A load prediction method for electric vehicle charging stations based on real-time user balancing

Abstract

The application relates to a real-time user equilibrium-based electric vehicle charging station load prediction method, which comprises the following steps: establishing an electric vehicle traffic network model and dividing regions, adopting an improved electric vehicle transfer node chain model to describe an electric vehicle travel process, using an improved Hough model to determine a charging station for the electric vehicle and constructing a micro traffic flow simulation framework, and establishing a real-time user equilibrium model to depict user path selection conditions in a short term and calculating the charging station load.

Description

Technical Field

[0001] This invention relates to the field of electric vehicle charging load forecasting technology, and in particular to a method for forecasting the load of electric vehicle charging stations based on real-time user balancing. Background Technology

[0002] With the global energy structure transformation and the advancement of "dual-carbon" goals, high-efficiency and low-carbon electric vehicles are gradually replacing traditional fuel vehicles as the new development direction, and the electric vehicle industry is experiencing explosive growth. This rapid growth has brought unprecedented challenges to urban power distribution networks, especially as the spatiotemporal uncertainty of charging loads exacerbates peak-valley differences in the power grid and leads to increasingly prominent problems such as local transformer overload. Accurately predicting the spatiotemporal distribution of charging station loads has become a key technical link in ensuring the safe operation of the power grid and improving the quality of charging services.

[0003] Currently, most functional zoning methods for electric vehicle load forecasting use grid partitioning. This method struggles to reflect the coupling between the divided functional zones and the road network, leading to deviations in electric vehicle travel forecasts. Current descriptions of electric vehicle travel chains often rely on a few typical travel chains or random sampling. This method yields limited travel chain types and fails to reflect regional differences, resulting in low accuracy and an inability to analyze different cities, leading to significant errors in load forecasting. Furthermore, electric vehicle load forecasting lacks algorithms that accurately reflect user charging station and route selection. Traditional shortest path algorithms cannot describe actual user choices and fail to account for the interactions between electric vehicles. These shortcomings of existing forecasting methods result in low accuracy in electric vehicle charging station load forecasting, significant deviations from reality, and limitations in their respective approaches. Summary of the Invention

[0004] This invention addresses the shortcomings of existing technologies by providing a load forecasting method for electric vehicle charging stations based on real-time user balancing. The specific technical solution is as follows:

[0005] A load forecasting method for electric vehicle charging stations based on real-time user balancing includes the following steps:

[0006] Step 1: Establish a traffic network model for electric vehicles and divide the area into regions;

[0007] Step 2: Describe the electric vehicle travel process using an improved electric vehicle transfer node chain model;

[0008] Step 3: Use the improved Hough model to determine charging stations for electric vehicles and construct a microscopic traffic flow simulation framework;

[0009] Step 4: Establish a real-time user balancing model to characterize user path selection in the short term and calculate the charging station load.

[0010] Furthermore, in step one, establishing the electric vehicle traffic network model and dividing the area includes:

[0011] Step 1.1: Model the road network topology using graph theory:

[0012]

[0013] In formula (1): To study the road network topology within the region; This is the set of all traffic network nodes within the region; This is a set of road segments within a road network, representing the number of roads within the network. This is the set of actual distances between roads within the road network. This is the set of actual travel times for roads within the road network. Number the road network nodes; Number the road segment;

[0014] Step 1.2: Construct a speed-flow BPR model based on real-time traffic flow on the road network:

[0015]

[0016] In formula (2): For road section Real-time speed; For road section The zero flow velocity, i.e., the initial design speed of the road segment; For road section Real-time traffic flow; For road section Maximum traffic capacity is directly proportional to the road grade; This is an empirical coefficient; For adaptive coefficients under different road grades;

[0017] Step 1.3: Considering changes in ambient temperature, construct a power consumption model per unit mileage for electric vehicles:

[0018]

[0019] In formula (3): Electricity consumption per unit distance; Power consumption per unit distance for air conditioning; This refers to the distance traveled. This refers to the air conditioner's cooling capacity. This refers to the heating capacity of the air conditioner. Outdoor temperature; This is the lower limit of the air conditioner's cooling temperature. This is the upper limit of the air conditioner's heating temperature.

[0020] Step 1.4: Use a Voronoi diagram to set the traffic nodes of the road network. Divide the road network into zones based on the central point:

[0021]

[0022] In equation (4): For road network nodes The Vino polygon formed with as the center point, For any point within this region, Represents the Vino polygon any point in the road network node shortest;

[0023] Step 1.5: Extract Points of Interest (POI) information from the region layer using the open map platform:

[0024]

[0025] In equation (5): Indicates the type of point of interest: 1 for residential point of interest, 2 for work point of interest, 3 for commercial point of interest, and 4 for other point of interest.

[0026] Step 1.6: Count the number of different types of POIs in the polygons formed by each road network node, and construct a map dividing residential areas, work areas, commercial areas, and other areas according to POI density based on the node polygons:

[0027]

[0028] In formula (6): express In the Vino polygon formed by the nodes The proportion of POIs express In the Vino polygon formed by the nodes The number of POIs.

[0029] Furthermore, in step two, the use of an improved electric vehicle transfer node chain model to describe the electric vehicle travel process includes:

[0030] Step 2.1: Based on electric vehicle travel data, construct a travel chain model describing the electric vehicle transfer process, calculate the travel frequency distribution of typical travel chains, and extract the electric vehicle travel frequency according to the probability distribution.

[0031] Step 2.2: Based on the typical weekday travel patterns of electric vehicles, the origin and destination of the travel chain are set to the residential area, and the first destination is set to the work area. The length of the travel chain can be determined by the number of trips. The functional area transfer process of the travel chain is described as follows:

[0032]

[0033] In equation (7): For the first The vehicle's travel chain; Indicates the residential area from which the journey begins; Indicates the work area; Indicates the type of functional area to be extracted;

[0034] Step 2.3, Calculation of Function Area Extraction Probability:

[0035]

[0036] In equation (8): The probability of selecting k-type functional areas for the next destination; This represents the total number of POIs in class k; This indicates the total number of POIs within the region;

[0037] Step 2.4: After determining the functional areas, the node selection model for electric vehicles is described as follows:

[0038]

[0039] In equation (9): Select nodes for class k functional areas The probability of;

[0040] Step 2.5: Determine the starting node and working node of electric vehicles based on the node selection model, and extract the starting time distribution of electric vehicle trips and the dwell time distribution of different types of areas through historical travel data;

[0041] Step 2.6: Set the initial SOC for each electric vehicle, set the SOC threshold, and set the charging judgment conditions:

[0042]

[0043] In formula (10): For the first The vehicle's current battery level; This refers to the charging threshold. This is for the power consumption during the next leg of the journey.

[0044] Furthermore, in step three, the step of using the improved Hough model to determine charging stations for electric vehicles and constructing a microscopic traffic flow simulation framework includes:

[0045] Step 3.1: For electric vehicles with charging needs, obtain the current node's location and the locations of candidate charging stations, and use the shortest path algorithm to calculate the time it takes for the electric vehicle to travel to the candidate charging station.

[0046]

[0047] In equation (11): The shortest time from the current location to the charging station; For road section Passage time; It is the sum of the time of each segment included in the shortest path;

[0048] Step 3.2: Based on the factors that electric vehicle users consider when choosing charging services, construct a comprehensive attractiveness assessment of candidate charging stations from aspects such as charging scale, charging capacity, and economic cost.

[0049]

[0050] In equation (12): For charging stations Overall attractiveness; These are the weighting coefficients for each indicator; For charging stations The number of charging stations; This represents the maximum number of charging piles among the candidate charging stations. For charging stations The rated charging power of a single charging pile; This is the maximum charging power; For charging stations The unit electricity price; The maximum electricity price;

[0051] Step 3.3: Based on the overall attractiveness of charging stations and the shortest arrival time, construct an improved Hough model for charging station selection:

[0052]

[0053] In equation (13): Select the probability for the charging station; The number of charging stations; Sensitivity coefficient;

[0054] Step 3.4: Determine the charging station selection based on the probability of each candidate charging station and the roulette wheel algorithm;

[0055] Step 3.5: Construct a micro-traffic flow simulation framework. The simulation framework performs vehicle-by-vehicle and second-by-second simulations of electric vehicles within the constructed road network area. Within each simulation step, it simulates the operating status of all electric vehicles, obtains the current and destination locations of electric vehicles, and calculates the SOC of electric vehicles and the charging status of charging stations in real time.

[0056] Furthermore, in step four, establishing a real-time user balancing model to characterize user path selection in the short term and calculating the charging station load includes:

[0057] Step 4.1: Based on the simulation framework, extract the current location and destination location of all electric vehicles to form a dynamic destination based on the progressively updated SOC and charging state of the electric vehicles. Within each simulation step, the feasible paths of the OD pairs satisfy flow conservation.

[0058]

[0059] In equation (14), For OD The set of feasible paths; For simulation step size Internal path selection Electric vehicle traffic; For simulation step size Internal OD The total flow of electric vehicles;

[0060] The traffic flow on this section is:

[0061]

[0062] In equation (15), For road section Traffic on the internet; Indicates whether the road segment is on the route. superior;

[0063] Step 4.2, User Equalization Objective Function:

[0064]

[0065] In equation (16), The summation of the integrals of the road resistance function for each road segment;

[0066] Step 4.3: Solve the objective function using the Frank-Wolfe iterative algorithm;

[0067] Step 4.3.1: For each OD pair, load the initial path flow using the shortest path. Calculate the initial road segment flow ;

[0068] Step 4.3.2: Calculate based on the velocity-flow model in Step 1.2. Road segment travel time under the next iteration ;

[0069] Step 4.3.3, based on In the next iteration, the road segment travel time is used to assign the shortest path to each OD pair, resulting in the road segment auxiliary flow. ;

[0070] Step 4.3.4: Find the optimal step size through linear search. And update the path flow:

[0071]

[0072] In equation (17), To assist in traffic flow on auxiliary road sections;

[0073] Step 4.3.5: If the objective function value or arc segment flow change in adjacent iterations meets the preset threshold, then it is considered that the simulation step size is complete. When UE equilibrium is achieved, let the equilibrium solution be denoted as . Otherwise, the iteration is repeated, thereby obtaining the equilibrium solution within each simulation step. Assign an actual driving path to each electric vehicle based on its different origin-destination (OD) pairs;

[0074] Step 4.4: Calculate the charging time for the electric vehicle.

[0075]

[0076]

[0077] In equation (18): For charging efficiency; Duration of stay; This is the slow charging power. This is for fast charging power; For the first The charging power of the vehicle;

[0078] In equation (19): The actual charging time of an electric vehicle at a charging station;

[0079] Step 4.5: Calculate the charging station load and total charging load within each simulation step:

[0080]

[0081] In equation (20): For charging stations The charging load within each simulation step is the sum of the charging power of each electric vehicle within that simulation step. This represents the total charging load of the charging station. Indicates the first The car is The charging time within the period.

[0082] The beneficial effects of this invention are:

[0083] 1. This invention addresses the problem that functional zone division in electric vehicle load forecasting fails to reflect the coupling between functional zones and road networks. It uses Voronoi diagrams and POI data to divide regions based on road network nodes, forming different types of region segmentation density maps. This effectively reflects the distribution of functional zones in a region. At the same time, it constructs a road network that considers dynamic traffic information, improving the problem that existing static road networks cannot reflect the impact of traffic flow.

[0084] 2. This invention addresses the problem of insufficient granularity in describing electric vehicle travel in existing models by adopting an improved electric vehicle transfer node chain model to describe the electric vehicle travel process. It refines the vehicle transfer from functional area to functional area-node transfer, while also optimizing the problem of limited existing travel chain patterns. Furthermore, it defines the probability of functional area selection in the travel chain based on the functional area division of different regions.

[0085] 3. This invention addresses the problem of electric vehicle charging station selection and travel route selection. In an improved Hough model, it introduces charging station size, charging power, real-time electricity price, and real-time travel time to describe users' charging station selection. Through iterative solution of user equilibrium model and micro traffic flow simulation framework, the travel route selection under real-time OD pair is obtained, which effectively improves the accuracy of electric vehicle charging station load prediction. Attached Figure Description

[0086] Figure 1 This is a flowchart of the electric vehicle charging station load prediction method based on real-time user balancing according to the present invention.

[0087] Figure 2 Logic diagram for determining the charging mode of an electric vehicle. Detailed Implementation

[0088] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0089] Example:

[0090] In this embodiment, a method for predicting the load of electric vehicle charging stations based on real-time user balancing includes: 1. Establishing a traffic network model considering dynamic traffic information, and using Voronoi diagrams and Point of Interest (POI) data to form multi-type functional area density maps; 2. Based on the actual regional division, using an improved electric vehicle transfer node chain model to describe the electric vehicle travel process; 3. Based on an improved Hough model, selecting charging stations for electric vehicles with charging needs, and constructing a micro-traffic flow simulation framework to characterize the operating status of electric vehicles; 4. Using a user balancing model to characterize user path selection in the short term, and finally using Monte Carlo simulation to predict the spatiotemporal distribution of electric vehicle charging station load. Specifically, it is carried out according to the following steps:

[0091] Step 1: Establish a traffic network model for electric vehicles and divide the area into regions;

[0092] Step 1.1: Use graph theory to model the road network topology, obtain the location and path length of traffic network nodes, which facilitates subsequent path selection calculations.

[0093]

[0094] In formula (1): To study the road network topology within the region; This is the set of all traffic network nodes within the region; This is a set of road segments within a road network, representing the number of roads within the network. This is the set of actual distances between roads within the road network. This is the set of actual travel times for roads within the road network. Number the road network nodes; This is the road segment number.

[0095] Step 1.2: Construct a speed-flow BPR model based on real-time traffic flow on the road network. This model describes the impact of traffic flow on road speed, and updates the actual travel time of each road segment based on the travel patterns of electric vehicles, thus determining the impact between electric vehicles.

[0096]

[0097] In formula (2): For road section Real-time speed; For road section The zero flow velocity, i.e., the initial design speed of the road segment; For road section Real-time traffic flow; For road section Maximum traffic capacity is directly proportional to the road grade; This is an empirical coefficient; These are adaptive coefficients for different road grades.

[0098] Step 1.3: Considering changes in ambient temperature, construct a power consumption model per unit mileage for electric vehicles. The energy consumption model is temperature-dependent and needs to consider the impact of air conditioning equipment on the power consumption per unit mileage of electric vehicles, as well as the travel speed of electric vehicles.

[0099]

[0100] In formula (3): Electricity consumption per unit distance; Power consumption per unit distance for air conditioning; This refers to the distance traveled. This refers to the air conditioner's cooling capacity. This refers to the heating capacity of the air conditioner. Outdoor temperature; This is the lower limit of the air conditioner's cooling temperature. This is the upper limit of the air conditioner's heating temperature.

[0101] Step 1.4: Use a Voronoi diagram to set the traffic nodes of the road network. By dividing the road network into regions around a central point, the Voronoi diagram can reflect the influence range of a node and has a strong coupling with the road network nodes.

[0102]

[0103] In equation (4): For road network nodes The Vino polygon formed with as the center point, For any point within this region, Represents the Vino polygon any point in the road network node Shortest.

[0104] Step 1.5: Extract Points of Interest (POI) information from the regional layer using the open map platform. This POI information reflects the functional area distribution within the region. POIs are categorized into four types, and their quantity and location distribution are statistically analyzed.

[0105]

[0106] In equation (5): Indicates the type of point of interest: 1 for residential point of interest, 2 for work point of interest, 3 for commercial point of interest, and 4 for other point of interest.

[0107] Step 1.6: Count the number of different types of POIs in the polygons formed by each road network node. Based on the node polygons, construct zoning maps of residential areas, work areas, commercial areas, and other areas according to POI density. This forms four functional zone polygon density maps, which can reflect the functional zone division within the area, thus constructing a node selection model.

[0108]

[0109] In formula (6): express In the Vino polygon formed by the nodes The proportion of POIs express In the Vino polygon formed by the nodes The number of POIs.

[0110] Step 2: Describe the electric vehicle travel process using an improved electric vehicle transfer node chain model:

[0111] Step 2.1: Based on electric vehicle travel data, construct a travel chain model describing the electric vehicle transfer process, calculate the travel frequency distribution of typical travel chains, extract the electric vehicle travel frequency according to the probability distribution, and obtain the number of destinations from the number of travels to facilitate subsequent destination extraction.

[0112] Step 2.2: Based on the typical weekday travel patterns of electric vehicles, the origin and destination of the travel chain are set to the residential area, and the first destination is set to the work area. The length of the travel chain can be determined by the number of trips. The functional area transfer process of the travel chain is described as follows:

[0113]

[0114] In equation (7): For the first The vehicle's travel chain; Indicates the residential area from which the journey begins; Indicates the work area; Indicates the type of functional area to be extracted.

[0115] Step 2.3: Calculate the probability of extracting a functional area. The probability of extracting a specific functional area is the number of POIs of that type divided by the total number of POIs in the area. This reflects the differences between different areas and increases the accuracy and diversity of travel chains.

[0116]

[0117] In equation (8): The probability of selecting k-type functional areas for the next destination; This represents the total number of POIs in class k; This indicates the total number of POIs within the region.

[0118] Step 2.4: After determining the functional areas, the node selection model for electric vehicles is described as follows: the node selection probability is consistent with the proportion of Points of Interest (POIs) of the node polygons in a certain type of functional area.

[0119]

[0120] In equation (9): Select nodes for class k functional areas The probability of.

[0121] Step 2.5: Determine the starting node and working node of the electric vehicle based on the node selection model. Extract the starting time distribution of electric vehicle trips and the dwell time distribution of different types of areas through historical travel data. From this, the time chain of electric vehicle transfer process and the time situation during the trip can be obtained.

[0122] Step 2.6: Set the initial SOC for each electric vehicle, set the SOC threshold, and set charging judgment conditions. If the current battery level of the electric vehicle is lower than the loss for the next leg of the journey and the SOC threshold, a charging demand is triggered, and the vehicle proceeds to a charging station to charge, ensuring destination accessibility.

[0123]

[0124] In formula (10): For the first The vehicle's current battery level; This refers to the charging threshold. This is for the power consumption during the next leg of the journey.

[0125] Step 3: Use the improved Hough model to determine charging stations for electric vehicles and construct a microscopic traffic flow simulation framework:

[0126] Step 3.1: For electric vehicles with charging needs, obtain the current node's location and the locations of candidate charging stations. Use the shortest path algorithm to calculate the time it takes for the electric vehicle to travel to the candidate charging station. The shortest arrival time is dynamically calculated based on real-time road segment travel time.

[0127]

[0128] In equation (11): The shortest time from the current location to the charging station; For road section Passage time; It is the sum of the time of each segment included in the shortest path.

[0129] Step 3.2: Based on the factors that electric vehicle users consider when choosing charging services, construct a comprehensive attractiveness index for candidate charging stations, considering aspects such as charging scale, charging capacity, and economic cost. The stronger the attractiveness index, the greater the probability of selecting the corresponding charging station.

[0130]

[0131] In equation (12): For charging stations Overall attractiveness; These are the weighting coefficients for each indicator; For charging stations The number of charging stations; This represents the maximum number of charging piles among the candidate charging stations. For charging stations The rated charging power of a single charging pile; This is the maximum charging power; For charging stations The unit electricity price; This represents the maximum electricity price.

[0132] Step 3.3: Based on the overall attractiveness of charging stations and the shortest arrival time, construct an improved Hough model for charging station selection:

[0133]

[0134] In equation (13): Select the probability for the charging station; The number of charging stations; The sensitivity coefficient is denoted as .

[0135] Step 3.4: Determine the charging station selection based on the probability of each candidate charging station and the roulette wheel algorithm. Accumulate the selection probabilities to obtain the probability accumulation matrix, and draw a random number to determine its interval to determine the selected charging station.

[0136] Step 3.5: Construct a micro-traffic flow simulation framework. The simulation framework performs vehicle-by-vehicle and second-by-second simulations of electric vehicles within the constructed road network area. Within each simulation step, it simulates the operating state of all electric vehicles, obtains the current and destination locations of electric vehicles, and calculates the SOC of electric vehicles and the charging status of charging stations in real time. The micro-traffic flow simulation framework can reflect the state of electric vehicles in the short term, thereby enabling refined calculation of the SOC of electric vehicles and generating charging demand. At the same time, it can extract the OD under the current state to provide a basis for subsequent calculations.

[0137] Step 4: Establish a user equilibrium model to characterize user path selection in the short term and calculate the charging station load:

[0138] Step 4.1: Based on the simulation framework, extract the current location and destination location of all electric vehicles to form a dynamic destination based on the progressively updated SOC and charging status of the electric vehicles. Within each simulation step, the feasible paths of the OD pairs satisfy flow conservation, and the number of vehicles from the origin to the destination is the sum of the number of vehicles that actually choose multiple routes.

[0139]

[0140] In equation (14), For OD The set of feasible paths; For simulation step size Internal path selection Electric vehicle traffic; For simulation step size Internal OD Total traffic flow of electric vehicles.

[0141] The flow of a road segment is incremented by 1 if a path passes through it. The flow of a road segment is the sum of the flow provided by all paths.

[0142]

[0143] In equation (15), For road section Traffic on the internet; Indicates whether the road segment is on the route. superior;

[0144] Step 4.2: User equilibrium objective function. The equilibrium objective is solved using an iterative method.

[0145]

[0146] In equation (16), The summation of the integrals of the road resistance function for each road segment;

[0147] Step 4.3: The Frank-Wolfe iterative algorithm is used to solve the objective function. By iterating the segment flow and auxiliary flow at the current step length, the equilibrium solution is finally obtained. The equilibrium solution can be used to determine the path selection of electric vehicles under different OD (origin and destination). In each iteration, the search direction and the optimal step length are found.

[0148] Step 4.3.1: For each OD pair, load the initial path flow using the shortest path. Calculate the initial road segment flow ;

[0149] Step 4.3.2: Calculate based on the velocity-flow model in Step 1.2. Road segment travel time under the next iteration ;

[0150] Step 4.3.3, based on In the next iteration, the road segment travel time is used to assign the shortest path to each OD pair, resulting in the road segment auxiliary flow. ;

[0151] Step 4.3.4: Find the optimal step size through linear search. And update the path flow:

[0152]

[0153] In equation (17), To assist in traffic flow on auxiliary road sections;

[0154] Step 4.3.5: If the objective function value or arc segment flow change in adjacent iterations meets the preset threshold, then it is considered that the simulation step size is complete. When UE equilibrium is achieved, let the equilibrium solution be denoted as . Otherwise, the iteration is repeated, thereby obtaining the equilibrium solution within each simulation step. Each electric vehicle is assigned an actual driving path based on its different origin-destination (OD) pairs.

[0155] Step 4.4: Calculate the charging time of the electric vehicle to obtain the charging status, charging power, and charging duration of a single electric vehicle within a day. The dwell time is derived from the dwell time distribution of different functional areas extracted in Step 2.5.

[0156]

[0157]

[0158] In equation (18): For charging efficiency; Duration of stay; This is the slow charging power. This is for fast charging power; For the first The charging power of the vehicle;

[0159] In equation (19): The actual charging time of an electric vehicle at a charging station.

[0160] Step 4.5: Calculate the charging station load and total charging load within each simulation step. The load generated by a single electric vehicle in the charging station within each simulation step is the average charging power, which is the actual charging amount within that step divided by the simulation step. The total charging station load is the sum of the charging station loads.

[0161]

[0162] In equation (20): For charging stations The charging load within each simulation step is the sum of the charging power of each electric vehicle within that simulation step. This represents the total charging load of the charging station. Indicates the first The car is The charging time within the period.

[0163] The beneficial effects of this invention are:

[0164] This invention addresses the problem that functional zone division in electric vehicle load forecasting fails to reflect the coupling between functional zones and road networks. It uses Voronoi diagrams and POI data to divide regions based on road network nodes, forming different types of region segmentation density maps. This effectively reflects the distribution of functional zones within a region and constructs a road network that considers dynamic traffic information, thus improving the problem that existing static road networks cannot reflect the impact of traffic flow.

[0165] Secondly, to address the problem of insufficient granularity in describing electric vehicle travel in existing models, this invention adopts an improved electric vehicle transfer node chain model to describe the electric vehicle travel process. This model refines the transfer of vehicles from functional areas into functional area-node transfers, while also optimizing the problem of limited existing travel chain patterns. Furthermore, it defines the probability of selecting functional areas in the travel chain based on the functional area division of different regions.

[0166] Furthermore, this invention addresses the issues of electric vehicle charging station selection and travel route selection by incorporating charging station size, charging power, real-time electricity price, and real-time travel time into an improved Hough model to describe users' charging station selection. Through iterative solutions using a user equilibrium model and a micro-traffic flow simulation framework, the travel route selection under real-time OD pairs is obtained, effectively improving the accuracy of electric vehicle charging station load prediction.

[0167] 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, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A load forecasting method for electric vehicle charging stations based on real-time user balancing, characterized in that, The steps include the following: Step 1: Establish a traffic network model for electric vehicles and divide the area into regions; Step 2: Describe the electric vehicle travel process using an improved electric vehicle transfer node chain model; Step 3: Use the improved Hough model to determine charging stations for electric vehicles and construct a microscopic traffic flow simulation framework; Step 4: Establish a real-time user balancing model to characterize user path selection in the short term and calculate the charging station load.

2. The method for load forecasting of electric vehicle charging stations based on real-time user balancing according to claim 1, characterized in that, Step one, establishing the electric vehicle traffic network model and dividing the area, includes: Step 1.1: Model the road network topology using graph theory: (1)； In formula (1): To study the road network topology within the region; This is the set of all traffic network nodes within the region; This is a set of road segments within a road network, representing the number of roads within the network. This is the set of actual distances between roads within the road network. This is the set of actual travel times for roads within the road network. Number the road network nodes; Number the road segment; Step 1.2: Construct a speed-flow BPR model based on real-time traffic flow on the road network: (2)； In formula (2): For road section Real-time speed; For road section The zero flow velocity, i.e., the initial design speed of the road segment; For road section Real-time traffic flow; For road section Maximum traffic capacity is directly proportional to the road grade; This is an empirical coefficient; For adaptive coefficients under different road grades; Step 1.3: Considering changes in ambient temperature, construct a power consumption model per unit mileage for electric vehicles: (3)； In formula (3): Electricity consumption per unit distance; Power consumption per unit distance for air conditioning; This refers to the distance traveled. This refers to the air conditioner's cooling capacity. This refers to the heating capacity of the air conditioner. Outdoor temperature; This is the lower limit of the air conditioner's cooling temperature. This is the upper limit of the air conditioner's heating temperature. Step 1.4: Use a Voronoi diagram to set the traffic nodes of the road network. Divide the road network into zones based on the central point: (4)； In equation (4): For road network nodes The Vino polygon formed with as the center point, For any point within this region, Representing the Vino polygon any point in the road network node shortest; Step 1.5: Extract Point of Interest (POI) information from the region layer using the open map platform: (5)； In equation (5): Indicates the type of point of interest: 1 for residential point of interest, 2 for work point of interest, 3 for commercial point of interest, and 4 for other point of interest. Step 1.6: Count the number of different types of POIs in the polygons formed by each road network node, and construct a map dividing residential areas, work areas, commercial areas, and other areas according to POI density based on the node polygons: (6)； In equation (6): express In the Vino polygon formed by the nodes The proportion of POIs express In the Vino polygon formed by the nodes The number of POIs.

3. The method for load forecasting of electric vehicle charging stations based on real-time user balancing according to claim 2, characterized in that, In step two, the use of an improved electric vehicle transfer node chain model to describe the electric vehicle travel process includes: Step 2.1: Based on electric vehicle travel data, construct a travel chain model describing the electric vehicle transfer process, calculate the travel frequency distribution of typical travel chains, and extract the electric vehicle travel frequency according to the probability distribution. Step 2.2: Based on the typical weekday travel patterns of electric vehicles, the origin and destination of the travel chain are set to the residential area, and the first destination is set to the work area. The length of the travel chain can be determined by the number of trips. The functional area transfer process of the travel chain is described as follows: (7)； In equation (7): For the first The vehicle's travel chain; Indicates the residential area from which the journey begins; Indicates the work area; Indicates the type of functional area to be extracted; Step 2.3, Calculation of Function Area Extraction Probability: (8)； In equation (8): The probability of selecting k-type functional areas for the next destination; This represents the total number of POIs in class k; This indicates the total number of POIs within the region; Step 2.4: After determining the functional areas, the node selection model for electric vehicles is described as follows: (9)； In equation (9): Select nodes for class k functional areas The probability of; Step 2.5: Determine the starting node and working node of electric vehicles based on the node selection model, and extract the starting time distribution of electric vehicle trips and the dwell time distribution of different types of areas through historical travel data; Step 2.6: Set the initial SOC for each electric vehicle, set the SOC threshold, and set the charging judgment conditions: (10)； In formula (10): For the first The vehicle's current battery level; This refers to the charging threshold. This is for the power consumption during the next leg of the journey.

4. The method for load forecasting of electric vehicle charging stations based on real-time user balancing according to claim 3, characterized in that, In step three, the step of using the improved Hough model to determine charging stations for electric vehicles and constructing a microscopic traffic flow simulation framework includes: Step 3.1: For electric vehicles with charging needs, obtain the current node's location and the locations of candidate charging stations, and use the shortest path algorithm to calculate the time it takes for the electric vehicle to travel to the candidate charging station. (11)； In equation (11): The shortest time from the current location to the charging station; For road section Passage time; It is the sum of the time of each segment included in the shortest path; Step 3.2: Based on the factors that electric vehicle users consider when choosing charging services, construct a comprehensive attractiveness assessment of candidate charging stations from aspects such as charging scale, charging capacity, and economic cost. (12)； In equation (12): For charging stations Overall attractiveness; These are the weighting coefficients for each indicator; For charging stations The number of charging stations; This represents the maximum number of charging piles among the candidate charging stations. For charging stations The rated charging power of a single charging pile; This is the maximum charging power; For charging stations The unit electricity price; The maximum electricity price; Step 3.3: Based on the overall attractiveness of charging stations and the shortest arrival time, construct an improved Hough model for charging station selection: (13)； In equation (13): Select the probability for the charging station; The number of charging stations; Sensitivity coefficient; Step 3.4: Determine the charging station selection based on the probability of each candidate charging station and the roulette wheel algorithm; Step 3.5: Construct a micro-traffic flow simulation framework. The simulation framework performs vehicle-by-vehicle and second-by-second simulations of electric vehicles within the constructed road network area. Within each simulation step, it simulates the operating status of all electric vehicles, obtains the current and destination locations of electric vehicles, and calculates the SOC of electric vehicles and the charging status of charging stations in real time.

5. The method for load forecasting of electric vehicle charging stations based on real-time user balancing according to claim 4, characterized in that, In step four, establishing a real-time user balancing model to characterize user path selection in the short term and calculating the charging station load includes: Step 4.1: Based on the simulation framework, extract the current location and destination location of all electric vehicles to form a dynamic destination based on the progressively updated SOC and charging state of the electric vehicles. Within each simulation step, the feasible paths of the OD pairs satisfy flow conservation. (14)； In equation (14), For OD The set of feasible paths; For simulation step size Internal path selection Electric vehicle traffic; For simulation step size Internal OD The total flow of electric vehicles; The traffic flow on this section is: (15)； In equation (15), For road section Traffic on the internet; Indicates whether the road segment is on the route. superior; Step 4.2, User Equalization Objective Function: (16)； In equation (16), The summation of the integrals of the road resistance function for each road segment; Step 4.3: Solve the objective function using the Frank-Wolfe iterative algorithm; Step 4.3.1: For each OD pair, load the initial path flow using the shortest path. Calculate the initial road segment flow ; Step 4.3.2: Calculate based on the velocity-flow model in Step 1.

2. Road segment travel time under the next iteration ; Step 4.3.3, based on In the next iteration, the road segment travel time is used to assign the shortest path to each OD pair, resulting in the road segment auxiliary flow. ; Step 4.3.4: Find the optimal step size through linear search. And update the path flow: (17)； In equation (17), To assist in traffic flow on auxiliary road sections; Step 4.3.5: If the objective function value or arc segment flow change in adjacent iterations meets the preset threshold, then it is considered that the simulation step size is complete. When UE equilibrium is achieved, let the equilibrium solution be denoted as . Otherwise, the iteration is repeated, thereby obtaining the equilibrium solution within each simulation step. Assign an actual driving path to each electric vehicle based on its different origin-destination (OD) pairs; Step 4.4: Calculate the charging time for the electric vehicle. (18)； (19)； In equation (18): For charging efficiency; Duration of stay; This is the slow charging power. This is for fast charging power; For the first The charging power of the vehicle; In equation (19): The actual charging time of an electric vehicle at a charging station; Step 4.5: Calculate the charging station load and total charging load within each simulation step: (20)； In equation (20): For charging stations The charging load within each simulation step is the sum of the charging power of each electric vehicle within that simulation step. This represents the total charging load of the charging station. Indicates the first The car is The charging time within the period.

Images