Electric vehicle fast charging pile load prediction method considering traffic flow estimation

Abstract

The application discloses a kind of electric vehicle fast charging pile load prediction method considering traffic flow estimation, constructs the model of electric vehicle and fuel vehicle in traffic network searching feasible path;By analyzing the energy use behavior of electric vehicle trip process, an electric vehicle battery power change model is constructed, a traffic flow distribution model containing electric vehicles and fuel vehicles is established to maximize user equilibrium state as the target, three sub-problems are proposed and the electric vehicle fast charging pile load prediction level is obtained by prediction algorithm.The present application comprehensively considers electric vehicles and fuel vehicles, can predict the load level of electric vehicle fast charging pile, reduce the adverse effects of electric vehicle charging load uncertainty and volatility on power grid, provide basis for power grid dispatching center to develop dispatching plan, and optimize power grid operation efficiency.

Description

Technical Field

[0001] This invention relates to a technology in the field of fast-charging piles, specifically a method for predicting the load of fast-charging piles for electric vehicles that takes into account traffic flow estimation. Background Technology

[0002] Current forecasting technologies for fast-charging electric vehicle charging stations employ probabilistic methods and artificial intelligence-based approaches. Probabilistic methods include Markov chains and Monte Carlo methods. The reliability parameters of probabilistic systems are calculated using mathematical formulas, thus offering unique advantages in forecasting speed and accuracy. Artificial intelligence-based methods are commonly used for short-term load forecasting; they do not require a precise model of the object during analysis and can represent complex nonlinear problems. However, current deep learning-based forecasting techniques largely rely on historical data and do not adequately consider traffic network theory. Summary of the Invention

[0003] This invention addresses the problem of high uncertainty and difficulty in predicting the load level of existing fast-charging electric vehicle charging piles. It proposes a load prediction method for fast-charging electric vehicle charging piles that takes into account traffic flow estimation. By comprehensively considering both electric vehicles and fuel vehicles, this method can predict the load level of fast-charging electric vehicle charging piles, reduce the adverse effects of the uncertainty and fluctuation of electric vehicle charging load on the power grid, provide a basis for the power grid dispatch center to formulate dispatch plans, and optimize the power grid operation efficiency.

[0004] This invention is achieved through the following technical solution:

[0005] This invention relates to a method for predicting the load of fast-charging stations for electric vehicles that takes into account traffic flow estimation. It constructs a model for electric vehicles and fuel vehicles to search for feasible paths in a traffic network. By analyzing the energy consumption behavior of electric vehicles during their travel process, it constructs a model for the change in electric vehicle battery charge. With the goal of maximizing user equilibrium, it establishes a traffic flow allocation model that includes electric vehicles and fuel vehicles. It proposes three sub-problems and obtains the predicted load level of fast-charging stations for electric vehicles through a prediction algorithm.

[0006] The aforementioned model for searching feasible paths for electric vehicles and gasoline vehicles in a transportation network refers to: constructing an optimal path selection problem based on the transportation network, including: a node set Θ consisting of a starting point, destinations such as garages and parking lots, and relay points such as intersections. N and the set of lines Θ connecting each point A The attributes of the participants in the operation of the traffic network include: origin (o), destination (d), and selected path (q). All traffic flows with the same origin and destination are considered as one Od pair, and electric vehicle charging stations are located at road intersections.

[0007] In the model described, each Od pair consists of cars that start from the starting point and travel to the destination, with the following constraints on the path:

[0008] ① A set of route selection schemes for traffic flow must be based on existing roads. in: For a binary variable, when When, it means that the traffic flow of this OOD pair has passed through the road between node i and node j, and is from node i to node j; when When this occurs, it means that the traffic flow of this set of Od pairs did not travel from node i to node j via the road between nodes i and j. As an auxiliary binary variable. The constant is used when a road exists between node i and node j. otherwise

[0009] ②Each road intersection (including the start and end points of the path) will either be avoided entirely or only visited once. in: For a binary variable, when When, it means that the traffic flow of this OOD pair has passed through the road between node i and node j, and is from node i to node j; when When this occurs, it means that the traffic flow of this set of Od pairs did not travel from node i to node j via the road between nodes i and j.

[0010] ③ The number of times each road intersection (excluding the start and end points of a path) serves as the start point of one road must be equal to the number of times it serves as the end point of another road. Where, k o and k d These are the starting node and the ending node, respectively.

[0011] ④ The starting point (end point) of a path must be the starting point (end point) of a particular road. in: For a binary variable, when When, the traffic flow of this OOD pair passes through the road between node i and node j, and is from node i to node j; when At that time, the traffic flow of this set of Od pairs did not travel from node i to node j through the road between nodes i and j; As an auxiliary binary variable; The constant is used when a road exists between node i and node j. otherwise k o and k dThese are the starting node and the ending node, respectively.

[0012] The electric vehicle battery charge variation model includes:

[0013] 1) The electric vehicle's charge when choosing path q at point i(j) satisfies: in: D represents the amount of electricity needed to select path q for the electric vehicle at point i(j). ij Let be the energy consumption of the electric vehicle on road segment (i,j), be the charging amount of the electric vehicle when choosing path q at point j, be an auxiliary variable, M is an infinite number, be the length of the road from node i to node j, and be the power consumption per unit length distance.

[0014] 2) The electric vehicle's charge at the starting point is equal to the sum of its initial charge and the charge it receives at the starting point. in: This represents the initial charge level of the electric vehicle.

[0015] 3) The upper and lower limits of electric vehicle battery capacity must meet the following requirements: Where: E min E is the lower limit of battery capacity. max This represents the upper limit of battery capacity.

[0016] 4) Considering the driver's range anxiety, i.e., the worry caused by low battery when driving an electric vehicle, the difference between the battery level of the electric vehicle when it chooses path q at point i(j) and the energy consumption of the electric vehicle on road segment (i,j) satisfies: in: To alleviate range anxiety for electric vehicle drivers, drivers must ensure the battery level is above a certain value.

[0017] 5) The relationship between the charging amount of electric vehicles and the charging stations satisfies: in: Let i be a binary variable, and when a charging station exists at point i, =1, otherwise It is 0.

[0018] In the electric vehicle battery charge change model described above, when the electric vehicle chooses to charge, the battery will be fully charged: in: Let the variables be binary variables, and when the electric vehicle decides to charge at node m, =1, otherwise Θ is 0 evcs This is the set of nodes where electric vehicle charging stations are located.

[0019] The traffic flow allocation model, which includes both electric and gasoline-powered vehicles, includes:

[0020] in: For the set of ods, W represents the traffic flow of electric vehicles choosing path q. rs Let rs be the total traffic flow of electric vehicles, x a Let x be the traffic flow of road segment a. m Let m be the vehicle flow rate at the charging station.

[0021] Θ evcs This is the set of nodes where electric vehicle charging stations are located.

[0022] Where: t a The time taken to travel through road a. For free travel time, u a For road capacity; eliminate quartic terms in the expression using piecewise linearization; Where: t m The time taken for an electric vehicle to reach the charging station at node m. This is the rated charging power. For fixed service hours, u m It is a constant coefficient used to characterize the congestion effect.

[0023] in: Let m be the charging load of the charging station. in: This is the maximum capacity of the charging station.

[0024] in: For binary variables, Let rs be a slack variable and a set of Od pairs. Let rs be the traffic flow of path q in the path set of the od pair. Let rs be the travel cost of electric vehicle choosing route q, and c be the cost of od. rs For the minimum travel cost of electric vehicles, Θ is the set of ods. rs Let Θ be the path set of the OOD pairs rs. OD Let Od be the set of Od pairs.

[0025] in: The cost of choosing path q, k is the coefficient for converting time cost into monetary cost, and ρ average The average electricity price for all charging stations. The charging electricity price for node m charging pile.

[0026] The decomposition refers to breaking down the entire problem into a simplified main problem, an optimal path generation subproblem, and a feasible path generation subproblem, specifically:

[0027] a) Simplified main problem: For all OD pairs, after the optimal path generation subproblem and feasible path generation subproblem have generated a path set and corresponding charging selection for each OD pair, the traffic flow for each path is calculated based on the approximate state of user equilibrium. The objective function is to minimize the slack variables, specifically: The restrictions include:

[0028] b) Optimal Path Generation Subproblem: For a given pair of OD (Operational Path) traffic flows, find the optimal path with the lowest overall cost. Specifically, this involves satisfying all constraints of the original problem. Where: t′ a and t′ m The values ​​of t are obtained by solving the simplified main problem. a and t m For the optimal path generation subproblem, these are all known constants.

[0029] c) Feasible Path Generation Subproblem: For a given pair of OD (Operational Traffic Flow) pairs, to overcome the limitation of charging station capacity which prevents a feasible solution to the simplified main problem, we seek the feasible path with the lowest overall cost. Specifically, this involves satisfying all constraints of the original problem. and in: This is the set of charging choices for the paths found in the previous calculations of the traffic flow by the OD (Optical Distribution System), γ q，before，i Let be a vector, representing all charging options for a given path in the charging option set found in the previous calculations of the traffic flow. Let be a binary variable, representing the charging selection of one of the charging stations for a given path in the set of traffic flow paths.

[0030] The prediction algorithm specifically includes:

[0031] Step 1) Set the relevant parameters for the charging pile and the transportation network, and initialize t′. a and t′ m , where: the superscript ' indicates the known quantities in the optimal path generation subproblem and the feasible path generation subproblem, and the unsuperscripted quantities are the results obtained after solving the subproblems;

[0032] Step 2) For each pair of ODs, solve the optimal path generation subproblem for traffic flow and update the path set for each pair of ODs.

[0033] Step 3) If no OD data for the vehicle flow is updated after step 2), the algorithm process terminates, and the calculation results such as the charging load of the charging pile are obtained.

[0034] Step 4) Solve the simplified main problem;

[0035] Step 5) When the simplified main problem has a feasible solution, then t is obtained. a and t m and use t a and t m Update t′ a and t m Then return to step 2) and continue;

[0036] Step 6) If the simplified main problem does not have a feasible solution, generate a subproblem for the feasible path of the traffic flow solution for each Od pair, update the path set of each Od pair, and then return to step 4) to continue.

[0037] Technical effect

[0038] This invention combines traffic network theory with consideration of the impact of traffic flow on the load level of fast charging stations. It estimates traffic flow using traffic network theory and then predicts the charging load of fast charging stations. The prediction results are more reasonable compared to existing technologies. Attached Figure Description

[0039] Figure 1 This is a flowchart of the present invention;

[0040] Figure 2 Here is a flowchart of the prediction algorithm;

[0041] Figure 3 The image shown is a rendering of an example. Detailed Implementation

[0042] like Figure 1 As shown in this embodiment, a method for predicting the load of fast charging stations for electric vehicles that takes into account traffic flow estimation is included:

[0043] Step 1: The traffic network topology used is based on the road network within the Second Ring Road of Xi'an City, Shaanxi Province, with 41 traffic network nodes and 72 bidirectional roads. The road network in the example is designed to simulate a real-world city. This method is suitable for large-scale road networks, and its operational model is adapted to highway systems. The example sets up 15 electric vehicle charging stations, located at nodes 1, 3, 4, 7, 10, 13, 17, 19, 20, 24, 28, 32, 33, 35, and 40. The travel demand of electric vehicle drivers is simulated using 50 OD pairs, with the traffic flow of each OD pair exhibiting a normal distribution over time. The coefficient k for converting time cost to monetary cost is set to 3, the battery capacity of the electric vehicle is set to 32 kWh, the constant coefficient used to characterize the congestion effect is set to 0.5, the upper limit of battery capacity is set to 32 kWh, the lower limit is set to 3.2 kWh, and the power consumption per unit distance is set to 0.15 kWh / km.

[0044] Step 2: Establish a model for electric vehicles and gasoline vehicles to search for feasible paths in the transportation network;

[0045] Step 3: Establish a model for changes in the electric vehicle's battery level;

[0046] Step 4: Establish a traffic network model for the vehicles;

[0047] Step 5: Decompose the entire problem into three subproblems: the simplified main problem, the optimal path generation subproblem, and the feasible path generation subproblem.

[0048] Step 6: Use a prediction algorithm to solve the three sub-problems obtained in Step 5, specifically including:

[0049] 6.1) Set the relevant parameters for the charging piles and the transportation network, and initialize t. ′ a and t ′ m ;

[0050] 6.2) For each pair of ODs, solve the optimal path generation subproblem for traffic flow and update the path set for each pair of ODs.

[0051] 6.3) If no OD (Operational Data Point) updates the vehicle flow after 6.2), the algorithm process terminates, and the calculation results such as the charging load of the charging pile are obtained.

[0052] 6.4) Solve the simplified principal problem;

[0053] 6.5) When the simplified main problem has a feasible solution, then t is obtained. a and t m and use t a and t m Update t′ a and t m Then return to 6.2) to continue;

[0054] 6.6) If the simplified main problem does not have a feasible solution, generate a subproblem for the feasible path of the traffic flow solution for each pair of Od pairs, update the path set of each pair of Od pairs, and then return to 6.4) to continue.

[0055] Through specific practical experiments, under the specific environment settings of calling the commercial solver gurobi in Python, the experimental data that can be obtained are as follows: The charging station load level within one hour of the simulation experiment is shown in Table 1 and... Figure 3 As shown:

[0056] Table 1

[0057] Charging station node number Charging station load (kWh) 1 545.733 3 637.5413 4 1056.773 7 698.1378 10 925.0046 13 1143.072 17 1999.968 19 1191.068 20 1245.29 24 1122.154 28 2000 32 1183.725 33 1028.535 35 736.2856 40 1011.845

[0058] Compared with existing technologies, this method takes into account the influence of traffic network factors and introduces Wardrop user equilibrium theory.

[0059] The above-described specific implementations can be partially adjusted by those skilled in the art in different ways without departing from the principles and purpose of the present invention. The scope of protection of the present invention is defined by the claims and is not limited to the above-described specific implementations. All implementation schemes within the scope of the claims are bound by the present invention.

Claims

1. A method for predicting the load of fast charging stations for electric vehicles that takes into account traffic flow estimation, characterized in that, Models for searching feasible paths in traffic networks for electric vehicles and fuel vehicles are constructed. By analyzing the energy consumption behavior of electric vehicles during their travel process, a model of electric vehicle battery charge change is constructed. With the goal of maximizing user equilibrium, a traffic flow distribution model including electric vehicles and fuel vehicles is established and then decomposed into a simplified main problem, an optimal path generation sub-problem, and a feasible path generation sub-problem. The load prediction level of electric vehicle fast-charging piles is obtained through a prediction algorithm. The aforementioned model for searching feasible paths for electric and gasoline vehicles in a transportation network refers to: constructing an optimal path selection problem based on the transportation network, including: a node set consisting of a starting point, destinations such as garages and parking lots, and relay points such as intersections. and the set of lines connecting each point The attributes of the participants in the operation of the traffic network include: origin (o), destination (d), and selected path (q). All traffic flows with the same origin and destination are considered as one od pair, and electric vehicle charging stations are located at road intersections. The electric vehicle battery charge variation model includes: 1) The electric vehicle's charge when choosing path q at point i(j) satisfies: , , ,in: ( Let q be the amount of electricity consumed when the electric vehicle selects path q at point i(j). Let be the energy consumption of the electric vehicle on road segment (i,j), be the charging amount of the electric vehicle when choosing path q at point j, be an auxiliary variable, M be an infinite number, be the length of the road from node i to node j, and be the power consumption per unit length distance. 2) The electric vehicle's charge at the starting point is equal to the sum of its initial charge and the charge it receives at the starting point. ,in: This is the initial charge level of the electric vehicle; 3) The upper and lower limits of electric vehicle battery capacity must meet the following requirements: ,in: This represents the lower limit of battery capacity. This represents the upper limit of battery capacity. 4) Considering the driver's range anxiety, i.e., the worry caused by low battery when driving an electric vehicle, the difference between the battery level of the electric vehicle when it chooses path q at point i(j) and the energy consumption of the electric vehicle on road segment (i,j) satisfies: ,in: To alleviate range anxiety for electric vehicle drivers, meaning drivers must ensure the battery level is above a certain value; 5) The relationship between the charging amount of electric vehicles and the charging stations satisfies: ,in: Let i be a binary variable, and when a charging station exists at point i, =1, otherwise =0; The traffic flow allocation model, which includes both electric and gasoline-powered vehicles, includes: i) , , ,in: For the rs set of od, the traffic flow is used to select the q path for electric vehicles. For the rs group, the total traffic flow of electric vehicles is represented. Let the traffic flow be the traffic volume of road segment a. Let m be the vehicle flow rate at the charging station. This is the set of nodes where electric vehicle charging stations are located. ii) ,in: The time taken to travel through road a. Free time for driving. For road capacity; eliminate quartic terms in the expression using piecewise linearization; ,in: The time taken for an electric vehicle to reach the charging station at node m. This is the rated charging power. For fixed service hours, is a constant coefficient used to characterize the congestion effect; iii) ,in: Let m be the charging load of the charging station. ,in: This represents the maximum capacity of the charging station. iv) , , ,in: For binary variables, Let rs be a slack variable and a set of Od pairs. Let rs be the traffic flow of path q in the path set of the od pair. Let rs represent the travel cost of the electric vehicle choosing route q. For the set of ODs, rs represents the minimum travel cost for electric vehicles. Let rs be the set of paths for the OD pairs. For the set of Od pairs; v) ,in: The cost of choosing path q, This is the coefficient for converting time cost into monetary cost. The average electricity price for all charging stations. The charging electricity price for node m charging pile.

2. The method for predicting the load of fast charging stations for electric vehicles that takes into account traffic flow estimation as described in claim 1, characterized in that, In the model for searching feasible paths in a traffic network for both electric and gasoline vehicles, each Od pair involves vehicles starting from a starting point and traveling to a destination. The constraints on the paths include: ① A set of route selection schemes for traffic flow must be based on existing roads. ,in: For a binary variable, when When, it means that the traffic flow of this OOD pair has passed through the road between node i and node j, and is from node i to node j; when When this occurs, it means that the traffic flow of this Od pair did not travel from node i to node j via the road between nodes i and j. As an auxiliary binary variable The constant is used when a road exists between node i and node j. ;otherwise ; ②Each road intersection (including the start and end points of the path) will either be avoided entirely or only visited once. ,in: For a binary variable, when When, it means that the traffic flow of this OOD pair has passed through the road between node i and node j, and is from node i to node j; when When this occurs, it means that the traffic flow of this set of Od pairs did not travel from node i to node j via the road between nodes i and j; ③ The number of times each road intersection (excluding the start and end points of a path) serves as the start point of one road must be equal to the number of times it serves as the end point of another road. ; ④ The starting point (end point) of a path must be the starting point (end point) of a particular road. ,in: For a binary variable, when When, the traffic flow of this OOD pair passes through the road between node i and node j, and is from node i to node j; when At that time, the traffic flow of this set of Od pairs did not travel from node i to node j through the road between nodes i and j; As an auxiliary binary variable; The constant is used when a road exists between node i and node j. ;otherwise ; and These are the starting node and the ending node, respectively.

3. The method for predicting the load of fast charging stations for electric vehicles that takes into account traffic flow estimation as described in claim 1, characterized in that, In the electric vehicle battery charge change model described above, when the electric vehicle chooses to charge, the battery will be fully charged: ,in: Let the variables be binary variables, and when the electric vehicle decides to charge at node m, =1, otherwise =0, This is the set of nodes where electric vehicle charging stations are located.

4. The method for predicting the load of fast charging stations for electric vehicles that takes into account traffic flow estimation as described in claim 1, characterized in that, The simplified main problem is as follows: For all OD pairs, after the optimal path generation subproblem and feasible path generation subproblem have generated a path set and corresponding charging selection for each OD pair, the traffic flow for each path is calculated based on the approximate state of user equilibrium. The objective function is to minimize the slack variables, specifically: The restrictions include: , , , , , , , , ; The optimal path generation subproblem refers to finding the optimal path with the lowest overall cost for a given pair of OD (Original Direction) pairs of traffic flows. Specifically, it involves satisfying all constraints of the original problem. ,in: and These are obtained by solving the simplified main problem. and For the optimal path generation subproblem, these are all known constants; The feasible path generation subproblem refers to finding the feasible path with the lowest overall cost for a given pair of OD (Operational Traffic Flow) pairs to overcome the lack of a feasible solution to the simplified main problem due to charging station capacity limitations. Specifically, it involves finding the feasible path while satisfying all constraints of the original problem. ,and ,in: This is the set of charging selections for all paths of the traffic flow by this OD (Operational Device). Let be a vector, representing all charging choices for a given path in the set of traffic flow paths. Let be a binary variable, representing the charging selection of one of the charging stations for a given path in the set of traffic flow paths.

5. The method for predicting the load of fast charging stations for electric vehicles that takes into account traffic flow estimation according to any one of claims 1-4, characterized in that, The prediction algorithm specifically includes: Step 1) Set the relevant parameters for the charging pile and the transportation network, and initialize them. and , where: the superscript ' indicates the known quantities in the optimal path generation subproblem and the feasible path generation subproblem, and the unsuperscripted quantities are the results obtained after solving the subproblems; Step 2) For each pair of ODs, solve the optimal path generation subproblem for traffic flow and update the path set for each pair of ODs. Step 3) If no OD data for the vehicle flow is updated after step 2), the algorithm process terminates, and the calculation results such as the charging load of the charging pile are obtained. Step 4) Solve the simplified main problem; Step 5) When the simplified main problem has a feasible solution, then we obtain... and and use and renew and Then return to step 2) and continue; Step 6) If the simplified main problem does not have a feasible solution, generate a subproblem for the feasible path of the traffic flow solution for each Od pair, update the path set of each Od pair, and then return to step 4) to continue.

Images