A highway network charging station optimization method based on charging demand space-time distribution

Abstract

The application discloses a highway network charging station optimization method based on charging demand space-time distribution, and belongs to the field of intelligent transportation, comprising the following steps: collecting road network information and highway toll records to generate road network vehicle driving track; combining electric vehicle energy consumption to establish an energy consumption model; simulating energy consumption conditions by using simulation software to obtain electric vehicle charging distribution, and establishing a site selection model for each service area number to obtain a light-storage-charging integrated charging station model; respectively calculating the satisfaction degree of charging demand of each charging station, the rational distribution of power resources, and the relief of road traffic congestion; based on the calculation results, a charging station fixed capacity model is established; and an optimal capacity scheme is solved. Through highway toll data, the application obtains effective electric vehicle paths and optimal charging station distribution, compared with the traditional technology, the application also considers factors such as weather, driver anxiety power, traffic density, motor efficiency and user behavior, and finally obtains more real and reliable results.

Description

Technical Field

[0001] This invention relates to the field of intelligent transportation, and specifically to an optimization method for charging stations in highway networks based on the spatiotemporal distribution of charging demand. Background Technology

[0002] Highway charging station planning is a complex system optimization problem involving multiple stakeholders, nonlinearity, multiple optimization objectives, and multiple constraints, influenced by environmental, policy, technological, economic, and transportation factors. The main approaches to highway charging station planning focus on regulating charging demand and the rational allocation of resources. Charging demand can be assessed using two main methods: one is from the electric vehicle perspective, simulating electric vehicle travel using Monte Carlo sampling and then combining this with energy consumption models to roughly estimate charging demand; the other is from the charging station perspective, analyzing traffic flow near charging stations and using queuing theory and time series analysis models to model the arrival rate and state of charge (historical data) of vehicles reaching service areas, thereby predicting charging demand.

[0003] However, existing technologies for electric vehicle (EV) research lack supporting and guiding data from actual EV usage. EVs are assumed to follow a uniform travel pattern, and calculations of energy consumption and charging volume are based on static data, leading to charging demand estimates that deviate from reality. Methods focusing on charging stations show good predictive performance for individual stations but lack consideration for the randomness of EV user travel and the heterogeneity of charging behavior on highways. This limits their application to provincial highway networks, and existing charging station-based methods lack estimations of operating costs. Furthermore, in practice, factors such as weather, driver anxiety about battery level, traffic density, motor efficiency, and user behavior must be considered. Therefore, a method combining EV and charging station approaches is urgently needed to analyze real-world scenarios. Summary of the Invention

[0004] To address the aforementioned issues, this invention provides an optimization method for charging stations in high-speed road networks based on the spatiotemporal distribution of charging demand.

[0005] To achieve the above technical solution, the specific steps are as follows:

[0006] S1. Collect and process road network data and highway toll records to obtain electric vehicle travel characteristic parameters and generate road network vehicle driving trajectories;

[0007] Electric vehicle travel characteristic parameters include: time data and location data;

[0008] The specific steps of S1 are as follows:

[0009] S1.1 Obtain the experimental highway network model using an open-source map; the steps are as follows:

[0010] S1.1.1 Obtain the road network model of the target province using the open-source map software OpenStreetMap;

[0011] S1.1.2 Modify and organize the road network model of the target province using the road network editing software Netdit;

[0012] The modification and reorganization involves retaining only the highway network model.

[0013] S1.1.3. Set the locations of toll stations, service areas, and parking areas on the expressway network, as well as the locations of detectors on road sections, as nodes and number them;

[0014] S1.2 Obtain the probability density curve of vehicles entering the highway network through kernel density estimation; the steps are as follows:

[0015] S1.2.1 Discretize the time data of vehicles entering the highway network by grid points and calculate the histogram of the time data;

[0016] The time data range is R = [0h, 24h], where R is MIN for 0h and MAX for 24h;

[0017] S1.2.2 Normalize the probability density of the histogram and perform a discrete cosine transform to obtain the frequency domain coefficients a. k The expression is as follows:

[0018]

[0019] In the formula, b k x represents the k-th interval; s Let A represent the s-th data sample; let A represent the total number of data samples; let b represent the total number of intervals; M represent the total number of intervals. k p represents the probability that a data sample falls within interval k; k p represents the normalized probability of the k-th interval; z This represents the normalized probability value of the z-th interval, where z∈k;

[0020] S1.2.3. Calculate the optimal bandwidth h iteratively using the fixed-point equation, and then smooth the histogram; the expression is as follows:

[0021]

[0022] R = MAX - MIN

[0023] In the formula, R represents the time data range; t * The optimal smoothing parameter is expressed as follows:

[0024]

[0025] In the formula, t represents the smoothing parameter; f(t) is the frequency domain smoothing function related to t, and its expression is as follows:

[0026]

[0027] In the formula, L represents the maximum order of iterations;

[0028] S1.2.4. The final probability density distribution is obtained through the smoothed frequency domain coefficients and the inverse discrete cosine transform; the expression is as follows:

[0029]

[0030] In the formula, Represents the smoothed frequency domain coefficients; Represents the inverse discrete cosine transform;

[0031] S1.3. Obtain the vehicle trajectories in the road network using the A* algorithm and path assignment model; the steps are as follows:

[0032] S1.3.1. Based on the vehicle location data in the provincial expressway toll record data, obtain the starting and ending points of the vehicle entering the expressway network;

[0033] S1.3.2 The A* algorithm calculates the shortest path from the starting point to the ending point. At the same time, considering the user's path selection along the way, random perturbations are applied to the road segment impedance in the mesoscopic traffic simulation software DynasTIM to expand the set of effective paths.

[0034] S2. Establish an energy consumption model by combining the factors that affect the energy consumption of electric vehicles;

[0035] Factors affecting the energy consumption of electric vehicles include: motor conversion efficiency, internal equipment operating energy consumption, and high-speed driving resistance (friction resistance, wind resistance, and gradient resistance).

[0036] The steps to establish an energy consumption model are as follows:

[0037] S2.1. Based on the law of conservation of energy, consider the energy loss of the vehicle from both vehicle dynamics and motor dynamics perspectives; firstly, establish the driving equation of the electric vehicle, as shown in the following expression:

[0038] F t =F m +F k +F p +F j

[0039] In the formula, F t For driving force; F m F represents frictional resistance. kFor air resistance; F p For ramp resistance; F j To increase resistance;

[0040] By organizing, we can obtain:

[0041]

[0042] In the formula, m is the vehicle mass; g is the acceleration due to gravity; f is the road friction resistance coefficient; α is the road slope angle; C D V is the drag coefficient; A is the vehicle's frontal area; v s δ represents the actual vehicle speed; δ is the rotational mass conversion factor.

[0043] S2.2 Establish driving energy consumption models under different driving conditions;

[0044] Different driving states include: constant speed driving, accelerating driving, and decelerating driving; that is, based on the vehicle's current speed v. s and target vehicle speed v m Determine different driving states of the vehicle;

[0045] Specifically as follows:

[0046] S2.2.1 When the vehicle is traveling at a constant speed (v) s =v m In this state, the vehicle experiences no acceleration resistance, therefore the energy consumption expression for the electric vehicle under these conditions is as follows:

[0047]

[0048] In the formula, η d For motor conversion efficiency;

[0049] S2.2.2, When the vehicle accelerates (v s <v m In the case of an uphill slope, the vehicle experiences acceleration resistance and requires a net driving force to accelerate. Therefore, the energy consumption expression for acceleration driving of an electric vehicle in this situation is as follows:

[0050]

[0051] In the formula, ΔT represents the torque added by the vehicle's acceleration motor, expressed as follows:

[0052]

[0053] In the formula, r is the wheel radius; i0 is the gear ratio of the reducer;

[0054] ω n The motor speed is expressed as follows:

[0055]

[0056] In the formula, i1 is the transmission ratio of the vehicle's transmission system;

[0057] S2.2.3, When the vehicle decelerates (v s >v m In the current state, the vehicle's regenerative braking system is activated, and there is no driving force. The expression for the energy recovered by braking in this state is as follows:

[0058]

[0059] In the formula, η r To improve regenerative braking efficiency;

[0060] S2.2.4 The energy consumption model for electric vehicles obtained by integration is as follows:

[0061] P = P x1 +P x2 +P i -P x3

[0062] In the formula, P x1 Energy consumption for electric vehicles; P x2 To accelerate the energy consumption of electric vehicles; P x3 Energy recovery during braking of electric vehicles; P i Energy consumption for the operation of internal equipment in electric vehicles;

[0063] S2.3, Target vehicle speed v in the driving energy consumption model under different driving conditions m Make corrections; the steps are as follows:

[0064] Specifically, the simulated vehicle speed in the highway network is not always constant, and when rain, snow, fog, or traffic congestion occur, it will affect the driver's visibility and the vehicle's grip. Therefore, the vehicle speed and the driver's driving anxiety need to be corrected.

[0065] In addition, considering that drivers are worried about extreme weather and heavy traffic, and that frequent starts, stops or decelerations may cause range anxiety, which could lead to incorrect charging decisions, drivers' anxiety about battery level also needs to be corrected.

[0066] The steps are as follows:

[0067] S2.3.1 The correction expression for vehicle speed based on traffic flow is as follows:

[0068]

[0069] In the formula, v′ m The target vehicle speed after traffic flow correction; vf k is the free-flow speed of the vehicle. m k represents the average density of the frontal influence area (SIR) of a vehicle. jam The congestion density in the area in front of the vehicle; k min The upper limit of density for vehicles operating at free-flow velocity; σ and β are the first and second parameters, respectively;

[0070] S2.3.2 The correction expression for vehicle speed due to weather conditions is as follows:

[0071]

[0072] In the formula, v′ m ′ represents the target speed after weather correction; The weather conditions from charging station i to i+1 are used as the correction factor for vehicle speed. Precipitation and snowfall are used as the scale for calibration. The target vehicle speed is reduced as precipitation and snowfall increase, with the reduction range being 4%-14%.

[0073] S2.4 Correct the anxiety level coefficient;

[0074] Considering that drivers are worried about extreme weather and heavy traffic, and that frequent starts, stops or decelerations can cause range anxiety, which may lead to incorrect charging decisions, drivers' anxiety about battery level also needs to be corrected.

[0075] The correction expression for the weather-related anxiety coefficient is as follows:

[0076]

[0077] In the formula, Let θ be the anxiety energy coefficient of an electric vehicle driver at time T at charging station i+1 on the highway network; θ is the basic anxiety energy coefficient under the condition of good weather and reasonable traffic flow. The weather and traffic flow for charging stations i to i+1 are used as correction factors for the anxiety-based battery coefficient, with a range of 6%-20%.

[0078] S3. Input the parameters from S1 into the mesoscopic traffic simulation software DynasTIM, and run the energy consumption model from S2 in the software to track the vehicle's state of charge in real time. Calculate the charging nodes selected by the vehicle and the amount of electricity replenished at the nodes through the constructed charging scenario, and finally obtain the spatiotemporal distribution of charging demand in various service areas of the highway network; the specific steps are as follows:

[0079] S3.1 Utilize a traffic simulator to track the traffic status of the highway network or customize road network information;

[0080] Traffic simulators include: the mesoscopic traffic simulation software DynasTIM;

[0081] Highway network traffic status or custom road network information includes: highway network toll node locations, lane speed limits, road gradient angles, air density, substation locations, service area locations, and the capacity of set charging stations;

[0082] S3.2 Simulation of vehicle driving on highway network;

[0083] S3.2.1 Set different vehicle parameters in the routing file, including: vehicle weight, frontal area, speed, rotational mass conversion factor, electric motor energy conversion efficiency, rolling resistance factor, energy consumption of internal equipment of electric vehicle and maximum state of charge;

[0084] S3.2.2 Based on the probability density curve of vehicles entering the highway network, vehicles are put into the road network for driving, and they travel according to the shortest path from the vehicle's origin to its destination and other effective paths.

[0085] S3.3 Calculate energy consumption using the established energy consumption model to obtain vehicle charging demand data;

[0086] Specifically, during the operation of the mesoscopic traffic simulation DynasTIM, the travel chain of vehicles in the highway network is as follows: Figure 4 The parameters (time from the previous node to this node, travel distance, and vehicle speed) of each node along the driving trajectory are input into the energy consumption model built in Python to calculate energy consumption, update the vehicle's remaining battery power, and generate correction factors based on the node parameters (weather information and traffic flow information). The vehicle's speed and anxiety battery coefficient are corrected according to S2.3 and S2.4. The driving speed of the simulated vehicle after leaving the node is updated through the TraCI interface. The simulation is updated in real time after each node is passed until the vehicle ends the simulation and leaves the road network.

[0087] S3.4. By obtaining the remaining battery power and the anxiety factor when the vehicle arrives at the node, the user's charging decision is formulated based on actual user driving habits and charging judgment, and the scenario is divided accordingly; as follows:

[0088] Scenario 1: When the driver reaches a certain node, the electric vehicle's state of charge is lower than the anxiety level, so the driver chooses to recharge at that node; the expression is as follows:

[0089]

[0090] Scenario 2: When the driver reaches a certain node, the electric vehicle's state of charge is higher than the anxiety level, and energy can still be replenished at the next node. The driver chooses to continue. The expression is as follows:

[0091]

[0092] Scenario 3: When the driver reaches a certain node, the electric vehicle's state of charge (SBC) is higher than the anxiety level, but there is no way to recharge at the next node, and the SBC is lower than the anxiety level upon reaching the next node, the driver will choose to recharge at that node; the expression is as follows:

[0093]

[0094] When the driver selects a charging node, charging is complete when the battery reaches 90% of its maximum state of charge. The formula for calculating the amount of battery charge replenished is as follows:

[0095] Q = 0.9SOC max -SOC i

[0096] In the power calculation formulas for scenarios one, two, three, and the supplement: SOC i State of charge (SOC) of the electric vehicle when it arrives at charging station i. max This represents the maximum state of charge of an electric vehicle. Let T be the anxiety coefficient of an electric vehicle driver at charging station i on the highway network at time T;

[0097] After the vehicle completes its operation in DynasTIM, the total charging amount of the electric vehicle at each node is calculated every hour. The charging demand of all nodes is the spatiotemporal distribution of charging demand in the study. At the same time, by setting up charging piles in some service areas, the charging service duration provided by the nodes and the waiting time for users to charge when the charging resources in the station are insufficient can also be obtained.

[0098] S4. Based on the spatiotemporal distribution of charging demand in each service area in S3 and numbering each service area, select the service area nodes where charging stations need to be built using the maximum coverage location model.

[0099] Specifically, the optimization objective is to maximize the total charging demand at the covered demand points:

[0100]

[0101] In the formula, Q represents the total number of demand points; d q The charging demand at demand point q;

[0102] Constraints are applied to coverage, number of sites, and decision variables; the expression is as follows:

[0103] Coverage constraints:

[0104] Site quantity constraints:

[0105] In the formula, N represents the total number of charging stations to be built; x pFor the first binary variable, y p For the second binary variable and a pq The third binary variable has the following expression:

[0106]

[0107]

[0108] The final output is the number of the selected service area, with a total of N.

[0109] S5. Establish a photovoltaic-storage-charging integrated charging station model. Using the charging load data of the selected charging station obtained in S3 and S4, obtain the output data of each unit when the photovoltaic-storage-charging station meets these loads.

[0110] Currently, the development prospects of charging stations are trending towards intelligent and digital multi-energy complementarity, achieving intelligent scheduling through technologies such as the Internet of Things, big data, and cloud computing. Among these, photovoltaic-storage charging station technology is relatively mature and has been widely put into use. The charging stations constructed in this article are all photovoltaic-storage charging stations, mainly composed of intelligent control systems, photovoltaic units, energy storage units, power distribution network interaction, and charging equipment.

[0111] Each unit operates based on its charging load status: when the photovoltaic power generation can meet the charging load, it prioritizes charging the energy storage system. If there is still surplus power, it sells electricity to the distribution network. Conversely, if the photovoltaic power generation is insufficient to support the charging load, the energy storage system discharges. If it still cannot meet the charging load or the energy storage capacity is lower than the set value, it purchases electricity from the distribution network. This invention requires the output data of each unit for operator revenue assessment.

[0112] The output models for each unit are established as follows:

[0113] S5.1 Establish a photovoltaic unit output model; details are as follows:

[0114] The output power of photovoltaic cells is mainly affected by the intensity of sunlight; therefore, the output model of a photovoltaic power unit is expressed as follows:

[0115]

[0116] In the formula, η pv denoted as photoelectric conversion efficiency; S is the area of ​​the photovoltaic unit; r(t) is the change in light intensity over time.

[0117] In this invention, the mean and variance are calculated based on some historical data. Latin hypercube stratified sampling is used to generate 500 photovoltaic scenario data points, which are then reduced to ten scenarios using k-means clustering to reflect the uncertainty of photovoltaics. The details are as follows:

[0118] The mean and variance are calculated based on historical photovoltaic data, as shown in the following expressions:

[0119]

[0120] σ t =u2(t)·μ t

[0121] In the formula, c represents the scaling factor used to control the calculation of the mean and variance, which is a constant; u2(t) represents the random factor generated by Latin hypercube sampling;

[0122] The photovoltaic output data for each photovoltaic-storage charging station, which follows a normal distribution, is generated using the following expression:

[0123]

[0124] In the formula, P pv1 (t) represents historical photovoltaic data; P pv2 (t) represents the photovoltaic data of each photovoltaic-storage charging station in the study;

[0125] S5.2 Establish the energy storage unit output model; details are as follows:

[0126] The output model of the energy storage unit includes two states: charging and discharging.

[0127] The expression for the discharge state is as follows:

[0128]

[0129] The expression for the charging state is as follows:

[0130]

[0131] In the formula, η z For energy storage unit conversion efficiency; η c η dis These are the charging efficiency and discharging efficiency, respectively; E h P represents the rated capacity of the energy storage unit. es (t) represents the charging and discharging power;

[0132] The battery model of the energy storage unit is established, and the expression is as follows:

[0133]

[0134] In the formula, SOC es (t) represents the state of charge of the energy storage unit during time period t;

[0135] S5.3 Establish the power output model of the distribution network; details are as follows:

[0136] Based on the charging load and the output of photovoltaic and energy storage units, a power distribution network output model is constructed; the expression is as follows:

[0137]

[0138] In the formula, η t For conversion efficiency; δ t Line loss rate; Let i be the charging demand of charging station i during time period t;

[0139] The constraints of this invention on energy storage units include:

[0140] Energy storage power constraint: -P es (t) max ≤P es (t)≤P es (t) max

[0141] Constraints on the state of charge of energy storage: 0.1E h ≤SOC es ≤0.9E h

[0142] Finally, the output results of each unit were obtained;

[0143] S6. Based on the results of S3, S4, and S5, calculate the degree to which the charging demand of each charging station is met, the rational allocation of power resources, and the degree of relief of road traffic congestion through the model.

[0144] Specifically, S4 selects N charging stations, and the set of the number of fast charging piles in each charging station is E = {m1, m2, ..., m n}, n∈N, and each fast charging pile in the charging station (slow charging piles are not considered in highway network service areas) can work 24 hours a day;

[0145] The calculation process is as follows:

[0146] S6.1 Calculate the degree to which charging demand is met; details are as follows:

[0147]

[0148] In the formula, The utilization rate of charging station i during time period t is specifically represented as follows:

[0149]

[0150] In the formula, Q represents the charging demand of charging station i during time period t (obtained from step S3.4); is The maximum charging resources that charging station i can provide are specifically represented as follows:

[0151] Q is =P f ·t is

[0152] In the formula, P f The charging power of the fast charging station; t is The duration of charging service provided by charging station i during time period t (obtained from step S3.4);

[0153] The average degree to which all charging stations meet the charging demand can be expressed as:

[0154]

[0155] S6.2 Calculate power resource allocation; details are as follows:

[0156] The formula for calculating the electricity consumption within the charging station is as follows:

[0157]

[0158] In the formula, η f Energy conversion efficiency of fast charging piles; E io For the electricity consumption of other electrical facilities within charging station i;

[0159] The entire photovoltaic-energy storage charging station is mainly powered by a photovoltaic system, an energy storage system, and a power distribution network. Therefore, the calculation formula for the power supply within the charging station is as follows:

[0160] E im =E i,pv +E i,es +E i,gr

[0161] In the formula, the power output model of the photovoltaic system in each charging station is expressed as:

[0162]

[0163] In the formula, E i,pv The power generation of the photovoltaic system within charging station i; η pv Photoelectric conversion efficiency; S i r is the area of ​​the photovoltaic units within the charging station i; i (t) represents the change in the solar radiation intensity on the horizontal surface of charging station i over time;

[0164] The power transmission model from the distribution network to each charging station can be approximated as follows:

[0165]

[0166] In the formula, η t For conversion efficiency; δ tGiven the line loss rate, the average distribution of power resources across all charging stations can be expressed as:

[0167]

[0168] S6.3 Calculate traffic mitigation measures; details are as follows:

[0169]

[0170] In the formula, U i The congestion index of the road segment between charging station i and the previous charging station; The actual traffic flow through the road segment during time period t (obtained through traffic monitoring equipment); The road capacity that can pass through the road segment during time period t (a fixed value set for each road in the traffic simulation road network model); V represents the average vehicle speed passing through charging station i during time period t; i,free The free-flow vehicle speed passing through charging station i under no congestion conditions;

[0171] S7. Based on the data from S5 and the calculation results from S6, establish a charging station capacity model to determine the daily net profit after all charging stations on the road network are put into operation; that is, calculate the entropy weights of three aspects—the matching degree between charging resources and charging demand, power resource allocation, and traffic mitigation—using the entropy weight method to establish a road performance evaluation model; establish an overall revenue model based on cooperative game theory after all charging stations are put into operation; and take maximizing road performance and maximizing revenue as the objective functions.

[0172] The steps are as follows:

[0173] S7.1 Establish a road performance evaluation model using the entropy weight method;

[0174] The expression for establishing the road performance evaluation model using the entropy weight method is as follows:

[0175] f1=ω1·v1(N)+ω2·v2(N)+ω3·v3(N)

[0176] In the formula, ω1, ω2, and ω3 are the weights of three indicators: the degree to which charging demand is met by the charging station, the average distribution of power resources, and the degree of traffic relief.

[0177] S7.2 Establish an operator revenue model through cooperative game theory;

[0178] A daily revenue function for all charging stations in the road network after they are put into use is established using a cooperative game theory model. The revenue expression for each charging station is as follows:

[0179]

[0180] In the formula, ct The unit price at which a charging station provides charging services to users during time period t;

[0181] The daily cost of each charging station includes four parts: photovoltaic system operation and maintenance cost, energy storage system operation and maintenance cost, electricity purchase cost from the distribution network, and on-site facility operation and maintenance cost.

[0182] The operation and maintenance cost of a photovoltaic system is expressed as follows:

[0183]

[0184] In the formula, c pv This refers to the unit operation and maintenance price of a photovoltaic system. Let Δt be the output power of the photovoltaic system at charging station i during time period t; Δt is the unit length of time period t (1 hour).

[0185] The operation and maintenance cost expression for an energy storage system is as follows:

[0186]

[0187] In the formula, c E E represents the unit capacity operation and maintenance cost of an energy storage system. im Configure the energy storage capacity for charging station i; c p The unit transmission cost of the energy storage system; Let t be the output power of the energy storage system at charging station i during time period t.

[0188] The interaction cost expression for the distribution network is as follows:

[0189]

[0190] In the formula, The unit price at which the charging station buys from and sells to the distribution network during time period t; For time period t, the power purchased and sold by charging station i from the distribution network;

[0191] The cost expression for the site's infrastructure is as follows:

[0192] C i,inst =c f ·m i +c o

[0193] In the formula, c f The maintenance cost for a single fast charging station; m i c is the number of fast charging stations within charging station i; o Fixed operation and maintenance costs for other facilities within the charging station (a set constant term);

[0194] Therefore, the daily net profit function after all charging stations in the entire road network are put into use is:

[0195]

[0196] S8. Solve the optimization objective of S7 using the NSGA-II algorithm, then obtain the ideal allocation scheme based on the Shapley value, and use the Topsis method to find the optimal solution in the Pareto solution set that best approximates the ideal allocation scheme. The decision variable corresponding to the optimal solution is the optimal capacity scheme; the steps are as follows:

[0197] S8.1 Obtain the spatiotemporal distribution map of charging demand in the highway network through sumo; the specific data has been obtained through step S3.4.

[0198] S8.2. Based on the average charging demand over 24 hours, set the ideal capacity range for each charging station, and define the form of each solution in the algorithm as x = {m1, m2, ..., m...} i ,…,m n}, where each value m i The number of charging piles installed at the corresponding node is an even number.

[0199] S8.3, According to the maximum population size N pop Generate an initial population, and check whether the generated initial solution satisfies the capacity interval constraint. If not, regenerate until the initial population size reaches N. pop ;

[0200] S8.4 Calculate the first fitness function f1 and the second fitness function f2 for all individuals, and obtain two sets. and The tournament selection algorithm is used to randomly select individuals from the initial population. The two individuals with the highest fitness functions win and are selected. Then, crossover and mutation operators are used to generate offspring individuals. The new individuals are checked again to see if they meet the capacity interval constraint. If they do, they are retained in the offspring population until the offspring population also reaches the maximum population size N. pop ;

[0201] S8.5 Merge the parent and child populations using an elite strategy, and use the fast non-dominated sorting algorithm to obtain the sorting level for the new population after the merger.

[0202] According to the sorting hierarchy, individuals are taken from the lowest to the highest level and added to the new population. If the number of individuals in the new population exceeds the maximum population size when a certain level is taken, the crowding degree of individuals in that level is calculated, and individuals are taken from the lowest to the highest crowding degree until the number of individuals in the new population reaches the maximum population size.

[0203] S8.6 If the number of iterations has not reached the maximum number of iterations (Gen)max If the condition is met, proceed to S7.4; otherwise, terminate the algorithm and output the Pareto front solution.

[0204] S8.7 All charging stations are members of the alliance s, N = {1, 2, ..., n}. A set of constant solutions x0 = {a, a, ..., a} is used to calculate the daily revenue function f2 as the characteristic function of the cooperative game among the alliance members, i.e.:

[0205] v(s)=f2(x0)

[0206] Then, the n charging stations engage in a game of strategy, and the allocation scheme is y = {y1, y2, ..., y}. n} satisfies the following formula:

[0207]

[0208]

[0209] In the formula, |s| is the number of elements in set s (equal to n); y i This represents the Shapley value of charging station i; v(s / i) is the revenue of consortium s excluding charging station i; [v(s)-v(s / i)] is the marginal contribution of charging station i to consortium s; based on the Shapley value of each charging station, a set of ideal solutions can be obtained. By combining the concept of topsis, the solution closest to the ideal solution is found from the Pareto front solution set, which is the optimal solution.

[0210] Beneficial effects of the present invention

[0211] This invention can obtain effective traffic routes and flow rates through highway entrance and exit toll data, which strongly guarantees the authenticity and reliability of the experiment.

[0212] Meanwhile, in energy consumption models, most studies consider that weather factors will exacerbate the reduction of the actual range of EVs. However, the energy consumption model of electric vehicles on highways is affected by many factors, such as the density of traffic ahead, motor efficiency, internal equipment operation, and the behavior of EV users. Therefore, this invention considers combining vehicle dynamics and motor dynamics to construct energy consumption models according to different situations, and also constructs charging behavior scenarios of drivers passing by charging stations.

[0213] This invention establishes a power output model for each unit in an integrated photovoltaic-storage-charging station under the background of smart grid and multi-energy complementary charging station technology. Based on the charging load and photovoltaic power generation, the power output results of each unit are calculated, which facilitates the evaluation of the operator's daily revenue in the subsequent optimization of the charging station capacity model.

[0214] Optimization studies of charging stations can be conducted experimentally based on the existing charging resources, adopting a strategy of charging less and stopping more frequently, which is beneficial for saving costs. Attached Figure Description

[0215] Figure 1 This is a flowchart of the steps of the present invention;

[0216] Figure 2 This is a schematic diagram of the probability density curve of a car entering a highway according to the present invention;

[0217] Figure 3 This is a schematic diagram of the shortest path and effective path set of the present invention;

[0218] Figure 4 This is a schematic diagram of the relationship curves of the present invention;

[0219] Figure 5 This is a spatiotemporal distribution diagram of charging demand according to the present invention; wherein, (a) is a schematic diagram of the spatiotemporal distribution path of charging demand; and (b) is a three-dimensional bar chart of the spatiotemporal distribution of charging demand.

[0220] Figure 6 This is a system structure block diagram of the photovoltaic-storage-charging integrated power station of the present invention;

[0221] Figure 7 This is a photovoltaic power output data diagram of the present invention; wherein, (a) is a schematic diagram of photovoltaic power generation under 500 scenarios; (b) is a schematic diagram of the number under each cluster; (c) is a schematic diagram of photovoltaic power generation under 10 reduced scenarios; (d) is a schematic diagram of the probability distribution under each scenario; and (e) is a schematic diagram of photovoltaic power output with uncertainty.

[0222] Figure 8 A diagram showing the power output of each unit in the photovoltaic-storage charging station;

[0223] Figure 9 This is a schematic diagram of the convergence results of the NSGA-II algorithm. Detailed Implementation

[0224] The present invention will be further described in detail below with reference to specific embodiments.

[0225] An optimization method for charging stations in a highway network based on the spatiotemporal distribution of charging demand includes the following steps:

[0226] S1. Collect and process road network data and highway toll records to obtain electric vehicle travel characteristic parameters and generate road network vehicle driving trajectories;

[0227] Electric vehicle travel characteristic parameters include: time data and location data;

[0228] The specific steps of S1 are as follows:

[0229] S1.1 Obtain the experimental highway network model using an open-source map; the steps are as follows:

[0230] S1.1.1 Obtain the road network model of the target province using the open-source map software OpenStreetMap;

[0231] In this experimental example, a road network map of Shandong Province was selected as the experimental object.

[0232] S1.1.2 Modify and organize the road network model of the target province using the road network editing software Netdit;

[0233] The modification and reorganization involves retaining only the highway network model.

[0234] S1.1.3. Set the locations of toll stations, service areas, and parking areas on the expressway network, as well as the locations of detectors on road sections, as nodes and number them;

[0235] S1.2 Obtain the probability density curve of vehicles entering the highway network through kernel density estimation; the steps are as follows:

[0236] S1.2.1 Discretize the time data of vehicles entering the highway network by grid points and calculate the histogram of the time data;

[0237] The time data range is R = [0h, 24h], where R is MIN for 0h and MAX for 24h;

[0238] S1.2.2 Normalize the probability density of the histogram and perform a discrete cosine transform to obtain the frequency domain coefficients a. k The expression is as follows:

[0239]

[0240] In the formula, b k x represents the k-th interval; s Let A represent the s-th data sample; let A represent the total number of data samples; let b represent the total number of intervals, which is 24; M represents the total number of intervals. k p represents the probability that a data sample falls within interval k; k p represents the normalized probability of the k-th interval; z This represents the normalized probability value of the z-th interval, where z∈k;

[0241] S1.2.3. Calculate the optimal bandwidth h iteratively using the fixed-point equation, and then smooth the histogram; the expression is as follows:

[0242]

[0243] R = MAX - MIN

[0244] In the formula, R represents the time data range; t * The optimal smoothing parameter is expressed as follows:

[0245]

[0246] In the formula, t represents the smoothing parameter; f(t) is the frequency domain smoothing function related to t, and its expression is as follows:

[0247]

[0248] In the formula, L represents the maximum order of iterations;

[0249] S1.2.4. The final probability density distribution is obtained through the smoothed frequency domain coefficients and the inverse discrete cosine transform; the expression is as follows: The result is as follows Figure 2 As shown;

[0250]

[0251] In the formula, Represents the smoothed frequency domain coefficients; Represents the inverse discrete cosine transform;

[0252] S1.3. Obtain the vehicle trajectories in the road network using the A* algorithm and path assignment model; the steps are as follows:

[0253] S1.3.1. Based on the vehicle location data in the provincial expressway toll record data, obtain the starting and ending points of the vehicle entering the expressway network;

[0254] S1.3.2, the A* algorithm calculates the shortest path from the starting point to the destination. Simultaneously, considering the user's route choices along the way, random perturbations are applied to the road segment impedance in the mesoscopic traffic simulation software DynasTIM to expand the set of effective paths; specific results are as follows... Figure 3 As shown;

[0255] S2. Establish an energy consumption model by combining the factors that affect the energy consumption of electric vehicles;

[0256] Factors affecting the energy consumption of electric vehicles include: motor conversion efficiency, internal equipment operating energy consumption, and high-speed driving resistance (friction resistance, wind resistance, and gradient resistance).

[0257] The steps to establish an energy consumption model are as follows:

[0258] S2.1. Based on the law of conservation of energy, consider the energy loss of the vehicle from both vehicle dynamics and motor dynamics perspectives; firstly, establish the driving equation of the electric vehicle, as shown in the following expression:

[0259] F t =F m +Fk +F p +F j

[0260] In the formula, F t For driving force; F m F represents frictional resistance. k For air resistance; F p For ramp resistance; F j To increase resistance;

[0261] By organizing, we can obtain:

[0262]

[0263] In the formula, m is the vehicle mass; g is the acceleration due to gravity; f is the road friction resistance coefficient; α is the road slope angle; C D V is the drag coefficient; A is the vehicle's frontal area; v s δ represents the actual vehicle speed; δ is the rotational mass conversion factor.

[0264] S2.2 Establish driving energy consumption models under different driving conditions;

[0265] Different driving states include: constant speed driving, accelerating driving, and decelerating driving; that is, based on the vehicle's current speed v. s and target vehicle speed v m Determine different driving states of the vehicle;

[0266] Specifically as follows:

[0267] S2.2.1 When the vehicle is traveling at a constant speed (v) s =v m In this state, the vehicle experiences no acceleration resistance, therefore the energy consumption expression for the electric vehicle under these conditions is as follows:

[0268]

[0269] In the formula, η d For motor conversion efficiency;

[0270] S2.2.2, When the vehicle accelerates (v s <v m In the case of an uphill slope, the vehicle experiences acceleration resistance and requires a net driving force to accelerate. Therefore, the energy consumption expression for acceleration driving of an electric vehicle in this situation is as follows:

[0271]

[0272] In the formula, ΔT represents the torque added by the vehicle's acceleration motor, expressed as follows:

[0273]

[0274] In the formula, r is the wheel radius; i0 is the gear ratio of the reducer;

[0275] ω n The motor speed is expressed as follows:

[0276]

[0277] In the formula, i1 is the transmission ratio of the vehicle's transmission system;

[0278] S2.2.3, When the vehicle decelerates (v s >v m In the current state, the vehicle's regenerative braking system is activated, and there is no driving force. The expression for the energy recovered by braking in this state is as follows:

[0279]

[0280] In the formula, η t To improve regenerative braking efficiency;

[0281] S2.2.4 The energy consumption model for electric vehicles obtained by integration is as follows:

[0282] P = P x1 +P x2 +P i -P x3

[0283] In the formula, P x1 Energy consumption for electric vehicles; P x2 To accelerate the energy consumption of electric vehicles; P x3 Energy recovery during braking of electric vehicles; P i Energy consumption for the operation of internal equipment in electric vehicles;

[0284] S2.3, Target vehicle speed v in the driving energy consumption model under different driving conditions m Make corrections; the steps are as follows:

[0285] Specifically, the simulated vehicle speed in the highway network is not always constant, and when rain, snow, fog, or traffic congestion occur, it will affect the driver's visibility and the vehicle's grip. Therefore, the vehicle speed and the driver's driving anxiety need to be corrected.

[0286] In addition, considering that drivers are worried about extreme weather and heavy traffic, and that frequent starts, stops or decelerations may cause range anxiety, which could lead to incorrect charging decisions, drivers' anxiety about battery level also needs to be corrected.

[0287] The steps are as follows:

[0288] S2.3.1 The correction expression for vehicle speed based on traffic flow is as follows:

[0289]

[0290] In the formula, v′ m The target vehicle speed after traffic flow correction; v f k is the free-flow speed of the vehicle. m k represents the average density of the frontal influence area (SIR) of a vehicle. jam The congestion density in the area in front of the vehicle; k min σ represents the upper limit of density when the vehicle is running at free-flow speed; σ and β are the first and second parameters, respectively; see Appendix Table 1 for detailed parameter settings;

[0291] Table 1. Parameters of the Density-Vehicle Speed ​​Model

[0292]

[0293] The density-vehicle speed result is as follows Figure 3 As shown;

[0294] S2.3.2 The correction expression for vehicle speed due to weather conditions is as follows:

[0295]

[0296] In the formula, v′ m ′ represents the target speed after weather correction; The weather conditions from charging station i to i+1 are used as the correction factor for vehicle speed. Precipitation and snowfall are used as the scale for calibration. The target vehicle speed is reduced as precipitation and snowfall increase, with the reduction range being 4%-14%.

[0297] S2.4 Correct the anxiety level coefficient;

[0298] Considering that drivers are worried about extreme weather and heavy traffic, and that frequent starts, stops or decelerations can cause range anxiety, which may lead to incorrect charging decisions, drivers' anxiety about battery level also needs to be corrected.

[0299] The correction expression for the weather-related anxiety coefficient is as follows:

[0300]

[0301] In the formula, Let θ be the anxiety energy coefficient of an electric vehicle driver at charging station i+1 on the highway network at time T; θ is the basic anxiety energy coefficient under the condition of good weather and reasonable traffic flow, with a value of 0.3. The weather and traffic flow for charging stations i to i+1 are used as correction factors for the anxiety-based battery coefficient, with a range of 6%-20%.

[0302] S3. Input the parameters from S1 into the mesoscopic traffic simulation software DynasTIM, and run the energy consumption model from S2 in the software to track the vehicle's state of charge in real time. Calculate the charging nodes selected by the vehicle and the amount of electricity replenished at the nodes through the constructed charging scenario, and finally obtain the spatiotemporal distribution of charging demand in various service areas of the highway network; the specific steps are as follows:

[0303] S3.1 Utilize a traffic simulator to track the traffic status of the highway network or customize road network information;

[0304] Traffic simulators include: the mesoscopic traffic simulation software DynasTIM;

[0305] Highway network traffic status or custom road network information includes: highway network toll node locations, lane speed limits, road gradient angles, air density, substation locations, service area locations, and the capacity of set charging stations;

[0306] S3.2 Simulation of vehicle driving on highway network;

[0307] S3.2.1 Set different vehicle parameters in the routing file, including: vehicle weight, frontal area, speed, rotational mass conversion factor, electric motor energy conversion efficiency, rolling resistance factor, energy consumption of internal equipment of electric vehicle and maximum state of charge. See Appendix Table 2 for specific parameter values.

[0308] Table 2 Energy Consumption Model Parameter Table

[0309]

[0310] S3.2.2 Based on the probability density curve of vehicles entering the highway network, vehicles are put into the road network for driving, and they travel according to the shortest path from the vehicle's origin to its destination and other effective paths.

[0311] S3.3 Calculate energy consumption using the established energy consumption model to obtain vehicle charging demand data;

[0312] Specifically, during the operation of the mesoscopic traffic simulation DynasTIM, the travel chain of vehicles in the highway network is as follows: Figure 4 The parameters (time from the previous node to this node, travel distance, and vehicle speed) of each node along the driving trajectory are input into the energy consumption model built in Python to calculate energy consumption, update the vehicle's remaining battery power, and generate correction factors based on the node parameters (weather information and traffic flow information). The vehicle's speed and anxiety battery coefficient are corrected according to S2.3 and S2.4. The driving speed of the simulated vehicle after leaving the node is updated through the TraCI interface. The simulation is updated in real time after each node is passed until the vehicle ends the simulation and leaves the road network.

[0313] S3.4. By obtaining the remaining battery power and the anxiety factor when the vehicle arrives at the node, the user's charging decision is formulated based on actual user driving habits and charging judgment, and the scenario is divided accordingly; as follows:

[0314] Scenario 1: When the driver reaches a certain node, the electric vehicle's state of charge is lower than the anxiety level, so the driver chooses to recharge at that node; the expression is as follows:

[0315]

[0316] Scenario 2: When the driver reaches a certain node, the electric vehicle's state of charge is higher than the anxiety level, and energy can still be replenished at the next node. The driver chooses to continue. The expression is as follows:

[0317]

[0318] Scenario 3: When the driver reaches a certain node, the electric vehicle's state of charge (SBC) is higher than the anxiety level, but there is no way to recharge at the next node, and the SBC is lower than the anxiety level upon reaching the next node, the driver will choose to recharge at that node; the expression is as follows:

[0319]

[0320] When the driver selects a charging node, charging is complete when the battery reaches 90% of its maximum state of charge. The formula for calculating the amount of battery charge replenished is as follows:

[0321] Q = 0.9SOC max -SOC i

[0322] In the power calculation formulas for scenarios one, two, three, and the supplement: SOC i State of charge (SOC) of the electric vehicle when it arrives at charging station i. max This represents the maximum state of charge of an electric vehicle. Let T be the anxiety coefficient of an electric vehicle driver at charging station i on the highway network at time T;

[0323] After the vehicles completed their operation in DynasTIM, the total charging amount of the electric vehicles at each node was calculated every hour. The charging demand of all nodes represents the spatiotemporal distribution of charging demand in this study. The results are shown in [link to results]. Figure 5 Parts (a) and (b) are included; at the same time, charging piles are set up and operated in some service areas, and the charging service time provided by the nodes and the user waiting time for charging when the charging resources in the station are insufficient can also be obtained.

[0324] Figure 5 The size of the circle in section (a) reflects the amount of charging required.

[0325] Depend on Figure 5 (b) It can be seen that the spatial and temporal distribution of charging demand in service areas is not uniform, showing a "peak and trough" driving pattern, which is consistent with the user's travel pattern.

[0326] S4. Based on the spatiotemporal distribution of charging demand in each service area in S3 and numbering each service area, select the service area nodes where charging stations need to be built using the maximum coverage location model.

[0327] Specifically, the optimization objective is to maximize the total charging demand at the covered demand points:

[0328]

[0329] In the formula, Q represents the total number of demand points; d q The charging demand at demand point q;

[0330] Constraints are applied to coverage, number of sites, and decision variables; the expression is as follows:

[0331] Coverage constraints:

[0332] Site quantity constraints:

[0333] In the formula, N represents the total number of charging stations selected for construction, which in this implementation is taken as 85% of the total number of service areas; x p For the first binary variable, y p For the second binary variable and a pq The third binary variable has the following expression:

[0334]

[0335] The final output is the number of the selected service area, with a total of N.

[0336] S5. Establish a photovoltaic-storage-charging integrated charging station model. Using the charging load data of the selected charging station obtained in S3 and S4, obtain the output data of each unit when the photovoltaic-storage-charging station meets these loads.

[0337] Currently, the development prospects of charging stations are trending towards intelligent and digital multi-energy complementarity, achieving intelligent scheduling through technologies such as the Internet of Things, big data, and cloud computing. Among these, photovoltaic-storage charging station technology is relatively mature and has been widely put into use. The charging stations constructed in this paper are all photovoltaic-storage charging stations, mainly composed of an intelligent control system, photovoltaic units, energy storage units, power distribution network interaction, and charging equipment. Their system block diagram is shown below. Figure 6 As shown.

[0338] Each unit operates based on its charging load status: when the photovoltaic power generation can meet the charging load, it prioritizes charging the energy storage system. If there is still surplus power, it sells electricity to the distribution network. Conversely, if the photovoltaic power generation is insufficient to support the charging load, the energy storage system discharges. If it still cannot meet the charging load or the energy storage capacity is lower than the set value, it purchases electricity from the distribution network. This invention requires the output data of each unit for operator revenue assessment.

[0339] The output models for each unit are established as follows:

[0340] S5.1 Establish a photovoltaic unit output model; details are as follows:

[0341] The output power of photovoltaic cells is mainly affected by the intensity of sunlight; therefore, the output model of a photovoltaic power unit is expressed as follows:

[0342]

[0343] In the formula, η pv denoted as photoelectric conversion efficiency; S is the area of ​​the photovoltaic unit; r(t) is the change in light intensity over time.

[0344] While this invention lacks data on solar irradiance for all regions of the province, the differences in solar irradiance between provinces are relatively small, and solar irradiance generally follows a normal distribution over time. Therefore, this invention calculates the mean and variance based on some historical data, uses Latin hypercube stratified sampling to generate data for 500 photovoltaic scenarios, and then reduces them to ten scenarios using k-means clustering to reflect the uncertainty of photovoltaics. The details are as follows:

[0345] The mean and variance are calculated based on historical photovoltaic data, as shown in the following expressions:

[0346]

[0347] σ t =u2(t)·μ t

[0348] In the formula, c represents the scaling factor used to control the calculation of mean and variance, which is a constant and is taken as 1 in this embodiment; u2(t) represents the random factor generated by Latin hypercube sampling;

[0349] The photovoltaic output data for each photovoltaic-storage charging station, which follows a normal distribution, is generated using the following expression:

[0350]

[0351] In the formula, P pv1 (t) represents historical photovoltaic data; P pv2 (t) represents the photovoltaic data of each photovoltaic-storage charging station in the study;

[0352] S5.2 Establish the energy storage unit output model; details are as follows:

[0353] The output model of the energy storage unit includes two states: charging and discharging.

[0354] The expression for the discharge state is as follows:

[0355]

[0356] The expression for the charging state is as follows:

[0357]

[0358] In the formula, η z For energy storage unit conversion efficiency; η c η dis These are the charging efficiency and discharging efficiency, respectively; E h P represents the rated capacity of the energy storage unit. es (t) represents the charging and discharging power;

[0359] The battery model of the energy storage unit is established, and the expression is as follows:

[0360]

[0361] In the formula, SOC es (t) represents the state of charge of the energy storage unit during time period t;

[0362] S5.3 Establish the power output model of the distribution network; details are as follows:

[0363] Based on the charging load and the output of photovoltaic and energy storage units, a power distribution network output model is constructed; the expression is as follows:

[0364]

[0365] In the formula, η t For conversion efficiency; δ t Line loss rate; Let i be the charging demand of charging station i during time period t;

[0366] The constraints of this invention on energy storage units include:

[0367] Energy storage power constraint: -P es (t) max ≤P es (t)≤P es (t) max

[0368] Constraints on the state of charge of energy storage: 0.1E h ≤SOC es ≤0.9E h

[0369] Finally, the output results of each unit were obtained, such as Figure 8 As shown;

[0370] S6. Based on the results of S3, S4, and S5, calculate the degree to which the charging demand of each charging station is met, the rational allocation of power resources, and the degree of relief of road traffic congestion through the model.

[0371] Specifically, S4 selects N charging stations, and the set of the number of fast charging piles in each charging station is E = {m1, m2, ..., m n}, n∈N, and each fast charging pile in the charging station (slow charging piles are not considered in highway network service areas) can work 24 hours a day;

[0372] The calculation process is as follows:

[0373] S6.1 Calculate the degree to which charging demand is met; details are as follows:

[0374]

[0375] In the formula, The utilization rate of charging station i during time period t is specifically represented as follows:

[0376]

[0377] In the formula, Q represents the charging demand of charging station i during time period t (obtained from step S3.4); is The maximum charging resources that charging station i can provide are determined by the number of fast charging piles within the station, as shown below:

[0378] Q is =P f ·t is

[0379] In the formula, P f The charging power of the fast charging station; t is The duration of charging service provided by charging station i during time period t (obtained from step S3.4);

[0380] The average degree to which all charging stations meet the charging demand can be expressed as:

[0381]

[0382] S6.2 Calculate power resource allocation; details are as follows:

[0383] The formula for calculating the electricity consumption within the charging station is as follows:

[0384]

[0385] In the formula, η f Energy conversion efficiency of fast charging piles; E io This refers to the electricity consumption of other electrical facilities within charging station i. The entire photovoltaic-storage charging station is mainly powered by the photovoltaic system, energy storage system, and power distribution network. Therefore, the calculation expression for the power supply within the charging station is as follows:

[0386] E im =E i,pv +E i,es +E i,gr

[0387] In the formula, the power output model of the photovoltaic system in each charging station is expressed as:

[0388]

[0389] In the formula, E i,pv The power generation of the photovoltaic system within charging station i; η pv Photoelectric conversion efficiency; S i r is the area of ​​the photovoltaic units within the charging station i; i (t) represents the change in the solar radiation intensity on the horizontal surface of charging station i over time;

[0390] The power transmission model from the distribution network to each charging station can be approximated as follows:

[0391]

[0392] In the formula, η t For conversion efficiency; δ t Given the line loss rate, the average distribution of power resources across all charging stations can be expressed as:

[0393]

[0394] S6.3 Calculate traffic mitigation measures; details are as follows:

[0395]

[0396] In the formula, U i The congestion index of the road segment between charging station i and the previous charging station; The actual traffic flow through the road segment during time period t (obtained through traffic monitoring equipment); The road capacity that can pass through the road segment during time period t (a fixed value set for each road in the traffic simulation road network model); V represents the average vehicle speed passing through charging station i during time period t; i,free The free-flow vehicle speed passing through charging station i under no congestion conditions;

[0397] S7. Based on the data from S5 and the calculation results from S6, establish a charging station capacity model to determine the daily net profit after all charging stations on the road network are put into operation; that is, calculate the entropy weights of three aspects—the matching degree between charging resources and charging demand, power resource allocation, and traffic mitigation—using the entropy weight method to establish a road performance evaluation model; establish an overall revenue model based on cooperative game theory after all charging stations are put into operation; and take maximizing road performance and maximizing revenue as the objective functions.

[0398] The steps are as follows:

[0399] S7.1 Establish a road performance evaluation model using the entropy weight method;

[0400] The expression for establishing the road performance evaluation model using the entropy weight method is as follows:

[0401] f1=ω1·v1(N)+ω2·v2(N)+ω3·v3(N)

[0402] In the formula, ω1, ω2, and ω3 are the weights of three indicators: the degree to which charging demand is met by the charging station, the average distribution of power resources, and the degree of traffic relief.

[0403] S7.2 Establish an operator revenue model through cooperative game theory;

[0404] A daily revenue function for all charging stations in the road network after they are put into use is established using a cooperative game theory model. The revenue expression for each charging station is as follows:

[0405]

[0406] In the formula, c t The unit price at which a charging station provides charging services to users during time period t;

[0407] The daily cost of each charging station includes four parts: photovoltaic system operation and maintenance cost, energy storage system operation and maintenance cost, electricity purchase cost from the distribution network, and on-site facility operation and maintenance cost.

[0408] The operation and maintenance cost of a photovoltaic system is expressed as follows:

[0409]

[0410] In the formula, c pv This refers to the unit operation and maintenance price of a photovoltaic system. Let Δt be the output power of the photovoltaic system at charging station i during time period t; Δt is the unit length of time period t (1 hour).

[0411] The operation and maintenance cost expression for an energy storage system is as follows:

[0412]

[0413] In the formula, c E E represents the unit capacity operation and maintenance cost of an energy storage system. im Configure the energy storage capacity for charging station i; c p The unit transmission cost of the energy storage system; Let t be the output power of the energy storage system at charging station i during time period t.

[0414] The interaction cost expression for the distribution network is as follows:

[0415]

[0416] In the formula, The unit price at which the charging station buys from and sells to the distribution network during time period t; For time period t, the power purchased and sold by charging station i from the distribution network;

[0417] The cost expression for the site's infrastructure is as follows:

[0418] C i,inst =c f ·m i +c o

[0419] In the formula, c f The maintenance cost for a single fast charging station; m i c is the number of fast charging stations within charging station i; o Fixed operation and maintenance costs for other facilities within the charging station (a set constant term);

[0420] Therefore, the daily net profit function after all charging stations in the entire road network are put into use is:

[0421]

[0422] S8. Solve the optimization objective of S7 using the NSGA-II algorithm, then obtain the ideal allocation scheme based on the Shapley value, and use the Topsis method to find the optimal solution in the Pareto solution set that best approximates the ideal allocation scheme. The decision variable corresponding to the optimal solution is the optimal capacity scheme; the steps are as follows:

[0423] S8.1 Obtain the spatiotemporal distribution map of charging demand in the highway network through sumo; the specific data has been obtained through step S3.4.

[0424] S8.2. Based on the average charging demand over 24 hours, set the ideal capacity range for each charging station, and define the form of each solution in the algorithm as x = {m1, m2, ..., m...} i ,…,m n}, where each value m iThe number of charging piles installed at the corresponding node is an even number.

[0425] S8.3, According to the maximum population size N pop Generate an initial population, and check whether the generated initial solution satisfies the capacity interval constraint. If not, regenerate until the initial population size reaches N. pop ;

[0426] S8.4 Calculate the first fitness function f1 and the second fitness function f2 for all individuals, and obtain two sets. and The tournament selection algorithm is used to randomly select individuals from the initial population. The two individuals with the highest fitness functions win and are selected. Then, crossover and mutation operators are used to generate offspring individuals. The new individuals are checked again to see if they meet the capacity interval constraint. If they do, they are retained in the offspring population until the offspring population also reaches the maximum population size N. pop ;

[0427] S8.5 Merge the parent and child populations using an elite strategy, and use the fast non-dominated sorting algorithm to obtain the sorting level for the new population after the merger.

[0428] According to the sorting hierarchy, individuals are taken from the lowest to the highest level and added to the new population. If the number of individuals in the new population exceeds the maximum population size when a certain level is taken, the crowding degree of individuals in that level is calculated, and individuals are taken from the lowest to the highest crowding degree until the number of individuals in the new population reaches the maximum population size.

[0429] S8.6 If the number of iterations has not reached the maximum number of iterations (Gen) max If the result is satisfactory, proceed to S7.4; otherwise, terminate the algorithm and output the Pareto front solution. The results are shown in Appendix Table 3, and the convergence effect is as follows: Figure 9 As shown;

[0430] Table 3 Pareto Front Solutions

[0431]

[0432]

[0433] S8.7 All charging stations are members of the alliance s, N = {1, 2, ..., n}. A set of constant solutions x0 = {a, a, ..., a} is used to calculate the daily revenue function f2 as the characteristic function of the cooperative game among the alliance members, i.e.:

[0434] v(s)=f2(x0)

[0435] Then, the n charging stations engage in a game of strategy, and the allocation scheme is y = {y1, y2, ..., y}. n} satisfies the following formula:

[0436]

[0437] In the formula, |s| is the number of elements in set s (equal to n); y i This is the Shapley value of i; v(s / i) is the revenue of consortium s excluding charging station i; [v(s)-v(s / i)] is the marginal contribution of charging station i to consortium s;

[0438] A set of ideal solutions can be obtained based on the Shapley value for each charging station. By combining the concept of topsis, the solution closest to the ideal solution is found from the Pareto front solution set, which is the optimal solution.

[0439] In summary, this invention obtains effective electric vehicle routes and optimal charging station distribution through highway toll data. Furthermore, considering the distribution of charging stations in the highway network, this invention also takes into account factors such as weather, driver anxiety about battery level, traffic density, motor efficiency, and user behavior, resulting in more accurate and reliable final results.

Claims

1. A method for optimizing charging stations in a highway network based on the spatiotemporal distribution of charging demand, characterized in that, Includes the following steps: S1. Collect and process road network data and highway toll records to obtain electric vehicle travel characteristic parameters and generate road network vehicle driving trajectories; S2. Establish an energy consumption model by considering the factors that affect the energy consumption of electric vehicles. The steps are as follows: S2.

1. Based on the law of conservation of energy, consider the energy loss of the vehicle from both vehicle dynamics and motor dynamics perspectives; firstly, establish the driving equation of the electric vehicle, as shown in the following expression: ； In the formula, As the driving force; Frictional resistance; For air resistance; For slope resistance; To increase resistance; By organizing, we can obtain: ；； In the formula, m is the mass of the vehicle; g is the acceleration due to gravity; The coefficient of friction resistance of the road; The road slope angle; Where A is the drag coefficient; A is the vehicle's frontal area. Current vehicle speed; This is the rotational mass conversion factor; S2.2 Establish driving energy consumption models under different driving conditions; Different driving states include: constant speed driving, accelerating driving, and decelerating driving; that is, based on the vehicle's current speed... and target speed Determine different driving states of the vehicle; Specifically as follows: S2.2.1 When the vehicle is traveling at a constant speed, there is no acceleration resistance. Therefore, the energy consumption expression for electric vehicle driving at this time is as follows: ； In the formula, For motor conversion efficiency; S2.2.2 When the vehicle is accelerating or going uphill, it experiences acceleration resistance and requires a net driving force to accelerate. Therefore, the energy consumption expression for acceleration driving of an electric vehicle is as follows: ； In the formula, The torque added to the vehicle's acceleration motor is expressed as follows: ； In the formula, r is the radius of the wheel; The gear ratio of the reducer; The motor speed is expressed as follows: ； In the formula, i1 is the transmission ratio of the vehicle's transmission system; S2.2.3 When the vehicle is decelerating and the regenerative braking system is activated, with no driving force, the expression for the regenerative braking energy of the electric vehicle is as follows: ； In the formula, To improve regenerative braking efficiency; S2.2.4 The energy consumption model for electric vehicles obtained by integration is as follows: ； In the formula, Energy consumption for electric vehicles; Accelerating the energy consumption of electric vehicles; Recover energy during braking of electric vehicles; Energy consumption for the operation of internal equipment in electric vehicles; S2.3 Target vehicle speed in the driving energy consumption model under different driving conditions Make corrections; the steps are as follows: S2.3.1 The correction expression for vehicle speed based on traffic flow is as follows: ； In the formula, The target vehicle speed after traffic flow correction; The free-flow vehicle speed; The average density of the area in front of the vehicle; The congestion density in the area in front of the vehicle; This represents the upper limit of density for vehicles operating at free-flow speeds. These are the first parameter and the second parameter, respectively. S2.3.2 The correction expression for vehicle speed due to weather conditions is as follows: ； In the formula, The target speed after weather correction; The weather conditions from charging station i to i+1 are used as the correction factor for vehicle speed. Precipitation and snowfall are used as the scale for calibration. The target vehicle speed is reduced as precipitation and snowfall increase, with the reduction range being 4%-14%. S2.4 Correct the anxiety level coefficient; Considering that drivers are worried about extreme weather and the range anxiety caused by frequent starts, stops or deceleration, which may lead to incorrect judgments in charging decisions, drivers' anxiety about battery level also needs to be corrected. The correction expression for the weather-related anxiety coefficient is as follows: ； In the formula, Let be the anxiety coefficient of an electric vehicle driver at time T at charging station i+1 on the highway network; To set a baseline anxiety level for battery capacity under favorable weather and reasonable traffic conditions; The weather and traffic flow conditions for charging stations i to i+1 are used as correction factors for the anxiety-based battery coefficient, with a range of 6%-20%. S3. Input the parameters of S1 into the mesoscopic traffic simulation software DynasTIM, and run the energy consumption model of S2 in the software to track the vehicle's state of charge in real time. Calculate the charging nodes selected by the vehicle and the amount of electricity replenished at the nodes through the constructed charging scenario, and finally obtain the spatiotemporal distribution of charging demand in each service area of ​​the highway network. S4. Based on the spatiotemporal distribution of charging demand in each service area in S3 and numbering each service area, select the service area nodes where charging stations need to be built using the maximum coverage location model. S5. Establish a photovoltaic-storage-charging integrated charging station model. Using the charging load data of the selected charging station obtained in S3 and S4, obtain the output data of each unit when the photovoltaic-storage-charging station meets these loads. S6. Based on the results of S3, S4, and S5, calculate the degree to which the charging demand of each charging station is met, the rational allocation of power resources, and the degree of relief of road traffic congestion through the model. S7. Based on the data from S5 and the calculation results from S6, establish a charging station capacity model to determine the daily net profit after all charging stations on the road network are put into use; Specifically, the entropy weights of three aspects—the matching degree between charging resources and charging demand, power resource allocation, and traffic mitigation—are calculated using the entropy weight method to establish a road performance evaluation model; an overall revenue model based on cooperative game theory is established after all charging stations are put into operation; and the goal functions of maximizing road performance and maximizing revenue are solved. S8. Solve the optimization objective of S7 using the NSGA-II algorithm, then obtain the ideal allocation scheme based on the Sharpe ratio, and use the Topsis method to find the optimal solution in the Pareto solution set that best approximates the ideal allocation scheme. The decision variable corresponding to this solution is the optimal capacity scheme.

2. The method for optimizing highway charging stations based on the spatiotemporal distribution of charging demand according to claim 1, characterized in that, The steps for collecting and processing road network data and highway toll records to obtain electric vehicle travel characteristic parameters and generate road network vehicle trajectories are as follows: S1.1 Obtain the experimental highway network model using an open-source map. The steps are as follows: S1.1.1 Obtain the road network model of the target province using the open-source map software OpenStreetMap; S1.1.2 Modify and organize the road network model of the target province using the road network editing software Netdit; The modification and reorganization involves retaining only the highway network model. S1.1.

3. Set the locations of toll stations, service areas, and parking areas on the expressway network, as well as the locations of detectors on road sections, as nodes and number them; S1.2 Obtain the probability density curve of vehicles entering the highway network through kernel density estimation; the steps are as follows: S1.2.1 Discretize the time data of vehicles entering the highway network by grid points and calculate the histogram of the time data; S1.2.2 Normalize the probability density of the histogram and perform a discrete cosine transform to obtain the frequency domain coefficients. The expression is as follows: ； ； ； In the formula, b k x represents the k-th interval; s Let A represent the s-th data sample; let A represent the total number of data samples; let b represent the total number of intervals; M represent the total number of intervals. k p represents the probability that a data sample falls within interval k; k This represents the normalized probability of the k-th interval; This represents the normalized probability value of the z-th interval, where z∈k; S1.2.3 Calculate the optimal bandwidth iteratively using the fixed-point equation. To smooth the histogram, the expression is as follows: ； In the formula, R represents the time data range; The optimal smoothing parameter is expressed as follows: ； In the formula, t represents the smoothing parameter; It is a frequency domain smoothing function related to t, and its expression is as follows: ； In the formula, Indicates the maximum order of iterations; S1.2.

4. The final probability density distribution is obtained through the smoothed frequency domain coefficients and the inverse discrete cosine transform; the expression is as follows: ； In the formula, Represents the smoothed frequency domain coefficients; Represents the inverse discrete cosine transform; S1.

3. Obtain the vehicle trajectories in the road network using the A* algorithm and path assignment model; the steps are as follows: S1.3.

1. Based on the vehicle location data in the provincial expressway toll record data, obtain the starting and ending points of the vehicle entering the expressway network; S1.3.2 The A* algorithm calculates the shortest path from the starting point to the destination. At the same time, considering the user's path selection along the way, random perturbations are applied to the road segment impedance in the mesoscopic traffic simulation software DynasTIM to expand the set of effective paths.

3. The method for optimizing highway charging stations based on the spatiotemporal distribution of charging demand according to claim 1, characterized in that, The parameters of S1 are input into the mesoscopic traffic simulation software DynasTIM, and the energy consumption model of S2 is run in the software to track the vehicle's state of charge in real time. The charging scenario is constructed to calculate the charging node selected by the vehicle and the amount of electricity replenished at the node, and finally obtain the spatiotemporal distribution of charging demand in each service area of ​​the highway network. The specific steps are as follows: S3.1 Utilize a traffic simulator to track the traffic status of the highway network or customize road network information; Traffic simulators include: the mesoscopic traffic simulation software DynasTIM; Highway network traffic status or custom road network information includes: highway network toll node locations, lane speed limits, road gradient angles, air density, substation locations, service area locations, and the capacity of set charging stations; S3.2 Simulation of vehicle driving on highway network; S3.2.1 Set different vehicle parameters in the routing file, including: vehicle weight, frontal area, speed, rotational mass conversion factor, electric motor energy conversion efficiency, rolling resistance factor, energy consumption of internal equipment of electric vehicle and maximum state of charge; S3.2.2 Based on the probability density curve of vehicles entering the highway network, vehicles are put into the road network for driving, and they travel according to the shortest path from the vehicle's origin to its destination and other effective paths. S3.3 Calculate energy consumption using the established energy consumption model to obtain vehicle charging demand data; S3.

4. By obtaining the remaining battery power and the anxiety factor when the vehicle arrives at the node, the user's charging decision is formulated based on actual user driving habits and charging judgment, and the scenario is divided accordingly; as follows: Scenario 1: When the driver reaches a certain node, the electric vehicle's state of charge is lower than the anxiety level, so the driver chooses to recharge at that node; the expression is as follows: ； Scenario 2: When the driver reaches a certain node, the electric vehicle's state of charge is higher than the anxiety level, and energy can still be replenished at the next node. The driver chooses to continue. The expression is as follows: ； Scenario 3: When the driver reaches a certain node, the electric vehicle's state of charge (SBC) is higher than the anxiety level, but there is no way to recharge at the next node, and the SBC is lower than the anxiety level upon reaching the next node, the driver will choose to recharge at that node; the expression is as follows: ； When the driver selects a charging node, charging is complete when the battery reaches 90% of its maximum state of charge. The formula for calculating the amount of battery charge replenished is as follows: ； In the power calculation formulas for scenarios one, two, three, and the supplement: The state of charge of the electric vehicle when it arrives at charging station i; This represents the maximum state of charge of an electric vehicle. Let denot be the anxiety coefficient of an electric vehicle driver at time T at charging station i on the highway network.

4. The method for optimizing highway charging stations based on the spatiotemporal distribution of charging demand according to claim 1, characterized in that, Based on the spatiotemporal distribution of charging demand in each service area in S3 and numbering each service area, the service area nodes where charging stations need to be built are selected using the maximum coverage location selection model, as detailed below: The optimization objective is to maximize the total charging demand at the covered demand points. ； In the formula, Q represents the total number of demand points; The charging demand at demand point q; Constraints are imposed on coverage, number of sites, and decision variables; the expression is as follows: Coverage constraints: ; Site quantity constraints: ; In the formula, N is the total number of charging stations to be built; For the first binary variable, For the second binary variable and The third binary variable has the following expression: ； 。 5. The method for optimizing highway charging stations based on the spatiotemporal distribution of charging demand according to claim 1, characterized in that, The establishment of the integrated photovoltaic-storage-charging station model involves using the charging load data of the selected charging station obtained through S3 and S4. When the photovoltaic-storage-charging station meets these loads, the output data of each unit is obtained, as detailed below: The output models for each unit are established as follows: S5.1 Establish a photovoltaic unit output model; details are as follows: The output power of photovoltaic cells is mainly affected by the intensity of sunlight; therefore, the output model of a photovoltaic power unit is expressed as follows: ； In the formula, S represents the photoelectric conversion efficiency; S represents the area of ​​the photovoltaic unit. The change in light intensity over time; S5.2 Establish the energy storage unit output model; details are as follows: The output model of the energy storage unit includes two states: charging and discharging. The expression for the discharge state is as follows: ； The expression for the charging state is as follows: ； In the formula, For energy storage unit conversion efficiency; These are charging efficiency and discharging efficiency, respectively. This refers to the rated capacity of the energy storage unit. This refers to the charging and discharging power. The battery model of the energy storage unit is established, and the expression is as follows: ； In the formula, This represents the state of charge of the energy storage unit during time period t. S5.3 Establish the power output model of the distribution network; details are as follows: Based on the charging load and the output of photovoltaic and energy storage units, a power distribution network output model is constructed; the expression is as follows: ； In the formula, For conversion efficiency; Line loss rate; Let i be the charging demand of charging station i during time period t; The constraints of energy storage units include: Energy storage power constraints: ; Constraints on the state of charge of energy storage: .

6. The method for optimizing highway charging stations based on the spatiotemporal distribution of charging demand according to claim 1, characterized in that, The calculation process for determining the degree to which charging demand is met, the rational allocation of power resources, and the mitigation of road traffic congestion at each charging station based on the results of S3, S4, and S5 is as follows: S6.1 Calculate the degree to which charging demand is met; details are as follows: ； In the formula, The utilization rate of charging station i during time period t is specifically represented as follows: ； In the formula, Let i be the charging demand of charging station i during time period t; The maximum charging resources that charging station i can provide are specifically represented as follows: ； In the formula, The charging power of the fast charging station; The charging service duration of charging station i during time period t; The average degree to which all charging stations meet charging demand is represented as follows: ； S6.2 Calculate power resource allocation; details are as follows: The formula for calculating the electricity consumption within the charging station is as follows: ； In the formula, Energy conversion efficiency of fast charging piles; For the electricity consumption of other electrical facilities within charging station i; The entire photovoltaic-energy storage charging station is mainly powered by a photovoltaic system, an energy storage system, and a power distribution network. Therefore, the calculation formula for the power supply within the charging station is as follows: ； In the formula, the power output model of the photovoltaic system in each charging station is expressed as: ； In the formula, The power generation of the photovoltaic system within charging station i; Photoelectric conversion efficiency; The area of ​​the photovoltaic units within charging station i; The change in sunlight intensity on the horizontal surface of charging station i over time; The power transmission model from the distribution network to each charging station can be approximated as follows: ； In the formula, For conversion efficiency; The line loss rate is used; therefore, the average distribution of power resources across all charging stations is expressed as follows: ； S6.3 Calculate traffic mitigation measures; details are as follows: ； ； In the formula, The congestion index of the road segment between charging station i and the previous charging station; The actual traffic flow through the road segment during time period t; Let t be the road capacity that the road segment can pass through during time period t; Let be the average vehicle speed passing through charging station i during time period t; Let be the free-flow speed of a vehicle passing through charging station i in the absence of congestion.

7. The method for optimizing charging stations in a highway network based on the spatiotemporal distribution of charging demand according to claim 6, characterized in that, Based on the data from S5 and the calculation results from S6, a charging station capacity model is established to determine the daily net profit after all charging stations in the road network are put into operation. The steps for determining the daily net profit after all charging stations in the road network are put into operation are as follows: S7.1 Establish a road performance evaluation model using the entropy weight method; The expression for establishing the road performance evaluation model using the entropy weight method is as follows: ； In the formula, , , The weights of three indicators are assigned to the charging station's ability to meet charging demand, the average distribution of power resources, and the degree of traffic relief. S7.2 Establish an operator revenue model through cooperative game theory; A daily revenue function for all charging stations in the road network after they are put into use is established using a cooperative game theory model. The revenue expression for each charging station is as follows: ； In the formula, The unit price at which a charging station provides charging services to users during time period t; The daily cost of each charging station includes four parts: photovoltaic system operation and maintenance cost, energy storage system operation and maintenance cost, electricity purchase cost from the distribution network, and on-site facility operation and maintenance cost. The operation and maintenance cost of a photovoltaic system is expressed as follows: ； In the formula, This refers to the unit operation and maintenance price of a photovoltaic system. Let t be the output power of the photovoltaic system at charging station i during the time period t. t represents the unit length of the time interval; The operation and maintenance cost expression for an energy storage system is as follows: ； In the formula, The unit capacity operation and maintenance cost of the energy storage system; Configure the energy storage capacity for charging station i; The unit transmission cost of the energy storage system; Let t be the output power of the energy storage system at charging station i during time period t. The interaction cost expression for the distribution network is as follows: ； In the formula, , The unit price at which the charging station buys from and sells to the distribution network during time period t; , For time period t, the power purchased and sold by charging station i from the distribution network; The cost expression for the site's infrastructure is as follows: ； In the formula, The maintenance cost for a single fast charging station; The number of fast charging stations in charging station i; Fixed maintenance costs for other facilities within the charging station; Therefore, the daily net profit function after all charging stations in the entire road network are put into use is: 。 8. The method for optimizing charging stations in a highway network based on the spatiotemporal distribution of charging demand according to claim 7, characterized in that, The steps for solving the optimization objective of S7 using the NSGA-II algorithm, obtaining the ideal allocation scheme based on the Sharpe ratio, and finding the optimal solution in the Pareto solution set that best approximates the ideal allocation scheme using the Topsis method, with the decision variable corresponding to the optimal solution being the optimal capacity scheme, are as follows: S8.1 Obtain the spatiotemporal distribution map of charging demand in the highway network through sumo; the specific data has been obtained through step S3.

4. S8.

2. Based on the average charging demand over 24 hours, determine the ideal capacity range for each charging station and define the form of each solution in the algorithm. , where each value The number of charging piles installed at the corresponding node is an even number. S8.3, According to the maximum population size Generate an initial population, and check whether the generated initial solution satisfies the capacity interval constraint. If not, regenerate until the initial population size reaches the required level. ; S8.4 Calculate the first fitness function for all individuals. Second function Two sets were obtained. and The tournament selection algorithm is used to randomly select individuals from the initial population. The two individuals with the highest fitness functions win and are selected. Then, crossover and mutation operators are used to generate offspring individuals. The new individuals are checked again to see if they meet the capacity interval constraint. If they do, they are retained in the offspring population until the offspring population also reaches the maximum population size. ; S8.5 Merge the parent and child populations using an elite strategy, and use the fast non-dominated sorting algorithm to obtain the sorting level for the new population after the merger. According to the sorting hierarchy, individuals are taken from the lowest to the highest level and added to the new population. If the number of individuals in the new population exceeds the maximum population size when a certain level is taken, the crowding degree of individuals in that level is calculated, and individuals are taken from the lowest to the highest crowding degree until the number of individuals in the new population reaches the maximum population size. S8.6 If the number of iterations has not reached the maximum number of iterations. If the condition is met, proceed to S7.4; otherwise, terminate the algorithm and output the Pareto front solution. S8.7, all charging stations are members of the alliance. Take a set of constant solutions Calculate the daily return function As a characteristic function of cooperative games among alliance members, that is: ； Then, the n charging stations engage in a game of strategy to determine the allocation scheme. Satisfy the following formula: ； ； In the formula, It is the number of elements in set s; That is, the Sharpe ratio of i; Remove the revenue from charging station i from the alliance s; The marginal contribution of charging station i to the alliance s.

Images