Electric vehicle charging load prediction method and system based on coupling evolution of traffic and soc

Abstract

The application discloses a traffic and SOC coupling evolution-based electric vehicle charging load prediction method and system, relates to the technical field of power system load prediction, and comprises the following steps: obtaining traffic network, travel demand and electric vehicle energy consumption parameter data of a target region; performing traffic flow distribution based on the traffic network and travel demand data to obtain dynamic space-time distribution of vehicles; establishing an initial state of charge probability model of vehicle clusters, calculating SOC consumption based on the dynamic space-time distribution, driving evolution of the probability model, determining a charging group based on the state of charge probability distribution when arriving, simulating a charging process to obtain charging probability time-varying characteristics; and aggregating the dynamic space-time distribution, the charging probability time-varying characteristics and charging power parameters to obtain total charging load. The application solves the technical problems that traditional methods cannot capture load mutation caused by traffic congestion and are difficult to simulate the evolution process of individual vehicle charging decisions, and significantly improves the space-time accuracy of charging load prediction.

Description

Technical Field

[0001] This invention relates to the field of power system load forecasting technology, and more specifically, to a method and system for forecasting electric vehicle charging load based on the coupled evolution of traffic and SOC. Background Technology

[0002] With the continuous advancement of transportation electrification and energy structure transformation, the number of electric vehicles (EVs) is growing rapidly, and their charging load has become one of the fastest-growing and most volatile load types in the power system. EV charging behavior is influenced by multiple factors, including traffic conditions, travel routes, driving habits, and battery state of charge (SOC), exhibiting significant spatiotemporal randomness and clustering. Accurate prediction of EV charging load is not only a crucial foundation for charging infrastructure planning, grid load balancing, and demand-side response, but also a key link in ensuring the safe operation of the distribution network and achieving coordinated and optimized scheduling of power sources, grid, and load. Therefore, establishing a high-precision charging load prediction method that reflects traffic dynamics and energy state evolution characteristics has significant engineering application value and practical significance.

[0003] However, existing charging load forecasting methods still have significant limitations. First, most methods treat the traffic system and battery state as static or isolated factors, failing to fully consider the real-time impact of dynamic traffic flow on vehicle battery state of charge. For example, in congested traffic, vehicle energy consumption increases sharply, significantly altering its state of charge and inducing temporary emergency charging demands, which existing models struggle to capture. Second, existing models are insufficient in describing the stochasticity of the complex "vehicle-road-electricity" system. They often rely on simple average speed or fixed path assumptions, failing to accurately simulate the evolution of individual vehicles' charging decisions based on real-time battery power and traffic conditions in complex road networks. This results in significant discrepancies between predicted and actual conditions, failing to meet the requirements of high-precision grid scheduling. Summary of the Invention

[0004] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide an electric vehicle charging load prediction method and system based on the coupled evolution of traffic and SOC. By establishing an initial state of charge probability model of the vehicle cluster and calculating SOC consumption based on dynamic spatiotemporal distribution, the method drives the evolution of the probability model. This addresses the problems of existing methods failing to capture load fluctuations caused by traffic congestion due to treating traffic systems and battery states as static and isolated factors, and having large prediction biases due to the inability to accurately simulate vehicle charging decisions by relying on simple average speed or fixed path assumptions.

[0005] To achieve the above objectives, the present invention provides the following technical solution: The electric vehicle charging load prediction method based on the coupled evolution of traffic and SOC includes the following steps: acquiring traffic network, travel demand, and electric vehicle energy consumption parameter data for the target area; allocating traffic flow based on the traffic network and travel demand data to obtain the dynamic spatiotemporal distribution of vehicles; establishing an initial state-of-charge probability model for the vehicle cluster, and calculating SOC consumption based on the dynamic spatiotemporal distribution, driving the evolution of the probability model to obtain the state-of-charge probability distribution upon arrival; determining the charging group based on the probability distribution, simulating the charging process to obtain the time-varying characteristics of charging probability; and aggregating the dynamic spatiotemporal distribution, the time-varying characteristics of charging probability, and the charging power parameters to obtain the total charging load.

[0006] In a preferred embodiment, the acquisition of traffic network, travel demand, and electric vehicle energy consumption parameter data for the target area specifically involves: constructing a dynamic traffic network that includes the free-flow speed and capacity of each link based on historical and real-time traffic data; identifying resident travel origin-destination (OD) pairs based on mobile signaling data, and combining the dynamic traffic network with gravity model inversion calculations to obtain the dynamic travel demand matrix for each OD pair within the predicted time period; acquiring vehicle battery capacity parameters, and establishing a dynamic energy consumption function per unit mileage based on historical driving data through data fitting.

[0007] In a preferred embodiment, the step of allocating traffic flow based on traffic network and travel demand data to obtain the dynamic spatiotemporal distribution of vehicles specifically involves: establishing a composite weighting function that considers real-time traffic impedance and expected energy consumption, wherein the expected energy consumption is determined based on a dynamic unit mileage energy consumption function; calculating the comprehensive toll cost of each alternative path based on the initial SOC of the vehicle and the composite weighting function; allocating the optimal path to the vehicle using a random user equilibrium allocation strategy according to the comprehensive toll cost; iteratively executing the comprehensive toll cost calculation and path allocation process until the road network traffic reaches a stable state, and outputting the dynamic spatiotemporal distribution of vehicles.

[0008] In a preferred embodiment, establishing the initial state of charge probability model for the vehicle cluster specifically involves: identifying key transportation hubs based on traffic network data; obtaining the SOC empirical probability distribution of each key hub at the initial prediction time of the same historical period based on historical data; determining the proportion of vehicle sources arriving at each key transportation hub through reverse traffic flow extrapolation based on the travel demand data of the current period; and weighting and fusing the SOC empirical probability distribution of the corresponding source areas according to the proportion of vehicle sources to generate the initial state of charge probability model of the key transportation hub in the current prediction period.

[0009] In a preferred embodiment, the step of calculating SOC consumption based on dynamic spatiotemporal distribution and driving the evolution of the probability model to obtain the probability distribution of state of charge upon arrival specifically involves: extracting micro-parameters reflecting traffic operation characteristics from the dynamic spatiotemporal distribution; calculating SOC consumption based on the micro-parameters and the dynamic energy consumption function per unit mileage and performing evolutionary deduction on the initial state of charge probability model; and aggregating the deduction results to generate the probability distribution of state of charge upon arrival.

[0010] In a preferred embodiment, the calculation of SOC consumption includes: analyzing the vehicle speed-time series based on the dynamic spatiotemporal distribution; inputting the speed-time series into the dynamic unit mileage energy consumption function to generate an instantaneous energy consumption pulse sequence; integrating the instantaneous energy consumption pulse sequence over the travel time, and converting the integration result into SOC consumption.

[0011] In a preferred embodiment, determining the charging group based on the probability distribution of the state of charge upon arrival and simulating the charging process to obtain the time-varying characteristics of the charging probability specifically involves: mapping the probability distribution of the state of charge upon arrival to a spatial pressure field of charging demand; spatiotemporally coupling the spatial pressure field with the service capacity of charging facilities to deduce the spatiotemporal propagation process of charging overflow; triggering preventive charging decisions based on the overflow propagation results and updating vehicle charging behavior; and iteratively executing the deduction and update process until the system reaches transient equilibrium, outputting the time-varying characteristics of the charging probability.

[0012] In a preferred embodiment, the aggregation of the dynamic spatiotemporal distribution, the time-varying characteristics of charging probability, and the charging power parameters to obtain the total charging load specifically involves: determining a spatiotemporal matrix of the number of vehicles based on the dynamic spatiotemporal distribution; fusing the time-varying characteristics of charging probability with the charging power parameters to obtain a time-varying function of the average charging power per vehicle; performing tensor operations on the spatiotemporal matrix and the time-varying function of the average charging power per vehicle to obtain the spatiotemporal distribution of the charging load, and then spatiotemporally aggregating it to obtain the total charging load.

[0013] A system for predicting electric vehicle charging load based on the coupled evolution of traffic and State of Charge (SOC) is characterized by comprising: a data acquisition module for acquiring traffic network, travel demand, and electric vehicle energy consumption parameter data for a target area; a traffic flow allocation module for allocating traffic flow based on traffic network and travel demand data to obtain the dynamic spatiotemporal distribution of vehicles; a coupled calculation module for establishing an initial state of charge probability model for a vehicle cluster, calculating SOC consumption based on the dynamic spatiotemporal distribution, driving the evolution of the probability model, and obtaining the state of charge probability distribution upon arrival; a charging simulation module for determining the charging group based on the probability distribution, simulating the charging process to obtain the time-varying characteristics of charging probability; and a load prediction module for aggregating the dynamic spatiotemporal distribution, the time-varying characteristics of charging probability, and charging power parameters to obtain the total charging load.

[0014] An electronic device includes: a memory and a processor; the memory is used to store a program; the processor is used to execute the program to implement the aforementioned electric vehicle charging load prediction method based on the coupled evolution of traffic and SOC.

[0015] The technical effects and advantages of the electric vehicle charging load prediction method and system based on the coupled evolution of traffic and SOC in this invention are as follows: 1. This invention establishes an initial state of charge probability model for vehicle clusters and calculates SOC consumption based on dynamic spatiotemporal distribution, driving the evolution of the probability model to achieve dynamic coupling between traffic flow and battery state of charge. It can accurately capture the accelerated battery consumption caused by traffic congestion and the resulting temporary charging demand, solving the technical problem of traditional methods that treat the traffic system as a static factor and cannot predict load changes.

[0016] 2. This invention determines the charging group based on the probability distribution, simulates the charging process to obtain the time-varying characteristics of the charging probability, and finely depicts the dynamic process of individual vehicles making charging decisions based on real-time power and traffic conditions in complex road networks. It overcomes the limitations of existing technologies that rely on simple average speed or fixed path assumptions, and significantly improves the spatiotemporal prediction accuracy of charging load. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating the electric vehicle charging load prediction method based on the coupled evolution of traffic and SOC of the present invention. Figure 2 This is a schematic diagram of the overall system architecture of the electric vehicle charging load prediction method based on the coupled evolution of traffic and SOC of the present invention; Figure 3 This is a structural block diagram of an exemplary electronic device provided for implementing embodiments of the present disclosure. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0019] Example 1, Figure 1 This invention presents a method for predicting electric vehicle charging load based on the coupled evolution of traffic and SOC, comprising the following steps: S1, acquire data on the traffic network, travel demand, and electric vehicle energy consumption parameters of the target area; In this embodiment, acquiring traffic network, travel demand, and electric vehicle energy consumption parameter data for the target area specifically involves: Static road network topology data for the target area (e.g., a city's main urban area) is obtained, sourced from OpenStreetMap or application programming interfaces (APIs) of commercial navigation map providers. This data includes at least the spatial geometry of road links, road classifications, and node connectivity. Historical traffic data is derived from floating car GPS trajectory data of the target area during the same time period (e.g., weekday morning rush hour 8:00-9:00) over the past few months. By performing map matching, speed calculation, and statistical analysis on the trajectory data, the free-flow speed of each road link under typical conditions is obtained. and traffic capacity Real-time traffic data comes from real-time traffic flow information provided by traffic management departments or real-time traffic services from online map platforms, and is used to obtain the real-time average speed of each link during the current forecast period. By associating static road network topology with dynamic traffic parameters, a dynamic traffic network is formed. This network can be represented as a directed graph. Where N is the set of nodes and L is the set of links. For each link... Its dynamic attribute set is .

[0020] The system processes anonymized mobile signaling data from several weeks prior to the forecast period. By identifying stable stops in user signaling location sequences, the origin (O) and destination (D) of each weekday's trips are extracted and aggregated at the traffic cell level to form a historical OD pair set. A dual-constraint gravity model is used to extrapolate dynamic travel demand. The basic form of this model is as follows:

[0021] in, It represents the trip volume from traffic zone i to traffic zone j. It represents the number of trips generated in traffic community i. It is the number of trips attracted by traffic in community j. This is the generalized travel cost from traffic zone i to traffic zone j. In this embodiment, it is calculated using the aforementioned dynamic traffic network and the shortest path algorithm, and its value is the time cost. , It is an impedance function. and It is a balance factor, which is ensured through iterative calculation. and Established at the same time.

[0022] By solving the above model, the dynamic travel demand matrix, refined to each traffic zone pair, can be obtained within the prediction period. .

[0023] Obtain historical driving data of a large number of electric vehicles within the target area from a vehicle monitoring platform or on-board terminal. Data entries should include: Vehicle Identification Number (VIN), trip segment ID, average speed, average acceleration, and energy consumption per unit mile corresponding to that segment. A multivariate nonlinear regression method is used to establish a function model with speed v and acceleration a as independent variables and energy consumption per unit mile e as the dependent variable, serving as the dynamic energy consumption per unit mile function.

[0024] in, , , , , These are the regression coefficients obtained by fitting using the least squares method. This item is used to distinguish the different energy consumption characteristics of acceleration (positive work) and deceleration (regenerative braking, negative work or zero work).

[0025] Unlike existing technologies that rely solely on static road network structures or average trip statistics for load forecasting, this embodiment constructs a dynamic traffic network by fusing historical and real-time traffic data. This network reflects the changing characteristics of road capacity and flow rate over time, thus providing a more realistic dynamic basis for subsequent traffic flow evolution.

[0026] Furthermore, this embodiment identifies travel origin-destination pairs based on mobile signaling data and obtains a dynamic travel demand matrix by combining it with gravity model inversion, thereby realizing the spatial correlation and temporal evolution modeling of travel behavior and capturing the changing patterns of vehicle flow direction and travel intensity between different regions.

[0027] Meanwhile, by establishing a dynamic energy consumption function per unit mileage based on historical driving data, energy consumption calculation can adapt to the dynamic influence of factors such as road congestion and speed fluctuations, thus avoiding the distortion problem of traditional fixed energy consumption coefficient models.

[0028] In summary, this embodiment enables collaborative modeling of traffic networks, travel behavior, and energy consumption characteristics, providing a data foundation with high spatiotemporal resolution, strong dynamic adaptability, and physical interpretability for subsequent SOC evolution and load forecasting.

[0029] S2, based on traffic network and travel demand data, allocates traffic flow to obtain the dynamic spatiotemporal distribution of vehicles; In this embodiment, the process of allocating traffic flow based on traffic network and travel demand data to obtain the dynamic spatiotemporal distribution of vehicles specifically involves: A composite weighting function considering real-time traffic impedance and expected energy consumption is established, specifically as follows:

[0030] in, This represents the total travel cost of choosing path p from starting point i to ending point j. This represents the real-time traffic impedance of the chosen path p from the starting point i to the ending point j. In this embodiment, it is characterized by the predicted travel time (in minutes), which is obtained by summing the results after dividing the length of each link in the path by its corresponding real-time average speed. This represents the expected energy cost of choosing path p from the starting point i to the ending point j. This cost is not a direct use of energy consumption values, but rather a dimensionless representation of the cost after associating it with the vehicle's battery state. and Let be the weighting coefficient, satisfying .

[0031] For each OD pair (i, j), use the K-shortest path algorithm to generate 3-5 alternative paths, forming a path set. The calculation process for the expected energy consumption is as follows: First, calculate the average speed of path p, and then substitute the calculated average speed into the dynamic energy consumption per unit distance function. Calculate energy consumption per unit distance acceleration Set it to 0, and then obtain the total energy consumption of the path. :

[0032] in, For link The length.

[0033] Total energy consumption Normalized to expected energy cost:

[0034] in, For battery capacity, The initial state of charge (SBC) of the vehicle is defined as [0,1], representing the percentage of remaining battery charge. The physical meaning of this formula is that for vehicles with a lower initial charge, the same energy consumption will result in a higher psychological cost, thus making them more inclined to choose the path with lower energy consumption.

[0035] The calculated real-time traffic impedance Compared with expected energy consumption Substituting into the composite weight function, we obtain the comprehensive toll cost for each candidate path p. .

[0036] Use the Logit model to calculate the probability that the vehicle chooses path p. :

[0037] in, It is a sensing parameter ( This metric is used to measure a driver's sensitivity to cost differences, and typically ranges from 0.5 to 3.0. Let q be the utility function for path p, and q be the identifier of the alternative path.

[0038] The dynamic travel demand of OD pair (i,j) is allocated to each alternative path according to the above probability.

[0039] Since path cost depends on network traffic (congestion level), and traffic is determined by path selection, an iterative process is needed to reach a balance, specifically: Assign an initial flow (e.g., free-flow state) to all links and set an iteration counter. Based on the current link traffic, the link travel time is updated using the BPR path resistance function, i.e.:

[0040] in, For free-flow time, For current traffic, For throughput, 0.15 and 4 are the standard parameters of the BPR function, derived through extensive empirical research.

[0041] Based on the updated road network status, a new round of comprehensive route cost calculation is performed, and new route allocation probabilities are calculated according to the Logit distribution. The traffic flow for each path and segment is updated according to the allocation probabilities to obtain the new traffic flow. The maximum relative rate of change of traffic on all links is calculated. If the maximum relative rate of change is less than a preset rate of change threshold, the system is considered to have reached a random user equilibrium state, and the iteration terminates; otherwise, [the process is terminated]. Then, a new round of iterations is carried out. After the iterations are completed, the system outputs the final dynamic spatiotemporal distribution.

[0042] This embodiment introduces a composite weighting function that considers real-time traffic impedance and expected energy consumption, making path selection not only influenced by road conditions but also dynamically driven by vehicle energy consumption constraints. This composite weighting achieves bidirectional coupling between traffic behavior and energy behavior, effectively improving the physical consistency and energy sensitivity of traffic flow assignment.

[0043] Furthermore, by introducing the initial SOC state of vehicles during the route allocation process, this embodiment can reflect the differences in route selection among vehicles with different battery levels, so that the traffic flow allocation results not only reflect macroscopic traffic patterns but also include microscopic energy consumption differences, thereby obtaining dynamic spatiotemporal distribution results with more realistic physical meaning.

[0044] By combining a stochastic user equilibrium strategy with an iterative solution mechanism, stable convergence of the road network can be achieved under multi-source constraints, enabling a dynamic balance between traffic impedance and energy consumption for vehicle flow. Compared with traditional single-objective shortest path algorithms, this embodiment can generate a vehicle distribution state that balances optimal energy consumption and road network efficiency, providing accurate input for subsequent SOC evolution and load forecasting.

[0045] S3. Establish an initial state of charge probability model for the vehicle cluster, and calculate the SOC consumption based on the dynamic spatiotemporal distribution to drive the evolution of the probability model and obtain the state of charge probability distribution at arrival. In this embodiment, the establishment of the initial state-of-charge probability model for the vehicle cluster specifically involves: Based on traffic network data, the betweenness centrality index is used to identify key nodes in the road network, and the top 20% of nodes in betweenness centrality are selected as key transportation hubs:

[0046] in, For the betweenness centrality of node v, This represents the number of shortest paths from node s to node t. This represents the number of shortest paths passing through node v.

[0047] Based on historical data, the empirical probability distribution of SOC for each key hub at the initial prediction time in the same historical period is obtained. Specifically, for each key hub k, SOC observation data of the same historical period (e.g., 8:00 AM on a weekday) is collected, and the empirical probability distribution is established using kernel density estimation:

[0048] in, Let k be the empirical probability density function of the SOC of the key transportation hub k. To observe the number of samples, This represents the k-th key transportation hub. For SOC values, For bandwidth parameters, For the i-th SOC observation, For kernel functions (such as Gaussian kernels).

[0049] Based on current travel demand data, the composition ratio of vehicle sources arriving at each key transportation hub is determined through reverse traffic flow simulation. For key transportation hub k, its vehicle source composition ratio is denoted as... , representing the proportion of vehicles from source region j, and satisfying .

[0050] Based on the proportion of vehicle origins, the empirical probability distributions of SOC in the corresponding origin regions are weighted and fused to generate an initial state-of-charge probability model for key transportation hubs in the current forecast period:

[0051] in, Let k be the initial SOC probability density function for hub k. Let be the historical SOC probability density function of the source region j.

[0052] The calculation of SOC consumption based on the dynamic spatiotemporal distribution drives the evolution of the probability model to obtain the probability distribution of the state of charge at arrival. Specifically: Microscopic parameters reflecting traffic operation characteristics, such as average speed, rate of change of acceleration, number of stops, and road slope, are extracted from the dynamic spatiotemporal distribution. The vehicle speed-time series is analyzed and input into the dynamic unit-mileage energy consumption function to generate an instantaneous energy consumption pulse sequence. This instantaneous energy consumption pulse sequence is integrated over the travel time, and the integration result is converted into State of Charge (SOC) consumption. Based on this SOC consumption, the initial state of charge probability model is dynamically evolved to obtain the evolved SOC probability distribution.

[0053] Indicates the initial SOC is The vehicles are consuming Afterwards, its SOC became .

[0054] Aggregate all paths and vehicle types to generate the overall state-of-charge probability distribution upon arrival. .

[0055] This embodiment, based on traffic networks and travel demand, generates an initial SOC probabilistic model reflecting current travel characteristics through reverse traffic flow extrapolation and weighted fusion. It then combines the dynamic spatiotemporal distribution of vehicles with a dynamic energy consumption function to simulate the continuous migration process of SOC in the spatiotemporal domain. This achieves a physical closed-loop coupling between SOC distribution and energy consumption processes, significantly improving the model's responsiveness and prediction accuracy to traffic fluctuations and energy consumption differences. It enables high-precision, interpretable modeling of the charging demand of a large-scale electric vehicle population.

[0056] S4. Based on the probability distribution, determine the charging group and simulate the charging process to obtain the time-varying characteristics of the charging probability. In this embodiment, the process of determining the charging group based on the probability distribution of the state of charge upon arrival and simulating the charging process to obtain the time-varying characteristics of the charging probability specifically includes: Probability distribution of the state of charge at arrival Mapped as a spatial pressure field of charging demand ,in, Let t represent spatial coordinates and t represent time. The mapping relationship is as follows:

[0057] in, The set charging trigger threshold.

[0058] This spatial pressure field is used to characterize the density distribution of potential charging demand at different times and locations. The higher the value, the stronger the charging driving force of vehicles in that area. Based on this pressure field, the set of vehicles with charging demand in the current predicted period can be identified, forming a charging group.

[0059] By spatiotemporally coupling the spatial pressure field of charging demand with the service capacity of charging facilities, a comprehensive service potential field is formed. :

[0060] in, This represents the distribution of service capacity of charging facilities in each region at time t.

[0061] Based on the gradient of the potential field difference, the overflow propagation process of charging demand in space and time can be deduced. This propagation process can be analogized to fluid diffusion behavior, and its rate of change satisfies:

[0062] in, The diffusion coefficient reflects the spatial migration rate of charging demand. The attenuation coefficient represents the probability that a vehicle will be able to receive or abandon charging while waiting or moving.

[0063] To deduce the spatiotemporal propagation process of charging overflow; When overflow propagation results indicate that charging demand in a certain area exceeds the local service capacity threshold, the system triggers a preventative charging decision. This decision mechanism alleviates peak load by selecting potential vehicle groups outside the overflow area to charge in advance. Strategies include, but are not limited to, issuing charging guidance signals in nearby areas in advance, adjusting the charging schedules of some vehicles, and implementing distributed scheduling delay strategies for some vehicles. By updating vehicle charging behavior data, the system recalculates the spatial pressure field. The system then proceeds to the next spatiotemporal propagation simulation. During this simulation and update process, the system continuously iterates through the cycle of "charging demand propagation—behavior update—pressure field reconstruction" until the transient equilibrium condition is met.

[0064] in, Indicates the number of iterations. This is the convergence threshold.

[0065] After reaching transient equilibrium, the proportion of vehicles actually entering the charging state at each time point is statistically analyzed, thus obtaining the time-varying characteristic function of the charging probability:

[0066] in, This represents the number of vehicles in the charging state at time t. This represents the total number of arriving vehicles.

[0067] This embodiment maps the probability distribution of the state of charge to a spatial pressure field and couples it with the service capacity of charging facilities in a spatiotemporal manner to deduce the propagation and dynamic feedback process of charging overflow. Through iterative updates, the system achieves transient equilibrium, obtaining time-varying characteristics that reflect the charging patterns of the population. This enables dynamic linkage modeling of charging behavior and facility load, significantly improving the spatial resolution, dynamic consistency, and interpretability of the predictions.

[0068] S5, aggregate the dynamic spatiotemporal distribution, time-varying characteristics of charging probability, and charging power parameters to obtain the total charging load.

[0069] In this embodiment, the aggregation of the dynamic spatiotemporal distribution, the time-varying characteristics of the charging probability, and the charging power parameters to obtain the total charging load specifically involves: Based on the aforementioned vehicle dynamic spatiotemporal distribution data, the number of vehicles remaining in each traffic zone at each time point within the prediction period is statistically analyzed to form a spatiotemporal matrix. The time resolution of this matrix can be set according to actual prediction needs, for example, with a step size of 5 minutes or 15 minutes, to incorporate the aforementioned time-varying characteristics of charging probability. With pre-acquired vehicle charging power parameters Multiplication yields the time-varying function of the average charging power per vehicle. Tensor operations are then performed between the spatiotemporal matrix and the time-varying function of the average charging power per vehicle to obtain the spatiotemporal distribution of the charging load. These tensor operations can be implemented through three-dimensional matrix multiplication or by using a block matrix parallel computing structure to improve computational efficiency. Double integration or summation is performed on all spatial units and time series of the charging load spatiotemporal distribution tensor to obtain the total charging load within the predicted time period.

[0070] This embodiment uses spatiotemporal tensor fusion and multidimensional aggregation to couple and model vehicle dynamic distribution, charging behavior, and power characteristics, achieving a continuous mapping of charging load from microscopic behavior to macroscopic demand. This method finely characterizes vehicle dwell states, integrates the time-varying characteristics of individual vehicle charging power, and generates the spatiotemporal distribution of charging load through tensor operations, ultimately aggregating the overall load. This process maintains both spatiotemporal continuity and physical consistency while improving prediction accuracy and interpretability, providing reliable support for power grid dispatching and planning.

[0071] Example 2, Figure 2 The present invention provides a system for predicting electric vehicle charging load based on the coupled evolution of traffic and SOC, comprising: The data acquisition module is used to acquire data on the traffic network, travel demand, and electric vehicle energy consumption parameters of the target area. The traffic flow allocation module is used to allocate traffic flow based on traffic network and travel demand data to obtain the dynamic spatiotemporal distribution of vehicles. The coupled calculation module is used to establish an initial state of charge probability model for the vehicle cluster, calculate the SOC consumption based on the dynamic spatiotemporal distribution, drive the evolution of the probability model, and obtain the state of charge probability distribution at arrival. The charging simulation module is used to determine the charging group based on the probability distribution and simulate the charging process to obtain the time-varying characteristics of the charging probability. The load prediction module is used to aggregate the dynamic spatiotemporal distribution, the time-varying characteristics of charging probability, and the charging power parameters to obtain the total charging load.

[0072] Example 3, an electronic device, such as Figure 3As shown, the device includes a memory and a processor. The memory stores a computer program, and when the processor calls the computer program in the memory, it can perform the steps provided in the above embodiments. Of course, the device may also include various necessary network interfaces, power supplies, and other components.

[0073] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0074] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0075] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0076] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0077] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0078] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for electric vehicle charging load forecasting based on coupling evolution of traffic and SOC, characterized in that, Includes the following steps: Acquire data on the target area's transportation network, travel demand, and electric vehicle energy consumption parameters; Traffic flow is allocated based on traffic network and travel demand data to obtain the dynamic spatiotemporal distribution of vehicles; An initial state-of-charge probability model for the vehicle cluster is established, and the SOC consumption is calculated based on the dynamic spatiotemporal distribution. This drives the evolution of the probability model to obtain the state-of-charge probability distribution upon arrival. The charging group is determined based on the probability distribution of the state of charge upon arrival, and the charging process is simulated to obtain the time-varying characteristics of the charging probability. Specifically, the probability distribution of the state of charge upon arrival is mapped to the spatial pressure field of charging demand; the spatial pressure field is spatiotemporally coupled with the service capacity of charging facilities to deduce the spatiotemporal propagation process of charging overflow; and preventive charging decisions are triggered based on the overflow propagation results to update vehicle charging behavior. By iteratively executing the deduction of the spatiotemporal propagation process and updating the vehicle charging behavior, the system reaches transient equilibrium and outputs the time-varying characteristics of the charging probability. The total charging load is obtained by aggregating the dynamic spatiotemporal distribution, the time-varying characteristics of charging probability, and the charging power parameters. Specifically, the spatiotemporal matrix of the number of vehicles is determined based on the dynamic spatiotemporal distribution; the time-varying characteristics of charging probability are fused with the charging power parameters to obtain the time-varying function of the average charging power per vehicle; tensor operations are performed on the spatiotemporal matrix and the time-varying function of the average charging power per vehicle to obtain the spatiotemporal distribution of the charging load, and the total charging load is obtained by spatiotemporal aggregation.

2. The traffic and SOC coupling evolution-based electric vehicle charging load prediction method according to claim 1, characterized in that, The acquisition of traffic network, travel demand, and electric vehicle energy consumption parameter data for the target area specifically includes: Based on historical and real-time traffic data, a dynamic traffic network is constructed that includes the free-flow speed and capacity of each link; Based on mobile signaling data, origin-destination pairs (OD pairs) of residents' trips are identified. Combined with dynamic traffic networks, dynamic travel demand matrices of each OD pair are obtained through gravity model inversion calculations during the prediction period. Obtain the vehicle's battery capacity parameters and, based on historical driving data, establish a dynamic energy consumption function per unit mileage through data fitting.

3. The electric vehicle charging load prediction method based on the coupled evolution of traffic and SOC as described in claim 2, characterized in that, The process of allocating traffic flow based on traffic network and travel demand data to obtain the dynamic spatiotemporal distribution of vehicles specifically involves: Establish a composite weighting function that considers real-time traffic impedance and expected energy consumption, where expected energy consumption is determined based on a dynamic energy consumption function per unit mileage; Based on the vehicle's initial SOC and composite weighting function, the comprehensive toll cost of each alternative route is calculated. Based on the overall toll cost, a random user equilibrium allocation strategy is used to assign the optimal route to the vehicle. The process of calculating comprehensive toll costs and allocating routes is iteratively executed until the road network traffic reaches a stable state, and the dynamic spatiotemporal distribution of vehicles is output.

4. The electric vehicle charging load prediction method based on the coupled evolution of traffic and SOC as described in claim 3, characterized in that, The initial state-of-charge probability model for establishing the vehicle cluster is specifically as follows: Identify key transportation hubs based on transportation network data; Based on historical data, obtain the SOC empirical probability distribution of each key hub at the initial prediction time in the same historical period; Based on the travel demand data for the current period, the composition ratio of vehicle sources arriving at each key transportation hub is determined by reverse traffic flow simulation. Based on the proportion of vehicle sources, the empirical probability distribution of SOC in the corresponding source areas is weighted and fused to generate an initial state-of-charge probability model for key transportation hubs in the current prediction period.

5. The electric vehicle charging load prediction method based on the coupled evolution of traffic and SOC as described in claim 4, characterized in that, The calculation of SOC consumption based on dynamic spatiotemporal distribution drives the evolution of the probability model to obtain the probability distribution of the state of charge at arrival, specifically as follows: Microscopic parameters reflecting traffic operation characteristics are extracted from the dynamic spatiotemporal distribution, including average speed, rate of change of acceleration, number of stops, and road slope. Based on the aforementioned micro parameters and dynamic unit mileage energy consumption function, the SOC consumption is calculated and the initial state of charge probability model is evolved and deduced. By aggregating the simulation results, a probability distribution of the state of charge at arrival is generated.

6. The electric vehicle charging load prediction method based on the coupled evolution of traffic and SOC as described in claim 5, characterized in that, The calculation of SOC consumption includes: Based on the dynamic spatiotemporal distribution, the vehicle speed-time series is analyzed; The speed-time sequence is input into the dynamic unit mileage energy consumption function to generate an instantaneous energy consumption pulse sequence; The instantaneous energy consumption pulse sequence is integrated over the travel time, and the integration result is converted into SOC consumption.

7. A system using the electric vehicle charging load prediction method based on traffic and SOC coupled evolution as described in any one of claims 1-6, characterized in that, include: The data acquisition module is used to acquire data on the traffic network, travel demand, and electric vehicle energy consumption parameters of the target area. The traffic flow allocation module is used to allocate traffic flow based on traffic network and travel demand data to obtain the dynamic spatiotemporal distribution of vehicles. The coupled calculation module is used to establish an initial state of charge probability model for the vehicle cluster, calculate the SOC consumption based on the dynamic spatiotemporal distribution, drive the evolution of the probability model, and obtain the state of charge probability distribution at arrival. The charging simulation module is used to determine the charging group based on the probability distribution and simulate the charging process to obtain the time-varying characteristics of the charging probability. The load prediction module is used to aggregate the dynamic spatiotemporal distribution, the time-varying characteristics of charging probability, and the charging power parameters to obtain the total charging load.

8. An electronic device, characterized in that, include: Memory and processor; The memory is used to store programs; The processor is used to execute a program to implement the electric vehicle charging load prediction method based on the coupled evolution of traffic and SOC as described in any one of claims 1-6.

Images