red line, indicating the presence of charging anomalies or boundary samples that can be distinguished from normal samples. Figure 5 On the far right, the data points are slightly above the red line, indicating that the high-scoring samples are more "extreme" than expected by a normal distribution. This aligns with the characteristic of "a few abnormal samples" in anomaly detection models. Therefore, the One-Class SVM model established in this embodiment successfully identified a small number of abnormal samples (low decision scores), and the "predicted probability" obtained through quantile mapping is highly positively correlated with the original decision scores, demonstrating that the model output has business interpretability.
[0116] Example 2:
[0117] Embodiment 2 of the present invention provides a charging selection probability prediction system for EV users on highways, comprising:
[0118] The data acquisition module is configured to acquire the personal attribute data and charging status data to be predicted, and perform preprocessing operations.
[0119] The model training module is configured to build a charging behavior prediction model using a single-class support vector machine and train the charging behavior prediction model using a multi-dimensional training sample set. Specifically, an objective function is designed based on the classification of user charging behavior, and the objective function is solved to obtain the decision value that characterizes the user's charging behavior features. The decision value is derived based on the density partitioning interval method to obtain the user's charging choice probability.
[0120] The probability prediction module is configured to use a trained charging behavior prediction model to predict the probability of charging choices based on personal attribute data and charging status data.
[0121] Example 3:
[0122] Embodiment 3 of the present invention provides a computer-readable storage medium storing a computer program adapted for loading by a processor and executing the steps in the charging selection probability prediction method for highway EV users as described in Embodiment 1 of the present invention.
[0123] Example 4:
[0124] Embodiment 4 of the present invention provides a computer device, the device comprising:
[0125] A processor, adapted to execute computer programs;
[0126] A computer-readable storage medium storing a computer program, which, when executed by the processor, implements the steps in the charging selection probability prediction method for highway EV users as described in Embodiment 1 of the present invention.
[0127] Example 5:
[0128] Embodiment 5 of the present invention provides a computer program product or computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the steps in the charging selection probability prediction method for highway EV users as described in Embodiment 1 of the present invention.
[0129] The steps and methods involved in Examples 2, 3, 4 and 5 above correspond to those in Example 1. For specific implementation methods, please refer to the relevant description section of Example 1.
[0130] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed in this application can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.
[0131] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. A computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the flow or function according to the embodiments of this application is generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in or transmitted through a computer-readable storage medium. The computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless means. The computer-readable storage medium can be any available medium that a computer can access or a data processing device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium, an optical medium, or a semiconductor medium, etc.
[0132] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.
Claims
1. A method for predicting the probability of charging choices for EV users on highways, characterized in that, Includes the following steps: Acquire the individual attribute data and charging status data to be predicted, and perform preprocessing operations; Personal attribute data includes gender and travel purpose, while charging status data includes remaining range, battery level when entering the highway, battery level when starting to charge, and remaining range. A charging behavior prediction model is constructed using a single-class support vector machine, and the model is trained using a multi-dimensional training sample set. Specifically, an objective function is designed based on the classification of user charging behavior, and the decision value representing the characteristics of user charging behavior is obtained by solving the objective function. The decision value is derived based on the density partitioning interval method to obtain the user's charging choice probability. The specific steps for deriving decision values based on the density-based interval method are as follows: Determine the range of decision values based on the decision value boundaries; Based on the density binning principle, intervals are divided according to the range of decision values; Assign a corresponding charging user sample probability to each interval; The specific steps for dividing intervals based on the decision value range according to the density binning principle are as follows: Sort the decision values from smallest to largest; The sorted decision values are divided into intervals with equal sample sizes based on the preset number of intervals. The trained charging behavior prediction model is used to predict the probability of charging choices based on personal attribute data and charging status data.
2. The charging selection probability prediction method for highway EV users as described in claim 1, characterized in that, The preprocessing operation involves cleaning the personal attribute data and charging status data, removing null and outlier values, and performing one-hot encoding on the personal attribute data.
3. The charging selection probability prediction method for highway EV users as described in claim 1, characterized in that, The specific steps for designing an objective function based on the classification of user charging behavior and solving the objective function to obtain decision values that characterize the features of user charging behavior are as follows: Based on the classification of user charging behavior, a hyperplane-based objective function is constructed. The inside of the hyperplane represents normal samples, i.e., EV users charging, while the outside represents abnormal samples, i.e., EV users not charging. The objective function is optimized by introducing dual transformation and kernel function; The decision value is obtained by solving the optimized objective function.
4. A charging selection probability prediction system for highway EV users, employing the charging selection probability prediction method for highway EV users as described in any one of claims 1-3, characterized in that, include: The data acquisition module is configured to acquire the personal attribute data and charging status data to be predicted, and perform preprocessing operations. The model training module is configured to build a charging behavior prediction model using a single-class support vector machine and train the charging behavior prediction model using a multi-dimensional training sample set. Specifically, an objective function is designed based on the classification of user charging behavior, and the objective function is solved to obtain the decision value that characterizes the user's charging behavior features. The decision value is derived based on the density partitioning interval method to obtain the user's charging choice probability. The probability prediction module is configured to use a trained charging behavior prediction model to predict the probability of charging choices based on personal attribute data and charging status data.
5. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the charging selection probability prediction method for highway EV users as described in any one of claims 1-3.
6. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program adapted to be loaded by a processor and executed as described in any one of claims 1-3, the method for predicting the probability of charging selection for highway EV users.
7. A computer device, characterized in that, include: A processor, adapted to execute computer programs; A computer-readable storage medium storing a computer program, which, when executed by the processor, implements the charging selection probability prediction method for highway EV users as described in any one of claims 1-3.