Charging pile management method and system based on data analysis

Abstract

The invention discloses a charging pile management method and system based on data analysis, and belongs to the field of charging management.The charging pile management method based on data analysis comprises the following steps that a time sequence prediction model is constructed based on historical charging session data, environmental factors, special time period information and the real-time occupancy state of a charging pile, predicting peak periods and valley periods of charging demands in different areas from a few hours to one day in the future; the method has the beneficial effects that the charging service price is dynamically adjusted by predicting the charging demand peak period and the charging demand valley period and combining the overall utilization rate prediction result of the charging piles in the region in the period, so that the regional charging price in the charging demand peak period is relatively high, and the regional charging price in the charging demand valley period is relatively low; therefore, the user is guided to actively perform peak shifting charging or select other areas for charging, the load distribution of the charging piles is optimized, the queuing pressure in the peak period of the charging demand is relieved, and the infrastructure utilization rate in the valley period of the charging demand is improved.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of charging management, and particularly relates to a charging pile management method and system based on data analysis. BACKGROUND

[0002] In the current charging pile management system, since the charging time of users is generally concentrated after work on weekdays (such as 18:00-21:00), the use of charging piles presents obvious peak-valley characteristics: charging piles are in short supply during peak periods, while a large number of charging piles are idle during valley periods (such as during the day on weekdays, at night, and during holidays). This imbalance between supply and demand periods causes double resource waste: it not only exacerbates the queuing pressure during peak periods, but also leads to low utilization of infrastructure during valley periods, which needs to be improved. SUMMARY

[0003] Therefore, it is necessary to provide a charging pile management method and system based on data analysis in view of the above problems.

[0004] The embodiment of the present application is implemented in the following manner: a charging pile management method based on data analysis comprises the following steps:

[0005] Based on historical charging session data (time stamp, geographic location, charging amount), environmental factors (weather conditions), special period information (holidays, surrounding large-scale activities), and real-time occupancy status of charging piles, a time series prediction model is constructed to predict the charging demand peak and valley periods in different regions in the next few hours to one day;

[0006] Based on the predicted charging demand peak or valley periods in each region, the overall utilization rate of charging piles in the region during the period (i.e., the proportion of charging piles in use) is predicted. The predicted utilization rate is compared with a preset effectiveness determination threshold. If the utilization rate is higher than the set upper limit (such as 80%) during the predicted charging demand valley period, or the utilization rate is still lower than the set lower limit (such as 70%) during the predicted charging demand peak period, it is determined that the supply and demand relationship of the current charging pile resources in the region during the period does not significantly deviate from the expected balanced state, and the price adjustment mechanism does not need to be triggered. Otherwise, dynamic price adjustment is performed.

[0007] When dynamic price adjustment is performed, dynamic charging service price strategies are implemented according to the predicted charging demand peak or valley periods. The price is reduced during the verified charging demand valley period to attract users to charge, and the price is increased during the verified charging demand peak period to suppress part of the non-emergency charging demand, thereby guiding users to actively stagger charging and change the region for charging, and optimizing the charging pile load distribution (the charging demand peak and valley periods may not be synchronized with the power grid itself).

[0008] In one embodiment, the application provides a data analysis-based charging pile management method, which further comprises:

[0009] Systematically aggregating charging pile operation data from different regions, and collecting charging process information (charging duration, real-time charging power, battery temperature, ambient temperature, etc.) of electric vehicles of different brands and models to build a standardized historical charging data set covering multiple vehicle models and multiple scenarios, thereby providing a benchmark reference for subsequent anomaly detection.

[0010] During the charging process of the target vehicle, real-time target vehicle charging information (power, temperature, etc.) is collected and compared with benchmark data of the same type of vehicle or similar working conditions in the historical charging data set. Through preset anomaly detection algorithms (such as threshold method based on statistical process control, isolation forest, or unsupervised clustering algorithm) or machine learning models (such as support vector machine SVM, one-class SVM), the current charging state of the target vehicle is analyzed to determine whether there is an anomaly (such as sudden power drop or rapid temperature rise). Once an anomaly is confirmed, the safety mechanism is automatically triggered to stop power supply to prevent potential risks.

[0011] In one embodiment, the application provides a data analysis-based charging pile management method, which further comprises:

[0012] The historical health data of each charging pile (failure frequency, average repair time, average failure-free time, power stability, efficiency decay, etc.) is continuously analyzed, and the real-time operation state (online state, module alarm) of each charging pile is combined to calculate and update the dynamic health index of each charging pile. Based on the health index, the parameters for judging vehicle charging anomalies in the anomaly detection algorithm or machine learning model are intelligently adjusted (for example, for a pile with poor health status, the anomaly judgment standard may be more sensitive).

[0013] In one embodiment, the application provides a data analysis-based charging pile management method, which further comprises:

[0014] When the charging process is interrupted due to an anomaly (such as triggered stop or failure to reach the expected target), the user is automatically and timely informed of the situation and possible reasons, and the notification methods include user-side APP notification and SMS notification. At the same time, according to the on-site resources (status of adjacent idle piles), the user is intelligently guided to take solutions, including replacing the charging pile and contacting customer service.

[0015] In one embodiment, the application provides a data analysis-based charging pile management method, which further comprises:

[0016] Comprehensive analysis of long-term accumulated regional historical charging data, demand prediction results and charging pile utilization rate index, accurate identification of long-term demand exceeding supply, data support for operators, guidance for scientific site selection of new charging stations and expansion of existing high-load sites (increase of pile number and upgrade of power), and relief of supply and demand contradiction.

[0017] In one of the embodiments, the application provides a charging pile management system based on data analysis, comprising:

[0018] A peak and valley prediction module is configured to construct a time series prediction model based on historical charging session data (timestamp, geographic location, charging amount), environmental factors (weather conditions), special period information (holidays, surrounding large-scale activities) and real-time occupancy status of charging piles, and predict charging demand peak and valley periods in different regions in the next few hours to one day.

[0019] A usage rate prediction and judgment module is configured to predict the overall usage rate (i.e. the proportion of charging piles in use) of charging piles in the region during the predicted charging demand peak or valley period based on the predicted charging demand peak or valley period of each region; compare the predicted usage rate with a preset effectiveness determination threshold; if the usage rate is higher than the set upper limit (e.g. 80%) during the predicted charging demand valley period, or the usage rate is still lower than the set lower limit (e.g. 70%) during the predicted charging demand peak period, it is determined that the supply and demand relationship of the current charging pile resources in the region during the period does not significantly deviate from the expected balanced state, and the price adjustment mechanism does not need to be triggered; otherwise, dynamic price adjustment is performed.

[0020] A price adjustment module is configured to implement dynamic charging service price strategy according to the predicted charging demand peak or valley period when dynamic price adjustment is performed, reduce the price during the verified charging demand valley period to attract users to charge, increase the price during the verified charging demand peak period to suppress part of the non-emergency charging demand, guide users to actively stagger peak and change regional charging, and optimize charging pile load distribution (the charging demand peak and valley period may not be synchronized with the power grid itself).

[0021] In one of the embodiments, the application provides a charging pile management system based on data analysis, further comprising:

[0022] A data set construction module is configured to systematically aggregate charging pile operation data from different regions, collect charging process information (charging duration, real-time charging power, battery temperature, ambient temperature, etc.) of different brands and models of electric vehicles, and construct a standardized historical charging data set covering multiple vehicle models and multiple scenarios, providing a benchmark reference for subsequent anomaly detection.

[0023] A charging status determination module is configured to collect target vehicle charging information (power, temperature, etc.) in real time during the charging process of the target vehicle, and match and compare the collected information with reference data of the same type of vehicle or similar working conditions in a historical charging data set; through a preset abnormality detection algorithm (such as a threshold method based on statistical process control, an isolated forest, or an unsupervised clustering algorithm) or a machine learning model (such as a support vector machine SVM, a single-class SVM), the current charging status of the target vehicle is determined to be abnormal (such as sudden power drop or rapid temperature rise); once the abnormality is confirmed, the safety mechanism is automatically triggered to stop power supply to prevent potential risks.

[0024] In one embodiment, the present application provides a data analysis-based charging pile management system, which further comprises:

[0025] A parameter adjustment module is configured to continuously analyze historical health data (failure frequency, average repair time, average failure-free time, power stability, efficiency decay, etc.) of each charging pile, combine real-time running status (online status, module alarm) of each charging pile, calculate and update a dynamic health index of each charging pile, and intelligently adjust parameters (for example, the abnormality determination standard of a pile with poor health status may be more sensitive) in the abnormality detection algorithm or the machine learning model for determining vehicle charging abnormality.

[0026] In one embodiment, the present application provides a data analysis-based charging pile management system, which further comprises:

[0027] A charging abnormality reminding module is configured to automatically and timely inform a user of a situation and possible reasons when a charging process is interrupted due to abnormality (such as triggered stop or failure to reach an expected target), and the notification mode includes user-side APP notification and SMS notification; at the same time, the user is intelligently guided to take a solution according to on-site resources (status of adjacent idle piles), and the solution includes replacing the charging pile and contacting customer service.

[0028] In one embodiment, the present application provides a data analysis-based charging pile management system, which further comprises:

[0029] An expansion guidance module is configured to comprehensively analyze long-term accumulated regional historical charging data, demand prediction results, and charging pile utilization rate indexes, accurately identify hotspots with long-term demand continuously exceeding supply, provide data support for operators, guide scientific site selection of new charging stations, and expansion (increase in pile number and upgrade in power) of existing high-load sites, and relieve the contradiction between supply and demand.

[0030] Compared with the prior art, the present application has the beneficial effects that: the present application dynamically adjusts the charging service price by predicting the charging demand peak period and the valley period, and combining the overall use rate prediction result of the charging pile in the region during the period, so that the regional charging price is higher during the charging demand peak period, and the regional charging price is lower during the charging demand valley period, thereby guiding the user to actively staggered charging or selecting other regions to charge, optimizing the charging pile load distribution, both relieving the queuing pressure during the charging demand peak period, and improving the infrastructure utilization rate during the charging demand valley period. BRIEF DESCRIPTION OF DRAWINGS

[0031] Figure 1 The first part flow diagram of a charging pile management method based on data analysis provided for the embodiment of the present application.

[0032] Figure 2 The second part flow diagram of a charging pile management method based on data analysis provided for the embodiment of the present application.

[0033] Figure 3 The third part flow diagram of a charging pile management method based on data analysis provided for the embodiment of the present application.

[0034] Figure 4 The fourth part flow diagram of a charging pile management method based on data analysis provided for the embodiment of the present application.

[0035] Figure 5 The fifth part flow diagram of a charging pile management method based on data analysis provided for the embodiment of the present application.

[0036] Figure 6 The first part schematic diagram of a charging pile management system based on data analysis provided for the embodiment of the present application.

[0037] Figure 7 The second part schematic diagram of a charging pile management system based on data analysis provided for the embodiment of the present application.

[0038] Figure 8 The third part schematic diagram of a charging pile management system based on data analysis provided for the embodiment of the present application.

[0039] Figure 9 The fourth part schematic diagram of a charging pile management system based on data analysis provided for the embodiment of the present application.

[0040] Figure 10 The fifth part schematic diagram of a charging pile management system based on data analysis provided for the embodiment of the present application. DETAILED DESCRIPTION

[0041] In order to make the purpose, technical scheme and advantages of the present application more clear, the present application is further described in detail below with reference to the drawings and examples. It should be understood that the specific examples described herein are only used to explain the present application and do not limit the present application.

[0042] It can be understood that the terms "first", "second" and the like used in the present application can be used herein to describe various elements, but unless specifically stated, these elements are not limited by these terms. These terms are only used to distinguish the first element from another element. For example, without departing from the scope of the present application, the first xx script can be referred to as the second xx script, and similarly, the second xx script can be referred to as the first xx script.

[0043] In one embodiment, as shown in Figure 1 A data analysis-based charging pile management method includes the following steps:

[0044] Step S1, based on historical charging session data (time stamp, geographic location, charging amount), environmental factors (weather conditions), special period information (holidays, surrounding large-scale activities) and real-time occupation state of charging piles, a time series prediction model is constructed to predict the charging demand peak period and trough period in different areas in the next few hours to one day;

[0045] Step S2, based on the predicted charging demand peak period or trough period of each area, the overall usage rate of the charging pile in that area during that period (i.e. the proportion of charging piles in use) is predicted; compare the predicted usage rate with the preset effectiveness determination threshold. If the usage rate is higher than the set upper limit (such as 80%) in the predicted charging demand trough period, or the usage rate is still lower than the set lower limit (such as 70%) in the predicted charging demand peak period, the current charging pile resource in that period in that area is determined to be significantly deviated from the expected balance state, and the price adjustment mechanism does not need to be triggered; otherwise, dynamic price adjustment is performed.

[0046] Step S3, when dynamic price adjustment is performed, according to the predicted charging demand peak period or trough period, dynamic charging service price strategy is implemented, the price is reduced in the verified charging demand trough period to attract users to charge, the price is increased in the verified charging demand peak period to suppress part of the non-emergency charging demand, and users are guided to actively stagger peak and change area charging to optimize the charging pile load distribution (the charging demand peak period and trough period may not be synchronized with the power grid itself. Peak period and trough period).

[0047] Step S1 predicts regional charging demand, accurately predicting future short-term charging load fluctuations. Charging demand is dynamically affected by time, location, weather, holidays, and other factors. By building a time series prediction model, historical patterns and environmental variables can be quantitatively analyzed to predict demand peaks / troughs in different regions in the next few hours to a day. This is the data basis for optimizing resource scheduling and price strategy, avoiding blind operation.

[0048] Step S2 checks the charging pile usage rate threshold to prevent invalid pricing and ensure that the price strategy is only triggered when the supply and demand are imbalanced. If the predicted usage rate is >80% during the trough period: even if the price is low, it cannot attract more users (resources are saturated), and price reduction will reduce revenue and exacerbate congestion. If the predicted usage rate is <70% during the peak period: resources are sufficient, and price increase will be meaningless to suppress user demand. Regional differentiated threshold: different regions have different supply and demand characteristics (e.g., city center vs. suburbs), and dynamic threshold improves flexibility. Core logic: only when the price lever can effectively regulate supply and demand (e.g., idle resources can attract users during the trough period, and congestion needs to be alleviated during the peak period) is the price adjustment started.

[0049] Step S3 implements a dynamic price strategy to balance real-time supply and demand through price leverage, improving overall resource utilization. Price reduction during the trough period: stimulates users to use idle resources for charging (e.g., at night), increasing pile utilization and revenue. Price increase during the peak period: suppresses non-urgent demand, alleviates congestion, and prioritizes high-demand users. Guiding cross-regional charging: combined with price differences, encourages users to flow to low-load areas (e.g., peripheral sites), achieving regional load balancing.

[0050] In one embodiment, as shown in Figure 2 the data analysis-based charging pile management method further includes:

[0051] Step S4 systematically aggregates charging pile operation data from different regions and collects charging process information (charging duration, real-time charging power, battery temperature, ambient temperature, etc.) for different brands and models of electric vehicles, building a standardized historical charging data set covering multiple vehicle models and scenarios, providing a benchmark reference for subsequent anomaly detection.

[0052] Step S5, during the charging process of the target vehicle, real-time acquisition of target vehicle charging information (power, temperature, etc.), and matching comparison with benchmark data in the historical charging data set under similar working conditions; through a pre-set anomaly detection algorithm (such as threshold method based on statistical process control, isolation forest, or unsupervised clustering algorithm) or machine learning model (such as support vector machine SVM, single-class SVM) analysis, to determine whether the current charging state of the target vehicle is abnormal (such as power drop, temperature rise too fast); once the anomaly is confirmed, the safety mechanism is automatically triggered to stop power supply, preventing potential risks.

[0053] Step S4 establishes a baseline reference system for anomaly detection. The charging power curve and temperature rise characteristics of different vehicle models (e.g., Tesla vs. BYD) differ. By aggregating standardized historical data (power, temperature, duration, etc.) from multiple regions, multiple vehicle models, and multiple scenarios, a standardized historical charging dataset covering mainstream vehicle models is constructed. This is a prerequisite for subsequent accurate identification of anomalies, avoiding misjudgment of vehicle differences as faults.

[0054] Step S5 is to ensure charging safety and prevent equipment damage or fire risk. Matching the same type of vehicle data: excluding vehicle differences and focusing on abnormal signals (such as a 80% power drop and an abnormal temperature rise rate). Multiple algorithms or models are analyzed in parallel: threshold method (such as temperature > 60°C to trigger an alarm), fast response to hard safety indicators; Isolation Forest / Clustering, unsupervised identification of outliers deviating from the normal mode (such as abnormal power fluctuations); SVM / Single-class SVM, supervised learning to distinguish complex abnormal patterns. Immediately cut off the current after confirming the anomaly to avoid accidents.

[0055] In one embodiment, as shown in Figure 3 a data analysis-based charging pile management method further includes:

[0056] Step S6: Continuously analyze the historical health data of each charging pile (fault frequency, average repair time, average fault-free time, power stability, efficiency decay, etc.), and combine the real-time running state of each charging pile (online state, module alarm) to calculate and update the dynamic health index of each charging pile. Based on the health index, intelligently adjust the parameters in the anomaly detection algorithm or machine learning model used to judge vehicle charging anomalies (for example, for piles with poor health status, the anomaly judgment standard may be more sensitive).

[0057] Step S6 optimizes the accuracy of anomaly detection according to the state of the equipment to reduce false positives and false negatives. Evaluate the reliability of the equipment based on indicators such as failure rate and stability (e.g., old piles with low health index) to calculate the health index. Dynamically adjust the detection parameters: high health index pile: relax the judgment threshold (avoid false positives for normal fluctuations). Low health index pile: tighten the threshold or increase sensitivity (e.g., power fluctuation > 10% to trigger an alarm). Avoid frequent false positives in old piles due to their own fluctuations, while ensuring that anomalies in faulty piles can be captured in a timely manner.

[0058] In one embodiment, as shown in Figure 4 a data analysis-based charging pile management method further includes:

[0059] Step S7, when the charging process is interrupted due to an abnormality (such as triggered stop or failure to reach the expected target), the user is automatically and timely informed of the situation and possible causes, and the notification methods include user-side APP notification and SMS notification; at the same time, according to the on-site resources (the state of the nearby idle piles), the user is intelligently guided to take solutions, and the solutions include replacing the charging pile and contacting customer service.

[0060] Step S7 To ensure user experience, quickly restore charging service. Multi-channel instant notification (APP / SMS) ensures that users are informed in the first time. Intelligent guidance solution recommends nearby idle piles, and based on real-time data, it guides the replacement of charging piles (reduces waiting time). Complex problems are quickly transferred to manual (such as battery failure requiring assistance). Through automated decision-making + humanized service, user satisfaction and platform trust are improved.

[0061] In one embodiment, as shown in Figure 5 A charging pile management method based on data analysis further includes:

[0062] Step S8, by comprehensively analyzing the long-term accumulated regional historical charging data, demand prediction results and charging pile utilization rate indicators, the hotspots with long-term demand continuously exceeding supply are accurately identified, data support is provided for operators, and scientific site selection of new charging stations and expansion (increasing the number of piles and upgrading power) of existing high-load sites are guided to alleviate the contradiction between supply and demand.

[0063] Step S8 fundamentally solves the contradiction between supply and demand and optimizes investment efficiency. Through long-term data analysis (such as 90% utilization rate in a certain area + queuing for more than 30 minutes), resource shortage sites are located. New site selection covers high-demand gap areas (such as highway service areas and newly built communities). When expanding existing sites, the number of piles is increased or high-power equipment is upgraded at overloaded sites. Resources are accurately invested in the most urgent areas to improve the return on investment.

[0064] In one embodiment, as shown in Figure 6 A charging pile management system based on data analysis includes:

[0065] The peak and valley prediction module 1 is configured to construct a time series prediction model based on historical charging session data (timestamp, geographic location, charging amount), environmental factors (weather conditions), special period information (holidays, surrounding large-scale activities) and real-time occupation state of charging piles, and predict the charging demand peak period and valley period of different regions in the next few hours to one day.

[0066] The use rate prediction judgment module 2 is used to predict the overall use rate (i.e. the proportion of charging piles being used) of the charging piles in the region during the predicted peak or valley period of the charging demand in each region; compare the predicted use rate with the preset effectiveness judgment threshold value; if the use rate is higher than the set upper limit (such as 80%, the set upper limit can be different in different regions because the use conditions of the charging piles are different in different regions) during the predicted valley period of the charging demand, or the use rate is still lower than the set lower limit (such as 70%, the set lower limit can be different in different regions) during the predicted peak period of the charging demand, it is determined that the supply and demand relationship of the current charging pile resources in the region during the period does not significantly deviate from the expected balanced state, and the price adjustment mechanism does not need to be triggered; otherwise, dynamic price adjustment is performed.

[0067] The price adjustment module 3 is used to implement a dynamic charging service price strategy according to the predicted peak or valley period of the charging demand when dynamic price adjustment is performed, reduce the price during the verified valley period of the charging demand to attract users to charge, increase the price during the verified peak period of the charging demand to suppress part of the non-emergency charging demand, guide users to actively stagger the peak, change the region for charging, and optimize the charging pile load distribution (the charging demand peak period and valley period can not be synchronized with the power grid itself).

[0068] When constructing the time series prediction model, first, the historical charging session data is cleaned (such as invalid timestamps are removed and missing charging amounts are filled), the geographical position is mapped to a grid region code, the environmental factors are quantified (such as the weather is converted to a numerical value: sunny = 0, rainy = 1, snowy = 2), and special periods are encoded with a one-hot encoding (holiday = 1 / 0, surrounding activity intensity is divided into 0-3 levels); then, LSTM (Long Short-Term Memory Network) is selected as the core model, the input dimension is [time series length x feature number], and the features include the average charging amount in the region in the past 2 hours, the real-time occupancy rate, the weather code, the period code, etc., and the output is the demand in the future 4 / 8 / 12 hours; 70% of the historical data is used to train the model (batch size = 32, iteration 50 times), the remaining 30% is used to verify the accuracy (MAE <10kW is required), and finally the time series prediction model is deployed to receive new data in real time to predict the demand peak and valley in each region within the next 24 hours (such as predicting that A region is in the peak period from 9:00 to 11:00, the demand is 120kW±5%, and B region is in the valley period from 14:00 to 16:00, the demand is 40kW±8%).

[0069] The dynamic charging service price strategy is illustrated. When it is determined that the price needs to be adjusted, the price reduction strategy is implemented in the low demand period (e.g., the predicted usage rate of 60% is lower than the threshold of 80% in the B area from 14:00 to 16:00), and the price increase strategy is implemented in the high demand period (e.g., the predicted usage rate of 85% is higher than the threshold of 70% in the A area from 9:00 to 11:00). In the low demand period, the base price of 1.2 yuan per degree is reduced to 0.9 yuan per degree (a reduction of 25%), and the preferential prompt “75% discount for charging in the current period” is pushed. In the high demand period, the base price of 1.2 yuan per degree is increased to 1.8 yuan per degree (a rise of 50%), and the APP end is simultaneously displayed with “high peak period surcharge, it is suggested to charge in the C area 2 kilometers away (the current price is 1.0 yuan per degree)”. The price adjustment range is based on the prediction level layering design (e.g., the price is increased by 30% when the demand exceeds the supply by 20%, and the price is increased by 80% when the demand exceeds the supply by 50%), so as to realize load shunting.

[0070] In one embodiment, as shown in FIG. 1, Figure 7 a data analysis-based charging pile management system further comprises:

[0071] A data set construction module 4 is configured to systematically aggregate charging pile operation data from different regions, collect charging process information (charging duration, real-time charging power, battery temperature, ambient temperature, etc.) of electric vehicles of different brands and models, and construct a standardized historical charging data set covering multiple vehicle models and multiple scenarios, thereby providing a reference for subsequent anomaly detection.

[0072] A charging state judgment module 5 is configured to collect target vehicle charging information (power, temperature, etc.) in real time during the charging process of the target vehicle, and match and compare the information with reference data of the same type of vehicle or similar working conditions in the historical charging data set. Through preset anomaly detection algorithms (such as threshold method based on statistical process control, isolation forest, or unsupervised clustering algorithm) or machine learning models (such as support vector machine SVM, single-class SVM), the current charging state of the target vehicle is analyzed to determine whether there is an anomaly (such as sudden power drop or rapid temperature rise). Once the anomaly is confirmed, the safety mechanism is automatically triggered to stop power supply to prevent potential risks.

[0073] The anomaly detection process is as follows: Real-time data streams of the target vehicle's charging power, battery temperature, and ambient temperature are collected. Simultaneously, a baseline charging curve (e.g., power variation range, temperature rise rate threshold) for the same brand and model of vehicle under similar operating conditions (e.g., ambient temperature ±3℃, initial SOC ±5%) is extracted from a standardized historical charging dataset. Parallel analysis using multiple algorithms is employed: First, a hard safety threshold is set through Statistical Process Control (SPC) (e.g., triggering a Level 1 alarm immediately when instantaneous power is 50% below the rated value or battery temperature >60℃). Simultaneously, the Isolation Forest algorithm is used to calculate the deviation score between real-time data points and the historical baseline set. If the deviation value is greater than 0.65, it is judged as an outlier. Unsupervised clustering (K-means) is used to classify the current charging stage (constant current / constant voltage) data into historical clusters (an alarm is triggered when the distance from the nearest cluster center exceeds 2 standard deviations). When any algorithm outputs an anomaly flag (such as SPC threshold breach, isolated forest score exceeding limit, or cluster outlier), the system automatically initiates a graded response: first, the power is reduced to 50% and monitored for 10 seconds. If the anomaly is not eliminated, the power supply is immediately cut off, and the anomaly characteristics are recorded (such as "power drops sharply by 72% in the 3rd minute + temperature rise rate of 0.8℃ / second exceeds the benchmark value by 3 times") for model iteration and optimization.

[0074] In one embodiment, such as Figure 8 As shown, a data analysis-based charging pile management system also includes:

[0075] The parameter adjustment module 6 is used to continuously analyze the historical health data of each charging pile (fault frequency, average repair time, average fault-free time, power stability, efficiency decay, etc.), and calculate and update the dynamic health index of each charging pile in combination with the real-time operating status of each charging pile (online status, module alarms). Based on the health index, the parameters used in the anomaly detection algorithm or machine learning model to judge vehicle charging anomalies are intelligently adjusted (for example, the anomaly judgment criteria may be more sensitive for charging piles with poor health status).

[0076] The calculation of the charging pile health index is quantified by fusing historical performance indicators and real-time state data: first, the historical health data of the charging pile in the past 30 days is extracted, including fault frequency (such as the number of faults per hundred hours = total number of faults / (total number of hours of operation / 100)), average fault repair time (MTTR, in minutes), average fault-free time (MTBF, in hours), power stability (standard deviation of output power during charging, such as ± 5% for excellent), efficiency decay rate (current output efficiency compared to initial efficiency decrease percentage); At the same time, access to real-time running state signals, including online status (0 / 1), module alarm number (such as temperature module alarm number ≥ 1 is marked); The above indicators are normalized (such as MTBF ≥ 500 hours for 100 points, ≤ 100 hours for 0 points; Power standard deviation ≤ 2% for 100 points, ≥ 10% for 0 points), and weighted sum is generated by preset weight distribution (fault frequency weight 40%, MTTR weight 20%, power stability weight 20%, efficiency decay weight 15%, real-time alarm weight 5%) Weighted sum of 0-100 points of initial health index; Finally, introduce time decay factor (such as the weight of the last 7 days data is 70%, and the earlier data is attenuated by 30% per week), combined with real-time alarm dynamic deduction (each active alarm deducts 5 points), Output dynamic health index (example: a pile of historical weighted score is 85 points, but there are currently 2 module alarms, the final health index = 85-2x5 = 75 points), The index is updated every hour and used to adjust the parameters for judging vehicle charging abnormalities in the abnormal detection algorithm or machine learning model (such as when the health index ≤ 60, the power fluctuation threshold is tightened from ± 15% to ± 8%).

[0077] In one embodiment, as shown in Figure 9 , a data analysis-based charging pile management system further comprises:

[0078] The charging abnormality reminding module 7 is used for automatically and timely informing the user of the situation and possible reasons when the charging process is interrupted due to abnormalities (such as triggered stop or failure to reach the expected target), and the notification method includes user-side APP notification and SMS notification; At the same time, according to the on-site resources (the state of the adjacent idle pile), the user is intelligently guided to take a solution, and the solution includes replacing the charging pile and contacting customer service.

[0079] The compensation strategy can be further automatically triggered according to the interruption reason (such as giving a discount coupon due to pile failure interruption; compensating points due to power grid fluctuation interruption).

[0080] In one embodiment, as shown in Figure 10 , a data analysis-based charging pile management system further comprises:

[0081] The expansion guidance module 8 is used for comprehensive analysis of long-term accumulated regional historical charging data, demand prediction results and charging pile utilization rate indexes, accurate identification of hot spots with long-term demand exceeding supply, data support for operators, guidance of scientific site selection of new charging stations and expansion (increase of pile number and upgrade of power) of existing high-load sites, and relief of supply and demand contradiction.

[0082] The map data can also be accessed with each charging station as the center, and it is scanned whether there is a residential area / office building within a radius of N kilometers (such as 3 kilometers). If there is no charging station in the covered area and the user density is > 5000 people / square kilometer, it is marked as a blind area, and considered as an expansion point.

[0083] It should be understood that although each step in the flowchart of each embodiment of the present application is displayed in sequence according to the arrow, these steps are not necessarily executed in the order indicated by the arrow. Unless otherwise stated herein, the execution of these steps is not strictly limited in sequence, and these steps can be executed in other orders. Moreover, at least part of the steps in each embodiment can include multiple sub-steps or multiple stages, which are not necessarily executed at the same time, but can be executed at different times, and the execution order of these sub-steps or stages is not necessarily sequential, but can be executed alternately or alternately with at least part of other steps or sub-steps or stages of other steps.

[0084] Each technical feature of the above-described embodiments can be combined arbitrarily. In order to make the description simple, each technical feature in the above-described embodiments is not described in all possible combinations, but as long as the combination of these technical features does not exist, it should be considered as the scope of the present disclosure.

[0085] The above-described embodiments only express several embodiments of the present application, and the description is more specific and detailed, but it should not be understood as limiting the scope of the present patent. It should be noted that for ordinary skilled persons in the art, without departing from the concept of the present application, a number of modifications and improvements can be made, which are within the scope of protection of the present application. Therefore, the protection scope of the present patent should be subject to the appended claims.

[0086] The above-described only the preferred embodiments of the present application, and does not limit the present application, any modification, equivalent replacement and improvement made within the spirit and principles of the present application, should be included in the protection scope of the present application.

[0087] Furthermore, it should be understood that although the specification is described in terms of embodiments, not every embodiment includes every feature described. The specification can include implicit combinations of explicitly mentioned features and / or explicit combinations of implicitely mentioned features. Each embodiment depends on the explicit combinations of features and / or the implicit combinations of features made specifically within that embodiment, and each such embodiment can be combined with every other such embodiment to create further embodiments.

Claims

1. A charging pile management method based on data analysis, characterized in that, This data-driven charging station management method includes the following steps: Based on historical charging session data, environmental factors, information on special time periods, and real-time occupancy status of charging piles, a time series prediction model is constructed to predict the peak and off-peak periods of charging demand in different areas within the next few hours to one day. Based on the predicted peak or off-peak charging demand periods for each region, the overall utilization rate of charging piles in that region during that period is predicted. The predicted utilization rate is compared with a preset validity threshold. If the utilization rate is higher than the set upper limit during the predicted off-peak charging demand period, or lower than the set lower limit during the predicted peak charging demand period, it is determined that the supply and demand relationship of charging pile resources in that region during that period has not significantly deviated from the expected equilibrium state, and there is no need to trigger the price adjustment mechanism; otherwise, dynamic price adjustment is performed. When making dynamic price adjustments, a dynamic charging service pricing strategy is implemented based on the predicted peak or off-peak charging demand periods. Prices are lowered during verified off-peak charging demand periods to attract users to charge, and prices are raised during verified peak charging demand periods to suppress some non-urgent charging demand. This guides users to actively avoid peak hours and change charging areas, thereby optimizing the load distribution of charging piles.

2. The charging pile management method based on data analysis according to claim 1, characterized in that, The data analysis-based charging pile management method also includes: The system systematically aggregates charging pile operation data from different regions and collects charging process information for electric vehicles of different brands and models, constructing a standardized historical charging dataset covering multiple vehicle models and scenarios to provide a benchmark reference for subsequent anomaly detection. During the charging process of the target vehicle, the charging information of the target vehicle is collected in real time and matched and compared with the benchmark data of the same type of vehicle or similar working conditions in the historical charging dataset; the target vehicle's current charging status is judged by a preset anomaly detection algorithm or machine learning model; once an anomaly is confirmed, the safety mechanism is automatically triggered to stop the power supply to prevent potential risks.

3. The charging pile management method based on data analysis according to claim 2, characterized in that, The data analysis-based charging pile management method also includes: Continuously analyze the historical health data of each charging pile, combine it with the real-time operating status of each charging pile, calculate and update the dynamic health index of each charging pile; based on the health index, intelligently adjust the parameters used in the anomaly detection algorithm or machine learning model to judge vehicle charging anomalies.

4. The charging pile management method based on data analysis according to claim 2 or 3, characterized in that, The data analysis-based charging pile management method also includes: When the charging process is interrupted due to an abnormality, the system will automatically and promptly notify the user of the situation and possible causes through user app notifications and SMS notifications. At the same time, based on on-site resources, the system will intelligently guide the user to take solutions, including replacing the charging station or contacting customer service.

5. The charging pile management method based on data analysis according to claim 1, characterized in that, The data analysis-based charging pile management method also includes: By comprehensively analyzing historical charging data, demand forecasts, and charging pile utilization rates accumulated over a long period, we can accurately identify hotspots where demand consistently exceeds supply. This provides data support for operators, guides the scientific site selection of new charging stations and the expansion of existing high-load sites, and alleviates the supply-demand imbalance.

6. A charging pile management system based on data analysis, characterized in that, include: The peak and off-peak prediction module is used to build a time series prediction model based on historical charging session data, environmental factors, special time period information and real-time occupancy status of charging piles, and predict the peak and off-peak charging demand periods in different areas in the next few hours to a day. The utilization rate prediction and judgment module is used to predict the overall utilization rate of charging piles in a region during the predicted peak or off-peak charging demand periods. The predicted utilization rate is compared with a preset validity judgment threshold. If the utilization rate is higher than the set upper limit during the predicted off-peak charging demand period, or lower than the set lower limit during the predicted peak charging demand period, it is determined that the supply and demand relationship of charging pile resources in the region during the current period has not deviated significantly from the expected equilibrium state, and there is no need to trigger the price adjustment mechanism; otherwise, dynamic price adjustment is performed. The price adjustment module is used to implement dynamic charging service pricing strategies based on the predicted peak or off-peak charging demand periods when making dynamic price adjustments. During verified off-peak charging demand periods, prices are lowered to attract users to charge, while prices are raised during verified peak charging demand periods to suppress some non-urgent charging demand. This guides users to actively avoid peak hours and change charging areas, thereby optimizing the load distribution of charging piles.

7. The charging pile management system based on data analysis according to claim 6, characterized in that, The data analysis-based charging pile management system also includes: The dataset construction module is used to systematically aggregate charging pile operation data from different regions and collect charging process information for electric vehicles of different brands and models to build a standardized historical charging dataset covering multiple models and scenarios, providing a benchmark reference for subsequent anomaly detection. The charging status judgment module is used to collect the charging information of the target vehicle in real time during the charging process and match and compare it with the benchmark data of the same type of vehicle or similar working conditions in the historical charging dataset; through the analysis of the preset anomaly detection algorithm or machine learning model, it determines whether there is any abnormality in the current charging status of the target vehicle; once an anomaly is confirmed, the safety mechanism is automatically triggered to stop the power supply to prevent potential risks.

8. The charging pile management system based on data analysis according to claim 7, characterized in that, The data analysis-based charging pile management system also includes: The parameter adjustment module is used to continuously analyze the historical health data of each charging pile, combine it with the real-time operating status of each charging pile, calculate and update the dynamic health index of each charging pile; based on the health index, it intelligently adjusts the parameters used in the anomaly detection algorithm or machine learning model to judge vehicle charging anomalies.

9. The charging pile management system based on data analysis according to claim 7 or 8, characterized in that, The data analysis-based charging pile management system also includes: The charging anomaly alert module is used to automatically and promptly inform the user of the situation and possible causes when the charging process is interrupted due to an anomaly. Notification methods include user-side APP notification and SMS notification. At the same time, based on on-site resources, it intelligently guides the user to take solutions, including replacing the charging station and contacting customer service.

10. The charging pile management system based on data analysis according to claim 6, characterized in that, The data analysis-based charging pile management system also includes: The extended guidance module is used to comprehensively analyze long-term accumulated regional historical charging data, demand forecast results, and charging pile utilization indicators to accurately identify hotspot areas where demand consistently exceeds supply. This provides data support for operators, guides the scientific site selection of new charging stations and the expansion of existing high-load sites, and alleviates the supply-demand imbalance.

Images