A simulation and scheduling optimization method for the spatiotemporal distribution of electric vehicle charging demand considering the travel patterns of vehicle owners

By simulating the travel laws and traffic road network of electric vehicles, optimizing the time and space distribution of electric vehicles, solving the problem of peak-to-valley difference in power grid load caused by disorderly charging of electric vehicles, and achieving the stability of the power system and the improvement of car owner satisfaction .

CN118410902BActive Publication Date: 2024-11-15ZHEJIANG JIACHEN NEW ENERGY CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410510965.9
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-04-26
Publication Date
2024-11-15
Estimated Expiration
2044-04-26

AI Technical Summary

Technical Problem

Disorderly charging of electric vehicles leads to an expansion of the peak-to-valley difference in the grid load, which may aggravate the imbalance of power supply and demand and affect the stability of the grid.

Method used

The space-time distribution simulation and scheduling optimization method of electric vehicle charging demand that considers the travel rules of car owners. By establishing a traffic road network model, using the Monte Carlo method and particle swarm algorithm, the space-time distribution and electricity price response of electric vehicles are optimized, and the charging space-time distribution and electricity price response of electric vehicles are formulated to reduce charging. Orderly charging scheduling scheme for cost and load peaks.

Benefits of technology

The orderly charging of electric vehicles is achieved, charging costs and power grid load peaks are reduced, and the stability of the power system and car owner satisfaction are improved.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118410902B_ABST
    Figure CN118410902B_ABST
Patent Text Reader

Abstract

The present invention provides a simulation and scheduling optimization method for the spatiotemporal distribution of electric vehicle charging demand considering the travel rules of car owners, including the following steps: considering the traffic network attributes, establishing a traffic network model, and considering the congested road conditions in the traffic network model; considering the travel rules of electric vehicles in the region, using the Monte Carlo method to establish a spatiotemporal distribution model of charging demand in the region; considering the demand response characteristics of electric vehicles, according to the time-of-use electricity price, establishing a scheduling model with the optimization goal of minimizing the charging cost of electric vehicles; considering the satisfaction of electric vehicle owners and the randomness of charging, adding time constraints and scheduling scale constraints to the established scheduling optimization model; finally, using a particle swarm algorithm with a penalty term to solve the scheduling optimization model with constraints. The present invention fully considers the traffic status, electric vehicle status, daily travel habits of car owners and charging needs, and the constructed charging scheduling optimization scheme can effectively reduce the peak load of the power grid while ensuring the satisfaction of car owners.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the technical field of electricity service, and relates to a method for simulating and optimizing the charging of electric vehicles, and in particular to a method for simulating and optimizing the temporal and spatial distribution of charging demands of electric vehicles taking into account the travel patterns of vehicle owners. Background Art

[0002] In recent years, electric vehicles have become increasingly popular in the field of modern transportation due to their green and pollution-free advantages. The global electric vehicle market has shown exponential growth, with sales exceeding 10 million units in 2022 and a penetration rate of 14%.

[0003] However, with the rapid growth of electric vehicle ownership, their electricity consumption and electricity load have also increased, which has a far-reaching impact on the power generation side. The potential impact of disorderly charging of electric vehicles on the power generation side includes increasing the electricity demand of the whole society and exacerbating the peak-to-valley difference of the power grid load. In terms of electricity consumption, one of the main driving forces for the rapid growth of electricity demand is the increase in the number of electric vehicles. In terms of power load, disorderly charging of electric vehicles will also increase the peak load of the entire network and aggravate the peak-to-valley difference of the load. Affected by residential electricity consumption and tertiary industry electricity consumption, the peak-to-valley difference of the national power grid is constantly expanding. Although the impact of electric vehicle charging on the power grid is not significant at present, if electric vehicles are charged disorderly during peak electricity consumption in the future, it may further aggravate the imbalance of power supply and demand. Therefore, it is necessary to dispatch electric vehicles to charge in an orderly manner to alleviate the peak-to-valley difference of the power generation side load, which is beneficial to the long-term development of electric vehicles. Summary of the invention

[0004] In order to solve the above problems, the present invention proposes a simulation and scheduling optimization method for the spatiotemporal distribution of electric vehicle charging demand taking into account the travel patterns of vehicle owners.

[0005] The simulation and scheduling optimization method of the spatiotemporal distribution of electric vehicle charging demand considering the travel patterns of vehicle owners is characterized by comprising the following steps:

[0006] S1: Establish a traffic network model for the planning area, calculate the section travel time according to the congestion situation in the traffic network model and set it as the section weight;

[0007] S2: Based on the traffic network model, considering the daily travel and charging demand characteristics of electric vehicles in the region, the Monte Carlo method is used to obtain the spatiotemporal distribution model of electric vehicle charging demand;

[0008] S3: Considering the demand response characteristics of electric vehicles, an optimization objective function is established based on time-of-use electricity price information to reduce the total charging cost of electric vehicles;

[0009] S4: Considering the randomness of electric vehicle owners’ charging, time constraints and scheduling scale constraints are added to the scheduling optimization model, and the established scheduling optimization model is solved by the particle swarm algorithm with penalty terms to obtain an orderly charging scheduling optimization plan for electric vehicles that minimizes charging costs and reduces the peak load of the power grid.

[0010] As a preferred embodiment, in S1, a directed traffic network diagram is established based on a graph theory method as follows:

[0011] G={N,E,W} (1)

[0012] Among them, N = {i = 1, ..., n} is a set of nodes, indicating that there are a total of n In the established model, charging stations are only built on the nodes of the road network, E = {<i,j> |i∈N,j∈N,i≠j} is a set of edges, representing road segments in the road network, W={w ij |<i,j> ∈E} is a set of road section weights, which represents the travel time of the road section between connected nodes. When the nodes are not connected, the road section weight is infinite.

[0013] Preferably, in S1, the driving time of the road is directly affected by the actual traffic conditions, and the road section weight calculated based on the traffic conditions is calculated as follows:

[0014]

[0015] In the formula, w ij , l ij Represents nodes in the road network i To Node j The driving time section weight and driving distance, ξ is a traffic information indicator, which represents different levels of traffic congestion, and the value range is [0,1]. When ξ=0, it means that the road traffic is normal. At this time, the driving time per kilometer of the electric vehicle is t z , when ξ is closer to 1, the driving time per kilometer of the electric vehicle on the road section is longer, and when ξ=1, the driving time approaches infinity, the weight of the road section between unconnected nodes is set to the infinite value inf, and the distance from the node to itself is 0.

[0016] As a preferred method, in S1, the Dijkstra algorithm based on greedy and breadth-first search is widely used in the shortest path search. The algorithm is used to plan the travel path with the shortest driving time for electric vehicles. The Dijkstra algorithm is used to search for the path with the smallest road section weight between all nodes in the road network, and the travel time and travel distance of the planned path between each node are obtained as follows:

[0017] T={T(i,j)|i,j=1,...,n} (3)

[0018] D={D(i,j)|i,j=1,...,n} (4)

[0019] T(i,j) and D(i,j) are any two nodes in the road network obtained after solving by Dijkstra algorithm. i , j The time and distance of the driving path with the smallest total weight.

[0020] Preferably, in S2, the Monte Carlo method based on randomness and probability is used to simulate the spatiotemporal distribution of charging demand in the planning area. A large number of electric vehicle travel records are obtained by random sampling based on the travel characteristics of electric vehicle owners in the planning area, and the spatiotemporal distribution of charging demand in the area is obtained through statistics. The activities of residents in the same area are roughly periodic, and the cycle is one day. Therefore, the charging demand in the area for 24 hours is simulated with one hour as the unit time.

[0021] As a preferred embodiment, in S2, the travel record of the electric vehicle includes the departure time. The departure time of the owner has a probability distribution characteristic and presents a bimodal trend in the morning and evening. Based on the probability distribution characteristics of the departure time in 24 hours, the hourly departure time t is randomly selected for the g-th electric vehicle. g .

[0022] As a preferred embodiment, in S2, the travel record of the electric vehicle includes the starting point and the end point of the trip. In order to accurately simulate the travel behavior of the electric vehicle, it is necessary to obtain the probability distribution characteristics of the starting point and the end point of the electric vehicle travel in the area. Define event A s The starting point of the trip is node s, event A s The probability of occurrence is as follows:

[0023]

[0024] Where U ij Represents a slave node i Departure to the node j The number of trip records, defining event A e The end point of the trip is node e. The end point of the trip is usually determined after the starting point is determined. After the starting point of the trip is determined to be s, the end point of the electric vehicle trip is node s according to the conditional probability formula. e The probability is:

[0025]

[0026] Among them, P(A e A s ) represents event A s and event A e The probability of simultaneous occurrence is calculated as follows:

[0027]

[0028] Based on the probability distribution feature P(A s ) and P(A e |A s ), are the starting and ending points of the trip randomly selected by the electric vehicle.

[0029] Preferably, in S2, the travel record of the electric vehicle includes an initial charge state, the initial charge state SOC of the electric vehicle ini The mean is μ and the variance is σ 2 The normal distribution N(μ,σ 2 ), its probability density function is as follows:

[0030]

[0031] In addition, overcharging and over-discharging of electric vehicles will cause imbalance in the chemical reaction inside the battery, aggravate battery aging, and lead to a reduction in battery capacity and life. In order to extend the service life of electric vehicle batteries and take into account the safety of electric vehicle battery operation, the charge state of electric vehicles should be within a reasonable range, as shown below:

[0032] SOC min ≤SOC g ≤SOC max ,g=1,2,...,G (9)

[0033] Among them, G is the total number of electric vehicles in the region, SOC g Indicates the charge state of the g-th electric vehicle. SOC min and SOC max Indicates the minimum and maximum values ​​for the battery's healthy state of charge.

[0034] As a preferred embodiment, in S2, it is assumed that the power consumption per kilometer of the electric vehicle is The battery capacity is Q. Based on the travel record of the electric vehicle, it is calculated to determine whether the remaining power of the electric vehicle meets the power consumption required for driving, as shown below:

[0035]

[0036] Where D(i,j) is the distance traveled by the g-th electric vehicle from the starting point i to the end point j following the planned path. is the initial charge state of the g-th electric vehicle at the starting point. The charging demand of the g-th electric vehicle is determined and calculated according to h(g), as shown below:

[0037]

[0038] If h(g)≥0, it means that the power required by the g-th electric vehicle to travel on the path is higher than the remaining power of the electric vehicle within the battery constraint range. At this time, it is considered that the g-th electric vehicle has a charging demand, and the charging demand q(g) is the maximum SOC required for the electric vehicle to be charged. max To meet the required electricity demand, only electric vehicles with charging needs are dispatched, and it is assumed that they can all be charged within a unit time of one hour.

[0039] As a preferred embodiment, in S2, by summing up the travel demands of all electric vehicles in the area, the spatiotemporal distribution demand of disorderly charging of electric vehicles in the planning area can be obtained as follows:

[0040]

[0041] In the formula, the set represents all electric vehicles with charging needs that depart from node i at time t. The charging needs of all electric vehicles in the set are summed to obtain the charging demand distributed at node i in the tth hour. It represents the total demand for disorderly charging of all electric vehicles at node i at time t, and q(k) represents the charging demand of the kth electric vehicle at node i.

[0042] As a preferred embodiment, in S3, the price-based demand response mechanism is used to optimize the charging demand of each node respectively, and the electric vehicles on the scheduling node are charged in an orderly manner. The objective function of the charging scheduling based on price response is to minimize the total charging cost of all electric vehicles distributed on the node:

[0043]

[0044] In the formula, i represents the i-th node in the road network, K i is the set of electric vehicles with charging demand distributed at node i, P i is the total charging cost of electric vehicles at node i. q(k) is the charging demand of the kth electric vehicle, t(t k ) is the electric vehicle at the tth k The electricity price per hour of charging.

[0045] As a preferred embodiment, in S4, it is assumed that the car owner reserves a φ time period to accept a reasonable charging schedule, and the constraint conditions are constructed as follows:

[0046]

[0047] In the formula, x k ,t kThey represent the ordered charging time and the disordered charging time of the kth electric vehicle respectively. The inequality constraint indicates that the charging time of the electric vehicle after being scheduled should not differ from the original charging time by more than the time expected by the owner.

[0048] As a preferred embodiment, in S4, in order to avoid the orderly charging method dispatching too many electric vehicles to cause a new load peak, the constraint conditions are constructed as follows:

[0049]

[0050] Among them, |K i | represents the total number of electric vehicles with charging demand at the i-th node, θ is a proportion factor, and the formula stipulates that the number of electric vehicles scheduled at the node cannot exceed the prescribed proportion of the total number of electric vehicles at the node, ρ k The dispatch flag is a bool variable indicating whether the kth electric car is dispatched. The details are as follows:

[0051]

[0052] If there is a deviation between the ordered charging time and the disordered charging time, it means that the electric vehicle is dispatched. k =1, if there is no time deviation, it means that the electric vehicle is not dispatched, then ρ k =0.

[0053] Preferably, in S4, the particle swarm algorithm is used to solve the established charging scheduling model. i When the charging scheduling problem of nodes is considered, the particles are set to |K i | dimensions, and one dimension represents a vehicle in the i The scheduling charging time of electric vehicles at each node. In order to solve the established problem while ignoring the negligible violation of the constraints by the excellent solution, the inequality constraints are constructed as penalty terms and added to the objective function, thus transforming the problem into the following unconstrained problem:

[0054]

[0055] Wherein, the first term is the value of the objective function constructed with the goal of minimizing the charging cost, the last two terms are penalty terms constructed by the time and dispatch scale inequality constraints, and λ1 and λ2 are penalty factors that determine the importance of the degree of constraint violation.

[0056] Preferably, in S4, the particle swarm algorithm specifically includes the following process:

[0057] S4.1: First, set the relevant parameters of the algorithm, randomly initialize the position and velocity of each particle in the population, and initialize the individual historical optimum and global historical optimum of the particle;

[0058] S4.2: Update the position and velocity of each dimension of the particle with reference to the individual historical optimum and the global historical optimum, as shown in the following formula:

[0059] v k (t+1)=ω·v k (t)+c1·rand·(pBes k (t)-x k (t))+c2·rand·(Leader k (t)-x k (t)) (18)

[0060] x k (t+1)=x k (t)+v k (t+1) (19)

[0061] In the formula, v k (t), x k (t) are the particle t The flight direction and position of the k-th dimension, pBes k (t), Leader k (t) are the individual historical optimal and global optimal positions of the kth dimension of the tth generation, ω, c1, c2 are the inertia weight, individual learning factor and global learning factor, rand is a random number, x k (t) represents the charging time of the kth electric vehicle after being scheduled at the current optimization node. Therefore, it is rounded after the position of formula (19) is updated and must be constrained to be within the range of [1,24];

[0062] S4.3: Calculate the fitness value of each particle using formula (17) as the optimization target. If the fitness value of the particle is better than the individual historical optimal pBes k (t) fitness value, then update the historical optimal pBes of the next generation individual k The position and fitness value of (t+1);

[0063] S4.4: If the current optimal fitness value of the population is better than the global historical optimal Leader k (t)’s fitness value, then update the next generation global historical optimal Leader k The position and fitness value of (t+1);

[0064] S4.5: When the number of population evolutions reaches the set upper limit, the algorithm iteration is completed and the current global historical optimal position is output, that is, the electric vehicle orderly charging scheduling optimization plan with the lowest total charging cost on the optimization node. If the number of population evolutions does not reach the upper limit, return to execute S4.2.

[0065] The beneficial effects of the present invention are:

[0066] The present invention fully considers the temporal and spatial distribution of charging demand due to traffic conditions and the travel patterns of car owners, as well as the satisfaction of car owners and the randomness of charging, and plans an orderly charging scheduling plan with the lowest charging cost for electric vehicles, thereby reducing the load peak while meeting the car owners' demand for reducing charging costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] Figure 1 The present invention is a flowchart of an embodiment of the present invention.

[0068] Figure 2 The traffic network diagram of the embodiment includes road congestion information.

[0069] Figure 3 This is a probability display diagram of road nodes in the experiment.

[0070] Figure 4 This is a graph showing the probability of departure time in the experiment.

[0071] Figure 5 This is a diagram showing the time-of-use electricity price in the experiment.

[0072] Figure 6 This is the heat map of the spatiotemporal distribution of the initial charging demand in the experiment.

[0073] Figure 7 This is a demand display diagram before and after the electric vehicle scheduling of node 1 in the experiment.

[0074] Figure 8 This is a diagram showing the electricity price before and after the electric vehicle dispatch at node 1 in the experiment.

[0075] Fig. 9 This is the heat map of the spatiotemporal distribution of charging demand after scheduling in the experiment. DETAILED DESCRIPTION

[0076] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0077] like Figure 1-9 As shown, the constructed traffic network diagram containing 36 nodes is used as an electric vehicle charging simulation and scheduling scenario to simulate the charging behavior of multiple groups of electric vehicles. Combined with the accompanying drawings, the present invention is further described in detail.

[0078] By combining the charging characteristics and driving characteristics of electric vehicles, a simulation and scheduling optimization method for the spatiotemporal distribution of electric vehicle charging demand that takes into account the travel patterns of car owners is implemented. First, a model is built for the electric vehicle charging scheduling problem, and a traffic network and electric vehicle charging demand model is established. Secondly, an optimization objective function is established with the minimization of the total charging cost of electric vehicles as the optimization goal, and time and scheduling scale constraints are established. Finally, the particle swarm optimization algorithm with a penalty term is used to solve the problem model, and the solved scheduling plan can reduce both charging costs and load peaks. For the specific workflow, please refer to Figure 1 The simulation and scheduling optimization method of the spatiotemporal distribution of electric vehicle charging demand considering the travel patterns of vehicle owners in this embodiment specifically includes the following steps:

[0079] (1) Considering the properties of the traffic network, a traffic network model of the planning area is established based on graph theory, and the weights of roads are set in the traffic network model based on congested road conditions;

[0080] (2) Based on the traffic network model, the daily travel characteristics of electric vehicles in the region are considered to simulate the travel records of electric vehicles and establish a spatiotemporal distribution model of charging demand;

[0081] (3) Considering the demand response characteristics of electric vehicles, an optimization objective function is established with the goal of minimizing the total charging cost of electric vehicles;

[0082] (4) Considering the randomness of electric vehicle owners' charging, time and scheduling scale constraints are added to the model, and a particle swarm algorithm with penalty terms is used to solve the constrained scheduling model to plan an electric vehicle charging scheduling plan that minimizes charging costs and reduces load peaks.

[0083] Specifically, in step (1), a directed traffic network graph is established based on graph theory as follows:

[0084] G={N,E,W} (1)

[0085] Among them, N = {i = 1, ..., n} is a set of nodes, indicating that there are a total of n In the established model, charging stations are only built at nodes of the road network. N = {i = 1, ..., n} E = {<i,j> |i∈N,j∈N,i≠j} is a set of edges, representing road segments in the road network. ij |<i,j> ∈E} is a set of road segment weights, which represents the travel time of the road segment between connected nodes. When the nodes are not connected, the road segment weight is infinite. The travel time of the road is directly affected by the actual traffic conditions. The calculation method of the road segment weight based on the traffic conditions is as follows:

[0086]

[0087] In the formula, w ij , l ij Represents nodes in the road network i To Node j The driving time section weight and driving distance. ξ is a traffic information indicator, which represents different levels of traffic congestion, and its value range is [0,1]. When ξ=0, it means that the road traffic is normal, and the driving time per kilometer of the electric vehicle is t z . The closer ξ is to 1, the longer the driving time per kilometer of the electric vehicle on the road section, and when ξ=1, the driving time approaches infinity. The weight of the road section between unconnected nodes is set to the infinite value inf, and the distance from the node to itself is 0. The Dijkstra algorithm based on greedy and breadth-first search is widely used in shortest path search. This algorithm is used to plan the travel path with the shortest driving time for electric vehicles. The Dijkstra algorithm is used to search for the path with the smallest road section weight between all nodes in the road network, and the driving time and driving distance of the planned path between each node are obtained as follows:

[0088] T={T(i,j)|i,j=1,...,n} (3)

[0089] D={D(i,j)|i,j=1,...,n} (4)

[0090] Specifically, in step (2), the Monte Carlo method based on randomness and probability is used to simulate the spatiotemporal distribution of charging demand in the planning area. A large number of electric vehicle travel records are obtained by random sampling based on the travel characteristics of electric vehicle owners in the planning area, and the spatiotemporal distribution of charging demand in the area is obtained by statistics. The activities of residents in the same area are roughly periodic, and the cycle is one day. Therefore, the charging demand in the area for 24 hours is simulated with one hour as the unit time.

[0091] The travel records of electric vehicles include the departure time. The departure time of the owner has a probability distribution characteristic, and shows a bimodal trend in the morning and evening. Based on the probability distribution characteristics of the departure time in 24 hours, g Electric vehicles are randomly selected to depart at the hourly time t g .

[0092] The travel records of electric vehicles include the starting point and the end point of the trip. In order to accurately simulate the travel behavior of electric vehicles, it is necessary to obtain the probability distribution characteristics of the starting point and the end point of electric vehicle travel in the region. Define event A s The starting point of the trip is node s. Event A s The probability of occurrence is as follows:

[0093]

[0094] Where Uij Represents the number of trip records from node i to node j. Define event A e The destination of the trip is node e. The destination of the trip is usually determined after the starting point is determined. After the starting point of the trip is determined to be s, the probability that the destination of the electric vehicle is node e according to the conditional probability formula is:

[0095]

[0096] Among them, P(A e A s ) represents event A s and event A e The probability of simultaneous occurrence is calculated as follows:

[0097]

[0098] Based on the probability distribution feature P(A s ) and P(A e |A s ), are the starting and ending points of the trip randomly selected by the electric vehicle.

[0099] The trip record of an electric vehicle includes the initial state of charge. The initial state of charge SOC of an electric vehicle ini The mean is μ and the variance is σ 2 The normal distribution N(μ,σ 2 ), and its probability density function is as follows:

[0100]

[0101] In addition, overcharging and over-discharging of electric vehicles will cause imbalance in the chemical reaction inside the battery, aggravate battery aging, and lead to a reduction in battery capacity and life. In order to extend the service life of electric vehicle batteries and take into account the safety of electric vehicle battery operation, the charge state of electric vehicles should be within a reasonable range, as shown below:

[0102] SOC min ≤SOC g ≤SOC max ,g=1,2,...,G (9)

[0103] Where G is the total number of electric vehicles in the region. g Indicates the charge state of the g-th electric vehicle. SOC min and SOC max Indicates the minimum and maximum values ​​for the battery's healthy state of charge.

[0104] Assume that the electric vehicle consumes The battery capacity is Q. Based on the travel record of the electric vehicle, it is calculated to determine whether the remaining power of the electric vehicle meets the power consumption required for driving, as shown below:

[0105]

[0106] Where D(i,j) is the distance traveled by the g-th electric vehicle from the starting point i to the end point j following the planned path. is the initial charge state of the g-th electric vehicle at the starting point. The charging demand of the g-th electric vehicle is determined and calculated according to h(g), as shown below:

[0107]

[0108] If h(g)≥0, it means that the power required by the g-th electric vehicle to travel on the path is higher than the remaining power of the electric vehicle within the battery constraint range. At this time, it is considered that the g-th electric vehicle has a charging demand, and the charging demand q(g) is the maximum SOC required for the electric vehicle to be charged. max Only electric vehicles with charging needs are dispatched, and it is assumed that they can be charged within one hour.

[0109] By summing up the travel demands of all electric vehicles in the region, the spatiotemporal distribution demand for disorderly charging of electric vehicles in the planning area can be obtained as follows:

[0110]

[0111] In the formula, the set represents all electric vehicles with charging needs that depart from node i at time t. The charging needs of all electric vehicles in the set are summed to obtain the charging demand distributed at node i at the tth hour.

[0112] Specifically, in step (3), the price-based demand response mechanism is used to optimize the charging demand of each node and schedule the electric vehicles on the node to charge in an orderly manner. The objective function of the price-response-based charging scheduling is to minimize the total charging cost of all electric vehicles distributed on the node:

[0113]

[0114] In the formula, i represents the i-th node in the road network, K i is the set of electric vehicles with charging demand distributed at node i, P i is the total charging cost of electric vehicles at node i. q(k) is the charging demand of the kth electric vehicle, t(t k ) is the electric vehicle at the tth k The electricity price per hour of charging.

[0115] Specifically, in step (4), it is assumed that the car owner reserves a time period of φ to accept a reasonable charging schedule, and the constraints are constructed as follows:

[0116]

[0117] In the formula, x k ,t k They represent the ordered charging time and the disordered charging time of the kth electric vehicle respectively. This inequality constraint indicates that the charging time of the electric vehicle after being scheduled should not differ from the original charging time by more than the time expected by the owner.

[0118] In order to avoid the orderly charging method dispatching too many electric vehicles and causing new load peaks, the constraints are constructed as follows:

[0119]

[0120] Among them, |K i | represents the total number of electric vehicles with charging demand at the i-th node, and θ is a ratio factor. The formula stipulates that the number of electric vehicles scheduled at a node cannot exceed the specified ratio of the total number of electric vehicles at the node. ρ k The dispatch flag is a bool variable indicating whether the kth electric car is dispatched. The details are as follows:

[0121]

[0122] If there is a deviation between the ordered charging time and the disordered charging time, it means that the electric vehicle is dispatched. k =1. If there is no time deviation, it means that the electric vehicle has not been dispatched. k =0.

[0123] The particle swarm algorithm is used to solve the established charging scheduling model. When solving the charging scheduling problem of the i-th node, the particles are set to |K i | dimensions, and one dimension represents the scheduled charging time of an electric vehicle at the i-th node. In order to solve the established problem while ignoring the negligible violation of the constraints by the excellent solution, the inequality constraints are constructed as penalty terms and added to the objective function, thus transforming the problem into the following unconstrained problem.

[0124]

[0125] In the formula, the first term is the value of the objective function constructed with the goal of minimizing the charging cost, and the last two terms are penalty terms constructed by the time and dispatch scale inequality constraints. λ1 and λ2 are penalty factors that determine the importance of the degree of constraint violation.

[0126] The particle swarm algorithm specifically includes the following processes:

[0127] S4.1: First, set the relevant parameters of the algorithm, randomly initialize the position and velocity of each particle in the population, and initialize the individual historical optimum and global historical optimum of the particle.

[0128] S4.2: Update the position and velocity of each dimension of the particle with reference to the individual historical optimum and the global historical optimum, as shown in the following formula:

[0129] v k (t+1)=ω·v k (t)+c1·rand·(pBes k (t)-x k (t))+c2·rand·(Leader k (t)-x k (t)) (18)

[0130] x k (t+1)=x k (t)+v k (t+1) (19)

[0131] In the formula, v k (t), x k (t) are the flight direction and position of the particle in the tth generation and kth dimension, pBes k (t), Leader k (t) are the individual historical optimal and global optimal positions of the kth dimension of the tth generation. ω, c1, c2 are the inertia weight, individual learning factor, and global learning factor, respectively. rand is a random number. x k (t) represents the charging time of the kth electric vehicle after being scheduled at the current optimization node. Therefore, it is rounded after the position of formula (19) is updated and must be constrained to be in the range of [1,24].

[0132] S4.3: Calculate the fitness value of each particle using formula (17) as the optimization target. If the fitness value of the particle is better than the individual historical optimal pBes k (t) fitness value, then update the historical optimal pBes of the next generation individual k The position and fitness value of (t+1).

[0133] S4.4: If the current optimal fitness value of the population is better than the global historical optimal Leader k (t)’s fitness value, then update the next generation global historical optimal Leader k The position and fitness value of (t+1).

[0134] S4.5: When the number of population evolutions reaches the set upper limit, the algorithm iteration is completed and the current global historical optimal position is output, that is, the orderly charging scheduling optimization plan for electric vehicles with the lowest total charging cost on the optimization node. If the number of population evolutions does not reach the upper limit, it returns to execute S4.2. The following is a further description of the effect of the present invention in combination with the experiment and the accompanying drawings of the experimental results:

[0135] In this embodiment, the car dispatching scenario uses a road network consisting of 36 nodes and 59 roads as the planning area. The specific topological structure of the road network is as follows: Figure 2 As shown in Table 1, the area is about 600 square kilometers. The road traffic status is divided into three types: normal traffic, moderate congestion and severe congestion. The corresponding traffic congestion status index ξ is shown in Table 1. Figure 2 In the figure, they are represented by light solid line, light dashed line and dark solid line respectively. When ξ=0, the driving time per kilometer of the road is t z =0.0125h / km.

[0136] Table 1

[0137] Congestion status normal Moderately crowded Severe congestion ξ ξ=0 ξ=0.5 ξ=0.8

[0138] Based on historical travel records, the probability of the starting point and the corresponding end point is obtained. Figure 3 The bar graph shows the probability distribution of the starting point of the electric vehicle, and the line graph shows the probability distribution of the destination corresponding to the starting point being 1. The departure time of the electric vehicle follows Figure 4 The time probability distribution of the display. The relevant parameters of EV and PSO are shown in Table 2. Figure 5 Shows time-of-use electricity prices.

[0139] Table 2

[0140]

[0141] According to the probability of the starting point and end point of the trip, the probability of the departure time, and the probability of the initial charge state distribution, the Monte Carlo method is used to simulate 20,000 electric vehicle driving records and determine the charging demand. Taking the g=9757th trip record as an example, the departure time of the electric vehicle is 8 am, starting from node 4 to node 9. The shortest driving path is calculated by the Dijsktra algorithm, and its driving distance is 28.27km, requiring 7.0675kWh of electricity to complete the trip. The initial charge state of the electric vehicle is The remaining safe available power is 5.1169 kWh. According to the formula, h(g) is greater than 0, and the charging demand is q(g) = 18.8831 kWh, which is the charging demand generated by node 4 at 8 o'clock. For the travel record without charging demand, take the g = 1 travel record as an example. The electric car starts from node 17 to node 19, the driving distance is 9.66 km, and the consumption is 2.415 kWh. The initial charge state of the electric car is The remaining available power is 4.3916 kWh, which meets the power required for the trip, so there is no need for charging. Among the 20,000 simulated travel records, there are 5,217 travel records with charging needs. All travel records with charging needs are counted to obtain the disordered spatiotemporal distribution of charging needs in the planning area, such as Figure 6 shown. Figure 6 The chart shows the disordered charging demand at different times at each node in the road network. The darker the color, the higher the demand load, and the lighter the color, the lower the demand load. The peak value of disordered charging demand is 818.9143 kWh. As can be seen from the figure, affected by the travel time pattern of electric vehicles, the peak values ​​of disordered charging demand are concentrated at 8:00 and 18:00, and the charging demand at other times is relatively low, showing a double peak feature overall.

[0142] Then, based on the initial charging demand, each node in the road network is charged and scheduled. The particle swarm algorithm randomly generates the initial population based on the probability of departure time, and iterates and evolves according to the formula to minimize the charging cost, and finally obtains the orderly charging scheduling plan for the node. Figure 7 The time distribution of charging demand before and after the scheduling of node 1 is shown. The peak of charging demand before scheduling occurs at 8 o'clock, and the overall fluctuation range is large. The peak load of node 1 before scheduling is calculated to be 410.284kWh, and the load standard deviation that measures the fluctuation range is 111.8946kWh. Figure 8 The figure shows the charging cost of electric vehicles in node 1. In the figure, the charging cost of a large number of electric vehicles has dropped significantly. These electric vehicles with reduced costs were initially charged mainly at the 8th, 9th, and 18th hours. The charging demand in these time periods is large and the electricity price is high. The scheduling optimization scheme reduces the charging cost while staying close to the original charging plan. For example, electric vehicles that were originally charged at 18 hours and 19 hours are scheduled to charge at 17 hours or 20 hours when the electricity price is lower. Electric vehicles that were originally charged at the 8th and 9th hours are scheduled to charge at the 6th and 10th hours when the electricity price is the lowest. At the same time, Figure 7As shown, the charging demand is transferred from the 8th, 9th and 18th hours to the 6th, 10th and 20th hours, reducing the charging demand during peak hours and ensuring more stable operation of the power system. In addition, for some electric vehicles whose charging costs remain unchanged, the new scheduling plan is consistent with the original charging plan. This is because it is assumed that not all car owners will charge according to the scheduling plan, and they will stick to their original plans. In the end, the standard deviation of the load after node 1 scheduling was 77.8713, a decrease of 30.41%, and the peak demand was 326.7722 kWh, a decrease of 20.35%. Combining all node scheduling plans, the spatiotemporal distribution of charging demand in the area after scheduling is as follows Fig. 9 shown. Fig. 9 This is the heat map of the time-space distribution of charging demand after scheduling. It can be seen that compared with Figure 6 The heat map of disordered charging demand shows that the color distribution of the heat map after scheduling is more uniform and uniform, and the fluctuation range of the spatiotemporal distribution of charging demand is smaller. As shown in Table 3, the total charging cost in the region after scheduling decreased by 19.61% compared with that before scheduling, the peak charging demand decreased by 12.94%, and the load standard deviation decreased by 7.06%. Therefore, the proposed orderly charging scheduling method for electric vehicles can reduce the charging cost of electric vehicles and improve the satisfaction of electric vehicle owners in the region under the premise of meeting the distribution law of charging time of electric vehicles in the region. At the same time, it can reduce the peak load of charging demand in the region, which is conducive to more stable operation of the power system.

[0143] Table 3

[0144]

Claims

1. A simulation and scheduling optimization method for the spatiotemporal distribution of electric vehicle charging demand considering the travel patterns of vehicle owners, characterized by: The steps include: S1: Establish a traffic network model for the planning area, calculate the section travel time according to the congestion situation in the traffic network model and set it as the section weight; S2: Based on the traffic network model, considering the daily travel and charging demand characteristics of electric vehicles in the region, the Monte Carlo method is used to obtain the spatiotemporal distribution model of electric vehicle charging demand; S3: Considering the demand response characteristics of electric vehicles, an optimization objective function is established based on time-of-use electricity price information to reduce the total charging cost of electric vehicles; S4: Considering the randomness of electric vehicle owners’ charging, time constraints and scheduling scale constraints are added to the scheduling optimization model, and the established scheduling optimization model is solved by using the particle swarm algorithm with penalty terms to obtain an orderly charging scheduling optimization plan for electric vehicles that minimizes charging costs while reducing the peak load of the power grid; In S1, the directed traffic network diagram is established based on the graph theory method as follows: G={N,E,W} (1) Among them, N = {i = 1, ..., n} is a set of nodes, indicating that there are n intersections in the road network. In the established model, charging stations are only built on the nodes of the road network. E={<i,j>|i∈N,j∈N,i≠j} is a set of edges, representing road segments in the road network, and W={w ij |<i,j>∈E} is a set of road segment weights, which represents the travel time of the road segment between connected nodes. When the nodes are not connected, the road segment weight is infinite; In S1, the driving time of the road is directly affected by the actual traffic conditions. The weight of the road section calculated based on the traffic conditions is calculated as follows: In the formula, w ij , l ij They represent the travel time, section weight and travel distance from node i to node j in the road network. ξ is a traffic information indicator that represents different levels of traffic congestion. Its value range is [0,1]. When ξ=0, it means that the road traffic is normal. At this time, the driving time per kilometer of the electric vehicle is t z , when ξ is closer to 1, the driving time of electric vehicles per kilometer on the road section is longer, and when ξ=1, the driving time approaches infinity, the weight of the road section between unconnected nodes is set to the infinite value inf, and the distance from the node to itself is 0; In S1, the Dijkstra algorithm based on greedy and breadth-first search is widely used in the shortest path search. This algorithm is used to plan the travel path with the shortest driving time for electric vehicles. The Dijkstra algorithm is used to search for the path with the smallest road section weight between all nodes in the road network, and the travel time and travel distance of the planned path between each node are obtained as follows: T={T(i,j)|i,j=1,...,n} (3) D={D(i,j)|i,j=1,...,n} (4) T(i,j) and D(i,j) are respectively the time and distance of the driving path with the minimum total weight between any two nodes i and j in the road network after solving by Dijkstra algorithm; In S2, the Monte Carlo method based on randomness and probability is used to simulate the spatiotemporal distribution of charging demand in the planning area. A large number of electric vehicle travel records are obtained by random sampling based on the travel characteristics of electric vehicle owners in the planning area, and the spatiotemporal distribution of charging demand in the area is obtained through statistics. The activities of residents in the same area are periodic, and the cycle is one day. Therefore, the charging demand in the area for 24 hours is simulated with one hour as the unit time; In S2, the travel records of electric vehicles include the departure time. The departure time of the owner has a probability distribution characteristic and presents a bimodal trend in the morning and evening. Based on the probability distribution characteristics of the departure time in 24 hours, the hourly departure time t is randomly selected for the g-th electric vehicle. g ; In S2, the travel records of electric vehicles include the starting point and the end point of the trip. In order to accurately simulate the travel behavior of electric vehicles, it is necessary to obtain the probability distribution characteristics of the starting point and the end point of the electric vehicle travel in the region. Define event A s The starting point of the trip is node s, event A s The probability of occurrence is as follows: Where U ij represents the number of trip records from node i to node j, and defines event A e The destination of the trip is node e. The destination of the trip is usually determined after the starting point is determined. After the starting point of the trip is determined to be s, the probability that the destination of the electric vehicle is node e according to the conditional probability formula is: Among them, P(A e A s ) represents event A s and event A e The probability of simultaneous occurrence is calculated as follows: Based on the probability distribution feature P(A s ) and P(A e |A s ), are the starting and ending points of the trip randomly selected by the electric vehicle.

2. The method for simulating and optimizing the time-space distribution of electric vehicle charging demand considering the travel patterns of vehicle owners according to claim 1 is characterized by: In S2, the travel record of the electric vehicle includes the initial charge state, the initial charge state SOC of the electric vehicle ini The mean is μ and the variance is σ 2 The normal distribution N(μ,σ 2 ), its probability density function is as follows: In addition, overcharging and over-discharging of electric vehicles will cause imbalance in the chemical reaction inside the battery, aggravate battery aging, and lead to a reduction in battery capacity and life. In order to extend the service life of electric vehicle batteries and take into account the safety of electric vehicle battery operation, the charge state of electric vehicles should be within a reasonable range, as shown below: SOC min ≤SOC g ≤SOC max ,g=1,2,...,G (9) Among them, G is the total number of electric vehicles in the region, SOC g Indicates the state of charge of the g-th electric vehicle, SOC min and SOC max Indicates the minimum and maximum values ​​for the battery's healthy state of charge.

3. The method for simulating and optimizing the time-space distribution of electric vehicle charging demand considering the travel patterns of vehicle owners according to claim 1 is characterized by: In S2, it is assumed that the power consumption per kilometer of electric vehicles is The battery capacity is Q. Based on the travel record of the electric vehicle, it is calculated to determine whether the remaining power of the electric vehicle meets the power consumption required for driving, as shown below: Where D(i,j) is the distance traveled by the g-th electric vehicle from the starting point i to the end point j following the planned path. is the initial charge state of the g-th electric vehicle at the starting point. The charging demand of the g-th electric vehicle is determined and calculated according to h(g), as shown below: If h(g)≥0, it means that the power required by the g-th electric vehicle to travel on the path is higher than the remaining power of the electric vehicle within the battery constraint range. At this time, it is considered that the g-th electric vehicle has a charging demand, and the charging demand q(g) is the maximum SOC required for the electric vehicle to be charged. max To meet the required electricity demand, only electric vehicles with charging needs are dispatched, and it is assumed that they can all be charged within a unit time of one hour.

4. The method for simulating and optimizing the time-space distribution of electric vehicle charging demand considering the travel patterns of vehicle owners according to claim 1 is characterized by: In S2, by summing up the travel demands of all electric vehicles in the area, the spatiotemporal distribution demand for disorderly charging of electric vehicles in the planning area can be obtained as follows: In the formula, the set represents all electric vehicles with charging needs that depart from node i at time t. The charging needs of all electric vehicles in the set are summed to obtain the charging demand distributed at node i in the tth hour. It represents the total demand for disorderly charging of all electric vehicles at node i at time t, and q(k) represents the charging demand of the kth electric vehicle at node i.

5. The method for simulating and optimizing the time-space distribution of electric vehicle charging demand considering the travel patterns of vehicle owners according to claim 1 is characterized by: In S3, the price-based demand response mechanism is used to optimize the charging demand of each node separately, and the electric vehicles on the scheduling nodes are charged in an orderly manner. The objective function of the charging scheduling based on price response is to minimize the total charging cost of all electric vehicles distributed on the nodes: In the formula, i represents the i-th node in the road network, K i is the set of electric vehicles with charging demand distributed at node i, P i is the total charging cost of electric vehicles at node i, q(k) is the charging demand of the kth electric vehicle, τ(t k ) is the electric vehicle at the tth k The electricity price per hour of charging.

6. The method for simulating and optimizing the time-space distribution of electric vehicle charging demand considering the travel patterns of vehicle owners according to claim 1 is characterized by: In S4, it is assumed that the car owner reserves a φ time period to accept a reasonable charging schedule, and the constraints are constructed as follows: In the formula, x k ,t k They represent the ordered charging time and the disordered charging time of the kth electric vehicle respectively. The inequality constraint indicates that the charging time of the electric vehicle after being scheduled should not differ from the original charging time by more than the time expected by the owner.

7. The method for simulating and optimizing the time-space distribution of electric vehicle charging demand considering the travel patterns of vehicle owners according to claim 1 is characterized by: In S4, in order to avoid the orderly charging method dispatching too many electric vehicles to cause a new load peak, the constraints are constructed as follows: Among them, |K i | represents the total number of electric vehicles with charging demand at the i-th node, θ is a proportion factor, and the formula stipulates that the number of electric vehicles scheduled at the node cannot exceed the prescribed proportion of the total number of electric vehicles at the node, ρ k The dispatch flag is a bool variable indicating whether the kth electric car is dispatched. The details are as follows: If there is a deviation between the ordered charging time and the disordered charging time, it means that the electric vehicle is dispatched. k =1, if there is no time deviation, it means that the electric vehicle is not dispatched, then ρ k =0.

8. The method for simulating and optimizing the time-space distribution of electric vehicle charging demand considering the travel patterns of vehicle owners according to claim 1 is characterized by: In S4, the particle swarm algorithm is used to solve the established charging scheduling model. When solving the charging scheduling problem of the i-th node, the particles are set to |K i | dimensions, and one dimension represents the scheduled charging time of an electric vehicle at the i-th node. In order to solve the established problem while ignoring the negligible violation of the constraints by the excellent solution, the inequality constraints are constructed as penalty terms and added to the objective function, thereby transforming the problem into the following unconstrained problem, Wherein, the first term is the value of the objective function constructed with the goal of minimizing the charging cost, the last two terms are penalty terms constructed by the time and dispatch scale inequality constraints, and λ1 and λ2 are penalty factors that determine the importance of the degree of constraint violation.

9. The method for simulating and optimizing the time-space distribution of electric vehicle charging demand considering the travel patterns of vehicle owners according to claim 1, characterized in that: In S4, the particle swarm algorithm specifically includes the following processes: S4.1: First, set the relevant parameters of the algorithm, randomly initialize the position and velocity of each particle in the population, and initialize the individual historical optimum and global historical optimum of the particle; S4.2: Update the position and velocity of each dimension of the particle with reference to the individual historical optimum and the global historical optimum, as shown in the following formula: v k (t+1)=ω·v k (t)+c1·rand·(pBes k (t)-x k (t))+c2·rand·(Leader k (t)-x k (t))(18) x k (t+1)=x k (t)+v k (t+1) (19) In the formula, v k (t), x k (t) are the flight direction and position of the particle in the tth generation and kth dimension, pBes k (t), Leader k (t) are the individual historical optimal and global optimal positions of the kth dimension of the tth generation, ω, c1, c2 are the inertia weight, individual learning factor and global learning factor, rand is a random number, x k (t) represents the charging time of the kth electric vehicle after being scheduled at the current optimization node. Therefore, it is rounded after the position of formula (19) is updated and must be constrained to be within the range of [1,24]; S4.3: Calculate the fitness value of each particle using formula (17) as the optimization target. If the fitness value of the particle is better than the individual historical optimal pBes k (t) fitness value, then update the historical optimal pBes of the next generation individual k The position and fitness value of (t+1); S4.4: If the current optimal fitness value of the population is better than the global historical optimal Leader k (t)’s fitness value, then update the next generation global historical optimal Leader k The position and fitness value of (t+1); S4.5: When the number of population evolutions reaches the set upper limit, the algorithm iteration is completed and the current global historical optimal position is output, that is, the orderly charging scheduling optimization plan for electric vehicles with the lowest total charging cost on the optimization node. If the number of population evolutions does not reach the upper limit, it returns to execute S4.2.