Electric vehicle charging station recommendation method and system

Abstract

An electric vehicle charging station recommendation method and system, applied to the related field of charging station recommendation. The method comprises: acquiring user charging features and influence degree values of user charging evaluation features on user charging preferences, wherein the user charging evaluation features are constructed on the basis of the user's evaluation information on charging stations (S1); calculating the user's comprehensive scores for different charging stations on the basis of the influence degree values, and constructing a user preference matrix by using the user's comprehensive scores for different charging stations as elements (S2); and decomposing the user preference matrix by means of a singular value decomposition method, acquiring potential matching relationships between the user and charging station features, and recommending a charging station to the user on the basis of the potential matching relationships (S3). The method solves the problem in the prior art that charging stations cannot be personalizedly recommended for different users, thereby improving user experience.

Description

Electric vehicle charging station recommendation method and system TECHNICAL FIELD

[0001] The present application relates to the field of charging station recommendation, in particular to an electric vehicle charging station recommendation method and system. BACKGROUND

[0002] With the popularity of electric vehicles, charging stations are increasingly used. The charging station recommendation system used by users when searching for charging stations is a key factor affecting user experience and charging station utilization.

[0003] Existing recommendation systems rely on simple rules or basic machine learning algorithms. For example, Chinese patent CN115345451A discloses an electric vehicle charging guidance method based on charging station recommendation strategy. The method is based on real-time related data of user charging demand and regular data required by the system. Based on time series analysis and machine learning algorithm, the future period of power grid load and charging station load is predicted. According to the load prediction result, combined with the charging demand of electric vehicles and the real-time state of charging stations, an optimization algorithm is used to provide charging station recommendations based on comprehensive time cost and economic cost. Chinese patent CN115577863A discloses an electric vehicle charging station recommendation method, system, device and medium. By obtaining electric vehicle charging parameters of the charging station, the minimum load difference value of the charging station is calculated based on the electric vehicle charging parameters to obtain the scheduling scheme under the condition of minimum cost. According to the scheduling scheme, the price sensitivity pricing method is used to guide electric vehicle charging, which minimizes the operation and maintenance cost of the charging station and realizes user charging cost compensation.

[0004] The above methods classify users and develop unified recommendation strategies based on statistical analysis of multi-user behavior, which cannot provide personalized recommendations for different user preferences, resulting in poor user experience. SUMMARY

[0005] To solve the problem that existing technologies cannot provide personalized recommendations for different users, the present application provides an electric vehicle charging station recommendation method, which comprises:

[0006] Obtain the influence degree value of user charging characteristics and user charging evaluation characteristics on user charging preferences, wherein the user charging evaluation characteristics are constructed based on user evaluation information of charging stations;

[0007] Calculate the comprehensive score of different charging stations for the user based on the influence degree value, and construct a user preference matrix with the comprehensive score of different charging stations for the user as an element.

[0008] The user preference matrix is decomposed by using a singular value decomposition method to obtain a latent matching relationship of user and charging station characteristics, and a charging station is recommended to the user based on the latent matching relationship.

[0009] Optionally, the obtaining of the influence degree value of the user charging characteristics and the user charging evaluation characteristics on the user charging preference comprises:

[0010] The user charging characteristics and the user charging evaluation characteristics are obtained.

[0011] The user charging characteristics and the user charging evaluation characteristics are input into a LightGBM model to obtain the influence degree value.

[0012] Optionally, the obtaining of the user charging characteristics and the user charging evaluation characteristics comprises:

[0013] User charging data is obtained.

[0014] The user charging data is filtered to obtain target data comprising a charging frequency greater than a preset value and corresponding evaluation information.

[0015] The target data is subjected to feature extraction to obtain the user charging characteristics and the user charging evaluation characteristics.

[0016] Optionally, the feature extraction of the target data to obtain the user charging characteristics and the user charging evaluation characteristics comprises:

[0017] The evaluation information data in the target data is subjected to one-hot encoding to obtain a user score, and the corresponding user charging evaluation characteristics are extracted based on the user score.

[0018] The non-evaluation information data in the target data is subjected to feature extraction to obtain the user charging characteristics.

[0019] Optionally, after the user charging data is obtained and before the user charging data is filtered, the method further comprises:

[0020] The user charging data is cleaned.

[0021] Optionally, the training step of the LightGBM model is:

[0022] User historical charging behavior data and corresponding evaluation information are collected.

[0023] The evaluation information is subjected to one-hot encoding to obtain a user score.

[0024] Training user charging characteristics are extracted based on the user historical charging behavior data, and training user charging evaluation characteristics are extracted based on the user score.

[0025] The LightGBM model is trained by using the training user charging feature and the training user charging evaluation feature, and during the training process, mean square error or mean absolute error is used to evaluate the training result of each step until the mean square error or mean absolute error is less than a preset threshold.

[0026] Optionally, the comprehensive score of the user to different charging stations based on the influence degree value comprises:

[0027] For each charging station, the user charging feature and the user charging evaluation feature corresponding to the charging station are weighted and summed based on the influence degree value to obtain a comprehensive score of the charging station.

[0028] Optionally, the recommendation of the charging station to the user based on the potential matching relationship comprises:

[0029] The potential score of the user to the to-be-selected charging station is obtained based on the potential matching relationship by using the singular value decomposition method, and the to-be-selected charging station comprises at least one or more of the following: a charging station with an existing potential score and a charging station without a score.

[0030] The charging station is recommended to the user based on the potential score.

[0031] Optionally, the recommendation of the charging station to the user based on the potential score comprises:

[0032] According to the high and low of the potential score, the corresponding charging stations are sorted from high to low, and the first preset number of charging stations in the sorting are recommended to the user.

[0033] In another aspect, the application also provides an electric vehicle charging station recommendation system, comprising:

[0034] An acquisition module is configured to acquire an influence degree value of a user charging feature and a user charging evaluation feature on user charging preferences, wherein the user charging evaluation feature is constructed based on user evaluation information of a charging station.

[0035] A construction module is configured to calculate a comprehensive score of the user to different charging stations based on the influence degree value, and construct a user preference matrix with the comprehensive score of the user to different charging stations as an element.

[0036] A recommendation module is configured to decompose the user preference matrix by using a singular value decomposition method, obtain a potential matching relationship between the user and the charging station feature, and recommend the charging station to the user based on the potential matching relationship.

[0037] Optionally, the acquisition module is specifically configured to:

[0038] obtaining the user charging feature and the user charging evaluation feature;

[0039] inputting the user charging feature and the user charging evaluation feature into a LightGBM model to obtain an influence degree value.

[0040] Optionally, the obtaining module obtains the user charging feature and the user charging evaluation feature, including:

[0041] obtaining user charging data;

[0042] filtering the user charging data to obtain target data including charging times greater than a preset value and corresponding evaluation information;

[0043] extracting features from the target data to obtain the user charging feature and the user charging evaluation feature.

[0044] Optionally, the obtaining module extracts features from the target data to obtain the user charging feature and the user charging evaluation feature, including:

[0045] performing one-hot encoding on evaluation information data in the target data to obtain a user score, and extracting corresponding user charging evaluation features based on the user score;

[0046] extracting features from non-evaluation information data in the target data to obtain user charging features.

[0047] Optionally, after the obtaining module obtains the user charging data and before the filtering of the user charging data, the method further includes:

[0048] cleaning the user charging data.

[0049] Optionally, the training step of the LightGBM model in the obtaining module is:

[0050] collecting user historical charging behavior data and corresponding evaluation information;

[0051] performing one-hot encoding on the evaluation information to obtain a user score;

[0052] extracting training user charging features based on the user historical charging behavior data, and extracting training user charging evaluation features based on the user score;

[0053] training the LightGBM model using the training user charging features and the training user charging evaluation features, and using mean square error or mean absolute error to evaluate the training result of each step in the training process until the mean square error or the mean absolute error is less than a preset threshold.

[0054] Optionally, the constructing module calculates a comprehensive score of the user to different charging stations based on the influence degree value, comprising:

[0055] For each charging station, the user charging feature and the user charging evaluation feature corresponding to the charging station are weighted and summed based on the influence degree value to obtain a comprehensive score of the charging station.

[0056] Optionally, the recommending module recommends the charging station to the user based on the potential matching relationship, comprising:

[0057] The singular value decomposition method is used to obtain a potential score of the user to the to-be-selected charging station based on the potential matching relationship, the to-be-selected charging station comprising at least one or more of the following: a charging station with an existing potential score and a charging station without a score;

[0058] The charging station is recommended to the user based on the potential score.

[0059] Optionally, the recommending module recommends the charging station to the user based on the potential score, comprising:

[0060] According to the high and low of the potential score, the corresponding charging stations are sorted from high to low, and the first preset number of charging stations in the sorting are recommended to the user.

[0061] Compared with the prior art, the present application has the following advantages:

[0062] The present application provides a method and system for recommending a charging station for an electric vehicle, comprising obtaining a user charging feature and an influence degree value of a user charging evaluation feature of the user to the charging station on the user preference; calculating a comprehensive score of the user to different charging stations based on the influence degree value to construct a user preference matrix; decomposing the user preference matrix by using a singular value decomposition method to obtain a potential matching relationship between the user and the charging station feature, and recommending the charging station to the user based on the potential matching relationship; the influence degree value representing the user preference is constructed based on the user charging evaluation feature of the user to the charging station, the different charging stations are scored based on the influence degree value to construct the user preference matrix, and the user preference matrix contains the influence of the user preference; the singular value decomposition is used to decompose the user preference matrix to obtain the potential matching relationship between the user preference and the charging station feature, the charging station is recommended to the user based on the matching relationship, and personalized recommendation considering the user preference is realized. BRIEF DESCRIPTION OF DRAWINGS

[0063] Fig. 1 is a main flowchart of the method for recommending a charging station for an electric vehicle according to the present application;

[0064] Fig. 2 is a detailed flowchart of the method for recommending a charging station for an electric vehicle according to the present application;

[0065] FIG. 3 is a detailed flowchart of step S1 in FIG. 2 according to the present application;

[0066] FIG. 4 is a detailed flowchart of step S11 in FIG. 3 according to the present application;

[0067] FIG. 5 is a flowchart of feature extraction of target data according to the present application;

[0068] FIG. 6 is a main flowchart of a training method of a LightGBM model according to the present application;

[0069] FIG. 7 is a step diagram of the training method of the LightGBM model according to the present application;

[0070] FIG. 8 is a flowchart of an application process of singular value decomposition according to the present application;

[0071] FIG. 9 is a structural diagram of an electric vehicle charging station recommendation system according to the present application. DETAILED DESCRIPTION

[0072] The existing electric vehicle charging station recommendation technology considers factors including distance, fast charging, discount, high quality, and idle. However, the subjective factors of user charging order evaluation are not considered. The existing recommendation method cannot meet the complex personal preferences of users, cannot make targeted personalized customized recommendations for the preferences of different users, and the user experience is not good. Based on this, the present application provides an electric vehicle charging station recommendation method and system. As shown in FIG. 1, it is the overall flowchart of the present application. The method mainly includes:

[0073] (1) obtaining related data of the charging station and user evaluation information based on the user historical order;

[0074] (2) cleaning and unifying the format of the data obtained in step (1), and performing statistics;

[0075] (3) performing feature extraction on the statistical results obtained in step (2) to obtain user charging features related to the objectivity of user charging behavior, and to obtain user charging evaluation features related to the subjectivity of user charging behavior;

[0076] (4) inputting the user charging features and the user charging evaluation features into a pre-trained LightGBM model to obtain the influence degree value of the user charging features and the user charging evaluation features on the user charging preference;

[0077] (5) performing weighted summation on the user charging features and the user charging evaluation features based on the influence degree value to obtain a comprehensive score of the charging station, and constructing a user preference matrix based on the comprehensive score;

[0078] (6) using singular value decomposition (SVD) method to decompose the user preference matrix to obtain the latent matching relationship;

[0079] (7) recommending the charging station to be selected based on the latent matching relationship.

[0080] Embodiment 1:

[0081] A method for recommending an electric vehicle charging station, as shown in FIG. 2, includes the following steps S1-S3.

[0082] S1: obtaining the influence degree value of the user charging feature and the user charging evaluation feature on the user charging preference, wherein the user charging evaluation feature is constructed based on the evaluation information of the user on the charging station.

[0083] In a further preferred scheme, as shown in FIG. 3, the step S1 includes the following steps S11 and S12.

[0084] S11: obtaining the user charging feature and the user charging evaluation feature.

[0085] The user charging feature is used to represent the objective feature in the user charging process, and the user charging evaluation feature is used to represent the subjective feature in the user charging process. As shown in FIG. 4, the process of obtaining the user charging feature and the user charging evaluation feature includes the following steps S111-S113.

[0086] S111: various user charging data can be obtained from the user and the charging station, for example, the charging record, the evaluation information of the user, the basic information of the charging station, etc.

[0087] S112: cleaning and screening the user charging data to obtain target data including charging times greater than a preset value and corresponding evaluation information.

[0088] The collected user charging data is cleaned and standardized to remove noise, outliers, etc. in the data, ensuring data quality and consistency. The basic information of the charging user, such as transaction amount, charging time, charging times, etc. is subjected to descriptive statistics to understand the distribution characteristics of the data and lay a foundation for subsequent analysis.

[0089] The target data useful for user preference is obtained from the cleaned data, for example, the user and the corresponding order containing user evaluation information are screened out, which have charging times greater than a preset number (for example, 10 times) in a preset period (for example, half a year). The user's evaluation is an important data source reflecting his preference and satisfaction. The existing system can collect the user's evaluation, but has not yet made in-depth analysis and utilized these data for personalized recommendation.

[0090] S113: Feature extraction is performed on the target data to obtain the user charging features and the user charging evaluation features. For example, 17 features are screened out in this embodiment, which can be divided into user charging features with objective characteristics and user charging evaluation features with subjective characteristics. For example, in this embodiment, the user charging features include transaction electricity, preferential amount, transaction amount, whether parking is free, charging time, one-time charging success rate, station position serious comment rate, abnormal order rate, station DC pile proportion, whether plug and play, whether charging in peak or peak period, charging method, order source, and transaction end reason; and the user charging evaluation features include fault score, order score, and excessive scanning score.

[0091] In a further preferred scheme, feature extraction is performed on the target data to obtain the user charging features and the user charging evaluation features. As shown in FIG. 5, the process includes the following steps 1 and 2:

[0092] Step 1: One-hot encoding is performed on the evaluation information data in the target data to obtain user scores, and the corresponding user charging evaluation features are extracted based on the user scores.

[0093] Step 2: Feature extraction is performed on the non-evaluation information data in the target data to obtain user charging features.

[0094] S12: The user charging features and the user charging evaluation features are input into a LightGBM model to obtain the influence degree value.

[0095] LightGBM is a machine learning algorithm based on a gradient boosting framework, which has efficient data processing capability and accurate prediction performance, and is particularly suitable for processing large-scale data sets. Through feature selection and parallel computing optimization, it can significantly reduce the calculation time and resource consumption while maintaining high accuracy. Therefore, the pre-trained LightGBM model is used to obtain the influence degree value of the user charging features and the user charging evaluation features on the user preference, which can realize accurate prediction of the influence degree value and efficiently process a large amount of data for each user.

[0096] As shown in FIG. 6, the training process of the LightGBM model mainly includes the following steps:

[0097] (1) The data of the user historical orders is cleaned;

[0098] (2) The cleaned data is screened to screen out orders with charging times greater than a preset number and evaluation information;

[0099] (3) One-hot encoding is performed on the evaluation information to obtain user scores;

[0100] (4) extracting user charging evaluation features based on user scores, and extracting user charging features based on user historical charging behavior data;

[0101] (5) inputting the user charging evaluation features and the user charging features into a LightGBM model to obtain a predicted influence degree value;

[0102] (6) evaluating the LightGBM model by using mean square error or mean absolute error, and optimizing hyperparameters of the LightGBM model to realize training of the LightGBM model.

[0103] In a further preferred scheme, as shown in FIG. 7, the training process of the above LightGBM model comprises the following steps S121-S124 in detail:

[0104] S121: collecting user historical charging behavior data and corresponding evaluation information;

[0105] As described above, various user charging data can be obtained from users and charging stations, such as user charging records, evaluation information, basic information of charging stations, etc. The user charging data is cleaned and selected to obtain user historical charging behavior data including charging times greater than a preset value and corresponding evaluation information, i.e., target data. A preset proportion of the target data can be set as a training set, and an exceptional proportion can be set as a test set, for example, 80% of the target data as the training set and 20% as the test set. Subsequently, the training set is used to train the LightGBM model, and the test set is used to test the trained LightGBM model.

[0106] S122: obtaining user scores by one-hot encoding the evaluation information.

[0107] S123: extracting training user charging features based on the user historical charging behavior data, and extracting training user charging evaluation features based on the user scores.

[0108] S124: training the LightGBM model by using the training user charging features and the training user charging evaluation features, and evaluating the training result of each step by using mean square error (MSE) or mean absolute error (MAE) until the mean square error or the mean absolute error is less than a preset threshold, so as to realize iterative optimization of the LightGBM model. The influence degree value of each feature in the user charging features and the user charging evaluation features on user preference is obtained by using the trained model, i.e., the importance degree of each feature to the user is obtained.

[0109] S2: calculating a comprehensive score of the user to different charging stations based on the influence degree value, and constructing a user preference matrix with the comprehensive score of the user to different charging stations as elements.

[0110] Based on the above influence degree value, the comprehensive score of user A to different charging stations (charging station 1, charging station 2, charging station 3, …) can be calculated.

[0111] Therefore, for each charging station, the corresponding user charging characteristics and each of the user charging evaluation characteristics are weighted and summed based on the size of the influence degree value. The influence degree value represents the influence degree of each characteristic on the user. Therefore, the greater the influence degree on the user, the higher the corresponding influence degree value, and the higher the weight in the corresponding weighted sum process. Thus, the comprehensive score that meets the customer preference can be obtained.

[0112] A user preference matrix is constructed with the user, the charging station, and the corresponding comprehensive score as elements.

[0113] S3: decomposing the user preference matrix by using a singular value decomposition method to obtain a latent matching relationship of the user and the charging station characteristics, and recommending a charging station to the user based on the latent matching relationship.

[0114] The singular value decomposition (SVD) method is used to decompose the user preference matrix in the present application. The SVD method is generally used in the fields of signal processing and statistics for extracting important characteristics in a matrix. SVD is used in collaborative filtering processes to extract latent characteristics of users and items by decomposing a user-item interaction matrix, thereby predicting scores or interests for uncontacted items.

[0115] As shown in FIG. 8, based on the above steps, the influence degree value of the user charging characteristics and the user charging evaluation characteristics on the user charging preference can be obtained. The characteristics are weighted and summed based on the influence degree value to obtain the comprehensive score of the user to each charging station. A user preference matrix is constructed with the comprehensive score as elements. The SVD method in the present application is used to solve the user preference matrix to obtain a latent matching relationship. Based on the latent matching relationship, the charging stations to be selected are scored, and the top-N charging stations are taken as the recommended objects.

[0116] In the present embodiment, SVD is used to obtain a latent matching relationship of the user and the charging station characteristics. Based on the latent matching relationship, the SVD method can calculate the latent score of the user to the charging station to be selected. At least one or more of the following: charging stations with existing latent scores and charging stations without scores. For example, when the user arrives in a new area, the charging stations to be selected are all charging stations without scores. When the user arrives in a region that has been visited, the region may have added some charging stations. Therefore, the charging stations to be selected include charging stations with existing latent scores or charging stations without scores.

[0117] According to the high and low of the potential score, the corresponding charging station is sorted from high to low, and the charging station in the front of the ranking is recommended to the user, thereby realizing the charging station recommendation method based on user preference, realizing personalized recommendation, and the recommendation result is more consistent with the customer expectation, and the user experience is greatly enhanced.

[0118] The results show that the accuracy, recall rate and F1 value of the LightGBM model combined with the SVD method are 0.5604, 0.6040 and 0.5527 respectively; while in the prior art, only the charging frequency is taken as the user preference influencing factor, and then the SVD method is used for decomposition, and the accuracy, recall rate and F1 value are 0.4003, 0.5477 and 0.4231 respectively, it can be seen that the recommended charging station is more in line with the "people's heart" and more in line with the user demand.

[0119] In summary, the application uses historical data and charging station information to obtain the influence degree of the evaluation features on user preference, determines the importance of each feature, constructs a user preference matrix, and then uses the SVD method to decompose the preference matrix to predict the potential score of the user to the un-rated charging station, thereby optimizing the recommendation sorting algorithm. By considering the charging preference of the user, the application can provide the user with a charging station recommendation that is more in line with his needs, thereby improving the user's charging experience. This personalized recommendation method helps to reduce the time and effort consumed by the user in finding a suitable charging station.

[0120] Embodiment 2:

[0121] The application based on the same inventive concept also provides a charging station recommendation system for electric vehicles, a structural schematic diagram of which is shown in FIG. 9, which comprises:

[0122] The acquisition module is configured to acquire the user charging features and the influence degree of the user charging evaluation features on user preference, wherein the user charging evaluation features are constructed based on the evaluation information of the user on the charging station.

[0123] The construction module is configured to calculate the comprehensive score of the user on different charging stations based on the influence degree value, and construct a user preference matrix with the comprehensive score of the user on different charging stations as elements.

[0124] The recommendation module is configured to decompose the user preference matrix by using the singular value decomposition method, acquire the potential matching relationship between the user and the charging station features, and recommend the charging station to the user based on the potential matching relationship.

[0125] In a possible implementation, the acquisition module is specifically configured to:

[0126] acquire the user charging features and the user charging evaluation features.

[0127] inputting the user charging feature and the user charging evaluation feature into the LightGBM model to obtain an influence degree value.

[0128] In a possible implementation, the acquisition module acquires the user charging feature and the user charging evaluation feature, including:

[0129] acquiring user charging data;

[0130] filtering the user charging data to obtain target data including charging times greater than a preset value and corresponding evaluation information;

[0131] extracting features from the target data to obtain the user charging feature and the user charging evaluation feature.

[0132] In a possible implementation, the acquisition module extracts features from the target data to obtain the user charging feature and the user charging evaluation feature, including:

[0133] performing one-hot encoding on evaluation information data in the target data to obtain a user score, and extracting corresponding user charging evaluation features based on the user score;

[0134] extracting features from non-evaluation information data in the target data to obtain user charging features.

[0135] In a possible implementation, after the acquisition module acquires the user charging data and before the acquisition module filters the user charging data, the acquisition module further includes:

[0136] cleaning the user charging data.

[0137] In a possible implementation, the training step of the LightGBM model in the acquisition module is:

[0138] collecting user historical charging behavior data and corresponding evaluation information;

[0139] performing one-hot encoding on the evaluation information to obtain a user score;

[0140] extracting training user charging features based on the user historical charging behavior data, and extracting training user charging evaluation features based on the user score;

[0141] training the LightGBM model using the training user charging features and the training user charging evaluation features, and using mean square error or mean absolute error to evaluate the training result of each step in the training process until the mean square error or the mean absolute error is less than a preset threshold.

[0142] In a possible implementation, the constructing module calculates a comprehensive score of the user to different charging stations based on the influence degree value, including:

[0143] For each charging station, the user charging feature and the user charging evaluation feature corresponding to the charging station are weighted and summed based on the size of the influence degree value to obtain a comprehensive score of the charging station.

[0144] In a possible implementation, the recommending module recommends the charging station to the user based on the potential matching relationship, including:

[0145] The singular value decomposition method is adopted to obtain a potential score of the user to a to-be-selected charging station based on the potential matching relationship, the to-be-selected charging station including at least one or more of the following: a charging station with an existing potential score and a charging station without a score;

[0146] The charging station is recommended to the user based on the potential score.

[0147] In a possible implementation, the recommending module recommends the charging station to the user based on the potential score, including:

[0148] According to the high and low of the potential score, the corresponding charging stations are sorted from high to low, and the first preset number of charging stations in the front of the sorting are recommended to the user.

[0149] The above are only embodiments of the present application and are not used to limit the present application, and any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application are included in the scope of the claims of the present application.

Claims

1. An electric vehicle charging station recommendation method characterized by, The method comprises the following steps: obtaining the influence degree value of the user charging feature and the user charging evaluation feature on the user charging preference, wherein the user charging evaluation feature is constructed based on the evaluation information of the user on the charging station; calculating the comprehensive score of the user on different charging stations based on the influence degree value, and constructing a user preference matrix with the comprehensive score of the user on different charging stations as elements; decomposing the user preference matrix by using the singular value decomposition method to obtain the latent matching relationship between the user and the charging station feature, and recommending the charging station to the user based on the latent matching relationship.

2. The electric vehicle charging station recommendation method of claim 1, wherein, The method comprises the following steps: obtaining the influence degree value of the user charging feature and the user charging evaluation feature on the user charging preference, wherein the user charging evaluation feature is constructed based on the evaluation information of the user on the charging station; calculating the comprehensive score of the user on different charging stations based on the influence degree value, and constructing a user preference matrix with the comprehensive score of the user on different charging stations as elements; 3. The electric vehicle charging station recommendation method of claim 2, wherein, decomposing the user preference matrix by using the singular value decomposition method to obtain the latent matching relationship between the user and the charging station feature, and recommending the charging station to the user based on the latent matching relationship. The method comprises the following steps: obtaining the influence degree value of the user charging feature and the user charging evaluation feature on the user charging preference, wherein the user charging evaluation feature is constructed based on the evaluation information of the user on the charging station; calculating the comprehensive score of the user on different charging stations based on the influence degree value, and constructing a user preference matrix with the comprehensive score of the user on different charging stations as elements; 4. The electric vehicle charging station recommendation method of claim 3, wherein, decomposing the user preference matrix by using the singular value decomposition method to obtain the latent matching relationship between the user and the charging station feature, and recommending the charging station to the user based on the latent matching relationship. The method comprises the following steps: obtaining the influence degree value of the user charging feature and the user charging evaluation feature on the user charging preference, wherein the user charging evaluation feature is constructed based on the evaluation information of the user on the charging station; 5. The electric vehicle charging station recommendation method according to claim 3 or 4, characterized by, calculating the comprehensive score of the user on different charging stations based on the influence degree value, and constructing a user preference matrix with the comprehensive score of the user on different charging stations as elements; decomposing the user preference matrix by using the singular value decomposition method to obtain the latent matching relationship between the user and the charging station feature, and recommending the charging station to the user based on the latent matching relationship.

6. The electric vehicle charging station recommendation method of claim 2, wherein, The method comprises the following steps: obtaining the influence degree value of the user charging feature and the user charging evaluation feature on the user charging preference, wherein the user charging evaluation feature is constructed based on the evaluation information of the user on the charging station; calculating the comprehensive score of the user on different charging stations based on the influence degree value, and constructing a user preference matrix with the comprehensive score of the user on different charging stations as elements; decomposing the user preference matrix by using the singular value decomposition method to obtain the latent matching relationship between the user and the charging station feature, and recommending the charging station to the user based on the latent matching relationship. The method comprises the following steps:

7. The electric vehicle charging station recommendation method of claim 1, wherein, obtaining the influence degree value of the user charging feature and the user charging evaluation feature on the user charging preference, wherein the user charging evaluation feature is constructed based on the evaluation information of the user on the charging station; calculating the comprehensive score of the user on different charging stations based on the influence degree value, and constructing a user preference matrix with the comprehensive score of the user on different charging stations as elements; 8. The electric vehicle charging station recommendation method of claim 1, wherein, decomposing the user preference matrix by using the singular value decomposition method to obtain the latent matching relationship between the user and the charging station feature, and recommending the charging station to the user based on the latent matching relationship. The method comprises the following steps: obtaining the influence degree value of the user charging feature and the user charging evaluation feature on the user charging preference, wherein the user charging evaluation feature is constructed based on the evaluation information of the user on the charging station; 9. The electric vehicle charging station recommendation method of claim 8, wherein, calculating the comprehensive score of the user on different charging stations based on the influence degree value, and constructing a user preference matrix with the comprehensive score of the user on different charging stations as elements; decomposing the user preference matrix by using the singular value decomposition method to obtain the latent matching relationship between the user and the charging station feature, and recommending the charging station to the user based on the latent matching relationship. According to the high and low of the potential score, the corresponding charging stations are sorted from high to low, and the top pre-set number of charging stations are recommended to the user.

10. An electric vehicle charging station recommendation system characterized by, Comprise: An acquisition module is configured to acquire a user charging feature and an influence degree value of a user charging evaluation feature on a user charging preference, wherein the user charging evaluation feature is constructed based on evaluation information of the user on the charging station; A construction module is configured to calculate a comprehensive score of the user on different charging stations based on the influence degree value, and construct a user preference matrix with the comprehensive score of the user on different charging stations as elements; A recommendation module is configured to decompose the user preference matrix by using a singular value decomposition method, acquire a potential matching relationship between the user and the charging station feature, and recommend the charging station to the user based on the potential matching relationship.

Images