Electric Vehicle User Preference Identification Method and Device Based on Historical Behavioral Data Analysis

Abstract

This invention discloses a method and apparatus for identifying electric vehicle user preferences based on historical behavioral data analysis. The method first collects historical charging order data from electric vehicle users, extracts statistical features of variables to generate a standardized dataset, then constructs a total utility expression using random utility theory, and builds a user choice probability model based on this expression using a binary logit model. Subsequently, based on historical order records in the standardized dataset, the charging selection probability is statistically calculated, and the user choice utility is calculated using a set formula based on the probability model. Finally, a Lasso regression model is trained using the standardized dataset and user choice utility to determine the undetermined parameters of the total utility expression. The influence of features is quantified through parameter weights, completing user preference identification and enabling charging choice prediction under a given environment. This invention solves the problems of existing technologies, such as reliance on questionnaire surveys, susceptibility to bias, lack of real historical data support, and difficulty in quantifying the influence weights of features.

Description

Technical Field

[0001] This invention relates to the field of intelligent transportation and energy interconnection technology, specifically to a method and apparatus for identifying electric vehicle user preferences based on historical behavioral data analysis. Background Technology

[0002] Currently, green transportation has become crucial for energy structure transformation. Electric vehicles, as the core carrier, can flexibly match their charging load with peak power generation from clean energy sources such as wind and solar power, effectively increasing the proportion of clean energy consumption, reducing the pressure on thermal power for peak shaving, and promoting the upgrading of the power system towards low carbon emissions. Against this backdrop, accurate identification of electric vehicle users' charging choices has become a core prerequisite for optimizing grid load distribution and unleashing the regulatory potential of electric vehicles. Related technologies have been widely applied in scenarios such as intelligent traffic scheduling, grid load forecasting, and charging operator operation optimization. Existing research often constructs user charging decision models to analyze observable factors such as state of charge, travel demand, and price costs, or combines latent variables such as users' socioeconomic attributes and psychological attitudes to explore key factors influencing users' charging choices.

[0003] However, existing technologies still have significant shortcomings: most studies rely on questionnaire surveys and statistical analysis methods, using charging scenarios and statistical attitude indicators to support model construction. The results are heavily influenced by factors such as questionnaire design quality and sample representativeness, making them prone to identification bias and unable to accurately reflect users' true charging preferences. While some studies have introduced theoretical frameworks such as discrete choice models, they have not fully integrated real-world user behavior data, remaining at the level of subjective assumptions and simulation analysis, lacking quantitative verification and practical application support. Furthermore, existing methods have not effectively addressed the issues of feature redundancy and parameter overfitting, making it difficult to quantify the influence weights of different factors on user preferences. This results in weak model generalization ability, failing to provide reliable technical support for charging guidance strategy formulation and grid load optimization.

[0004] Therefore, there is an urgent need for a method for identifying electric vehicle user preferences based on historical behavioral data analysis to address the problems of existing technologies, such as reliance on questionnaires, susceptibility to bias, lack of real historical data support, and difficulty in quantifying the influence weight of features. Summary of the Invention

[0005] To address these issues, the present invention provides a method and apparatus for identifying electric vehicle user preferences based on historical behavioral data analysis, which solves the problems of existing electric vehicle user preference identification methods that rely on questionnaire surveys, are prone to bias, lack support from real historical data, and are difficult to quantify the influence weight of features.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for identifying electric vehicle user preferences based on historical behavior data analysis, characterized in that it includes: By collecting and acquiring historical charging order data from electric vehicle users; and extracting statistical features from the historical charging order data to generate a standardized dataset; Based on the statistical characteristics of the variables in the standardized dataset, and combined with random utility theory, a total utility expression is constructed. Based on the total utility expression, a user choice probability model is constructed using a binary logit model. Based on historical charging order records in the standardized dataset, the probability of user charging selection under a set combination of features is statistically obtained; based on the user charging selection probability and the user selection probability model, the user selection utility is calculated by a set formula. The Lasso regression model is trained using the standardized dataset and the user choice utility to determine the undetermined parameters of the total utility expression. The influence of each feature on the user choice is quantified by the parameter weights in the total utility expression, thereby completing user preference identification and realizing charging choice prediction under the given environment.

[0007] As a preferred embodiment of the electric vehicle user preference identification method based on historical behavioral data analysis, the statistical features of the variables include user ID, age, annual income, distance to charging station, charging time, waiting time, initial SOC, and charging amount.

[0008] As a preferred embodiment of the electric vehicle user preference identification method based on historical behavior data analysis, the total utility expression is: ; In the formula, The total utility of choosing option i for user n; For observable representative utility; These are unobservable random terms; ; In the formula, , ... These are parameters to be determined. , ... Statistical characteristics of variables.

[0009] As a preferred solution for electric vehicle user preference identification methods based on historical behavior data analysis, in the process of constructing the user selection probability model through the binary logit model, the expression of the user selection probability model is: ; In the formula, Let I be the probability that decision-maker n chooses option i; I is the set of options available to the user; and k is any option in the set.

[0010] As a preferred embodiment of the electric vehicle user preference identification method based on historical behavioral data analysis, in the process of calculating the user selection utility using the set formula, the set formula is a logarithmic transformation formula, expressed as: ; In the formula, Select utility for the user; Select the probability for the user.

[0011] The present invention also provides an electric vehicle user preference recognition device based on historical behavior data analysis, which employs the above-mentioned electric vehicle user preference recognition method based on historical behavior data analysis, including: The variable statistical feature extraction module is used to collect historical charging order data of electric vehicle users; extract variable statistical features from the historical charging order data; and generate a standardized dataset. The user selects a utility expression construction module, which is used to construct a total utility expression based on the statistical characteristics of the variables in the standardized dataset and in combination with random utility theory. The user choice probability model construction module is used to construct a user choice probability model based on the total utility expression using a binary logit model. The user choice utility acquisition module is used to statistically obtain the user charging choice probability under a set feature combination based on historical charging order records in the standardized dataset; and to calculate the user choice utility based on the user charging choice probability and the user choice probability model through a set formula. The parameter weight determination module in the user choice utility expression is used to train the Lasso regression model using the standardized dataset and the user choice utility to determine the undetermined parameters of the total utility expression; and to quantify the influence of each feature on the user choice by using the parameter weights in the total utility expression, thereby completing user preference identification and realizing charging choice prediction under a given environment.

[0012] As a preferred embodiment of an electric vehicle user preference identification device based on historical behavioral data analysis, the variable statistical feature extraction module includes user ID, age, annual income, charging station distance, charging time, waiting time, initial SOC, and charging amount.

[0013] As a preferred embodiment of the electric vehicle user preference recognition device based on historical behavior data analysis, the total utility expression in the user selection utility expression construction module is: ; In the formula, The total utility of choosing option i for user n; For observable representative utility; These are unobservable random terms; ; In the formula, , ... These are parameters to be determined. , ... Statistical characteristics of variables.

[0014] As a preferred embodiment of an electric vehicle user preference recognition device based on historical behavior data analysis, in the user selection probability model construction module, during the process of constructing the user selection probability model using the binary logit model, the expression of the user selection probability model is: ; In the formula, Let I be the probability that decision-maker n chooses option i; I is the set of options available to the user; and k is any option in the set.

[0015] As a preferred embodiment of an electric vehicle user preference recognition device based on historical behavior data analysis, in the user selection utility acquisition module, during the calculation of the user selection utility using the set formula, the set formula is a logarithmic transformation formula, expressed as: ; In the formula, Select utility for the user; Select the probability for the user.

[0016] The present invention has the following advantages: First, this invention is based on real historical charging order data, eliminating reliance on questionnaires and avoiding subjective bias, thus effectively reflecting users' true charging preferences.

[0017] Second, this invention integrates random utility theory and binary logit model to construct a scientific quantitative relationship between probability and utility, and clarifies the mechanism of preference influence.

[0018] Third, this invention uses the Lasso regression model to effectively compress redundant features, avoid overfitting, and accurately quantify the influence weight of each feature on the user's choice.

[0019] Fourth, the model has strong generalization ability and high recognition accuracy, which can provide reliable technical support for power grid load forecasting and charging guidance strategy formulation. Attached Figure Description

[0020] To more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings in the following description are merely exemplary, and those skilled in the art can derive other embodiments based on the provided drawings without creative effort.

[0021] The structures, proportions, sizes, etc. illustrated in this specification are only for the purpose of assisting those skilled in the art in understanding and reading the content disclosed herein, and are not intended to limit the conditions under which the present invention can be implemented. Therefore, they have no substantial technical significance. Any modifications to the structure, changes in the proportions, or adjustments to the size, without affecting the effects and objectives that the present invention can produce, should still fall within the scope of the technical content disclosed in the present invention.

[0022] Figure 1 This is a flowchart illustrating the electric vehicle user preference identification method based on historical behavior data analysis provided in Embodiment 1 of the present invention. Figure 2 This is a schematic diagram illustrating the specific implementation process of the electric vehicle user preference identification method based on historical behavior data analysis provided in Embodiment 1 of the present invention. Figure 3 This is a schematic diagram of the regression coefficients of the user choice utility model in one possible embodiment provided in Embodiment 1 of the present invention; Figure 4 This is a schematic diagram of the architecture of the electric vehicle user preference recognition device based on historical behavior data analysis provided in Embodiment 2 of the present invention. Detailed Implementation

[0023] The following specific embodiments illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. 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.

[0024] Example 1 See Figure 1 and Figure 2 Embodiment 1 of the present invention provides a method for identifying electric vehicle user preferences based on historical behavior data analysis, comprising the following steps: S1. Collect and obtain historical charging order data of electric vehicle users; extract variable statistical features from the historical charging order data to generate a standardized dataset; S2. Based on the statistical characteristics of the variables in the standardized dataset, and combined with random utility theory, construct the total utility expression; S3. Based on the total utility expression, construct a user choice probability model using a binary logit model; S4. Based on the historical charging order records in the standardized dataset, statistically obtain the user charging selection probability under a set feature combination; based on the user charging selection probability and the user selection probability model, calculate the user selection utility using a set formula; S5. Using the standardized dataset and the user choice utility, train the Lasso regression model to determine the undetermined parameters of the total utility expression; quantify the influence of each feature on the user choice by using the parameter weights in the total utility expression, complete user preference identification, and achieve charging choice prediction under the given environment.

[0025] In this embodiment, in step S1, historical charging order data of electric vehicle users is collected and obtained; statistical features of variables are extracted from the historical charging order data to generate a standardized dataset.

[0026] Specifically, firstly, historical charging order data from electric vehicle users is collected through legal and compliant means. Data sources may include the Evwatts public dataset or operational record datasets from charging operators, ensuring data coverage of different user groups, charging scenarios, and time periods to guarantee data representativeness and completeness. Secondly, key variable statistical features are extracted from the collected raw order data. These statistical features must comprehensively reflect user charging behavior and related influencing factors, specifically including user socioeconomic attributes such as user ID, age, and annual income; charging scenario features such as charging station distance, charging time, and waiting time; and vehicle-related features such as initial SOC (State of Charge) and charging amount. Finally, the extracted variable statistical features are standardized. A normalizer eliminates differences in dimensions between different features, and the data is split into training and testing sets according to a preset ratio, such as 8:2. The data format is standardized to CSV, with the first row containing the feature name and target variable name, and subsequent rows containing the specific sample data, ultimately generating a standardized dataset with a standardized structure that can be directly used for model training.

[0027] In this embodiment, in step S2, based on the statistical characteristics of the variables in the standardized dataset and combined with random utility theory, a total utility expression is constructed.

[0028] Specifically, based on the statistical characteristics of variables in the standardized dataset generated in step S1, and taking the theory of random utility as the core basis, it is assumed that users always follow the principle of "utility maximization" in the charging selection process, that is, users will choose the charging solution that brings them the greatest utility.

[0029] The specific assumption is as follows: Under certain conditions, decision-makers will face different choices, assuming there are I such choices. Given these conditions, each choice has a specific utility. If there is n decision-makers, the utility they can obtain by choosing option i can be denoted as... Let i = 1, 2, ..., I. This utility is influenced by environmental conditions and depends on the decision-maker's own psychological preferences; different decision-makers may derive different utilities from choices made under the same conditions.

[0030] The choice made by the decision-maker is the choice that maximizes their own utility, therefore, if and only if > When i≠j, the decision-maker will choose option i to maximize their utility.

[0031] Based on this assumption, a total utility expression is constructed, which decomposes the total utility chosen by the user into observable and unobservable components: ; In the formula, The total utility of choosing option i for user n; For observable representative utility; These are unobservable random terms used to characterize unquantifiable factors such as individual preferences that are not measured, temporary decision-making interference, and data measurement errors. Random vectors The joint density function is f( By constructing a stochastic utility theory, we can obtain a concrete representation of the utility that decision-makers experience when making choices, and use this as the basis for subsequent choice processes. Among these, the stochastic vector... The form of the distribution function will be a key factor influencing the subsequent selection process.

[0032] in, ; In the formula, , ... These are parameters to be determined. , ... Statistical characteristics of variables.

[0033] In this embodiment, in step S3, a user choice probability model is constructed based on the total utility expression using a binary logit model.

[0034] Specifically, a specific choice model is constructed based on the representation of the decision-maker's utility. This part is based on the theory of Discrete Choice Models (DCM). In discrete choice models, the distribution assumption of the random error term is the core of model construction, and its choice directly determines the model form, computational complexity, and empirical applicability. The random error term (usually denoted as ε) represents the unobservable part of the decision-maker's utility. The distribution assumption of the random error term originates from statistical inference of unobservable factors, and its setting must meet two basic requirements: First, ensure that the selection probability is within the interval [0, 1]. Second, the sum of the probabilities of satisfying all options is 1.

[0035] This invention considers both mathematical convenience and empirical applicability. Based on the total utility expression constructed in step S2, and combined with discrete choice model theory, a binary logit (BL) model is selected to construct the user choice probability model. This model has the advantages of strong mathematical convenience and wide empirical applicability. In the model construction process, it is assumed that the random term ε in the total utility expression follows an independent and identically distributed Type I extreme value distribution. This distribution assumption ensures that the user choice probability meets the basic statistical requirements of "taking values ​​in the interval [0,1]" and "the sum of the probabilities of all alternatives is 1".

[0036] The expression for the user selection probability model is as follows: ; In the formula, Let I be the probability that decision-maker n chooses option i; I be the set of options available to the user; and k be any option in the set. This model clarifies the representativeness utility. There is a positive correlation between the probability of selection and the representative utility of a charging solution; that is, the higher the representative utility of a certain charging solution, the greater the probability that the user will choose that solution.

[0037] This invention will select utility for the user. Consider it a decision-making tendency, choose The value of can be 0 or 1, and we have: ; ; In the formula, , ... These are parameters to be determined. , ... Statistical characteristics of variables.

[0038] Assume random error term The distribution function is F Under the binary logit model, F ; ; The formula indicates that under the condition of a binary logit model, since correspond Therefore, the probability of users choosing to charge After equivalent transformation, we obtain In the formula, Choose a result variable for charging; its value can be 0 or 1. For observable representative utility; for An equivalent transformation of this, the probability is used to characterize the likelihood of a user choosing to charge in the corresponding state.

[0039] The above calculation process yields a quantitative relationship between the probability of decision-maker n choosing option i and the observable utility of that option for the decision-maker. In other words, given a specific option, the probability of the decision-maker making a choice can be calculated based on its observable utility value. Conversely, if the user's choice probability is known, the observable utility of that option for the decision-maker can be calculated based on this relationship. Further data processing and analysis are then performed based on this foundation.

[0040] In this embodiment, in step S4, based on the historical charging order records in the standardized dataset, the user charging selection probability under a set feature combination is statistically obtained; based on the user charging selection probability and the user selection probability model, the user selection utility is calculated by a set formula.

[0041] Specifically, firstly, based on the historical charging order records in the standardized dataset generated in step S1, statistical classification is performed according to different combinations of variable statistical features. For example, "charging station distance 2km + waiting time 10 minutes + initial SOC 30%" is one feature combination. The frequency of users choosing to charge under each feature combination is counted, and combined with the total number of orders under that feature combination, the probability of users choosing to charge under the set feature combination is calculated. This probability directly reflects a user's charging behavior tendency under specific conditions. Subsequently, based on the quantitative relationship established by the user choice probability model constructed in step S3, the calculation formula for user choice utility is derived through inverse operation, i.e., the formula is set as follows: ; In the formula, Select utility for the user; Select the probability for the user.

[0042] The statistically obtained user charging choice probability Substituting into this formula completes the transformation from selection probability to user selection utility, ultimately yielding the user selection utility corresponding to the statistical characteristics of the variable, providing a target variable for the subsequent training of the Lasso regression model.

[0043] In this embodiment, in step S5, the Lasso regression model is trained using the standardized dataset and the user choice utility to determine the undetermined parameters of the total utility expression; the influence of each feature on the user choice is quantified by the parameter weights in the total utility expression, thereby completing user preference identification and realizing charging choice prediction under the given environment.

[0044] Specifically, firstly, the training set from the standardized dataset generated in step S1 is used as the input features, and the user choice utility calculated in step S4 is used as the target variable, together forming the training data for the Lasso regression model. Before model training, the input features are standardized to eliminate the influence of dimensions. Then, the Lasso regression model parameters are initialized, with the regularization coefficient α = 0.001 and the maximum number of iterations set to 10000. The training data is then input into the model for fitting training. The Lasso regression model achieves parameter compression through the L1 norm penalty term, shrinking the parameters corresponding to irrelevant or minimally influential features to 0, thereby accurately identifying the parameters that have a significant impact on user choice utility. The regression coefficients output by the training are the parameters to be determined in the total utility expression. , ... Next, the weights of these parameters are interpreted: positive parameters indicate that the corresponding feature has a promoting effect on user selection, while negative parameters indicate an inhibiting effect. The larger the absolute value of the parameter, the more significant the influence of the feature on user preference, thus completing the quantitative identification of user preferences. Finally, the test set from the standardized dataset is input into the trained model, and the model performance is verified using mean squared error (MSE) and coefficient of determination (R²). The verified model can predict users' charging choices under a given environment, providing support for the formulation of charging guidance strategies and grid load optimization.

[0045] In one possible implementation, a verification example is provided as follows: I. Reading the dataset Table 1 shows some of the user charging information collected in this embodiment: Table 1 User Charging Information (First 15 Rows)

[0046] II. Lasso Regression Model Processing The Lasso regression model processing algorithm is as follows: / / Data loading and initialization / / Read "evwatts2.2.csv" Dataset = Read the CSV file "evwatts2.2.csv" / / Data splitting: Feature variable X; Target variable y; / / Feature variable X = dataset[all rows, all columns except the last column] Target variable y = dataset[all rows, last column] / / Extract data metadata Feature name list = column names of the dataset [excluding the last column] Target variable name = Column name of the dataset [last column] / / Organization Dataset Structure Custom dataset = { "Data": Feature variable X; "Target value": Target variable y; "Feature Name": A list of feature names; "Target variable name": Target variable name; Description: Charging order information; } / / Output data information Output "Feature data shape:" + number of rows and columns Output "Feature Name:" + custom dataset["Feature Name"] Output "Features of the first two samples:" + the first two rows of data from the custom dataset ["data"]. Output "target values ​​of the first two samples:" + the first two elements of the custom dataset ["target values"]. Output "first five rows of the dataset:" + the first 5 rows of data in the dataset. Output "Feature shape:" + the number of rows and columns of feature variable X / / Split the training set / test set + standardize / / Split at an 8:2 ratio; Random seed = 42; Training features X_train, testing features X_test, training target y_train, testing target y_test = Split dataset(feature variable X, target variable y, test set proportion = 0.2, random seed = 42) / / Standardization processing Create a standardizer Training set feature standardization = standardizer fit(training features X_train) and transformation Test set feature standardization = Standardizer transformation(test feature X_test) / / Lasso regression model training / / Initialize the Lasso model: { The regularization coefficient α = 0.001; Random seed = 42; Maximum number of iterations = 10000 } / / Create a Lasso model: { α=0.001,; Random seed = 42; Maximum number of iterations = 10000 } / / Training set feature standardization, training target y_train / / Output model coefficients Output "Lasso regression coefficients:" For each feature name, iterate through the pairs of "corresponding model coefficients": Output "feature name" + "":" + "coefficient" / / Model Prediction and Performance Evaluation / / Predict the target value using the standardized test set Test set predicted value y_pred = Model prediction (test set feature standardization) / / Calculate the evaluation indicators: Mean Squared Error (MSE), Coefficient of Determination R² Mean Squared Error (MSE) = Calculated as Mean Squared Error(Test Target y_test, Test Set Predicted Values ​​y_pred) Coefficient of determination R² = (Test target y_test, Test set predicted values ​​y_pred) / (Test set predicted values) / / Output evaluation results Output "test set MSE:"+MSE (rounded to 4 decimal places) Output "test set R²:"+R² (rounded to 4 decimal places) / / Visualization: Feature Coefficient Bar Chart Create a canvas (width=10, height=6). Draw a bar chart (x-axis = list of feature names, y-axis = model coefficients) Add a horizontal reference line (y=0, red dashed line). Set the chart title to: "Lasso Regression Coefficients" Rotate the x-axis label by 45 degrees Adjust layout Show chart

[0047] III. Results User choice utility model regression coefficients, such as Figure 3 As shown.

[0048] from Figure 3 As can be seen from the data, the Lasso regression coefficients are shown in Table 2: Table 2 Lasso regression coefficients

[0049] The model calculation results show that, for decision-makers' utility, the distance to the charging station and waiting time have relatively obvious negative coefficients, while the charging time, initial charging capacity and other characteristic variables show weak correlation under the dataset conditions used in this study.

[0050] The application scenarios of this invention are as follows: In the scenario of power grid load forecasting, this invention can accurately predict the charging load demand in different time periods and regions by identifying the charging preferences of electric vehicle users, thus providing data support for power grid dispatching and clean energy consumption.

[0051] In the scenario of optimizing the operation of charging operators, this invention can clearly identify users' core needs such as waiting time and distance to the station, helping operators to optimize the layout of charging stations, adjust charging pricing and service strategies, and improve user satisfaction and operational efficiency.

[0052] In the context of intelligent charging guidance, this invention can predict charging choices under specific conditions based on user preferences, and push suitable charging solutions to users through navigation apps and in-vehicle systems, guiding users to charge during off-peak hours and choose the optimal charging station.

[0053] In the context of new energy vehicle industry planning, this invention can quantify user charging behavior preferences, providing a basis for automakers to develop charging-related functions that meet user needs and for policymakers to formulate targeted industry support policies.

[0054] In the context of electricity market demand response, this invention can accurately identify user groups that are sensitive to factors such as price and time of day, helping power companies to formulate differentiated demand response strategies and stimulating users' enthusiasm for participating in grid regulation.

[0055] It should be noted that the method of this disclosure embodiment can be executed by a single device, such as a computer or server. The method of this embodiment can also be applied to a distributed scenario, where multiple devices cooperate to complete the task. In such a distributed scenario, one of these devices may execute only one or more steps of the method of this disclosure embodiment, and the multiple devices will interact with each other to complete the method described.

[0056] It should be noted that the above description describes some embodiments of this disclosure. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in a different order than that shown in the above embodiments and still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0057] Example 2 See Figure 4 Embodiment 2 of the present invention also provides an electric vehicle user preference recognition device based on historical behavior data analysis, comprising: The variable statistical feature extraction module 001 is used to collect and obtain historical charging order data of electric vehicle users; extract variable statistical features from the historical charging order data, and generate a standardized dataset. The user selects the utility expression construction module 002, which is used to construct the total utility expression based on the statistical characteristics of the variables in the standardized dataset and in combination with random utility theory. User selection probability model construction module 003 is used to construct a user selection probability model based on the total utility expression using a binary logit model; The user choice utility acquisition module 004 is used to statistically obtain the user charging choice probability under a set feature combination based on the historical charging order records in the standardized dataset; and to calculate the user choice utility based on the user charging choice probability and the user choice probability model through a set formula. The parameter weight determination module 005 in the user choice utility expression is used to train the Lasso regression model using the standardized dataset and the user choice utility to determine the undetermined parameters of the total utility expression; and to quantify the influence of each feature on the user choice by using the parameter weights in the total utility expression, thereby completing user preference identification and realizing charging choice prediction under the given environment.

[0058] In this embodiment, the variable statistical feature extraction module 001 includes user ID, age, annual income, distance to charging station, charging time, waiting time, initial SOC, and charging amount.

[0059] In this embodiment, in the user-selected utility expression construction module 002, the total utility expression is: ; In the formula, The total utility of choosing option i for user n; For observable representative utility; These are unobservable random terms; ; In the formula, , ... These are parameters to be determined. , ... Statistical characteristics of variables.

[0060] In this embodiment, in the user selection probability model construction module 003, during the process of constructing the user selection probability model using the binary logit model, the expression of the user selection probability model is: ; In the formula, Let I be the probability that decision-maker n chooses option i; I is the set of options available to the user; and k is any option in the set.

[0061] In this embodiment, in the user selection utility acquisition module 004, during the process of calculating the user selection utility using the set formula, the set formula is a logarithmic transformation formula, and its expression is: ; In the formula, Select utility for the user; Select the probability for the user.

[0062] It should be noted that the information interaction and execution process between the modules of the above system are based on the same concept as the method embodiment in Embodiment 1 of this application, and the resulting technical effects are the same as those in the method embodiment of this application. For details, please refer to the description in the method embodiment shown above in this application, and it will not be repeated here.

[0063] Example 3

[0064] Embodiment 3 of the present invention provides a non-transitory computer-readable storage medium storing program code for an electric vehicle user preference identification method based on historical behavior data analysis. The program code includes instructions for executing the electric vehicle user preference identification method based on historical behavior data analysis of Embodiment 1 or any possible implementation thereof.

[0065] Computer-readable storage media can be any available medium that a computer can access, or a data storage device such as a server or data center that integrates one or more available media. The available medium can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media (e.g., solid-state drives (SSDs)).

[0066] Example 4

[0067] Embodiment 4 of the present invention provides an electronic device, including: a memory and a processor; The processor and the memory communicate with each other via a bus; the memory stores program instructions that can be executed by the processor, and the processor can execute the electric vehicle user preference recognition method based on historical behavior data analysis according to Embodiment 1 or any possible implementation thereof by calling the program instructions.

[0068] Specifically, a processor can be implemented in hardware or software. When implemented in hardware, the processor can be a logic circuit, an integrated circuit, etc. When implemented in software, the processor can be a general-purpose processor that reads software code stored in memory. This memory can be integrated into the processor or located outside the processor and exist independently.

[0069] 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. The 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 processes or functions described in the embodiments of the present invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable system. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means.

[0070] It is obvious to those skilled in the art that the modules or steps of the present invention described above can be implemented using general-purpose computing systems. They can be centralized on a single computing system or distributed across a network of multiple computing systems. Optionally, they can be implemented using program code executable by a computing system, thereby storing them in a storage system for execution by the computing system. In some cases, the steps shown or described can be performed in a different order than those presented herein, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, the present invention is not limited to any particular combination of hardware and software.

[0071] Although the present invention has been described in detail above with general descriptions and specific embodiments, modifications or improvements can be made to it, which will be obvious to those skilled in the art. Therefore, all such modifications or improvements made without departing from the spirit of the present invention fall within the scope of protection claimed by the present invention.

Claims

1. A method for identifying preferences of an electric vehicle user based on historical behavior data analysis, characterized by, include: By collecting and acquiring historical charging order data from electric vehicle users; Statistical features of variables are extracted from the historical charging order data to generate a standardized dataset; Based on the statistical characteristics of the variables in the standardized dataset, and combined with random utility theory, a total utility expression is constructed. Based on the total utility expression, a user choice probability model is constructed using a binary logit model. Based on historical charging order records in the standardized dataset, the probability of user charging selection under a set combination of features is statistically obtained; based on the user charging selection probability and the user selection probability model, the user selection utility is calculated by a set formula. The Lasso regression model is trained using the standardized dataset and the user choice utility to determine the undetermined parameters of the total utility expression. The influence of each feature on the user choice is quantified by the parameter weights in the total utility expression, thereby completing user preference identification and realizing charging choice prediction under the given environment.

2. The electric vehicle user preference identification method based on historical behavior data analysis according to claim 1, characterized in that, The statistical characteristics of the variables include user ID, age, annual income, distance to charging station, charging time, waiting time, initial SOC, and charging amount.

3. The electric vehicle user preference identification method based on historical behavior data analysis according to claim 2, characterized in that, The total utility expression is: ； In the formula, The total utility of choosing option i for user n; For observable representative utility; These are unobservable random terms; ； In the formula, , ... These are parameters to be determined. , ... Statistical characteristics of variables.

4. The electric vehicle user preference identification method based on historical behavior data analysis according to claim 3, characterized in that, In constructing the user choice probability model using the binary logit model, the expression for the user choice probability model is: ； In the formula, Let I be the probability that decision-maker n chooses option i; I is the set of options available to the user; and k is any option in the set.

5. The electric vehicle user preference identification method based on historical behavior data analysis according to claim 4, characterized in that, In the process of calculating the user's chosen utility using the set formula, the set formula is a logarithmic transformation formula, expressed as: ； In the formula, Select utility for the user; Select the probability for the user.

6. An electric vehicle user preference recognition device based on historical behavior data analysis, employing the electric vehicle user preference recognition method based on historical behavior data analysis as described in any one of claims 1-5, characterized in that, include: The variable statistical feature extraction module is used to collect and obtain historical charging order data of electric vehicle users; Statistical features of variables are extracted from the historical charging order data to generate a standardized dataset; The user selects a utility expression construction module, which is used to construct a total utility expression based on the statistical characteristics of the variables in the standardized dataset and in combination with random utility theory. The user choice probability model construction module is used to construct a user choice probability model based on the total utility expression using a binary logit model. The user choice utility acquisition module is used to statistically obtain the user charging choice probability under a set feature combination based on historical charging order records in the standardized dataset; and to calculate the user choice utility based on the user charging choice probability and the user choice probability model through a set formula. The parameter weight determination module in the user choice utility expression is used to train the Lasso regression model using the standardized dataset and the user choice utility to determine the undetermined parameters of the total utility expression; and to quantify the influence of each feature on the user choice by using the parameter weights in the total utility expression, thereby completing user preference identification and realizing charging choice prediction under a given environment.

7. The electric vehicle user preference recognition device based on historical behavior data analysis according to claim 6, characterized in that, In the variable statistical feature extraction module, the variable statistical features include user ID, age, annual income, distance to charging station, charging time, waiting time, initial SOC, and charging amount.

8. The electric vehicle user preference recognition device based on historical behavior data analysis according to claim 7, characterized in that, In the user-selected utility expression construction module, the total utility expression is: ； In the formula, The total utility of choosing option i for user n; For observable representative utility; These are unobservable random terms; ； In the formula, , ... These are parameters to be determined. , ... Statistical characteristics of variables.

9. The electric vehicle user preference recognition device based on historical behavior data analysis according to claim 8, characterized in that, In the user selection probability model construction module, during the process of constructing the user selection probability model using the binary logit model, the expression of the user selection probability model is: ； In the formula, Let I be the probability that decision-maker n chooses option i; I is the set of options available to the user; and k is any option in the set.

10. The electric vehicle user preference recognition device based on historical behavior data analysis according to claim 9, characterized in that, In the user-selected utility acquisition module, during the process of calculating the user-selected utility using the set formula, the set formula is a logarithmic transformation formula, and its expression is: ； In the formula, Select utility for the user; Select the probability for the user.

Images