Electric vehicle charging load prediction method based on vehicle network interaction

Abstract

The invention discloses an electric vehicle charging load prediction method based on vehicle network interaction. The method comprises the following steps: obtaining bidirectional information flow generated in a vehicle network interaction process; based on the bidirectional information flow, a dynamic coupling model is constructed, and the dynamic coupling model is used for representing the dynamic coupling relation between the traffic network and the power network and simulating the space-time distribution characteristics of the charging load; performing initial prediction on the charging load of the electric vehicle by using the dynamic coupling model to obtain an initial load prediction value; according to a power grid dispatching instruction and a charging electricity price excitation signal which are obtained in real time, the initial load prediction value is dynamically corrected in combination with vehicle response characteristic data, and a final charging load prediction value is generated. According to the method, the problems that a traditional prediction model cannot describe the dynamic coupling relation sufficiently and lacks a bidirectional information fusion mechanism are solved, and the charging load prediction accuracy, adaptability and responsiveness are improved.

Description

Technical Field

[0001] This invention relates to the field of electric vehicle and smart grid collaborative optimization technology, and in particular to a method for predicting electric vehicle charging load based on vehicle-grid interaction. Background Technology

[0002] With the large-scale integration of electric vehicles into the power grid, vehicle-to-grid (V2G) interaction technology has become a research hotspot. Existing charging load forecasting methods are mainly based on historical charging data statistical analysis or simple time-series models, failing to fully exploit the bidirectional information flow generated during V2G interaction. On the one hand, traditional forecasting models are insufficient in characterizing the dynamic coupling relationship between vehicle travel patterns, user charging behavior, and grid status, making it difficult to accurately reflect the spatiotemporal distribution characteristics of load under the deep integration of transportation and power networks, resulting in significant deviations between forecast results and actual conditions. On the other hand, existing technologies lack an effective mechanism for bidirectional V2G information fusion, unable to dynamically correct forecast results in real time using multi-source heterogeneous information such as grid dispatch instructions, charging price incentives, and vehicle response characteristics. This leads to significantly insufficient adaptability and responsiveness of forecasting models when facing complex operating conditions such as grid load fluctuations and traffic changes, making it difficult to meet the needs of smart grids for refined control of charging load.

[0003] Therefore, there is an urgent need to provide a technical solution to address the above problems. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides a method for predicting electric vehicle charging load based on vehicle-to-grid interaction. The technical solution of this method is as follows:

[0005] The system acquires bidirectional information streams generated during vehicle-to-grid interaction, including vehicle travel pattern data, user charging behavior data, power grid status data, power grid dispatch instructions, charging price incentive signals, and vehicle response characteristic data.

[0006] Based on the bidirectional information flow, a dynamic coupling model is constructed. The dynamic coupling model is used to characterize the dynamic coupling relationship between the transportation network and the power network, and to simulate the spatiotemporal distribution characteristics of the charging load.

[0007] Using the aforementioned dynamic coupling model, an initial prediction of the electric vehicle charging load is made to obtain the initial load prediction value;

[0008] Based on the real-time acquired power grid dispatch instructions and charging price incentive signals, combined with the vehicle response characteristic data, the initial load forecast value is dynamically corrected to generate the final charging load forecast value.

[0009] The technical solution of this invention obtains bidirectional information flow between vehicles and the power grid and constructs a dynamic coupling model between the transportation network and the power grid. It then dynamically corrects the prediction results by combining power grid dispatch instructions, charging price incentives, and vehicle response characteristics. This solves the problems of insufficient characterization of dynamic coupling relationships and lack of bidirectional information fusion mechanism in traditional prediction models, thereby improving the accuracy, adaptability, and responsiveness of charging load prediction.

[0010] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, and in order to make the above and other objects, features and advantages of the present invention more apparent and understandable, specific embodiments of the present invention are described below. Attached Figure Description

[0011] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0012] The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:

[0013] Figure 1 This is a flowchart illustrating an embodiment of the electric vehicle charging load prediction method based on vehicle-to-grid interaction according to the present invention. Detailed Implementation

[0014] Exemplary embodiments of the invention will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the invention are shown in the drawings, it should be understood that the invention can be implemented in various forms and should not be limited to the embodiments set forth herein.

[0015] Figure 1 This diagram illustrates a flowchart of an embodiment of an electric vehicle charging load prediction method based on vehicle-to-grid interaction provided by the present invention, executed by a control terminal. Figure 1 As shown, it includes the following steps:

[0016] S1. Acquire the bidirectional information flow generated during the vehicle-to-grid interaction process. The bidirectional information flow includes vehicle travel pattern data, user charging behavior data, power grid status data, power grid dispatch instructions, charging price incentive signals, and vehicle response characteristic data.

[0017] The vehicle-to-grid (V2G) interaction process refers to the two-way interaction between electric vehicles and the power grid through information and energy exchange. For example, when an electric vehicle connects to a charging station in area A, its battery status and charging needs are uploaded to the power grid management center. Simultaneously, the power grid sends an instruction to adjust the charging time based on the area's load conditions. Two-way information flow refers to the data set generated during the V2G interaction process and transmitted bidirectionally between the transportation and power systems. For example, real-time location data uploaded by the electric vehicle constitutes the information flow from vehicle to grid, while time-of-use pricing signals from the power grid constitute the information flow from grid to vehicle. Vehicle travel pattern data refers to characteristic data reflecting the travel time, route, frequency, and destination of electric vehicles. For example, analyzing data recorded by a fleet management platform reveals that vehicle EV-001 departs from area B at 8:00 AM on a weekday, traveling via a main road to the business district in area C. User charging behavior data refers to characteristic data reflecting the charging habits of electric vehicle users. For example, user A typically chooses a DC fast charging station near their workplace between 5:00 PM and 7:00 PM on weekdays when the vehicle's remaining battery level is below 30%. Power grid status data refers to a set of parameters reflecting the real-time operating conditions of the power network; for example, a substation in a certain area monitors that the load rate of the distribution network lines at 14:30 is 85%, the node voltage deviation is -5%, and the frequency fluctuation range is ±0.02Hz of the normal value. Power grid dispatch instructions refer to load control commands issued by the power grid control center to maintain system stability; for example, when the peak electricity consumption in a region approaches, the dispatch center sends an instruction to all charging stations in area A to reduce the total charging power by 20% between 18:00 and 20:00. Charging price incentive signals refer to dynamic price signals set to guide charging behavior; for example, the electricity price publishing system publishes an evening off-peak electricity price to charging stations in area B: charging fees from 22:00 on the same day to 6:00 the next day are settled at 60% of the standard electricity price. Vehicle response characteristic data refers to parameters reflecting the responsiveness of electric vehicles to power grid control instructions; for example, statistics show that the average response delay of a certain model of electric vehicle to peak-shaving instructions is 3 minutes, and the response sensitivity coefficient to electricity price incentives is 0.8.

[0018] S2. Based on the bidirectional information flow, a dynamic coupling model is constructed. The dynamic coupling model is used to characterize the dynamic coupling relationship between the transportation network and the power network, and to simulate the spatiotemporal distribution characteristics of the charging load.

[0019] The dynamic coupling model refers to a mathematical model characterizing the interaction between the transportation network and the power network. For example, the established model can simulate a sudden increase in the load of charging stations in adjacent areas due to traffic congestion on main roads and calculate the impact of this event on the voltage of the surrounding distribution network. The transportation network refers to the physical system composed of roads, vehicles, and transportation facilities; for example, the transportation infrastructure system including urban main roads, expressways, and intersection signal control systems. The power network refers to the power supply system composed of power generation, transmission, and distribution equipment; for example, the distribution network architecture consisting of substations, feeders, and low-voltage distribution areas covering a certain area. The dynamic coupling relationship refers to the correlation between changes in traffic flow and fluctuations in power grid load; for example, congestion on main roads during the evening peak hours leads to a delay in charging demand in adjacent areas, thus causing a change in the shape of the evening peak load curve of the power grid in that area. Charging load refers to the power demand generated in the power grid by the charging behavior of electric vehicles; for example, 10 electric vehicles charging simultaneously at 60kW at a charging station generate a total power demand of 600kW at 18:00. Spatiotemporal distribution characteristics refer to the variation patterns of charging load in time and space. For example, the charging load in business districts peaks between 9:00 and 11:00 on weekdays, while the charging load in residential areas shows a clustering characteristic between 20:00 and 22:00.

[0020] S3. Using the dynamic coupling model, make an initial prediction of the electric vehicle charging load to obtain the initial load prediction value.

[0021] The initial load forecast refers to the uncorrected load forecast result calculated by the dynamic coupling model; for example, the baseline value of the charging load in area A at 14:00 tomorrow is predicted to be 850kW based on historical data.

[0022] S4. Based on the real-time acquired power grid dispatch instructions and the charging electricity price incentive signal, and combined with the vehicle response characteristic data, dynamically correct the initial load forecast value to generate the final charging load forecast value.

[0023] The final charging load forecast refers to the load forecast result after being corrected by grid dispatch instructions and electricity price incentives; for example, after considering the impact of peak shaving instructions, the initial forecast value of 850kW is corrected to the actual forecast value of 720kW.

[0024] The technical solution of this embodiment obtains the bidirectional information flow of vehicle-to-grid interaction and constructs a dynamic coupling model of traffic network and power network. It combines power grid dispatch instructions, charging price incentives and vehicle response characteristics to dynamically correct the prediction results, which solves the problems of insufficient characterization of dynamic coupling relationship and lack of bidirectional information fusion mechanism in traditional prediction models, and improves the accuracy, adaptability and responsiveness of charging load prediction.

[0025] In one alternative approach, S1 specifically includes:

[0026] Through the vehicle-to-grid interaction platform, the system obtains real-time data on vehicle travel patterns and vehicle response characteristics from vehicle terminals, real-time data on user charging behavior from charging facilities, real-time data on grid status and grid dispatch instructions from the grid monitoring system, and real-time charging price incentive signals from the electricity price publication system.

[0027] Among them, the vehicle-to-grid (V2G) interaction platform refers to a software system that enables V2G data exchange and business collaboration; for example, a V2G interaction management platform simultaneously accesses electric vehicle fleet data and power distribution automation system data. Vehicle terminals refer to data acquisition and communication devices installed on electric vehicles; for example, a certain type of electric vehicle onboard terminal can collect vehicle location information, remaining battery power, and charging interface status in real time. Charging facilities refer to the infrastructure that provides power replenishment for electric vehicles; for example, a DC fast charging pile installed in a certain industrial park is equipped with charging status monitoring and billing modules. Power grid monitoring systems refer to systems that monitor and control the operating status of the power grid; for example, a regional energy management system collects load data and switch status of substations within its jurisdiction in real time. Electricity price publishing systems refer to computer systems that generate and publish dynamic electricity price information; for example, an electricity price publishing platform operated by a power trading center updates time-of-use electricity price data every 15 minutes.

[0028] Among the above-mentioned optional methods, the real-time acquisition path of multi-source data is further clarified, ensuring the timeliness and accuracy of information such as vehicle travel patterns, user behavior, power grid status, dispatch instructions and electricity price signals, and providing comprehensive data support for the subsequent construction of dynamic coupling models.

[0029] In one alternative approach, S2 specifically includes:

[0030] The vehicle travel pattern data, the user charging behavior data, and the power grid status data are subjected to multi-source data fusion processing to generate a fusion feature set.

[0031] Among them, the fusion feature set refers to the set of feature vectors after multi-source data has been fused; for example, vehicle trajectory data, charging records and power grid operation data are fused to generate a multi-dimensional feature vector containing spatiotemporal features, behavioral features and power grid features.

[0032] Based on the fused feature set, a traffic network sub-model and a power network sub-model are established, and the traffic network sub-model and the power network sub-model are connected by a coupling function to form the dynamic coupling model.

[0033] The traffic network sub-model refers to the algorithm module that simulates vehicle movement and traffic conditions; for example, a road network speed prediction model built based on travel data can calculate the travel time of each road segment during a specific time period. The power network sub-model refers to the algorithm module that simulates power flow and electrical conditions in the power grid; for example, a distribution network model based on power flow calculation can solve for the voltage distribution at each node and the power distribution in each branch. The coupling function refers to the mathematical expression that establishes the relationship between the traffic model and the power grid model; for example, the charging demand allocation function is defined as the product of the probability of vehicle arrival in the traffic network and the charging availability at power grid nodes.

[0034] The dynamic coupling model is used to simulate the distribution characteristics of charging load in time and space.

[0035] In the above-mentioned optional methods, a fusion feature set is further generated by multi-source data fusion, a traffic network sub-model and a power network sub-model are established, and a coupling function is used to connect them to realize the quantitative representation of the dynamic relationship between traffic and power networks, thereby improving the simulation accuracy of the spatiotemporal distribution characteristics of charging load.

[0036] In one alternative approach, S3 specifically includes:

[0037] The real-time collected vehicle travel pattern data, user charging behavior data, and power grid status data are input into the dynamic coupling model, and the spatiotemporal distribution of electric vehicle charging demand within the predicted time period is calculated through the dynamic coupling model.

[0038] The forecast period refers to the time range covered by the load forecast; for example, 24 consecutive time periods are set when making a day-ahead forecast, with each time period lasting 1 hour. The spatiotemporal distribution of charging demand refers to the quantitative distribution of charging demand in time and space within the forecast period; for example, the forecast results show that the business district will generate a total charging demand of 350kW during the morning peak from 8:00 to 9:00 tomorrow.

[0039] Based on the spatiotemporal distribution of charging demand, the initial load forecast value is obtained by aggregation.

[0040] In the above-mentioned optional methods, real-time data is further input into the dynamic coupling model to calculate the spatiotemporal distribution of charging demand during the prediction period, and aggregate to generate initial load prediction values, thereby realizing the transformation from single-vehicle demand to overall load and providing benchmark prediction results for dynamic correction.

[0041] In one optional approach, the step of inputting the real-time collected vehicle travel pattern data, user charging behavior data, and power grid status data into the dynamic coupling model, and calculating the spatiotemporal distribution of electric vehicle charging demand within a predicted time period using the dynamic coupling model, further includes:

[0042] Based on the historical trajectory set and real-time traffic flow data in the vehicle travel pattern data, the time probability distribution of each electric vehicle arriving at different charging areas during the prediction period is calculated through the traffic network sub-model in the dynamic coupling model.

[0043] Historical trajectory sets refer to the collection of data on past vehicle routes; for example, the trajectory records of all transportation tasks undertaken by an electric vehicle fleet over the past 30 days, including timestamps and location coordinate sequences. Real-time traffic flow data refers to dynamic data reflecting the current road traffic conditions; for example, real-time traffic information provided by a traffic information platform, including average vehicle speeds and congestion levels for each road segment. Charging areas refer to charging service areas defined by geographical location; for example, dividing the main urban area into multiple grid areas with sides of 2km, each grid being defined as a charging area. Temporal probability distribution refers to the probability distribution of a vehicle arriving at a specific location at a specific time; for example, statistical analysis shows that the probability of a ride-hailing vehicle arriving in a business district between 8:00 and 9:00 on a weekday is 0.15.

[0044] Based on the user's state of charge preference and charging time period preference in the charging behavior data, the time probability distribution is modified according to behavioral preferences to generate charging decision probabilities.

[0045] Here, "state of charge preference" refers to a user's psychological expectation threshold for the remaining battery power of a vehicle; for example, surveys show that most taxi drivers are willing to charge when the battery level is below 25%. "Charging time preference" refers to the distribution characteristics of users' preferred charging times; for example, private car users mainly charge between 18:00 and 22:00, while commercial vehicle users mainly charge between 23:00 and 4:00 the next day. "Charging decision probability" refers to the probability value of a user choosing to charge after considering various factors; for example, the probability of a user choosing to charge a vehicle with 30% battery power, located in a commercial area, and experiencing electricity price discounts is calculated to be 0.82.

[0046] Based on the node voltage and line capacity in the power grid status data, the dynamic carrying weight of each charging area is calculated through the power network sub-model in the dynamic coupling model.

[0047] Node voltage refers to the effective voltage value at a measurement point in the power network; for example, the voltage value recorded by a voltage monitoring device at a distribution node is maintained within ±5% of the rated value. Line capacity refers to the maximum current that a power line is allowed to carry continuously; for example, a distribution line with a rated current carrying capacity of 400A corresponds to a theoretical transmission capacity of approximately 6.9 MVA. Dynamic load weight refers to a dynamic indicator reflecting the charging acceptance capacity of a power grid node; for example, based on real-time calculations, the weighting coefficient for adding charging load to a node under its current operating state is 0.75.

[0048] The spatiotemporal distribution of charging demand is calculated by fusing the charging decision probability and the dynamic load weight through the coupling function in the dynamic coupling model.

[0049] Among the above-mentioned optional methods, the arrival time probability is further calculated based on vehicle trajectory and real-time traffic flow, combined with user state of charge and charging time preference correction, and dynamic carrying weight is calculated considering grid node voltage and line capacity. The spatiotemporal distribution of charging demand is then generated to improve the accuracy of spatial prediction.

[0050] In one alternative approach, the formula for calculating the spatiotemporal distribution of charging demand is: ;in, This represents the charging load at time t and spatial location s. This represents the total number of active electric vehicles during the forecast period. This indicates the standard charging power of vehicle v. Indicates a given set of traffic conditions The vehicle v's state of charge is below the charging threshold at time t. The conditional probability, This indicates that vehicle v is in time segment The spatial distribution probability of reaching position s Represents the node voltage based on location s and line current The dynamic load-bearing weight function is calculated. This represents a time selection function. This represents the total number of time segments within the predicted period.

[0051] It should be noted that the formula for calculating the spatiotemporal distribution of charging demand is based on a probabilistic model and power grid physical constraints. It calculates the spatiotemporal distribution of charging demand by aggregating charging behavior data of all active electric vehicles, traffic conditions, and power grid status data within the prediction period. This formula integrates the vehicle's standard charging power, the conditional probability of the state of charge being below a threshold, the probability distribution of the vehicle's arrival location, the power grid's dynamic load weighting function, and the time selection function, reflecting the dynamic coupling relationship between the transportation network and the power network in the spatiotemporal dimension. By accurately quantifying the temporal and spatial distribution characteristics of electric vehicle charging demand, this formula provides fine-grained data input for subsequent load aggregation, thereby addressing the problem of insufficient spatiotemporal characterization in traditional prediction models.

[0052] Among the above-mentioned optional methods, a quantitative formula is further used to comprehensively consider vehicle charging power, state of charge threshold probability, spatial distribution probability, dynamic load weight, and time selection function to achieve accurate calculation of charging load in the spatiotemporal dimension, providing a refined data foundation for load aggregation.

[0053] In one alternative approach, the step of aggregating the initial load forecast value based on the spatiotemporal distribution of charging demand includes:

[0054] Based on the power grid topology, the electrical correlation between nodes at various spatial locations is determined.

[0055] In this context, the power grid topology refers to the connection relationships between various components in a power network; for example, a distribution network using a ring network design includes the connection methods of multiple substations, feeders, and sectionalizing switches. A spatial node refers to a location point in the power grid topology that possesses independent electrical characteristics; for example, node N01 defined in a distribution network model corresponds to the actual physical location of a regional distribution box. Electrical connectivity refers to the tightness of the electrical connection between power grid nodes; for example, impedance matrix calculations show an electrical distance of 0.35 between node A and node B, resulting in a connectivity index of 0.78.

[0056] Based on the spatiotemporal distribution of charging demand and the electrical correlation, spatial load weighted aggregation is performed, taking into account line capacity constraints and node voltage stability.

[0057] Among these, line capacity constraint refers to the limitation that the power transmitted by a power grid line must not exceed its rated capacity; for example, if the maximum allowable load rate of a cable line is 90%, the corresponding power limit is 5.6MW. Node voltage stability refers to the operational requirement that the voltage at power grid nodes be maintained within the allowable range; for example, the acceptable voltage range for a low-voltage distribution area is specified as ±7% of the rated voltage. Spatial load weighted aggregation refers to a method of summing regional loads considering electrical interdependencies; for example, the charging demand of multiple adjacent nodes is weighted and summed according to the inverse of electrical distance to obtain the predicted total regional load.

[0058] The weighted aggregated spatial load is subjected to time series convolution smoothing to generate the initial load prediction value.

[0059] The weighted aggregated spatial load refers to the spatial load distribution after electrical correlation weighting. For example, if the original loads of three nodes in a certain zone are 100kW, 150kW, and 200kW respectively, the total regional load after weighted aggregation is 420kW.

[0060] In the above-mentioned optional methods, the electrical correlation between nodes is further determined based on the power grid topology, spatial load weighting aggregation is performed by combining line capacity and voltage stability, and smooth initial load forecast values ​​that conform to power grid constraints are generated through time series convolution smoothing.

[0061] In one alternative approach, the initial load forecast is calculated using the following formula: ;in, This represents the initial load forecast at time t. Represents the total number of nodes in spatial location. This represents the maximum allowed capacity of node s. This represents the current at node s at time t. This represents the voltage at node s at time t. This represents the electrical impedance between node s and node j. This represents the load factor from node s to node j at time t.

[0062] It should be noted that the initial load forecast calculation formula spatially weights and aggregates the spatiotemporal distribution of charging demand through grid topology and electrical connectivity, and incorporates node capacity margin, electrical impedance, and load factor influences to reflect grid operation constraints. This formula adjusts load allocation based on the ratio of the maximum allowable capacity of a node to its current power, and quantifies the mutual influence between nodes through electrical impedance and load factor, thereby ensuring that the aggregation process meets line capacity and node voltage stability requirements. By generating initial load forecasts that not only integrate spatial load distribution but also consider grid safety operation boundaries, this formula improves the engineering practicality and reliability of the forecast results.

[0063] Among the above-mentioned optional methods, the load aggregation calculation considering the grid security constraints is further realized by integrating the node capacity margin, electrical impedance and load rate influence factors through quantitative formulas, ensuring that the initial prediction value is within the grid carrying capacity range and improving the engineering practicality of the prediction results.

[0064] In one alternative approach, S4 specifically includes:

[0065] The load adjustment instruction value is parsed and extracted from the power grid dispatch instruction.

[0066] Among them, the load adjustment instruction value refers to the quantitative indicator of load adjustment specified in the power grid dispatch instruction; for example, if the dispatch instruction requires a certain area to reduce the load by 150kW during the evening peak period, this value is the load adjustment instruction value.

[0067] The price incentive intensity value is extracted from the charging price incentive signal.

[0068] Among them, the price incentive strength value refers to the quantitative value of the amplitude of the electricity price incentive signal; for example, the peak electricity price increases by 0.8 yuan / kWh, the normal electricity price remains unchanged, and the off-peak electricity price decreases by 0.5 yuan / kWh. These values ​​constitute the price incentive strength value.

[0069] The vehicle's response rate to dispatch instructions and its response rate to electricity price incentives are obtained from the vehicle response characteristic data.

[0070] The vehicle response rate to dispatch instructions refers to the proportion of vehicles that actually respond to dispatch instructions. For example, if a fleet receives a peak-shaving instruction and 82% of its vehicles actually adjust their charging power, this proportion is the response rate. The vehicle response rate to electricity price incentives refers to the proportion of vehicles that adjust their charging behavior due to changes in electricity prices. For example, after the implementation of off-peak electricity pricing, if the proportion of vehicles choosing to charge during off-peak hours increases from 30% to 65%, this 35% increase is the response rate.

[0071] A dynamic correction factor is calculated based on the load adjustment instruction value, the price incentive intensity value, the vehicle's response rate to the dispatch instruction, and the vehicle's response rate to the electricity price incentive.

[0072] The dynamic correction factor refers to the dynamic adjustment coefficient used to correct the initial forecast value. For example, after taking into account dispatch instructions and electricity price incentives, the load forecast correction factor for the period at 14:00 tomorrow is calculated to be 0.85.

[0073] The initial load forecast value is corrected using the dynamic correction factor to generate the final charging load forecast value.

[0074] In the above-mentioned optional methods, the adjustment amount is further extracted from the power grid dispatch instructions and electricity price incentive signals. The dynamic correction factor is calculated in combination with the vehicle's response rate to dispatch and electricity price, and the initial prediction value is corrected in real time so that the prediction result can reflect the dynamic impact of external incentives and changes in vehicle behavior.

[0075] In one alternative approach, the formula for calculating the final charging load prediction is:

[0076] ;in, This represents the final predicted charging load at time t. This represents the load adjustment instruction value extracted from the power grid dispatch instruction. This represents the price incentive strength value extracted from the charging electricity price incentive signal. This represents the vehicle's response rate to dispatch instructions, obtained from vehicle response characteristic data. This represents the vehicle's response rate to electricity price incentives, obtained from vehicle response characteristic data. , and This indicates the preset correction weight coefficient.

[0077] It should be noted that the formula for calculating the final charging load forecast utilizes a dynamic correction factor to adjust the initial load forecast in real time. This correction factor is calculated based on the load adjustment value of the grid dispatch command, the intensity of electricity price incentives, vehicle response rates, and their interactions. This formula balances the impact of dispatch commands and electricity price incentives through preset weighting coefficients and introduces cross-terms to capture the coupling effect of multiple factors, thereby achieving refined correction of the initial forecast value. Generating the final charging load forecast using this formula enables the forecast results to dynamically respond to grid control strategies and market signals, improving the adaptability and accuracy of the forecast model under complex operating conditions.

[0078] Among the above-mentioned optional methods, a quantitative formula is further used to integrate load adjustment instructions, price incentive intensity and corresponding response rate, and cross-action terms and weighting coefficients are introduced to achieve refined dynamic correction of the initial forecast value, thereby improving the accuracy of charging load forecast in response to grid regulation and market incentives.

[0079] The above description is merely a preferred embodiment of the present invention and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of disclosure in this invention is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-disclosed concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features with similar functions disclosed in this invention.

[0080] It should be noted that the terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and represent a limitation on a specific order or sequence. Where appropriate, the order of use for similar objects can be interchanged so that the embodiments of this application described herein can be implemented in an order other than that shown or described.

[0081] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A method for predicting electric vehicle charging load based on vehicle-to-grid interaction, characterized in that, The method includes: The system acquires bidirectional information streams generated during vehicle-to-grid interaction, including vehicle travel pattern data, user charging behavior data, power grid status data, power grid dispatch instructions, charging price incentive signals, and vehicle response characteristic data. Based on the bidirectional information flow, a dynamic coupling model is constructed. This dynamic coupling model is used to characterize the dynamic coupling relationship between the transportation network and the power network, and to simulate the spatiotemporal distribution characteristics of the charging load. Using the aforementioned dynamic coupling model, an initial prediction of the electric vehicle charging load is made to obtain the initial load prediction value; Based on the real-time acquired power grid dispatch instructions and charging price incentive signals, combined with the vehicle response characteristic data, the initial load forecast value is dynamically corrected to generate the final charging load forecast value.

2. The electric vehicle charging load prediction method based on vehicle-to-grid interaction according to claim 1, characterized in that, The step of acquiring the bidirectional information flow generated during the vehicle-to-network interaction process further includes: Through the vehicle-to-grid interaction platform, the system obtains real-time data on vehicle travel patterns and vehicle response characteristics from vehicle terminals, real-time data on user charging behavior from charging facilities, real-time data on grid status and grid dispatch instructions from the grid monitoring system, and real-time charging price incentive signals from the electricity price publication system.

3. The electric vehicle charging load prediction method based on vehicle-to-grid interaction according to claim 1, characterized in that, The step of constructing a dynamic coupling model based on the bidirectional information flow further includes: The vehicle travel pattern data, the user charging behavior data, and the power grid status data are subjected to multi-source data fusion processing to generate a fusion feature set; Based on the fused feature set, a traffic network sub-model and a power network sub-model are established, and the traffic network sub-model and the power network sub-model are connected by a coupling function to form the dynamic coupling model; The dynamic coupling model is used to simulate the distribution characteristics of charging load in time and space.

4. The electric vehicle charging load prediction method based on vehicle-to-grid interaction according to claim 3, characterized in that, The step of using the dynamic coupling model to make an initial prediction of the electric vehicle charging load and obtain an initial load prediction value further includes: The real-time collected vehicle travel pattern data, user charging behavior data, and power grid status data are input into the dynamic coupling model, and the spatiotemporal distribution of electric vehicle charging demand within the predicted period is calculated through the dynamic coupling model. Based on the spatiotemporal distribution of charging demand, the initial load forecast value is obtained by aggregation.

5. The electric vehicle charging load prediction method based on vehicle-to-grid interaction according to claim 4, characterized in that, The step of inputting the real-time collected vehicle travel pattern data, user charging behavior data, and power grid status data into the dynamic coupling model, and calculating the spatiotemporal distribution of electric vehicle charging demand within the predicted time period using the dynamic coupling model, further includes: Based on the historical trajectory set and real-time traffic flow data in the vehicle travel pattern data, the time probability distribution of each electric vehicle arriving at different charging areas during the prediction period is calculated through the traffic network sub-model in the dynamic coupling model. Based on the state of charge preference and charging time preference in the user charging behavior data, the time probability distribution is corrected for behavioral preferences to generate charging decision probabilities. Based on the node voltage and line capacity in the power grid status data, the dynamic carrying weight of each charging area is calculated through the power network sub-model in the dynamic coupling model. The spatiotemporal distribution of charging demand is calculated by fusing the charging decision probability and the dynamic load weight through the coupling function in the dynamic coupling model.

6. The electric vehicle charging load prediction method based on vehicle-to-grid interaction according to claim 5, characterized in that, The formula for calculating the spatiotemporal distribution of charging demand is: ;in, This represents the charging load at time t and spatial location s. This represents the total number of active electric vehicles during the forecast period. This indicates the standard charging power of vehicle v. Indicates that in a given set of traffic conditions The vehicle v's state of charge is below the charging threshold at time t. The conditional probability, This indicates that vehicle v is in time segment The spatial distribution probability of reaching position s Represents the node voltage based on location s and line current The dynamic load-bearing weight function is calculated. This represents a time selection function. This represents the total number of time segments within the predicted period.

7. The electric vehicle charging load prediction method based on vehicle-to-grid interaction according to claim 6, characterized in that, The step of aggregating the initial load forecast value based on the spatiotemporal distribution of charging demand includes: Based on the power grid topology, determine the electrical correlation between nodes at various spatial locations; Based on the spatiotemporal distribution of charging demand and the electrical correlation, spatial load weighting aggregation is performed, taking into account line capacity constraints and node voltage stability. The weighted aggregated spatial load is subjected to time series convolution smoothing to generate the initial load prediction value.

8. The electric vehicle charging load prediction method based on vehicle-to-grid interaction according to claim 7, characterized in that, The formula for calculating the initial load forecast value is as follows: ;in, This represents the initial load forecast at time t. Represents the total number of nodes in spatial location. This represents the maximum allowed capacity of node s. This represents the current at node s at time t. This represents the voltage at node s at time t. This represents the electrical impedance between node s and node j. This represents the load factor from node s to node j at time t.

9. The electric vehicle charging load prediction method based on vehicle-to-grid interaction according to claim 8, characterized in that, The step of dynamically correcting the initial load forecast value based on the real-time acquired power grid dispatch instructions and the charging electricity price incentive signal, combined with the vehicle response characteristic data, to generate the final charging load forecast value, further includes: The load adjustment instruction value is parsed and extracted from the power grid dispatch instruction; The price incentive intensity value is extracted from the charging electricity price incentive signal; The vehicle response rate to dispatch instructions and the vehicle response rate to electricity price incentives are obtained from the vehicle response characteristic data. Based on the load adjustment instruction value, the price incentive intensity value, the vehicle's response rate to the dispatch instruction, and the vehicle's response rate to the electricity price incentive, a dynamic correction factor is calculated. The initial load forecast value is corrected using the dynamic correction factor to generate the final charging load forecast value.

10. The electric vehicle charging load prediction method based on vehicle-to-grid interaction according to claim 9, characterized in that, The formula for calculating the final charging load forecast is as follows: ;in, This represents the final predicted charging load at time t. This represents the load adjustment instruction value extracted from the power grid dispatch instruction. This represents the price incentive strength value extracted from the charging electricity price incentive signal. This represents the vehicle's response rate to dispatch instructions, obtained from vehicle response characteristic data. This represents the vehicle's response rate to electricity price incentives, obtained from vehicle response characteristic data. , and This indicates the preset correction weight coefficient.

Images