Aggregation display method and display platform of charging pile information

Abstract

The present invention proposes an aggregated display method and display platform for charging pile information, which belongs to the field of charging piles. The method includes: obtaining user behavior data and environmental data of the charging pile platform to obtain user needs; generating a scheduling strategy for the charging pile, and providing a personalized user demand display for the charging pile platform; weighted fusion of the state data of the charging pile and the environmental data; using the time series data as input to obtain a load forecast value; for each user, generating a personalized recommendation list for each user, collecting real-time user feedback on the charging piles recommended by the user after using the personalized recommendation list, and combining the personalized recommendation list to generate user satisfaction with the recommended charging piles. The present invention solves several core problems of the traditional charging pile information platform by fully combining artificial intelligence and distributed network technology, and greatly improves the intelligence, real-time performance and user experience of the platform.

Description

Technical Field

[0001] The present invention belongs to the field of charging piles, and in particular relates to an aggregation display method and a display platform for charging pile information. Background Art

[0002] With the rapid growth of the number of electric vehicles (EVs), charging piles, as the infrastructure for charging EVs, have become an indispensable part of the modern urban transportation system. In order to cope with the growing demand for charging, the number of charging piles has increased rapidly, especially in large cities and along highways, where charging piles have been distributed in many public areas. However, despite the increase in the number of charging piles, there are still many problems with their efficiency and the level of intelligence in the display of charging pile information, which need to be solved urgently.

[0003] Traditional charging pile information display platforms usually only display basic information such as the location and current status of charging piles through static maps, and do not fully consider the personalized needs of users, real-time environmental changes, and dynamic changes in charging pile resources. These traditional platforms are relatively weak in user experience and resource management. First, most of these platforms simply display the status of all charging piles in a certain area, without providing personalized recommendations based on user needs, behaviors, charging habits, etc., resulting in users often needing to spend a lot of time to screen suitable targets when looking for charging piles. Secondly, the load information of charging piles is often not fed back to users in a timely manner. Many platforms do not display the real-time load information of charging piles during peak hours, resulting in users often choosing charging piles with heavy loads, or even giving up due to long waiting times. Once a charging pile fails or is occupied, the existing platform lacks intelligent scheduling functions and cannot quickly guide users to other idle charging piles, resulting in a waste of charging resources and user distress.

[0004] In addition, existing charging pile information platforms usually rely on a centralized data management model, where all data is centrally managed and displayed by a central server. However, as the scale of the charging pile network expands, this centralized management model will not only face the pressure of processing a large amount of data, but may also cause information updates to lag due to network delays and data transmission bottlenecks, affecting user experience. Many platforms also lack real-time perception of the environment around the charging pile, such as the battery charging status of the charging pile, whether there are obstacles, temperature, etc. The lack of this information often affects the efficiency and safety of the charging pile.

[0005] Currently, the technology to solve these problems has not been fully studied and applied. Most charging pile information aggregation and display platforms remain at the primary stage of information display and cannot truly achieve intelligent charging pile management and dynamic scheduling. The matching of real-time dynamic information of charging pile resources and personalized needs of users is still a difficult problem that has not been effectively solved. Summary of the invention

[0006] The purpose of the present invention is to propose an aggregation display method and display platform for charging pile information. By fully combining artificial intelligence and distributed network technology, several core problems of traditional charging pile information platforms are solved, greatly improving the intelligence, real-time performance and user experience of the platform.

[0007] In order to achieve the above object, a first aspect of the present invention provides a method for displaying charging pile information, the method comprising:

[0008] S1. Obtain user behavior data and environmental data of the charging pile platform, pre-process the user behavior data and environmental data, build a user demand prediction model based on the long short-term memory network based on the pre-processed data, obtain user demand, perform demand clustering analysis on the users, obtain grouping processing results of different users, and finally generate a scheduling strategy for charging piles using a multi-objective optimization algorithm based on the user demand and the grouping processing results;

[0009] S2. Weighted fusion of the state data of the charging pile and the environmental data to form a state feature vector, and a set of state feature vectors as time series data;

[0010] S3. Based on the time series data, combined with historical load data, long short-term memory network is used to predict future The load at the moment is predicted based on the future The load forecast value at each moment is designed based on a scheduling strategy based on an optimization algorithm, which is used to achieve load balance and maximize the working efficiency of the charging pile;

[0011] S4. For each user, a charging demand model is established based on the user's historical user behavior data, and the user demand vector is output. , which is used to reflect the user's charging demand trend, and then according to the user demand vector Evaluate the suitability of each charging station and generate a personalized recommendation list for each user;

[0012] S5. Real-time user feedback on the charging piles recommended by users after using the personalized recommendation list is collected, and the user's satisfaction with the recommended charging piles is generated in combination with the personalized recommendation list. The charging demand model is updated based on the user's satisfaction with the recommended charging piles. At the same time, a periodic feedback integration mechanism is designed to combine historical data with real-time data by combining time windows, and long-term trends are obtained through periodic feedback. Finally, a final personalized recommended charging pile list is generated based on real-time user feedback and long-term trends.

[0013] Furthermore, the aggregate display method further includes designing an adaptive sampling mechanism to dynamically adjust the sampling frequency of the charging pile data according to the real-time load condition and environmental data of the charging pile, and the sampling frequency is also corrected and adjusted according to the ratio between the current load and the maximum load of the charging pile;

[0014] Among them, when the load is high or the grid load is large, the sampling frequency is increased to ensure the real-time nature of the data; and when the load is low, the sampling frequency is reduced to save computing resources and storage space.

[0015] Furthermore, the scheduling strategy includes the prediction of charging demand in each time period and the load scheduling arrangement of each charging pile; the user demand display includes the prediction of the time point and duration of future charging demand, the changing trend of user charging behavior and the dynamic scheduling strategy based on demand prediction; the user behavior data includes the charging time period, charging frequency, charging duration, charging location and charging type; the environmental data includes power load and environmental data;

[0016] The pre-processing comprises:

[0017] Z-score standardization is used to eliminate the influence of different feature dimensions;

[0018] A time period weight mapping method is designed to divide the charging period and map the weighted charging period value to obtain; wherein, the mapping rule is based on the weighted summation of the indicator function of the peak period and the indicator function of the non-peak period, wherein the weight of the peak period is greater than the weight of the non-peak period, which is used to enhance the characteristics of the charging period value.

[0019] Furthermore, the user demand prediction model is predicted by an adaptive regularization term, wherein the adaptive regularization term is an adaptive weight term, which is weighted according to the user behavior frequency;

[0020] The cluster analysis performs classification by minimizing the sum of the squares of the distances from each point to its cluster center, and combines the user's historical behavior frequency through adaptive weighted K-means to apply different weights to each point to reflect the intensity of his charging behavior.

[0021] Furthermore, the scheduling strategy of charging piles is generated by using a multi-objective optimization algorithm according to the user needs and the grouping processing results, specifically:

[0022] Based on the user demand and the grouping processing result, combined with the power grid load imbalance, a weighted sum of demand prediction error and load imbalance is minimized through a multi-objective optimization algorithm.

[0023] Furthermore, the S3 further includes introducing a dynamic adjustment constraint of electric power into the scheduling strategy:

[0024] If the grid load is too high, the charging power of the charging pile will be reduced; if the grid load is low, the charging pile will be set to increase the charging power without overloading;

[0025] The optimization goal of the scheduling strategy is to adjust the charging power of each charging pile , which minimizes the difference between the actual load and the predicted load of each charging pile, thereby improving the overall scheduling efficiency of the platform.

[0026] Furthermore, the charging piles in the personalized recommendation list are arranged in descending order according to the fitness scores, and the charging piles with the highest fitness are recommended first. If the user's needs change, the recommendation list will be adjusted according to the new user feature vector to ensure that the recommendation results are more personalized and accurate;

[0027] The charging demand model includes the user's charging frequency, preferred charging time period and charging duration;

[0028] The fitness score is calculated by the fitness function based on the current idle state of the charging pile, the load condition of the charging pile, the user demand vector and the distance between the user and the charging pile;

[0029] Among them, the idle state of the charging pile It directly determines whether the user can use the charging pile, and the load of the charging pile It reflects its current working status. The charging pile with the lowest load is more suitable for the user demand with the highest frequency. The user demand vector Then, by comparing the characteristic vector of the charging pile The inner product is used to measure the fitness, and the distance This affects the accessibility between the charging pile and the user; the fitness function is calculated by 、Charging pile load , User demand vector and distance The influence degree is weighted and calculated.

[0030] Furthermore, in S5, the real-time user feedback is smoothed using a weighted sliding average to ensure that the latest data has a larger weight.

[0031] Furthermore, the user's satisfaction with the recommended charging pile is obtained by weighted summing up the recommendation suitability and user feedback, wherein the weight of the weighted summing up is dynamically adjusted to reflect the importance of feedback and recommendation in different situations;

[0032] The charging demand model is updated based on the satisfaction at the same time, and the minimization of the loss function of the charging demand model is performed by the gradient descent method, wherein the charging demand model is continuously adjusted through model parameters to ensure that the predicted user satisfaction is closest to the actual user feedback.

[0033] In a second aspect of the present invention, a charging pile information aggregation display platform is provided, which is used to implement any of the above charging pile information aggregation display methods, and the platform includes:

[0034] A charging pile platform data acquisition unit is used to acquire user behavior data and environmental data of the charging pile platform, and pre-process the user behavior data and environmental data, build a user demand prediction model based on a long short-term memory network based on the pre-processed data, obtain user demand, and then perform demand clustering analysis on the user to obtain grouping processing results of different users, and finally generate a scheduling strategy for the charging piles using a multi-objective optimization algorithm according to the user demand and the grouping processing results;

[0035] A state feature vector acquisition unit is used to perform weighted fusion of the state data of the charging pile and the environmental data to form a state feature vector, and to form a state feature vector set as time series data;

[0036] A time series analysis unit is used to analyze the future load based on the time series data and the historical load data using a long short-term memory network. The load at the moment is predicted based on the future The load forecast value at each moment is designed based on a scheduling strategy based on an optimization algorithm, which is used to achieve load balance and maximize the working efficiency of the charging pile;

[0037] The user personalized analysis unit is used to establish a charging demand model for each user through the user's historical user behavior data and output the user demand vector , which is used to reflect the user's charging demand trend, and then according to the user demand vector Evaluate the suitability of each charging station and generate a personalized recommendation list for each user;

[0038] The charging pile information recommendation aggregation display unit is used to collect real-time user feedback on the charging piles recommended by the user after using the personalized recommendation list, and generate user satisfaction with the recommended charging piles in combination with the personalized recommendation list, and update the charging demand model based on the user's satisfaction with the recommended charging piles. At the same time, a periodic feedback integration mechanism is designed to combine historical data with real-time data by combining time windows, obtain long-term trends through periodic feedback, and finally generate a final personalized recommended charging pile list based on real-time user feedback and long-term trends.

[0039] The beneficial technical effects of the present invention are at least as follows:

[0040] (1) One of the core innovations of the present invention is the introduction of an adaptive user behavior analysis module, which builds a personalized user behavior model by analyzing the user's historical charging data, driving trajectory, charging time and other behavioral characteristics. The system can dynamically adjust the display content according to the user's charging habits, preferences and needs, and recommend the charging piles and charging time periods that best meet their needs. Unlike the traditional static display method, this method can prioritize the user's frequently used charging piles or recommend idle charging piles according to the user's real-time demand changes, avoiding users wasting time looking for suitable charging piles. This innovation solves the problem that personalized recommendations cannot be achieved in existing platforms, and improves the user's charging efficiency and experience.

[0041] (2) Another key innovation is that the present invention combines distributed sensor networks and edge computing technologies. By deploying a variety of sensors (such as temperature and humidity sensors, current sensors, obstacle monitoring, etc.) around each charging pile, the working status, surrounding environment and other key data of the charging pile are collected in real time. These sensor information is processed locally by edge computing devices, reducing the delay in data transmission and processing, and improving the real-time and accuracy of information. This innovation can perceive key factors such as the load status, charging efficiency, and ambient temperature of the charging pile in real time, and intelligently schedule the use of the charging pile based on this information, avoiding users from choosing overloaded charging piles during peak hours, thereby improving the efficiency of charging resource utilization and reducing user waiting time.

[0042] (3) The present invention also uses a deep learning algorithm to predict the load conditions of charging piles. Combined with data from a distributed sensor network, the platform can predict the load changes of charging piles in real time and perform dynamic scheduling. When the platform predicts that some charging piles are about to reach a high load, it will provide users with advance warnings and recommend charging piles with lower loads. Through this mechanism, the platform can intelligently schedule charging piles based on the real-time load and environmental conditions of the charging piles, so that charging pile resources can be used more efficiently, avoiding resource waste, and improving the user's charging experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] The present invention is further described using the accompanying drawings, but the embodiments in the accompanying drawings do not constitute any limitation to the present invention. A person skilled in the art can obtain other drawings based on the following drawings without creative work.

[0044] Figure 1 This is a flow chart of the method for displaying charging pile information in an aggregated manner according to the present invention.

[0045] Figure 2 This is a framework diagram of the charging pile information aggregation display platform of the present invention. DETAILED DESCRIPTION

[0046] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.

[0047] like Figure 1 As shown, the method for displaying the electric pile information provided by the embodiment of the present invention includes the following steps S1-S5:

[0048] S1. Obtain user behavior data and environmental data of the charging pile platform, and pre-process the user behavior data and environmental data. Based on the pre-processed data, a user demand prediction model based on a long short-term memory network is constructed to obtain user demand. Then, a demand clustering analysis is performed on the users to obtain grouping processing results of different users. Finally, a scheduling strategy for charging piles is generated using a multi-objective optimization algorithm based on the user demand and the grouping processing results.

[0049] Specifically, the user behavior data and environmental data of the charging pile platform are obtained as the basis for model training. User behavior data includes:

[0050] Charging period: The charging start and end time selected by the user.

[0051] Charging frequency: The periodicity of the user's charging (how many times a week, whether it is regular).

[0052] Charging time: The duration of each charge.

[0053] Charging location: The location of charging stations or charging piles frequently selected by users.

[0054] Equipment Type: The model and specifications of the electric vehicle.

[0055] Environmental data includes:

[0056] Power load: Real-time load conditions of the power grid.

[0057] Weather data: external factors that affect charging behavior, such as temperature, humidity, and wind speed.

[0058] Furthermore, to ensure that features of different dimensions are on the same scale, the original data is first standardized. For different input features, Z-score standardization can be used to eliminate the influence of different feature dimensions. The standardization formula is as follows:

[0059] ;

[0060] in, is the original data, is the mean, is the standard deviation.

[0061] Furthermore, the charging time periods are divided and mapped. In order to further improve the prediction accuracy of the model, especially in distinguishing between peak and off-peak periods, a time period weight mapping method can be designed. The specific mapping rules are as follows:

[0062] ;

[0063] in, is the indicator function, and is the weight coefficient of the time period, , which is used to reflect the different importance of peak and non-peak periods.

[0064] Furthermore, this step uses a long short-term memory network (LSTM) to model time series data. LSTM has a strong ability to model time series data and can capture the time series relationship in charging behavior. During training, the hidden state of LSTM is updated based on past information to provide a basis for predicting charging demand at the next moment. In order to further improve the performance of the model, an adaptive regularization term is introduced. This regularization term is dynamically adjusted according to the behavior of different users, thereby avoiding overfitting of a small number of high-frequency behavior users. The loss function formula is as follows:

[0065] ;

[0066] in, is the predicted value, is the true value, is the regularization coefficient, are the parameters of the LSTM model, It is an adaptive weight item, which is weighted according to the frequency of user behavior.

[0067] Furthermore, the adaptive weight term Defined as:

[0068] ;

[0069] in, is a hyperparameter, Is the user at time The frequency of behavior. High-frequency users will have smaller weights, reducing overfitting of their models.

[0070] Furthermore, based on the user needs obtained by the prediction model, demand clustering is performed so that different users can be grouped during display. Adaptive weighted K-means clustering is used here to cluster users with different needs to improve the accuracy of clustering. The traditional K-means clustering algorithm performs classification by minimizing the sum of the squares of the distance from each point to its cluster center. However, adaptive weighted K-means applies different weights to each point to reflect the intensity of its charging behavior. The objective function of weighted K-means is:

[0071] ;

[0072] in, is the adaptive weight, is the cluster center. Adaptive weight The design takes into account the historical behavior frequency of users and imposes smaller weights on high-frequency users to avoid excessive influence on clustering results.

[0073] Furthermore, a scheduling strategy for charging piles is generated based on the clustering results and the user's personalized needs. The strategy includes:

[0074] Forecast of charging demand for each period.

[0075] Load scheduling arrangement for each charging pile.

[0076] Furthermore, in order to further optimize the scheduling strategy, a multi-objective optimization algorithm is introduced to consider the charging demand while taking into account the balance of the grid load and the user's charging waiting time. The optimization objective function can be expressed by the following formula:

[0077] ;

[0078] in, For the Charging stations at the time Forecast demand, For actual needs, and To adjust the parameters, represents the load imbalance of the power grid, and the optimization objective is to minimize the weighted sum of demand forecast error and load imbalance.

[0079] Furthermore, through clustering and prediction results, the charging pile platform is provided with personalized user demand display, including:

[0080] Prediction of the timing and duration of future charging needs.

[0081] Changing trends in user charging behavior.

[0082] Dynamic scheduling strategy based on demand forecast.

[0083] At the same time, the platform will provide real-time feedback to the platform based on changes in actual user behavior, adjust the prediction model and clustering algorithm to improve the accuracy of future predictions. The feedback mechanism is implemented through the following formula:

[0084] ;

[0085] in, is the feedback coefficient, To predict demand, To meet actual needs, the feedback mechanism can adjust the prediction results of future models according to users' actual behavior and enhance the platform's adaptive capabilities.

[0086] By combining the time series model (LSTM) and the adaptive weighted K-means clustering algorithm, this solution can accurately capture the user's charging demand pattern and optimize the scheduling strategy of charging piles through a multi-objective optimization algorithm.

[0087] S2. Perform weighted fusion on the state data of the charging pile and the environmental data to form a state feature vector, and form a state feature vector set as time series data.

[0088] Specifically, various parameters of charging piles are monitored in real time through integrated sensors and smart devices to ensure timely acquisition of all data. Sensors continuously collect status information such as battery power and temperature, and environmental sensors obtain meteorological data, surrounding power grid load and other information. These real-time data will be provided to the platform for further analysis and processing.

[0089] Furthermore, the data sampling frequency is dynamically adjusted according to the real-time load of the charging pile and the external environment (such as grid load and meteorological conditions). When the load is high or the grid load is large, the sampling frequency is increased to ensure the real-time nature of the data; and when the load is low, the sampling frequency is reduced to save computing resources and storage space. Sampling frequency According to the current load of the charging pile With maximum load The specific formula is as follows:

[0090] ;

[0091] in, and are the minimum and maximum values ​​of the sampling frequency, respectively. is the current charging pile load, The maximum load of the charging pile.

[0092] Furthermore, the operation of the charging pile is not only affected by the load, but also by external factors such as grid load, temperature and meteorological conditions. Therefore, it is necessary to weight the external environmental data. (such as grid load, temperature, etc.) and a weighting factor Multiply them together to get the weighted value of the impact of the environmental feature on the charging pile. Weighting coefficient The calculation formula is as follows:

[0093] ;

[0094] in, For the The weighting coefficient of each environmental feature, and is the adjustment coefficient, is the environmental characteristic at the moment This formula can dynamically adjust the weight of each feature according to environmental changes.

[0095] Furthermore, the state data of the charging pile (such as power, battery temperature, and load) is weighted and fused with the external environmental characteristic data (such as grid load and meteorological data) to form a unified state feature vector , which is used for subsequent scheduling optimization. The calculation formula of the composite state eigenvector is as follows:

[0096] ;

[0097] in, Indicates characteristics (such as charging pile load, temperature, etc.), is the weight coefficient of the feature, By using this weighted fusion method, the present invention can dynamically adjust the feature vector to adapt to the current actual situation of the charging pile.

[0098] Furthermore, at each moment, by comparing the predicted load and actual load , if there is a deviation, the platform will send a feedback adjustment signal. The calculation formula of the feedback mechanism is as follows:

[0099] ;

[0100] in, is the feedback signal, is the predicted load, is the actual load, is the feedback adjustment coefficient. The feedback signal will be used to adjust future load forecasts or charging scheduling strategies to improve scheduling accuracy.

[0101] Furthermore, all collected data and generated feature vectors are stored in real time and synchronized to the cloud for subsequent historical analysis and long-term scheduling optimization. Through multi-point storage and synchronization, the integrity and availability of data are ensured, while ensuring the real-time nature of the data.

[0102] In this step, the present invention ensures that the charging pile status and environmental data can accurately and in real time reflect the current operating conditions by designing an adaptive sampling mechanism, environmental weighted feature mapping, composite state feature vectors, and real-time feedback mechanisms. These innovative methods not only improve the efficiency of data collection, but also dynamically adapt to different environmental conditions and load states, providing high-quality data input for subsequent scheduling optimization and demand forecasting. In this way, the present invention can effectively improve the accuracy and responsiveness of charging pile scheduling.

[0103] S3. Based on the time series data, combined with historical load data, long short-term memory network is used to predict future The load at the moment is predicted based on the future The load forecast value at each moment is designed based on a scheduling strategy of an optimization algorithm, which is used to achieve load balance and maximize the working efficiency of the charging pile.

[0104] Specifically, the present invention uses the charging pile status data obtained in step S2 The load forecasting and scheduling decision are made based on the characteristics of the environmental data and historical load data. First, the load forecasting is based on the time series data of the charging piles, so the present invention adopts the time series model to predict the future load. Given the time series data obtained in step 2 (including charging pile status and environmental data), the present invention uses a trained LSTM (long short-term memory network) model to predict future Through the time-dependent characteristics of the LSTM model, the model can effectively capture the patterns and trends of historical loads and generate future load forecasts. , specifically:

[0105] ;

[0106] in, Represents the mapping function obtained by the LSTM model after training, is the historical data input, and is the predicted future load. Here, The time step for prediction is usually 1 hour or can be selected according to actual needs.

[0107] Next, use this load forecast value As input, the present invention enters the link of intelligent scheduling. The scheduling goal is to adjust the power output of the charging pile according to the real-time load of the power grid to avoid the situation where the power grid is overloaded, and at the same time to improve the utilization rate of the charging pile. To this end, the present invention designs a scheduling strategy based on an optimization algorithm. Specifically, the charging power of each charging pile needs to be adjusted in real time during the scheduling process. ( represents the charging pile number) to achieve load balance across the entire platform and maximize the efficiency of the charging piles. This problem can be modeled as a constrained optimization problem, the goal of which is to minimize the total load deviation of the platform. The optimization problem can be expressed as:

[0108] ;

[0109] in, It is a charging station The charging power, It is The load forecast value at the time, is the number of charging piles. The optimization goal is to adjust the charging power of each charging pile , so that the actual load of each charging pile is consistent with the predicted load The differences between them are minimized, thereby improving the overall scheduling efficiency of the platform.

[0110] At the same time, in order to ensure the safe operation of the power grid during periods of high or low load, the present invention introduces dynamic adjustment constraints on charging power. If the power grid load is too high, the dispatching platform will reduce the charging power of the charging pile; if the power grid load is low, the charging pile can increase the charging power without overloading, thereby improving charging efficiency.

[0111] Furthermore, in order to further improve the accuracy and efficiency of scheduling, the present invention also designs a feedback scheduling mechanism. Specifically, after each scheduling, the platform will monitor the actual load of each charging pile. and predicted load This evaluation process can be done by using the error To express:

[0112] ;

[0113] When the error When the power output is too high, the platform will adjust the power output of the charging pile. To correct the load forecasting model and dispatching strategy. This adjustment is dynamic, and the dispatching plan is continuously optimized based on real-time data to ensure that each dispatch can better adapt to changes in grid load and fluctuations in charging demand.

[0114] Through this dynamic feedback scheduling mechanism, the platform can flexibly respond to changes in demand under different grid load conditions, improve the efficiency of charging piles, ensure grid load balance, and further enhance the overall intelligent level of charging services. This solution fully combines charging pile load prediction, intelligent scheduling and real-time feedback mechanism, so that charging piles can be flexibly scheduled according to predicted and real-time load conditions, so that they can still operate stably and efficiently when the grid is under great pressure or charging demand surges.

[0115] S4. For each user, a charging demand model is established based on the user's historical user behavior data, and the user demand vector is output. , which is used to reflect the user's charging demand trend, and then according to the user demand vector Evaluate the suitability of each charging station and generate a personalized recommendation list for each user.

[0116] Specifically, in the step of personalized charging pile recommendation, the goal is to recommend the most suitable charging pile for each user to ensure the efficiency and convenience of the charging process. The input of this step comes from the load forecast data in step 3. , and the user's historical behavior data ,This information will be combined with the real-time status of the charging pile and a personalized recommendation list will be generated through the ,recommendation algorithm.

[0117] First, for each user, the present invention uses historical behavior data Building a charging demand model , which includes information such as the user's charging frequency, preferred charging time, and charging duration. After analysis, these historical data will be converted into a user feature vector , which can accurately reflect the user's charging demand trend. On this basis, the present invention evaluates the adaptability of each charging pile to provide personalized recommendations for users.

[0118] Furthermore, during the recommendation process, the charging pile's suitability score The calculation of adaptability requires a combination of multiple factors, including the current idle state of the charging station. , Load conditions of charging piles , User demand vector And the distance between the user and the charging station These factors are weighted by a function to integrate, where the weight of each factor It is dynamically adjusted by the platform during actual operation. The calculation formula for the fitness score is:

[0119] ;

[0120] in, is a function that represents the impact of different features on the adaptability of the charging pile. The present invention weights these features according to their importance. For example, the idle state of the charging pile It directly determines whether the user can use the charging pile, and the load of the charging pile It reflects its current working status. Charging piles with lower load are more suitable for high-frequency user needs. User demand vector Then, by comparing the characteristic vector of the charging pile The inner product is used to measure the fitness, and the distance This affects the accessibility between the charging pile and the user.

[0121] Furthermore, in order to further refine the recommendation results, the fitness function It can be formalized as:

[0122] ;

[0123] In this formula, The influence of idle state, load, user demand matching and distance are controlled respectively. By weighting and synthesizing these factors, the present invention obtains the adaptability score of each charging pile. For example, when When it is higher, it means that the charging pile is relatively idle, so the adaptability score is higher; When it is higher, it means that the charging pile is heavily loaded and the adaptability score will be lower.

[0124] Furthermore, based on the suitability scores of charging piles, the platform generates a personalized recommendation list for each user. The charging piles in this recommendation list are ranked by suitability scores. In descending order, the platform recommends charging piles with higher adaptability first. If the user's needs change, such as the charging time or location, the platform will recommend charging piles based on the new demand vector. Adjust the recommendation list to ensure that the recommendation results are more personalized and accurate.

[0125] In addition, in order to continuously optimize the recommendation effect, the platform also adopts a real-time learning mechanism. Specifically, every time a user makes a choice (for example, choosing a charging station for charging), the platform will feed this choice back to the model to update the charging demand model. Hyperparameters in the calculation formula for charging pile compatibility This adaptive adjustment mechanism enables the platform to continuously improve the accuracy of recommendations over time.

[0126] Ultimately, based on load forecasts, historical behavior data, and real-time status of charging piles, the recommendation platform can generate the most suitable charging pile recommendation list for each user, providing a personalized and efficient charging experience.

[0127] S5. Real-time user feedback on the charging piles recommended by users after using the personalized recommendation list is collected, and the user's satisfaction with the recommended charging piles is generated in combination with the personalized recommendation list. The charging demand model is updated based on the user's satisfaction with the recommended charging piles. At the same time, a periodic feedback integration mechanism is designed to combine historical data with real-time data by combining time windows, and long-term trends are obtained through periodic feedback. Finally, a final personalized recommended charging pile list is generated based on real-time user feedback and long-term trends.

[0128] Specifically, the input data comes from the personalized recommendation results in the previous stage And user feedback .here, It is a recommendation list generated by the charging pile recommendation platform based on user needs. It is the actual feedback from users on the charging piles, including charging time, power, satisfaction and other data. Feedback data Smoothing is required to remove the impact of outliers and bursty data and to ensure that the latest feedback information can quickly update the platform.

[0129] Use a weighted sliding average to smooth user feedback, ensuring that the latest data has a larger weight:

[0130] ;

[0131] in, is the smoothing factor, is the original feedback data, Control the update rate of feedback. This method ensures that the platform's response speed to feedback data is balanced with historical data.

[0132] Furthermore, the weights of user satisfaction and recommendation suitability are calculated:

[0133] The platform combines real-time user feedback with personalized recommendation results to calculate user satisfaction with the recommended charging piles. . The user's feedback is weighted by the following formula: and recommended results The degree of influence on final satisfaction is adjustable:

[0134] ;

[0135] in, and The weights of recommendation fitness and user feedback are controlled respectively. As the platform runs, and The value of will be adjusted dynamically to reflect the importance of feedback and recommendations in different situations. For example, when there is a charging pile failure or frequent user complaints, the platform will automatically increase The weight of the system is given, and more emphasis is placed on user feedback.

[0136] Furthermore, the model is updated based on satisfaction:

[0137] In order to continuously optimize the charging pile recommendation platform in actual operation, the present invention updates the model parameters by minimizing the following loss function, so that the charging piles recommended by the platform are more in line with the actual needs of users. The loss function consists of two parts: recommendation accuracy and feedback satisfaction:

[0138] ;

[0139] in, is the satisfaction predicted based on the current model, is the weight matrix of the model, is the regularization coefficient, which is used to avoid overfitting. The loss function is minimized by the gradient descent method, where the model parameters Continuously adjust to ensure predicted user satisfaction Try to get as close to real user feedback as possible .

[0140] Furthermore, periodic feedback integration and model stability:

[0141] In long-term operation, in order to prevent the model from over-relying on short-term feedback, this paper designs a periodic feedback integration mechanism. By introducing a time window, historical data is combined with real-time data to obtain more accurate long-term trends. The specific periodic feedback model is:

[0142] ;

[0143] in, is the time window size, Indicates the past The model can effectively capture the changes in user behavior over a long period of time, such as the impact of seasonal fluctuations or special events (such as holidays) on the frequency of charging pile usage.

[0144] When updating the model, the model is periodically fed back With real-time data Combined, the next step is to optimize the recommendation model:

[0145] ;

[0146] in, and They are the weighted coefficients of real-time data and long-term trends. By integrating the two, the recommendation platform can avoid over-reliance on short-term fluctuations while ensuring the guiding role of long-term trends.

[0147] Furthermore, the final optimization and personalized charging pile recommendation:

[0148] Based on real-time feedback and periodic integration model optimization, the platform will eventually generate a personalized list of recommended charging stations. , dynamically adjust the recommendation strategy based on the user’s needs and historical behavior. For example, if a user frequently charges in a specific area and has a high rating, the platform will recommend charging piles in that area and avoid recommending users to charging piles with low usage rates.

[0149] At this time, through the platform's continuous feedback mechanism, the charging pile recommendation model will be continuously updated to reflect changes in user behavior and the usage status of the charging piles in real time, thereby maximizing the efficiency of charging pile use and user satisfaction.

[0150] like Figure 2 As shown, in another embodiment of the present invention, a charging pile information aggregation display platform is provided, which is used to implement any of the above charging pile information aggregation display methods, and the platform includes:

[0151] The charging pile platform data acquisition unit 301 is used to acquire user behavior data and environmental data of the charging pile platform, and pre-process the user behavior data and environmental data, and build a user demand prediction model based on the long short-term memory network based on the pre-processed data to obtain user demand, and then perform demand clustering analysis on the user to obtain the grouping processing results of different users, and finally generate a scheduling strategy for the charging piles using a multi-objective optimization algorithm according to the user demand and the grouping processing results;

[0152] The state feature vector acquisition unit 302 is used to perform weighted fusion of the state data of the charging pile and the environmental data to form a state feature vector, and to form a state feature vector set as time series data;

[0153] The time series analysis unit 303 is used to use a long short-term memory network to analyze the future based on the time series data and historical load data. The load at the moment is predicted based on the future The load forecast value at each moment is designed based on a scheduling strategy based on an optimization algorithm, which is used to achieve load balance and maximize the working efficiency of the charging pile;

[0154] The user personalized analysis unit 304 is used to establish a charging demand model for each user through the user's historical user behavior data and output a user demand vector , which is used to reflect the user's charging demand trend, and then according to the user demand vector Evaluate the suitability of each charging station and generate a personalized recommendation list for each user;

[0155] The charging pile information recommendation aggregation display unit 305 is used to collect real-time user feedback on the charging piles recommended by the user after using the personalized recommendation list, and generate user satisfaction with the recommended charging piles in combination with the personalized recommendation list, and update the charging demand model based on the user's satisfaction with the recommended charging piles. At the same time, a periodic feedback integration mechanism is designed to combine historical data with real-time data by combining time windows, obtain long-term trends through periodic feedback, and finally generate a final personalized recommended charging pile list based on real-time user feedback and long-term trends.

[0156] It should be noted that the workflow described above is merely illustrative and does not limit the scope of protection of the present invention. In practical applications, technicians in this field can select part or all of them according to actual needs to achieve the purpose of the present embodiment, and no limitation is made here.

[0157] In addition, for technical details that are not described in detail in this embodiment, reference can be made to the parameter operation method provided in any embodiment of the present invention, and will not be repeated here.

[0158] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or system including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or system. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or system including the element.

[0159] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0160] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as a read-only memory / random access memory, a magnetic disk, or an optical disk), and includes a number of instructions for enabling a terminal device (which can be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to execute the methods described in each embodiment of the present invention.

[0161] The above are only preferred embodiments of the present invention, and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A method for displaying charging pile information, characterized in that: The method comprises: S1. Obtain user behavior data and environmental data of the charging pile platform, pre-process the user behavior data and environmental data, build a user demand prediction model based on the long short-term memory network based on the pre-processed data, obtain user demand, perform demand clustering analysis on the users, obtain grouping processing results of different users, and finally generate a scheduling strategy for charging piles using a multi-objective optimization algorithm based on the user demand and the grouping processing results; S2. Weighted fusion of the state data of the charging pile and the environmental data to form a state feature vector, and a set of state feature vectors as time series data; S3. Based on the time series data, combined with historical load data, long short-term memory network is used to predict future The load at the moment is predicted based on the future The load forecast value at each moment is designed based on a scheduling strategy based on an optimization algorithm, which is used to achieve load balance and maximize the working efficiency of the charging pile; S4. For each user, a charging demand model is established based on the user's historical user behavior data, and the user demand vector is output. , which is used to reflect the user's charging demand trend, and then according to the user demand vector Evaluate the suitability of each charging station and generate a personalized recommendation list for each user; S5. Real-time user feedback on the charging piles recommended by the user after using the personalized recommendation list is collected in real time, and the user's satisfaction with the recommended charging piles is generated in combination with the personalized recommendation list, and the charging demand model is updated based on the user's satisfaction with the recommended charging piles. At the same time, a periodic feedback integration mechanism is designed to combine historical data with real-time data by combining time windows, and long-term trends are obtained through periodic feedback. Finally, a final personalized recommended charging pile list is generated based on real-time user feedback and long-term trends; The aggregate display method also includes designing an adaptive sampling mechanism to dynamically adjust the sampling frequency of the charging pile data according to the real-time load condition and environmental data of the charging pile, and the sampling frequency is also corrected and adjusted according to the ratio between the current load and the maximum load of the charging pile; Among them, when the load is high or the grid load is large, the sampling frequency is increased to ensure the real-time nature of the data; while when the load is low, the sampling frequency is reduced to save computing resources and storage space; The scheduling strategy includes the prediction of charging demand in each time period and the load scheduling arrangement of each charging pile; the user demand display includes the prediction of the time point and duration of future charging demand, the changing trend of user charging behavior and the dynamic scheduling strategy based on demand prediction; the user behavior data includes the charging time period, charging frequency, charging duration, charging location and charging type; the environmental data includes power load and environmental data; The pre-processing comprises: Z-score standardization is used to eliminate the influence of different feature dimensions; A time period weight mapping method is designed to divide the charging period and map the weighted charging period value to obtain; wherein the mapping rule is based on the weighted sum of the indicator function of the peak period and the indicator function of the non-peak period, wherein the weight of the peak period is greater than the weight of the non-peak period, which is used to enhance the characteristics of the charging period value; The user demand prediction model is predicted by an adaptive regularization term, wherein the adaptive regularization term is an adaptive weight term, which is weighted according to the user behavior frequency; The cluster analysis performs classification by minimizing the sum of the squares of the distances from each point to its cluster center, and at the same time, combines the user's historical behavior frequency through adaptive weighted K-means to apply different weights to each point to reflect the intensity of his charging behavior; The S3 also includes introducing a dynamic adjustment constraint of electric power into the scheduling strategy: If the grid load is too high, the charging power of the charging pile will be reduced; if the grid load is low, the charging pile will be set to increase the charging power without overloading; The optimization goal of the scheduling strategy is to adjust the charging power of each charging pile , which minimizes the difference between the actual load and the predicted load of each charging pile, thereby improving the overall scheduling efficiency of the platform.

2. The method for displaying charging pile information according to claim 1, characterized in that: The scheduling strategy of the charging piles is generated by using a multi-objective optimization algorithm according to the user needs and the grouping processing results, specifically: Based on the user demand and the grouping processing result, combined with the power grid load imbalance, a weighted sum of demand prediction error and load imbalance is minimized through a multi-objective optimization algorithm.

3. The method for displaying charging pile information according to claim 1, characterized in that: The charging piles in the personalized recommendation list are arranged in descending order according to the adaptability score, and the charging piles with the highest adaptability are recommended first. If the user's needs change, the recommendation list will be adjusted according to the new user feature vector to ensure that the recommendation results are more personalized and accurate; The charging demand model includes the user's charging frequency, preferred charging time period and charging duration; The fitness score is calculated by the fitness function based on the current idle state of the charging pile, the load condition of the charging pile, the user demand vector and the distance between the user and the charging pile; Among them, the idle state of the charging pile It directly determines whether the user can use the charging pile, and the load of the charging pile It reflects its current working status. The charging pile with the lowest load is more suitable for the user needs with the highest frequency. User demand vector Then, by comparing the characteristic vector of the charging pile The inner product is used to measure the fitness, and the distance This affects the accessibility between the charging pile and the user; the fitness function is calculated by 、Charging pile load , User demand vector and distance The influence degree is weighted and calculated.

4. The method for displaying charging pile information according to claim 1, characterized in that: In S5, the real-time user feedback is smoothed using a weighted sliding average to ensure that the latest data has a larger weight.

5. The method for displaying charging pile information according to claim 3 or 4, characterized in that: The user's satisfaction with the recommended charging pile is obtained by weighted summing up the recommendation suitability and user feedback, wherein the weight of the weighted summing up is dynamically adjusted to reflect the importance of feedback and recommendations in different situations; The charging demand model is updated based on the satisfaction at the same time, and the minimization of the loss function of the charging demand model is performed by the gradient descent method, wherein the charging demand model is continuously adjusted through model parameters to ensure that the predicted user satisfaction is closest to the actual user feedback.

6. The charging pile information aggregation display platform is characterized by: The platform is used to implement the charging pile information aggregation display method according to any one of claims 1 to 5, comprising: A charging pile platform data acquisition unit is used to acquire user behavior data and environmental data of the charging pile platform, and pre-process the user behavior data and environmental data, build a user demand prediction model based on a long short-term memory network based on the pre-processed data, obtain user demand, and then perform demand clustering analysis on the user to obtain grouping processing results of different users, and finally generate a scheduling strategy for the charging piles using a multi-objective optimization algorithm according to the user demand and the grouping processing results; A state feature vector acquisition unit is used to perform weighted fusion of the state data of the charging pile and the environmental data to form a state feature vector, and to form a state feature vector set as time series data; A time series analysis unit is used to analyze the future load based on the time series data and the historical load data using a long short-term memory network. The load at the moment is predicted based on the future The load forecast value at each moment is designed based on a scheduling strategy based on an optimization algorithm, which is used to achieve load balance and maximize the working efficiency of the charging pile; The user personalized analysis unit is used to establish a charging demand model for each user through the user's historical user behavior data and output the user demand vector , which is used to reflect the user's charging demand trend, and then according to the user demand vector Evaluate the suitability of each charging station and generate a personalized recommendation list for each user; The charging pile information recommendation aggregation display unit is used to collect real-time user feedback on the charging piles recommended by the user after using the personalized recommendation list, and generate the user's satisfaction with the recommended charging piles in combination with the personalized recommendation list, and update the charging demand model based on the user's satisfaction with the recommended charging piles. At the same time, a periodic feedback integration mechanism is designed to combine historical data with real-time data by combining time windows, obtain long-term trends through periodic feedback, and finally generate a final personalized recommended charging pile list based on real-time user feedback and long-term trends; The aggregate display method also includes designing an adaptive sampling mechanism to dynamically adjust the sampling frequency of the charging pile data according to the real-time load condition and environmental data of the charging pile, and the sampling frequency is also corrected and adjusted according to the ratio between the current load and the maximum load of the charging pile; Among them, when the load is high or the grid load is large, the sampling frequency is increased to ensure the real-time nature of the data; while when the load is low, the sampling frequency is reduced to save computing resources and storage space; The scheduling strategy includes the prediction of charging demand in each time period and the load scheduling arrangement of each charging pile; the user demand display includes the prediction of the time point and duration of future charging demand, the changing trend of user charging behavior and the dynamic scheduling strategy based on demand prediction; the user behavior data includes the charging time period, charging frequency, charging duration, charging location and charging type; the environmental data includes power load and environmental data; The pre-processing comprises: Z-score standardization is used to eliminate the influence of different feature dimensions; A time period weight mapping method is designed to divide the charging period and map the weighted charging period value to obtain; wherein the mapping rule is based on the weighted sum of the indicator function of the peak period and the indicator function of the non-peak period, wherein the weight of the peak period is greater than the weight of the non-peak period, which is used to enhance the characteristics of the charging period value; The user demand prediction model is predicted by an adaptive regularization term, wherein the adaptive regularization term is an adaptive weight term, which is weighted according to the user behavior frequency; The cluster analysis performs classification by minimizing the sum of the squares of the distances from each point to its cluster center, and at the same time, combines the user's historical behavior frequency through adaptive weighted K-means to apply different weights to each point to reflect the intensity of his charging behavior; The timing analysis unit further includes introducing a dynamic adjustment constraint of electric power into the scheduling strategy: If the grid load is too high, the charging power of the charging pile will be reduced; if the grid load is low, the charging pile will be set to increase the charging power without overloading; The optimization goal of the scheduling strategy is to adjust the charging power of each charging pile , which minimizes the difference between the actual load and the predicted load of each charging pile, thereby improving the overall scheduling efficiency of the platform.

Images