Flexible resource prediction method for vehicle-pile-grid considering large-scale electric vehicle access

Abstract

The application relates to the technical field of power system dispatching, in particular to a car-pile-grid flexible resource prediction method considering the access of large-scale electric vehicles, which comprises the following steps: acquiring SOC data and expected remaining charging time of each charging vehicle at different time points in each control period of a charging station, and acquiring charging demand load, traffic flow and average driving speed at each time point, and extracting a charging peak period of the charging station; calculating a first evaluation value, a second evaluation value and an evaluation coefficient, adjusting a smoothing parameter of a prediction algorithm, and predicting the charging demand load of the charging station in the next control period; based on the prediction result, evaluating the adjustable flexible resource margin of the charging station in the next control period, so as to perform cluster division and charging-discharging scheduling of the charging station. The application significantly improves the load prediction accuracy, and improves the effectiveness and safety of peak load shifting and cluster scheduling of the power grid.

Description

Technical Field

[0001] This application relates to the field of power system dispatching technology, specifically to a flexible vehicle-pile-grid resource prediction method that takes into account the large-scale access of electric vehicles. Background Technology

[0002] With the large-scale development of new energy vehicles, electric vehicles, as massive distributed mobile energy storage units, have become an important trend to participate in power grid peak shaving and valley filling and demand response through vehicle-charging-grid collaboration. Accurately predicting the adjustable and flexible resource margin of charging stations is the core prerequisite for realizing large-scale vehicle-grid interaction and cluster charging and discharging scheduling.

[0003] Currently, the assessment of flexible resources for charging stations mainly relies on the prediction of baseline load, typically forecasting future load to estimate the adjustable space of flexible resources. However, existing forecasting methods usually employ fixed smoothing parameters, which can only better fit load evolution under steady-state conditions. They generally ignore the coupled effects of complex internal and external sudden environmental factors faced by charging stations. For example, in actual operation, the load of charging stations is easily affected by the concentrated charging behavior of vehicles within the station, leading to regular clustered surges in load. At the same time, influenced by the dynamics of the surrounding road network, traffic flow outside the station may also surge in large quantities at specific times, transforming into sudden charging demand. When the load surge caused by the superposition of the above internal and external factors occurs, the fixed smoothing parameters cannot amplify the sensitivity to the latest data in a timely manner, exhibiting a serious response lag defect. This results in a severe distortion in the prediction of charging load, leading to a large deviation in the assessed adjustable flexible resource margin, and affecting the accuracy and effectiveness of vehicle-charging-grid charging and discharging scheduling instructions. Summary of the Invention

[0004] To address the aforementioned technical issues, a flexible vehicle-charging-network resource prediction method that considers large-scale electric vehicle access is provided to resolve existing problems.

[0005] The solution to the technical problem presented in this application is to provide a flexible vehicle-charging-network resource prediction method that takes into account large-scale electric vehicle access, including the following steps: The system acquires the SOC data and estimated remaining charging time of each charging vehicle at different times within each control cycle, as well as the charging demand load, traffic flow and average driving speed at each time, and extracts the peak charging period of the charging station. For each control cycle, the discrete distribution characteristics of SOC data and expected remaining charging time of different charging vehicles are analyzed, the concentration distribution degree that characterizes the consistency of vehicle charging status in the station is calculated, and the first evaluation value that characterizes the regularity of charging load aggregation in the station is determined by combining the periodic characteristics of the number of vehicles charging in the station and the number of vehicles queuing for charging. By analyzing the traffic flow density represented by traffic flow and average vehicle speed within each control cycle, the correlation between traffic flow and charging demand load, and the proximity of the control cycle to peak charging periods, a second assessment value representing the risk of traffic influx from outside the station is calculated. This second assessment value is combined with the first assessment value to evaluate the assessment coefficient representing the sudden change trend of charging load. Based on this, the smoothing parameters of the prediction algorithm are adjusted. Using the adjusted smoothing parameters, the prediction algorithm is used to predict the charging demand load of the charging station in the next control cycle. Based on the prediction results, the adjustable and flexible resource margin of the charging station in the next control cycle is evaluated to implement the clustering and charging / discharging scheduling of the charging station.

[0006] Preferably, the calculation process for the concentration distribution degree is as follows: Each control cycle and the multiple control cycles preceding it are defined as local time periods; the dispersion of the SOC data of all charging vehicles at the same moment is positively fused with the dispersion of the expected remaining charging time of all charging vehicles to form the instantaneous dispersion. The degree of concentration distribution is negatively correlated with the degree of instantaneous dispersion.

[0007] Preferably, the specific process of the forward fusion is as follows: the product of the two discretenesses is used as the instantaneous discreteness.

[0008] Preferably, the process of obtaining the first evaluation value is as follows: The number of vehicles charging at each charging station at each time point is counted as the number of vehicles charging at each time point; the number of vehicles that have arrived at the charging station at each time point and are waiting for an available charging pile is counted as the number of vehicles in queue at each time point. The number of charging vehicles at all times within a local time period is decomposed into a trend and the periodic intensity is calculated. The number of queuing vehicles at all times within a local time period is also decomposed into a trend and the periodic intensity is calculated. The sum of these two periodic intensities is used as the charging cycle index. The first evaluation value is positively correlated with both the charging cycle index and the concentration distribution.

[0009] Preferably, the calculation process for the second evaluation value is as follows: The ratio of the mean traffic flow at all times within a local time period to the mean average vehicle speed at all times is calculated as the traffic flow density. Calculate the correlation between traffic flow and charging demand load at all times within a local time period, and perform a positive mapping on it as the correlation coefficient; The interval between each regulation cycle and the peak charging period is statistically analyzed, and the peak interval duration is calculated. The second assessment value is positively correlated with traffic density and correlation coefficient, but negatively correlated with peak interval duration.

[0010] Preferably, the process of obtaining the peak interval duration is as follows: if the last moment in each control cycle is not within the peak charging period, then the time interval between the last moment in each control cycle and the start moment of the peak charging period is counted and used as the peak interval duration; otherwise, the peak interval duration is assigned a preset value.

[0011] Preferably, the evaluation coefficient is positively correlated with both the first evaluation value and the second evaluation value.

[0012] Preferably, the charging station is at the first Each regulation cycle corresponds to the adjusted smoothing parameter. The calculation formula is: ,in, The preset initial smoothing parameters, For the first The evaluation coefficient for each regulatory cycle, This is the preset adjustment range.

[0013] Preferably, the step of predicting the charging demand load of the charging station in the next control cycle includes: using the EMA algorithm based on the adjusted smoothing parameters to predict the charging demand load of the charging station in the next control cycle, and obtaining the predicted charging demand load at each time point in the next control cycle.

[0014] Preferably, the assessment of the adjustable and flexible resource margin of the charging station in the next control cycle includes: For the last moment of each control cycle, obtain the maximum theoretical charging load and maximum reverse discharge load of the charging station at that moment; The adjustable flexible resource margin includes upward flexible resource margin and downward flexible resource margin. The upward flexible resource margin of the charging station is the difference between the maximum theoretical charging load and the predicted charging demand load. The downward flexible resource margin of the charging station is the sum between the predicted charging demand load and the maximum reverse discharge load.

[0015] This application has at least the following beneficial effects: This application calculates the concentration distribution degree, which has the advantage of accurately identifying the concentrated charging behavior characteristics of vehicles with highly consistent states, providing micro-level data support for exploring sudden load aggregation within charging stations; determining the first evaluation value has the advantage of integrating the consistency of vehicle charging states with the periodic characteristics of traffic queuing, accurately quantifying the load aggregation pattern caused by in-station scheduling behavior, and effectively characterizing the load mutation tendency caused by inherent factors within the station; determining the second evaluation value has the advantage of quantifying the driving risk intensity of the external environment of the charging station on charging demand, accurately filtering out the interference of pure passing traffic in the road network, and predicting and locking in advance the risk of external charging load mutation caused by the influx of real traffic; determining the evaluation coefficient has the advantage of considering the aggregation pattern within the station and the risk of influx from outside the station, comprehensively and dynamically quantifying the overall load mutation trend faced by the charging station, and providing a precise quantitative adjustment benchmark for subsequent smoothing parameter adjustments; adjusting the smoothing parameters of the prediction algorithm has the advantage of dynamically adjusting... The smoothing parameters enable the algorithm to automatically increase its sensitivity to the latest data to eliminate response lag during load surges, while maintaining historical data memory to filter noise during stable load periods. This significantly improves the algorithm's adaptive tracking capability in complex dynamic scenarios. The algorithm predicts the charging demand load of charging stations in the next control cycle. Based on the prediction results, it assesses the adjustable and flexible resource margin of charging stations in the next control cycle to execute charging station cluster partitioning and charging / discharging scheduling. Its beneficial effect lies in the fact that prediction based on the adjusted smoothing parameters can output a high-precision demand load baseline that eliminates lag errors, significantly improving load prediction accuracy. This provides a reliable benchmark for accurately assessing the physical adjustable space of charging stations, ensuring that subsequent margin calculations will not suffer from "artificially high" or "artificially low" distortions due to prediction deviations. It provides virtual power plants or aggregators with quantifiable and reliable capability boundaries, enabling spatiotemporal aggregation and collaborative optimization control of charging pile clusters, and significantly improving the effectiveness and security of power grid peak shaving and valley filling and cluster scheduling. Attached Figure Description

[0016] Figure 1 A flowchart illustrating the steps of a flexible vehicle-charging-network resource prediction method considering large-scale electric vehicle access provided in this application embodiment; Figure 2 A flowchart illustrating the steps of a method for obtaining a first evaluation value provided in an embodiment of this application. Detailed Implementation

[0017] The following, in conjunction with the accompanying drawings and embodiments, provides a more detailed description of the vehicle-charging-network flexible resource prediction method proposed in this application that takes into account large-scale electric vehicle access.

[0018] Please see Figure 1The diagram illustrates a flowchart of a flexible vehicle-charging-network resource prediction method considering large-scale electric vehicle access, according to an embodiment of this application. The method includes the following steps: Step 1: Obtain the SOC data and estimated remaining charging time of each charging vehicle at different times during each control cycle of the charging station, as well as the charging demand load, number of charging vehicles, number of queuing vehicles, traffic flow and average driving speed at each time, and extract the peak charging period of the charging station.

[0019] With the large-scale integration of electric vehicles, charging load exhibits high spatiotemporal randomness. The schedulable and flexible resources of a single charging station are coupled with the influence of vehicle access status, user behavior, and the external traffic environment. To achieve effective prediction of flexible resources, it is first necessary to establish a high-resolution, multi-dimensional data acquisition mechanism to obtain real-time vehicle parameters within the charging station and dynamic information on the external road network, providing basic data support for subsequent resource assessment and prediction.

[0020] Based on the above analysis, the integrated digital resource system of the city district can be used to obtain real-time information on charging stations and road network information of the roads where the entrances and exits of charging stations are located. The charging station information includes the SOC data and estimated remaining charging time of each charging vehicle at each time, as well as the charging demand load, number of charging vehicles and number of vehicles in queue at each time. The total number of vehicles charging at each charging station at each time point is counted as the total number of vehicles charging at each time point. The number of vehicles that have arrived at the charging station and are waiting for an available charging pile at each time point is counted as the number of vehicles in the queue at each time point. The road network information includes traffic flow and average vehicle speed at various times; In this embodiment, the data acquisition time interval is 30 seconds. As for other implementation methods, the implementer can set it according to the actual situation.

[0021] Based on the charging demand load of the charging station in the past day, extract the peak charging time of the charging station; In this embodiment, the process of obtaining peak charging periods is as follows: the entire day is divided into multiple short periods; the average charging demand load at all times within each short period is calculated as the average load; the average charging demand load at all times throughout the day is calculated as the daily average load; and multiple consecutive short periods with an average load greater than or equal to a preset multiple of the daily average load are marked as peak charging periods; wherein, the duration of a single short period is 15 minutes, and the preset multiple is set to 1.5. As other implementation methods, the implementer can set it according to the actual situation.

[0022] Multiple time points are defined as a control cycle. The SOC data and estimated remaining charging time of each charging vehicle at different time points within each control cycle are obtained, as well as the number of charging vehicles, the number of vehicles in queue, the charging demand load, traffic flow and average driving speed at different time points. In this embodiment, the duration of the control cycle is 5 minutes. As for other implementation methods, the implementer can set it according to the actual situation.

[0023] Thus, the SOC data and estimated remaining charging time of each charging vehicle at different times within each control cycle of the charging station are obtained, as well as the number of charging vehicles, the number of vehicles in queue, the charging demand load, traffic flow and average driving speed at different times; and the peak charging period of the charging station is extracted.

[0024] Step 2: For each control cycle, analyze the discrete distribution characteristics of SOC data and expected remaining charging time of different charging vehicles, calculate the concentration distribution degree that characterizes the consistency of vehicle charging status in the station, and combine the periodic characteristics of the number of vehicles charging in the station and the number of vehicles queuing for charging to determine the first evaluation value that characterizes the regularity of charging load aggregation in the station.

[0025] In the vehicle-charging-grid interactive system, electric vehicles within charging stations act as distributed energy storage units. Their flexible resources, which can be utilized by the grid, are essentially the adjustable portion of the real-time charging demand load at the charging station. The charging demand load not only reflects the immediate intensity of the vehicles' demand for electricity within the station but also contains information on the margin of flexible resources. It is the direct target of the virtual power plant's power allocation and dispatch commands. The level of the charging demand load directly determines the adjustable direction and margin of the flexible resources. Specifically: when the charging demand load is low, it indicates weak charging demand from vehicles within the station, with most vehicles having sufficient battery energy storage or low urgency for charging. In this case, the flexible resources mainly exhibit downward adjustment capabilities, meaning the grid can dispatch vehicles within the station to suspend charging or discharge via V2G to participate in peak shaving or provide ancillary services. When the charging demand load is high, it indicates strong charging demand from vehicles within the station, with generally low battery energy storage. In this case, the flexible resources mainly exhibit upward adjustment capabilities, meaning the grid can dispatch vehicles within the station to increase charging power to absorb wind and solar power curtailment or participate in valley filling. Therefore, by understanding the future evolution trend of charging demand load, we can indirectly infer the upward / downward adjustment capacity that charging stations can provide.

[0026] Based on this, the charging demand load of charging stations is used as the core indicator of flexible resource status. The EMA (Exponential Moving Average) algorithm is employed to predict it. The prediction results will serve as the future flexible resource dispatchability of charging stations, allowing virtual power plants or aggregators to perform unified scheduling. Specifically, when performing rolling predictions, the value of the smoothing parameter in the EMA algorithm determines the sensitivity of the prediction results to both new and old data. A larger smoothing parameter results in a faster response speed to the latest data and more sensitive tracking of load fluctuations; a smaller smoothing parameter results in a stronger memory of historical data and a smoother prediction trajectory. Therefore, the smoothing parameter needs to be dynamically adjusted based on the internal and external data characteristics of the charging station to achieve a balance between prediction accuracy and response sensitivity.

[0027] Secondly, when the SOC distribution of vehicles in a charging station is concentrated, the remaining charging time is similar, and the number of charging vehicles and the number of vehicles in the queue show strong periodic fluctuations, it indicates that the charging behavior of vehicles in the station is highly synchronous. The station may face a scenario of concentrated vehicle charging scheduling. In this scenario, the charging load will show step-like jumps or periodic pulse characteristics. If the smoothing parameter is too small, the prediction results will rely too much on the stable historical data before scheduling, resulting in a lag in the response to concentrated charging events and a large prediction deviation. At this time, the smoothing parameter should be increased to enable the algorithm to respond to new data faster and thus more accurately capture the load changes brought about by concentrated charging events.

[0028] Based on the above analysis, a first evaluation value is calculated by considering the SOC distribution characteristics of vehicles within the charging station, the remaining charging time distribution characteristics, and the periodic fluctuation characteristics of the number of charging vehicles and the number of vehicles in the queue. The flowchart of the method for obtaining the first evaluation value provided in this application embodiment is shown below. Figure 2 As shown, it specifically includes: Each control cycle and the multiple control cycles preceding it are defined as local time periods; In this embodiment, the duration of the local time period is 2 hours. As for other implementation methods, the implementer can set it according to the actual situation.

[0029] The dispersion of SOC data of all charging vehicles at the same moment is positively fused with the dispersion of the expected remaining charging time of all charging vehicles to form the instantaneous dispersion. In this embodiment, the maximum normalization method is used to normalize the estimated remaining charging time at all times. The dispersion is measured by calculating the variance of the SOC data of all charging vehicles at the same time and the variance of the normalized estimated remaining charging time of all charging vehicles at the same time. The maximum normalization method and the calculation of variance are well-known techniques and will not be described in detail here. It should be noted that forward fusion specifically represents an additive relationship, a multiplicative relationship, etc. In this embodiment, the product of the two dispersions is used as the instantaneous dispersion.

[0030] The concentration distribution of each regulation cycle is negatively correlated with the instantaneous dispersion. It should be noted that a negative correlation means that the dependent variable decreases as the independent variable increases, and increases as the independent variable decreases.

[0031] In this embodiment, the mean of the instantaneous dispersion of all moments within a local time period is calculated and negatively mapped to it, serving as the concentration distribution of each control cycle. The negative mapping process is as follows: the reciprocal of the mean of the instantaneous dispersion of all moments within the local time period is used as the result of the negative mapping. In order to avoid the denominator being 0 when calculating the reciprocal, a parameter adjustment factor is added to the denominator. In this embodiment, the parameter adjustment factor is set to 0.01. In other implementation methods, the implementer can set it according to the actual situation.

[0032] It should be noted that the smaller the instantaneous dispersion, the more similar the SOC and remaining charging time of the vehicles charging in the station at the same moment, reflecting that they are in a concentrated charging state at this time; the larger the concentration distribution, the more consistent the SOC level of the vehicles in the station and the more concentrated the charging time, and the more similar the vehicle behavior.

[0033] The number of charging vehicles at all times within a local time period is decomposed into a trend and the periodic intensity is calculated. The number of queuing vehicles at all times within a local time period is also decomposed into a trend and the periodic intensity is calculated. The sum of these two periodic intensities is used as the charging cycle index. In this embodiment, the STL (Seasonal and Trend decomposition using Loess) algorithm is used for trend decomposition and the periodicity strength is calculated. Both the STL algorithm and the calculation of periodicity strength are well-known techniques and will not be elaborated upon here. The calculation process for periodicity strength is as follows: the STL algorithm is used to decompose the seasonal term and the residual term. The formula for periodicity strength is: ,in, For periodic intensity, Let Variance be the variance of the residual term. The variances of the seasonal term and the residual term are... This is the function for finding the maximum value.

[0034] The first assessment value of the charging station in each control cycle is positively correlated with the charging cycle index and the concentration distribution degree. It should be noted that a positive correlation means that the dependent variable increases as the independent variable increases and decreases as the independent variable decreases.

[0035] In this embodiment, the product of the charging cycle index and the concentration distribution degree is used as the first evaluation value; It should be noted that the larger the charging cycle index, the stronger the regularity of the changes in the number of charging vehicles and the number of vehicles in the queue, and the more obvious the batch and periodic characteristics of vehicles entering and leaving the station; the larger the first evaluation value, the more likely the charging station has both high regularity and high state concentration, and the more likely there are scenarios of concentrated vehicle scheduling for charging. The smoothing parameter should be increased to improve the sensitivity to new data, so as to accurately capture the load changes caused by concentrated charging events.

[0036] Thus, the first evaluation value of the charging station in each control cycle is obtained.

[0037] Step 3: Calculate the second assessment value representing the risk of traffic influx outside the station by considering the traffic flow density represented by traffic flow and average driving speed within each control cycle, the correlation between traffic flow and charging demand load, and the proximity of the control cycle to the peak charging period.

[0038] Furthermore, when predicting the charging demand load of charging stations, the traffic flow status of the surrounding road network is an important leading indicator reflecting future changes in charging demand. Specifically, as a charging station approaches its peak charging period, the number of vehicles heading there gradually increases, traffic flow on surrounding roads rises accordingly, and average vehicle speed decreases. The charging station transitions from a low-load to a high-load state. If this trend can be captured in advance, and the smoothing parameters of the prediction algorithm are adjusted accordingly, the response speed and prediction accuracy for future high-fluctuation states can be significantly improved. However, in actual road conditions, the increase in traffic flow around charging stations may stem from various factors, such as through traffic or vehicles attracted by nearby commercial activities. These passing vehicles do not necessarily enter the charging station to generate charging demand. Simply increasing the smoothing parameters based on the increase in traffic flow may introduce noise interference and amplify prediction errors. Therefore, it is essential to discern the true correlation between traffic flow and charging demand.

[0039] To effectively identify this correlation, a comprehensive assessment was made from three dimensions: the relationship between traffic flow and vehicle speed, the relationship between traffic flow and charging demand load, and the proximity to peak charging times. When traffic flow increases and vehicle speed decreases significantly, it indicates that vehicles are slowing down and queuing to enter the charging station, suggesting a strong directional focus towards charging. Conversely, if traffic flow is high but vehicle speed remains at a high level, vehicles are more likely to be passing through quickly, unrelated to charging demand. Secondly, the correlation between traffic flow and charging demand load reflects the charging conversion efficiency of the road network. If changes in traffic flow and changes in charging demand load at charging stations show a significant positive correlation, it indicates that increased traffic flow... If the traffic flow is effectively converted into charging demand, the smoothing parameter adjustment range should be increased. If the correlation between the two is low, then the increase in traffic flow has not brought about actual charging demand. In this case, the smoothing parameter should be suppressed to avoid the algorithm over-responding to irrelevant traffic disturbances. In addition, the proximity to the charging peak period provides important prior information: the closer to the charging peak period, the higher the probability of a rapid increase in charging demand even if the current load has not yet increased significantly. The algorithm's response sensitivity should be increased in advance. Conversely, if the current period is in or about to enter a low period, even if the traffic flow is large, the smoothing parameter should be suppressed to avoid misjudgment.

[0040] Based on the above analysis, the second evaluation value is calculated by considering the relationship between vehicle speed and traffic flow, the correlation between traffic flow and charging demand load, and the proximity to peak charging periods. Specifically: The ratio of the mean traffic flow at all times within a local time period to the mean average vehicle speed at all times is calculated as the traffic flow density. It should be noted that, in order to avoid the denominator being 0 when calculating the ratio, a parameter adjustment factor is added to the denominator. In this embodiment, the parameter adjustment factor is set to 0.01. As for other implementation methods, the implementer can set it according to the actual situation. Secondly, the higher the traffic flow density, the higher the traffic flow per unit speed, which usually means that the road congestion is high, the vehicles are moving slowly, and the surrounding roads may be saturated or close to saturation. The possibility of vehicles slowing down to enter the station increases, and the probability of increased charging demand is high.

[0041] Calculate the correlation between traffic flow and charging demand load at all times within a local time period, and perform a positive mapping on it as the correlation coefficient; In this embodiment, the correlation is measured by calculating the Pearson correlation coefficient between traffic flow and charging demand load at all times within a local time period. The Pearson correlation coefficient is a well-known technique and will not be elaborated upon here. Secondly, the positive mapping process involves selecting the maximum value between the correlation and 0, and summing it with the value 1, as the correlation coefficient. The calculation formula is: ,in, For the degree of relevance, This represents the function that takes the maximum value; by taking the maximum value, only positive correlations are retained, so that only traffic flow that is positively correlated with charging demand will have a positive contribution to the adjustment of the smoothing parameters.

[0042] If the last moment of each control cycle is not within the peak charging period, the time interval between the last moment of each control cycle and the start moment of the peak charging period is counted and used as the peak interval duration; otherwise, the peak interval duration is assigned a preset value. The second evaluation value of the charging station in each control cycle is positively correlated with the traffic flow density and the correlation coefficient, but negatively correlated with the peak interval duration. In this embodiment, the preset value is set to 0. In other implementations, the implementer can set it according to the actual situation. The product of traffic flow density and correlation coefficient is calculated, and the ratio of it to the peak interval duration is used as the second evaluation value. In order to avoid the denominator being 0 when calculating the ratio, a parameter adjustment factor is added to the denominator. In this embodiment, the parameter adjustment factor is set to 0.01. In other implementations, the implementer can set it according to the actual situation.

[0043] It should be noted that a larger correlation coefficient indicates a strong positive correlation between traffic flow and charging demand load, reflecting a high degree of synchronization between changes in traffic flow and changes in charging demand load within the station, indicating that vehicles in the current road network are indeed heading to the station to charge. The smaller the peak interval duration, the closer to or into the peak charging demand period, and the more significantly the smoothing parameter should be adjusted to quickly respond to the upcoming load changes. The larger the obtained second evaluation value, the larger the traffic flow and the lower the vehicle speed on the roads where the charging station entrance and exit are located. At the same time, the stronger the positive correlation between traffic flow and charging demand load within the charging station, and the closer to the peak charging period, the higher the probability and urgency of the large amount of traffic flow converging in the external road network transforming into charging load at the station. This reflects that the charging station is about to or is facing a sudden surge in load caused by the influx of external traffic. In subsequent predictions, the impact of external data on flexible resource fluctuations is extremely significant, and the smoothing parameter of the algorithm must be increased to enhance the sensitivity to new load change trends and avoid prediction lag.

[0044] This yields the second evaluation value for the charging station in each control cycle.

[0045] Step 4: Based on the first and second evaluation values, evaluate the evaluation coefficients that characterize the sudden change trend of charging load. Adjust the smoothing parameters of the prediction algorithm accordingly. Using the adjusted smoothing parameters, use the prediction algorithm to predict the charging demand load of the charging station in the next control cycle. Based on the prediction results, evaluate the adjustable and flexible resource margin of the charging station in the next control cycle to perform cluster partitioning and charging / discharging scheduling of the charging station.

[0046] Furthermore, based on the first and second evaluation values, the evaluation coefficients are determined as follows: The evaluation coefficients of charging stations in each control cycle are positively correlated with both the first and second evaluation values; In this embodiment, the maximum value normalization method is used to normalize the first and second evaluation values ​​of all control cycles in the past day. Based on the preset first weight and the preset second weight, the normalized first evaluation value and the normalized second evaluation value are weighted and summed as the evaluation coefficient. The sum of the preset first weight and the preset second weight is 1. In this embodiment, the preset first weight is set to equal the preset second weight, both of which are set to 0.5. In other implementation methods, the implementer can set them according to the actual situation.

[0047] It should be noted that the larger the evaluation coefficient, the more obvious the vehicle centralized scheduling characteristics are inside the charging station, and the more a large number of actual charging vehicles are about to flood into the external road network. This reflects that the flexible resources of the charging station are about to undergo a drastic and substantial change. This requires that when predicting the flexible resources of the charging station, the smoothing parameter of the algorithm must be increased simultaneously to improve the algorithm's sensitivity to the latest load data, thereby accurately tracking load changes and eliminating prediction lag.

[0048] Furthermore, the smoothing parameters of the EMA algorithm are adjusted using evaluation coefficients, specifically as follows: The formula for calculating the smoothing parameters of the charging station after adjustment in each control cycle is as follows: in, For the charging station Each regulation cycle corresponds to the adjusted smoothing parameter. The preset initial smoothing parameters, For the first The evaluation coefficient for each regulatory cycle, The preset adjustment range; In this embodiment, the preset initial smoothing parameter is set to 0.1. The preset adjustment range is used to control the upper limit of the amplification of the smoothing parameter by the evaluation coefficient, ensuring that the smoothing parameter can provide sufficient response sensitivity when dealing with sudden load changes, while avoiding the smoothing parameter from exceeding the reasonable limit, which would cause the algorithm to completely lose its memory of historical data and anti-interference filtering ability. Therefore, this value is set to 0.7. As other implementation methods, implementers can set it according to the actual situation.

[0049] Based on the adjusted smoothing parameters, the EMA algorithm is used to predict the charging demand load of the charging station in the next control cycle, and the predicted charging demand load at each time in the next control cycle is obtained. It should be noted that the EMA algorithm is a well-known technique and will not be elaborated upon here.

[0050] Based on the prediction results, the adjustable and flexible resource margin of the charging station in the next control cycle is evaluated and input to the vehicle-charging-grid collaborative scheduling terminal, so that the scheduling terminal can generate charging and discharging power scheduling instructions for the charging vehicles connected to the grid within the station according to the real-time load status of the power grid; specifically: For the last moment of each control cycle, obtain the maximum theoretical charging load and maximum reverse discharge load of the charging station at that moment; In this embodiment, BMS data of all charging vehicles and the rated parameters of the corresponding charging piles are acquired. For a vehicle that is currently charging, the maximum allowable charging power of the vehicle's BMS is extracted and compared with the rated output power of the corresponding charging pile. The minimum value between the two is taken as the maximum charging load of the charging vehicle. The maximum charging loads of all charging vehicles are summed to obtain the maximum theoretical charging load. Charging vehicles that support V2G reverse discharge and whose current SOC is higher than the preset discharge lower limit are selected. The maximum allowable discharge power of their BMS is extracted and compared with the rated reverse discharge power of the corresponding charging pile. The minimum value between the two is taken as the maximum discharge load of the vehicle. The maximum discharge loads of all selected charging vehicles are summed to obtain the maximum reverse discharge load. The preset discharge lower limit is set to 20%. As another implementation method, the implementer can set it according to the actual situation.

[0051] The difference between the maximum theoretical charging load and the predicted charging demand load is calculated as the upward adjustment flexible resource margin for the charging station. The sum of the predicted charging demand load and the maximum reverse discharge load is calculated as the downward flexible resource margin for the charging station. The combination of the upward and downward adjustments to the flexible resource margin is used as the flexible resource prediction result for the charging station in the next control cycle. It should be noted that the upward adjustment of flexible resource margin represents the maximum amount of electricity that this charging station can absorb during off-peak periods; the downward adjustment of flexible resource margin represents the maximum amount of charging demand that this charging station can cut during peak periods, and even its ability to send electricity back to the grid. Thus, when the grid is in a low-load period, a dispatching instruction to increase the charging power of vehicles in the station is generated based on the upward adjustment of flexible resource margin; when the grid is in a high-load period, a dispatching instruction to reduce the charging power of vehicles in the station or control vehicles to discharge electricity back to the grid is generated based on the downward adjustment of flexible resource margin, so as to achieve power supply balance in the grid.

[0052] The flexible resource prediction results are input into the vehicle-charging-grid collaborative scheduling terminal, enabling the scheduling terminal to generate peak-shaving and valley-filling charging and discharging scheduling instructions for vehicles connected to the grid at charging stations, based on the real-time load status of the power grid. Specifically, the scheduling terminal dynamically divides multiple charging stations within its jurisdiction into several clusters based on the upward and downward adjustment of flexible resource margins, combined with the distribution network topology. This is abstracted into a large-scale mobile energy storage aggregation model and transmitted to the virtual power plant or aggregator. The virtual power plant or aggregator, combined with the real-time peak and valley status of the power grid, formulates a system that balances fluctuation smoothing, economic efficiency, and user convenience. The system receives global power commands and decomposes them into sub-commands for each cluster, simultaneously sending optimized time-of-use pricing signals to each charging station. Based on real-time grid conditions, redundant capacity is dynamically allocated between clusters to achieve spatiotemporal energy interaction between feeders and efficient absorption of new energy sources. Local controllers within each cluster decompose the received power sub-commands to each charging pile using model predictive control (MPC) algorithms or rule-based strategies, controlling power electronic devices such as charging piles supporting V2G functionality to perform charging and discharging operations. Simultaneously, the smart charging app pushes real-time electricity prices to users, guiding them to participate in demand response.

Claims

1. A flexible resource prediction method for vehicle-charging station-network considering large-scale electric vehicle access, characterized in that, The method includes the following steps: The system acquires the SOC data and estimated remaining charging time of each charging vehicle at different times within each control cycle, as well as the charging demand load, traffic flow and average driving speed at each time, and extracts the peak charging period of the charging station. For each control cycle, the discrete distribution characteristics of SOC data and expected remaining charging time of different charging vehicles are analyzed, the concentration distribution degree that characterizes the consistency of vehicle charging status in the station is calculated, and the first evaluation value that characterizes the regularity of charging load aggregation in the station is determined by combining the periodic characteristics of the number of vehicles charging in the station and the number of vehicles queuing for charging. By analyzing the traffic flow density represented by traffic flow and average vehicle speed within each control cycle, the correlation between traffic flow and charging demand load, and the proximity of the control cycle to peak charging periods, a second assessment value representing the risk of traffic influx from outside the station is calculated. This second assessment value is combined with the first assessment value to evaluate the assessment coefficient representing the sudden change trend of charging load. Based on this, the smoothing parameters of the prediction algorithm are adjusted. Using the adjusted smoothing parameters, the prediction algorithm is used to predict the charging demand load of the charging station in the next control cycle. Based on the prediction results, the adjustable and flexible resource margin of the charging station in the next control cycle is evaluated to implement the clustering and charging / discharging scheduling of the charging station.

2. The flexible vehicle-charging-network resource prediction method considering large-scale electric vehicle access as described in claim 1, characterized in that, The calculation process for the concentration distribution degree is as follows: Each control cycle and the multiple control cycles preceding it are defined as local time periods; the dispersion of the SOC data of all charging vehicles at the same moment is positively fused with the dispersion of the expected remaining charging time of all charging vehicles to form the instantaneous dispersion. The degree of concentration distribution is negatively correlated with the degree of instantaneous dispersion.

3. The flexible vehicle-charging-network resource prediction method considering large-scale electric vehicle access as described in claim 2, characterized in that, The specific process of the forward fusion is as follows: the product of the two discretenesses is taken as the instantaneous discreteness.

4. The flexible vehicle-charging-network resource prediction method considering large-scale electric vehicle access as described in claim 2, characterized in that, The process of obtaining the first evaluation value is as follows: The number of vehicles charging at each charging station at each time point is counted as the number of vehicles charging at each time point; the number of vehicles that have arrived at the charging station at each time point and are waiting for an available charging pile is counted as the number of vehicles in queue at each time point. The number of charging vehicles at all times within a local time period is decomposed into a trend and the periodic intensity is calculated. The number of queuing vehicles at all times within a local time period is also decomposed into a trend and the periodic intensity is calculated. The sum of these two periodic intensities is used as the charging cycle index. The first evaluation value is positively correlated with both the charging cycle index and the concentration distribution.

5. The flexible vehicle-charging-network resource prediction method considering large-scale electric vehicle access as described in claim 2, characterized in that, The calculation process for the second evaluation value is as follows: The ratio of the mean traffic flow at all times within a local time period to the mean average vehicle speed at all times is calculated as the traffic flow density. Calculate the correlation between traffic flow and charging demand load at all times within a local time period, and perform a positive mapping on it as the correlation coefficient; The interval between each regulation cycle and the peak charging period is statistically analyzed, and the peak interval duration is calculated. The second assessment value is positively correlated with traffic density and correlation coefficient, but negatively correlated with peak interval duration.

6. The flexible vehicle-charging-network resource prediction method considering large-scale electric vehicle access as described in claim 5, characterized in that, The process of obtaining the peak interval duration is as follows: if the last moment of each control cycle is not within the peak charging period, then the time interval between the last moment of each control cycle and the start moment of the peak charging period is counted and used as the peak interval duration; otherwise, the peak interval duration is assigned a preset value.

7. The flexible vehicle-charging-network resource prediction method considering large-scale electric vehicle access as described in claim 1, characterized in that, The evaluation coefficient is positively correlated with both the first evaluation value and the second evaluation value.

8. The flexible vehicle-charging-network resource prediction method considering large-scale electric vehicle access as described in claim 1, characterized in that, Charging station at Each regulation cycle corresponds to the adjusted smoothing parameter. The calculation formula is: ,in, The preset initial smoothing parameters, For the first The evaluation coefficient for each regulatory cycle, This is the preset adjustment range.

9. The flexible vehicle-charging-network resource prediction method considering large-scale electric vehicle access as described in claim 1, characterized in that, The method of predicting the charging demand load of the charging station in the next control cycle includes: using the EMA algorithm based on the adjusted smoothing parameters to predict the charging demand load of the charging station in the next control cycle, and obtaining the predicted charging demand load at each time point in the next control cycle.

10. The flexible vehicle-charging-network resource prediction method considering large-scale electric vehicle access as described in claim 9, characterized in that, The assessment of the adjustable and flexible resource margin of the charging station in the next control cycle includes: For the last moment of each control cycle, obtain the maximum theoretical charging load and maximum reverse discharge load of the charging station at that moment; The adjustable flexible resource margin includes upward flexible resource margin and downward flexible resource margin. The upward flexible resource margin of the charging station is the difference between the maximum theoretical charging load and the predicted charging demand load. The downward flexible resource margin of the charging station is the sum between the predicted charging demand load and the maximum reverse discharge load.

Images