An intelligent management system for electric vehicle charging based on multi-source data analysis

Abstract

The present invention discloses an intelligent management system for electric vehicle charging based on multi-source data analysis, which relates to the technical field of electric vehicle charging. By extracting information from multi-source data including historical charging data HC, driving behavior data Db, geographical location information GP, and seasonal factors TS, and using PCA feature extraction technology, the system can generate more personalized and accurate charging plans according to the actual situations and preferences of different users. The system can dynamically adjust the charging time, charging amount, and location according to the specific charging habits, driving routes, and seasonal factors of users, making the charging experience more intelligent. Through PCA technology, the charging time T, charging amount Q, and charging station selection D of historical charging data H C are subjected to feature extraction. The system converts high-dimensional data into multi-dimensional features Z and generates a feature matrix HZ, reducing data redundancy and retaining key features, thereby greatly improving the efficiency of data analysis.

Description

Technical Field

[0001] The present invention relates to the technical field of electric vehicle charging, and particularly to an intelligent management system for electric vehicle charging based on multi-source data analysis. Background Art

[0002] As an important part of ITS, the intelligent charging management system mainly aims to provide personalized charging solutions through intelligent algorithms and multi-source data analysis to improve the charging efficiency of electric vehicles. Electric vehicle charging involves multiple dimensions of data sources, such as historical charging data, driving behavior, geographical location, and seasonal factors. These data can be analyzed and processed by the intelligent management system to provide charging solutions that better meet the individual needs of users.

[0003] In the invention patent with the application publication number CN215552653U, an intelligent management system for electric vehicle charging piles based on big data is disclosed. By giving a maintenance warning for the charging pile to alert maintenance personnel, after the charging pile is damaged, the charging head on it cannot charge. When the current vehicle owner connects the charging head to the electric vehicle, no current passes through the charging head. After the current detection system detects that no current passes through the charging head, it will simultaneously transmit this signal to the warning system, the maintenance reminder system, and the on-site reminder system. The warning system can warn the current vehicle owner that this charging pile cannot be used.

[0004] However, it is very difficult for the driver to find the most suitable charging station with the above functions. Moreover, at present, the current situation of the electric vehicle charging system mainly relies on the user's manual decision-making or a simple charging station management system. There are many deficiencies, and the personalized needs, driving habits, and driving routes of users are not effectively considered, resulting in an overly mechanical charging process and an inability to flexibly respond to actual needs. This charging method lacking intelligent analysis easily causes problems such as too long charging time, unreasonable distribution of charging stations, and large fluctuations in charging costs for users during the charging process. Summary of the Invention

[0005] Aiming at the deficiencies of the prior art, the present invention provides an intelligent management system for electric vehicle charging based on multi-source data analysis, which solves the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention is realized through the following technical solutions: An intelligent management system for electric vehicle charging based on multi-source data analysis includes a data acquisition module, a data preprocessing module, a feature extraction and weight assignment module, a personalized charging plan generation module, a user feedback and learning module, and a system monitoring and maintenance module;

[0007] The data acquisition module collects multi-source data in real time through sensors and in-vehicle devices, including the user's historical charging data H C, driving behavior data D b , geographical location information G P and seasonal factor T S , which form the original data set W X ;

[0008] The data preprocessing module cleans, denoises and normalizes the original data set W X to obtain the multi-source data set W;

[0009] The feature extraction and weight assignment module extracts features including charging time T, charging amount Q and charging station selection D from the multi-source data set W by using PCA feature extraction technology;

[0010] The personalized charging scheme generation module generates a charging scheme according to the output of the weighted recommendation model and in combination with the user's charging requirements, including the optimal charging time T C , the optimal charging amount Q C and the optimal charging location D LOC ;

[0011] The user feedback and learning module records the user's feedback on the charging scheme and adjusts the weight assignment strategy;

[0012] The system monitoring and maintenance module monitors the status of the vehicle during charging in real time, detects anomalies and faults, and conducts regular evaluations.

[0013] Preferably, the data acquisition module includes a sensor data acquisition unit and a data integration and storage unit;

[0014] The sensor data acquisition unit acquires historical charging data H C , driving behavior data D b , geographical location information G P and seasonal factor T S . The historical charging data H C is composed of fixed-period extraction of the interaction records of the charging pile communication equipment. The driving behavior data D b is acquired by the in-vehicle ECU, and the geographical location information G P is acquired by the vehicle's GPS module. The seasonal factor T S is obtained through the time and date information in the system;

[0015] Among them, the historical charging data H C includes charging time T, charging amount Q, charging power P C and charging station selection D; The driving behavior data D b includes average vehicle speed, driving duration and driving mode;

[0016] The data integration and storage unit processes the historical charging data H collected by the sensor data acquisition unit C , driving behavior data D b , geographical location information G P and seasonal factors T S to perform preliminary integration, forming the original data set W X .

[0017] Preferably, the data preprocessing module includes a data cleaning and denoising unit and a data normalization unit;

[0018] The data cleaning and denoising unit cleans and denoises the original data set W X , where cleaning includes missing value processing and denoising includes using a smoothing filter to remove noise;

[0019] Missing value processing performs missing value filling on the original data set W X to obtain the original filled data set W Xf , and then the original filled data set W Xf is denoised using a smoothing filter to obtain the original denoised data set W XS ;

[0020] The data normalization unit normalizes the original denoised data set W XS to obtain the multi-source data set W.

[0021] Preferably, the feature extraction and weight assignment module includes a feature extraction unit and a weight assignment unit;

[0022] The feature extraction unit extracts the historical charging data H from the multi-source data set W C , and extracts the historical charging data H in the multi-source data set W by using PCA C to obtain the features of the user's charging time T, charging amount Q, and charging station selection D, forming the multi-dimensional feature Z , and performs feature matrix transformation to obtain the feature matrix HZ:

[0023] The feature matrix HZ is transformed through the following transformation formula:

[0024] ;

[0025] In the formula, P represents the transformation matrix of PCA;

[0026] After the feature matrix HZ is linearly combined, the weighted feature matrix HZ is obtained:

[0027] HZ ;

[0028] In the formula, respectively represent the preset weight values of the charging time T, the charging amount Q, and the charging station selection D, and simultaneously preset the weight value of the charging time T , the preset weight value of the charging amount Q and the preset weight value of the charging station selection D , which form a weight value group .

[0029] Preferably, the personalized charging scheme generation module includes a scheme generation unit and a scheme optimization unit;

[0030] The scheme generation unit calculates the comprehensive score S of each region according to the weighted feature matrix HZ and the weight value group obtained by the feature extraction and weight assignment module , and obtains the optimal charging time T C , the optimal charging amount Q C and the optimal charging location D LOC ;

[0031] The comprehensive score S is obtained through the following formula:

[0032] S ;

[0033] In the formula, represents the i-th eigenvalue of the j-th region, and represent the parameters of the non-linear contribution of the feature, represents the influence of the square term describing the feature, represents the logarithmic change of the feature;

[0034] According to the comprehensive score S, obtain the time period with the highest score , the charging amount with the highest score and the charging location with the highest score :

[0035] , , ;

[0036] In the formula, respectively represent the scores of the charging time, the charging amount, and the charging location;

[0037] According to the time period with the highest score obtain the optimal charging time T C ;

[0038] The optimal charging time T C is obtained through the following formula:

[0039] TC ;

[0040] Where Pr represents the remaining power of the vehicle, and Ce represents the charging efficiency. represents the influence of seasons on the charging efficiency, with higher efficiency in summer and lower efficiency in winter.

[0041] According to the charging amount with the highest score Obtain the optimal charging amount Q C ;

[0042] Q C ;

[0043] Where represents the total capacity of the vehicle battery, soc represents the percentage of the remaining power, and its value range is from 0 to 1. represents the reserved power required for future driving.

[0044] According to the charging location with the highest score Obtain the optimal charging location D LOC ;

[0045] The said optimal charging location D LOC Is obtained through the following formula:

[0046] D LOC ;

[0047] Where D LOC Represents the optimal charging location, specifically referring to the recommended charging station. represents the distance between the user's location and the charging station. represents the load factor of the charging station. represents the adjustment factor. represents the minimum value function.

[0048] Preferably, the said scheme optimization unit adjusts the scheme according to the user's current requirements, external conditions, and the current battery state of the vehicle. The said scheme optimization unit adjusts the optimal charging time T C And the optimal charging location D LOC ;

[0049] The constraint formula is:

[0050] Optimal charging time T C ;

[0051] Optimal charging location D LOC ;

[0052] Where Represents the charging time range accepted by the user, Represents the charging location accepted by the user.

[0053] Preferably, the user feedback and learning module includes a feedback recording unit and a reinforcement learning optimization unit;

[0054] The feedback recording unit collects the user's feedback data and records the user's satisfaction after the user's charging is completed;

[0055] The user's satisfaction is represented by using the defined feedback score F;

[0056] The user's satisfaction F is obtained through the following formula:

[0057] F ;

[0058] In the formula, respectively represent the weight values of the charging time, the charging amount, and the charging location, R L represents the user's satisfaction score for the charging location, represents the difference between the optimal charging time T C and the actual charging time, represents the optimal charging amount Q C and the difference between the actual charging amount, t max represents the peak value of the charging time, Q max represents the peak value of the charging amount.

[0059] Preferably, the reinforcement learning optimization unit learns according to the feedback data by using the Q-learning algorithm and gradually adjusts the weight allocation strategy;

[0060] Using the Q-learning algorithm, define as the value of performing action a in state s, and increase the user's satisfaction F;

[0061] The formula of the Q-learning algorithm is:

[0062] ;

[0063] In the formula, s represents the current user charging demand and historical behavior, a represents the charging scheme recommended by the system, represents the learning rate, represents the discount factor, represents the next state after performing action a.

[0064] Preferably, the system monitoring and maintenance module includes a real-time monitoring unit and a maintenance optimization unit;

[0065] The real-time monitoring unit is responsible for real-time monitoring of parameters during the charging process, including current, voltage, charging power, charging speed, and temperature changes;

[0066] When the charging power and charging speed show abnormal fluctuations within a certain period of time, the alarm mechanism is triggered to send the abnormal information to the user.

[0067] A smart management system for electric vehicle charging based on multi-source data analysis. The maintenance and optimization unit responds according to the abnormal situations detected by the real-time monitoring unit, including adjusting the charging power, pausing the charging, and switching the charging mode.

[0068] The present invention provides a smart management system for electric vehicle charging based on multi-source data analysis, having the following beneficial effects:

[0069] (1) By extracting information from multi-source data including historical charging data H C , driving behavior data D b , geographical location information G P , and seasonal factors T S , and using the PCA feature extraction technology, the system can generate more personalized and accurate charging schemes according to the actual situations and preferences of different users. The system can dynamically adjust the charging time, charging amount, and location according to the specific charging habits, driving routes, and seasonal factors of the users, making the charging experience more intelligent and convenient, and improving the user satisfaction and usage experience.

[0070] (2) Through the sensor data acquisition unit, the system realizes the comprehensive acquisition of multi-dimensional data, including historical charging data H C , driving behavior data D b , geographical location information G P , and seasonal factors T S . Through the rich data dimensions, the system can more accurately understand the charging needs of users and dynamically adjust the charging scheme according to the actual environment and user behavior, significantly improving the accuracy and effectiveness of personalized recommendations. The data integration and storage unit preliminarily integrates the acquired multi-source data to form the original data set WX, laying a foundation for subsequent data analysis and processing. The introduction of the data preprocessing module, especially the design of the data cleaning and denoising unit and the data normalization unit, solves the problems of data noise interference and inconsistent data formats in traditional systems.

[0071] (3) Through the PCA technology for historical charging data H CFeature extraction is performed on the charging time T, charging quantity Q, and charging station selection D. The system converts high-dimensional data into multi-dimensional features Z and generates a feature matrix HZ. This method reduces data redundancy, removes irrelevant noise information, and retains key features, thus greatly improving the efficiency of data analysis. Compared with the traditional analysis mode based on raw data, PCA reduces the computational complexity and ensures more accurate charging scheme recommendations; the weight allocation unit linearly combines and weights the feature matrix HZ, enabling the charging time T, charging quantity Q, and charging station selection D to dynamically adjust the weight values according to the historical behaviors and preferences of different users.

[0072] (4)Calculate the comprehensive score S for different regions, enabling the system to provide personalized charging suggestions in different areas. In the formula, the non-linear contribution parameter and square term influence of eigenvalues are utilized, complex environmental factors during charging are considered, and the trend of charging behavior is accurately described through the logarithmic change of features. Compared with traditional methods, the comprehensive score S can reflect factors such as the real-time power load, weather, and user preferences in the region where the user is located, enhancing the adaptability and accuracy of the charging scheme in each region. Through the optimization of the charging time T C not only considers the battery capacity and charging efficiency but also combines the impact of seasonal factors on charging efficiency, avoiding power waste in extreme weather. At the same time, by calculating the remaining battery percentage and reserved power requirements of the vehicle, the system can recommend an optimal charging quantity Q that not only meets the driving requirements but also does not waste charging time C . Description of the Drawings

[0073] Figure 1 It is a schematic diagram of the block diagram process of an intelligent management system for electric vehicle charging based on multi-source data analysis according to the present invention. Detailed Embodiments

[0074] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0075] Embodiment 1

[0076] The present invention provides an intelligent management system for electric vehicle charging based on multi-source data analysis. Please refer to Figure 1 , which includes a data acquisition module, a data preprocessing module, a feature extraction and weight allocation module, a personalized charging scheme generation module, a user feedback and learning module, and a system monitoring and maintenance module;

[0077] The data acquisition module collects multi-source data in real time through sensors and in-vehicle devices, including the user's historical charging data H C , driving behavior data D b , geographical location information G P and seasonal factors T S , to form the original data set W X ;

[0078] The data preprocessing module cleans, denoises, and normalizes the original data set W X to obtain the multi-source data set W;

[0079] The feature extraction and weight assignment module extracts features including charging time T, charging amount Q, and charging station selection D features from the multi-source data set W by using PCA feature extraction technology;

[0080] The personalized charging scheme generation module generates a charging scheme according to the output of the weighted recommendation model and in combination with the user's charging requirements, including the optimal charging time T C , optimal charging amount Q C and optimal charging location D LOC ;

[0081] The user feedback and learning module records the user's feedback on the charging scheme and adjusts the weight assignment strategy;

[0082] The system monitoring and maintenance module monitors the status of the vehicle in real time during the charging process, detects abnormalities and faults, and conducts regular evaluations.

[0083] In this embodiment, by extracting information from multi-source data including historical charging data H C , driving behavior data D b , geographical location information G P and seasonal factors T S , and using PCA feature extraction technology, the system can generate more personalized and accurate charging schemes according to the actual situations and preferences of different users. The system can dynamically adjust the charging time, charging amount, and location according to the user's specific charging habits, driving routes, and seasonal factors, making the charging experience more intelligent and convenient, and improving the user's satisfaction and usage experience.

[0084] Through the user feedback and learning module, the system uses the user's feedback on the generated solutions to continuously adjust the weight allocation strategy and continuously optimize the charging solution recommendation using the reinforcement learning algorithm. The system monitoring and maintenance module ensures the stability of the entire system during operation, promptly discovers and solves potential anomalies or faults, and ensures that users can obtain stable services. The system comprehensively scores S for the charging solutions in each region. The system can combine multi-dimensional factors such as charging time T, charging quantity Q, and charging station selection D to provide personalized charging solutions for users in different regions, maximize the utilization of charging resources, avoid charging congestion during peak periods, improve charging efficiency, and save charging costs.

[0085] Embodiment 2

[0086] This embodiment is an explanatory description based on Embodiment 1. Please refer to Figure 1 , specifically: The data acquisition module includes a sensor data acquisition unit and a data integration and storage unit;

[0087] The sensor data acquisition unit acquires historical charging data H C , driving behavior data D b , geographical location information G P and seasonal factors T S . The historical charging data H C is composed of fixed-period extraction of the interaction records of the charging pile communication equipment. The driving behavior data D b is acquired by the in-vehicle ECU. The geographical location information G P is acquired by the vehicle's GPS module. The seasonal factors T S are obtained through the time and date information in the system;

[0088] Among them, the historical charging data H C includes charging time T, charging quantity Q, charging power P C and charging station selection D; The driving behavior data D b includes average vehicle speed, driving duration, and driving mode;

[0089] The data integration and storage unit preliminarily integrates the historical charging data H C , driving behavior data D b , geographical location information G P and seasonal factors T S collected by the sensor data acquisition unit to form the original data set W X . The data preprocessing module includes a data cleaning and denoising unit and a data normalization unit;

[0090] The data cleaning and denoising unit cleans and denoises the original data set W XPerform cleaning and denoising operations. Cleaning includes handling missing values, and denoising includes using a smoothing filter to remove noise;

[0091] The missing value handling is performed on the original dataset W X to perform missing value filling to obtain the original filled dataset W Xf and then, the original filled dataset W Xf is denoised using a smoothing filter to obtain the original denoised dataset W XS ;

[0092] The data normalization unit normalizes the original denoised dataset W XS to obtain the multi-source dataset W.

[0093] In this embodiment, through the sensor data acquisition unit, the system realizes the comprehensive acquisition of multi-dimensional data, including historical charging data H C , driving behavior data D b , geographical location information G P and seasonal factors T S . Through rich data dimensions, the system can more accurately understand the user's charging needs and dynamically adjust the charging plan according to the actual environment and user behavior, significantly improving the accuracy and effectiveness of personalized recommendations. The data integration and storage unit preliminarily integrates the collected multi-source data to form the original dataset WX, laying a foundation for subsequent data analysis and processing. The introduction of the data preprocessing module, especially the design of the "data cleaning and denoising unit" and the "data normalization unit", solves the problems of data noise interference and inconsistent data formats in traditional systems. Handling missing values and using a smoothing filter to remove noise ensure the integrity and reliability of the original data. Data normalization further eliminates the dimensional differences between different data types, enabling multi-source data to be analyzed and processed under the same standard, and improving the accuracy of data analysis results.

[0094] Embodiment 3

[0095] This embodiment is an explanatory description based on Embodiment 2. Please refer to Figure 1 , specifically: The feature extraction and weight assignment module includes a feature extraction unit and a weight assignment unit;

[0096] The feature extraction unit extracts the historical charging data H from the multi-source dataset W C , and by using PCA to extract the historical charging data H in the multi-source dataset W C , the features of the user's charging time T, charging amount Q, and charging station selection D are obtained, which form the multi-dimensional feature Z , and a feature matrix conversion is performed to obtain the feature matrix HZ:

[0097] The feature matrix HZ is transformed by the following transformation formula:

[0098] ;

[0099] where P represents the transformation matrix of PCA;

[0100] After the feature matrix HZ is linearly combined, the weighted feature matrix HZ is obtained:

[0101] HZ ;

[0102] where respectively represent the preset weight values of the charging time T, the charging amount Q, and the charging station selection D. The preset weight value of the charging time T is synchronized , the preset weight value of the charging amount Q and the preset weight value of the charging station selection D form a weight value group .

[0103] In this embodiment, the charging time T, the charging amount Q, and the charging station selection D of the historical charging data H C are subjected to feature extraction by PCA technology, and the system converts high-dimensional data into multi-dimensional features Z and generates a feature matrix HZ. This method reduces data redundancy, removes irrelevant noise information, and retains key features, thus greatly improving the efficiency of data analysis. Compared with the traditional analysis mode based on original data, PCA reduces the computational complexity and ensures more accurate charging scheme recommendations; the weight allocation unit makes the charging time T, the charging amount Q, and the charging station selection D dynamically adjust the weight values according to the historical behaviors and preferences of different users through linear combination and weighted processing of the feature matrix HZ. The feature matrix HZ is weighted by preset weight values to generate the weighted feature matrix HZ. This dynamic adjustment mechanism of the feature matrix can automatically optimize the recommendation model as the user behavior data changes, making the charging scheme generation more intelligent and adaptive.

[0104] Embodiment 4

[0105] This embodiment is an explanatory description based on Embodiment 3. Please refer to Figure 1 , specifically: the personalized charging scheme generation module includes a scheme generation unit and a scheme optimization unit;

[0106] The scheme generation unit calculates the comprehensive score S of each region according to the weighted feature matrix HZ and the weight group obtained by the feature extraction and weight allocation module, and obtains the optimal charging time T for different regions, and obtains the optimal charging amount Q C and the optimal charging location D C ​LOC ;

[0107] The comprehensive score S is obtained by the following formula:

[0108] S ;

[0109] In the formula, represents the i-th eigenvalue of the j-th region, and represent the parameters of the non-linear contribution of the feature, represents the influence of the square term of the feature description, represents the logarithmic change of the feature;

[0110] According to the comprehensive score S, obtain the time period with the highest score , the charging amount with the highest score and the charging location with the highest score :

[0111] , , ;

[0112] In the formula, respectively represent the scores of the charging time, the charging amount and the charging location;

[0113] According to the time period with the highest score obtain the optimal charging time T C ;

[0114] The optimal charging time T C is obtained by the following formula:

[0115] T C ;

[0116] In the formula, Pr represents the remaining battery power of the vehicle, Ce represents the charging efficiency, represents the influence of the season on the charging efficiency;

[0117] According to the charging amount with the highest score obtain the optimal charging amount Q C ;

[0118] Q C ;

[0119] In the formula, represents the total capacity of the vehicle battery, soc represents the percentage of the remaining battery power, with a value range of 0 to 1, represents the reserved battery power required for future driving;

[0120] According to the charging location with the highest score Obtain the best charging location D LOC ;

[0121] The said best charging location D LOC Is obtained through the following formula:

[0122] D LOC ;

[0123] In the formula, D LOC Represents the best charging location, specifically representing the recommended charging station, Represents the distance between the user's location and the charging station, Represents the load factor of the charging station, Represents the adjustment factor, Represents the minimum value function

[0124] In this embodiment, the comprehensive score S is calculated for different regions, enabling the system to provide personalized charging recommendations in different regions. In the formula, by using the non-linear contribution parameters and square term effects of the eigenvalues, the complex environmental factors during charging are considered, and the trend of charging behavior is accurately described through the characteristic logarithmic change. Compared with the traditional method, the comprehensive score S can reflect factors such as the real-time power load, weather, and user preferences in the user's area, improving the adaptability and accuracy of the charging scheme in each region. Through the optimization of the charging time T C by the formula, not only the battery power and charging efficiency are considered, but also the influence of seasonal factors on the charging efficiency is combined, avoiding power waste under extreme weather conditions. At the same time, by calculating the remaining battery percentage of the vehicle and the demand for reserved battery power , the system can recommend an optimal charging amount Q to the user that not only meets the driving demand but also does not waste charging time C . Compared with the traditional fixed charging amount recommendation, this embodiment is more flexible and energy-saving.

[0125] In terms of charging location recommendation, the system considers the distance between the user's current location and the charging station, the load factor of the charging station, and the adjustment factor through the D LOC formula, ensuring that the charging station selected by the user is both convenient and not overly crowded. At the same time, the introduction of the minimum value function ensures that the user can always select a charging station with the lowest load, avoiding the problem of excessive charging waiting time. Compared with the ordinary charging station recommendation method, this embodiment greatly improves the rationality of charging station selection and the user experience through the comprehensive calculation of the load factor and distance.

[0126] Embodiment 5

[0127] An intelligent management system for electric vehicle charging based on multi-source data analysis, please refer toFigure 1 , specifically: the solution optimization unit adjusts the solution according to the user's current requirements, external conditions, and the current battery state of the vehicle. The solution optimization unit adjusts the optimal charging time T C and the optimal charging location D LOC by using a constrained optimization algorithm;

[0128] The constraint formula is:

[0129] The optimal charging time T C ∈[T min , T max ;

[0130] The optimal charging location D LOC ∈[D LOC,min , D LOC,max ;

[0131] In the formula, T min , T max represents the range of charging times acceptable to the user, and D LOC,min , D LOC,max represents the charging locations acceptable to the user.

[0132] The user feedback and learning module includes a feedback recording unit and a reinforcement learning optimization unit;

[0133] The feedback recording unit collects the user's feedback data and records the user's satisfaction after the user completes charging;

[0134] The user's satisfaction is represented by defining a feedback score F;

[0135] The user's satisfaction F is obtained through the following formula:

[0136] F ;

[0137] In the formula, respectively represent the weight values of the charging time, charging amount, and charging location. R L represents the user's satisfaction score for the charging location, represents the difference between the optimal charging time T C and the actual charging time, represents the difference between the optimal charging amount Q C and the actual charging amount. t max represents the peak value of the charging time, and Q max represents the peak value of the charging amount.

[0138] The reinforcement learning optimization unit learns according to the feedback data by using the Q-learning algorithm and gradually adjusts the weight allocation strategy;

[0139] Using the Q-learning algorithm, define the value of executing action a in state s, and increase the user satisfaction F;

[0140] The formula of the Q-learning algorithm is:

[0141] ;

[0142] In the formula, s represents the current user charging demand and historical behavior, a represents the charging scheme recommended by the system, represents the learning rate, represents the discount factor, represents the next state after executing action a.

[0143] The system monitoring and maintenance module includes a real-time monitoring unit and a maintenance optimization unit;

[0144] The real-time monitoring unit is responsible for real-time monitoring of parameters during the charging process, including current, voltage, charging power, charging speed, and temperature changes;

[0145] When the charging power and charging speed fluctuate abnormally within a period of time, an alarm mechanism is triggered to send the abnormal information to the user.

[0146] The maintenance optimization unit responds according to the abnormal conditions detected by the real-time monitoring unit, including adjusting the charging power, pausing the charging, and switching the charging mode.

[0147] In this embodiment, through the solution optimization unit, the constraint optimization algorithm is used to adjust the optimal charging time T C and the charging location D LOC to ensure that the charging scheme better fits the user's current needs and acceptance range. This constraint optimization can dynamically balance the user's expectations and actual conditions, ensuring the efficiency and flexibility of the charging process. Through the user feedback and learning module, the system collects the user's feedback data after each charging, and uses the satisfaction score F to quantitatively evaluate the user's charging experience. The system not only records the user's feedback on the charging time, charging amount, and charging location, but also gradually adjusts the system's weight allocation strategy through the Q-learning algorithm in reinforcement learning.

[0148] The system monitoring and maintenance module monitors key parameters during the charging process through the real-time monitoring unit, including current, voltage, charging power, charging speed, and temperature changes, etc., and can detect abnormalities in a timely manner during the charging process and give early warnings. For example, when there are abnormal fluctuations in the charging power or speed, the system will automatically trigger the alarm mechanism and notify the user to ensure the safety of the charging process. In addition, based on the abnormal detection results of the real-time monitoring unit, the maintenance and optimization unit automatically executes corresponding response measures, including adjusting the charging power, pausing the charging, or switching the charging mode, greatly improving the safety and stability of the system and avoiding potential charging failures and risks.

[0149] Although the embodiments of the present invention have been shown and described, those of ordinary skill in the art can understand that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An intelligent management system for electric vehicle charging based on multi-source data analysis, characterized in that: It includes a data acquisition module, a data preprocessing module, a feature extraction and weight assignment module, a personalized charging plan generation module, a user feedback and learning module, and a system monitoring and maintenance module; The data acquisition module collects multi-source data in real time through sensors and in-vehicle devices, including the user's historical charging data H C , driving behavior data D b , geographical location information G P and seasonal factors T S , forming the original data set W X ; The data preprocessing module performs cleaning, denoising, and normalization on the original dataset W X to obtain a multi-source dataset W; The feature extraction and weight assignment module extracts features including charging time T, charging quantity Q, and charging station selection D from the multi-source data set W by using PCA feature extraction technology; The feature extraction and weight assignment module includes a feature extraction unit and a weight assignment unit; The feature extraction unit extracts historical charging data H from the multi-source dataset W C , and extracts the historical charging data H in the multi-source dataset W by using PCA C to obtain the features of the user's charging time T, charging amount Q, and charging station selection D, and form a multi-dimensional feature Z , and perform feature matrix conversion to obtain the feature matrix HZ: The feature matrix HZ is transformed by the following transformation formula: ； In the formula, P represents the transformation matrix of PCA; After the feature matrix HZ is linearly combined, the weighted feature matrix HZ is obtained: HZ ; Wherein, respectively represent the preset weight values of the charging time T, the charging amount Q, and the charging station selection D, and synchronously preset the weight value of the charging time T , the preset weight value of the charging amount Q and the preset weight value of the charging station selection D , to form a weight value group ; The personalized charging plan generation module generates a charging plan, including the optimal charging time T, according to the output of the weighted recommendation model and in combination with the user's charging requirements. C The optimal charging amount Q C and the optimal charging location D LOC ; The personalized charging plan generation module includes a plan generation unit and a plan optimization unit; The solution generation unit calculates the comprehensive score S for each region according to the weighted feature matrix HZ and the weight group obtained by the feature extraction and weight assignment module, and obtains the optimal charging time T , optimal charging amount Q C and optimal charging location D C for different regions; LOC ​ The comprehensive score S is obtained by the following formula: S ； In the formula, represents the i-th eigenvalue of the j-th region, and represents the parameter of the non-linear contribution of the feature, represents the influence of the square term describing the feature, represents the logarithmic change of the feature; Obtain the time period with the highest score, the charging amount with the highest score, and the charging location with the highest score according to the comprehensive score S , the charging amount with the highest score and the charging location with the highest score : ， ， ； In the formula, respectively represent the score of charging time, the score of charging amount, and the score of charging location; According to the time period with the highest score Obtain the optimal charging time T C ; The optimal charging time T C is obtained by the following formula: T C ； where Pr represents the remaining power of the vehicle, and Ce represents the charging efficiency, represents the influence of seasons on the charging efficiency; According to the charging amount of the highest score Obtain the optimal charging amount Q C ; Q C ； In the formula, represents the total capacity of the vehicle battery, soc represents the percentage of the remaining power, and the value range is from 0 to 1. represents the reserved power required for future driving. According to the charging location with the highest score Obtain the best charging location D LOC ; The optimal charging location D LOC is obtained through the following formula: D LOC ； Wherein, D LOC represents the optimal charging location, specifically representing the recommended charging station, represents the distance between the user's location and the charging station, represents the load factor of the charging station, represents the adjustment factor, represents the minimum value function; The user feedback and learning module records the user's feedback on the charging plan and adjusts the weight assignment strategy; The system monitoring and maintenance module monitors the state of the vehicle during charging in real time, detects abnormalities and faults, and conducts regular evaluations.

2. The intelligent management system for electric vehicle charging based on multi-source data analysis according to claim 1, wherein: The data acquisition module includes a sensor data acquisition unit and a data integration and storage unit; The sensor data acquisition unit collects historical charging data H C , driving behavior data D b , geographical location information G P and seasonal factor T S . The historical charging data H C is composed of fixed-period extraction of the interaction records of the charging pile communication equipment. The driving behavior data D b is collected by the in-vehicle ECU. The geographical location information G P is collected by the vehicle's GPS module. The seasonal factor T S is obtained through the time and date information in the system; Among them, the historical charging data H C includes charging time T, charging quantity Q, and charging power P C and charging station selection D; driving behavior data D b includes average vehicle speed, driving duration, and driving mode; The data integration and storage unit performs preliminary integration on the historical charging data H collected by the sensor data acquisition unit C , driving behavior data D b , geographical location information G P and seasonal factor T S to form an original data set W X .

3. The intelligent management system for electric vehicle charging based on multi-source data analysis according to claim 1, characterized in that: The data preprocessing module includes a data cleaning and denoising unit and a data normalization unit; The data cleaning and denoising unit performs cleaning and denoising processing on the original dataset W X The cleaning includes handling missing values, and the denoising includes removing noise using a smoothing filter; Missing value processing for the original dataset W X Perform missing value filling processing to obtain the original filled dataset W Xf , and then for the original filled dataset W Xf Use a smoothing filter for denoising processing to obtain the original denoised dataset W XS ; The data normalization unit normalizes the original denoised dataset W XS to obtain a multi-source dataset W.

4. The intelligent management system for electric vehicle charging based on multi-source data analysis according to claim 1, characterized in that: The solution optimization unit adjusts the solution according to the user's current requirements, external conditions, and the current battery state of the vehicle. The solution optimization unit adjusts the optimal charging time T C and the optimal charging location D LOC by using a constrained optimization algorithm; The constraint formula is: Optimal charging time T C ; Optimal charging location D LOC ; In the formula, represents the charging time range acceptable to the user, represents the charging location acceptable to the user.

5. The intelligent management system for electric vehicle charging based on multi-source data analysis according to claim 1, characterized in that: The user feedback and learning module includes a feedback recording unit and a reinforcement learning optimization unit; The feedback recording unit collects the user's feedback data and records the user's satisfaction after the user completes charging; The user's satisfaction is represented by using the defined feedback score F; The user's satisfaction F is obtained by the following formula: F ； Wherein, respectively represent the weight values of the charging time, the charging amount, and the charging location, R L represents the user's satisfaction score for the charging location, represents the difference between the optimal charging time T C and the actual charging time, represents the difference between the optimal charging amount Q C and the actual charging amount, t max represents the peak value of the charging time, Q max represents the peak value of the charging amount.

6. The intelligent management system for electric vehicle charging based on multi-source data analysis according to claim 5, characterized in that: The reinforcement learning optimization unit uses the Q-learning algorithm to learn according to the feedback data and gradually adjusts the weight assignment strategy; Using the Q-learning algorithm, define as the value of executing action a in state s, and increase the user satisfaction F; The formula of the Q-learning algorithm is: ； Where s represents the current user charging demand and historical behavior, and a represents the charging scheme recommended by the system. represents the learning rate, represents the discount factor, represents the next state after performing action a.

7. The intelligent management system for electric vehicle charging based on multi-source data analysis according to claim 1, characterized in that: The system monitoring and maintenance module includes a real-time monitoring unit and a maintenance optimization unit; The real-time monitoring unit is responsible for real-time monitoring of the parameters during charging, including current, voltage, charging power, charging speed, and temperature change; When the charging power and charging speed fluctuate abnormally within a period of time, the alarm mechanism is triggered and the abnormal information is sent to the user.

8. The intelligent management system for electric vehicle charging based on multi-source data analysis according to claim 7, characterized in that: The maintenance optimization unit responds according to the abnormalities detected by the real-time monitoring unit, including adjusting the charging power, pausing charging, and switching the charging mode.

Images