An electric vehicle user incentive method, computer storage medium and program product for new energy power transmission

Abstract

The present invention discloses an electric vehicle user incentive method for new energy power transmission, formalizes the electric vehicle scheduling problem, takes maximizing the total remaining power of all electric vehicles as the goal, ensures that electric vehicles arrive at the destination before the deadline under energy constraints, and obtains an electric vehicle path allocation plan based on deep reinforcement learning; formalizes the social optimal selection problem of electric vehicle users, and obtains a social optimal electric vehicle user incentive method based on greedy thinking. The present invention solves the problem that large-scale new energy power directly connected to the power grid will have a significant impact on the safe and stable operation and economic benefits of the power grid system. The present invention shows advantages in maximizing social utility and the completion rate of new energy power transmission tasks.

Description

Technical Field

[0001] The present invention belongs to the technical field of new energy power transmission, and specifically relates to an electric vehicle user incentive method based on deep reinforcement learning for new energy power transmission, a computer storage medium and a program product. Background Art

[0002] The development of new energy power helps promote environmental protection and energy conservation and emission reduction. As the cleanest energy, wind power and photovoltaics have been developed and applied on a large scale around the world. The Global Wind Energy Council (GWEC) released a report titled "Global Wind Power Report (2022)", which shows that from 2011 to 2021, the global cumulative installed capacity of wind power reached 837GW, with a compound annual growth rate of 13.40%. The International Energy Agency (IEA) released the "Renewable Energy Analysis and Forecast 2022 to 2027" report, which predicts that between 2022 and 2027, the global installed capacity of renewable energy will increase by nearly 2,400GW.

[0003] Due to the deployment of large-scale renewable energy collection equipment, such as wind turbines or solar panels, renewable energy power stations are usually deployed in suburban areas far away from cities. In addition, renewable energy power generation has significant intermittent, volatile and random characteristics. Therefore, large-scale grid connection of renewable energy not only affects the safe and stable operation of urban power grid systems, but also significantly affects the economic benefits of urban power grid system operation. Summary of the invention

[0004] In view of the above problems, the present invention proposes an electric vehicle user incentive method for renewable energy power transmission, which aims to improve the social utility, task completion rate and reward-cost ratio of electric vehicles in renewable energy power transmission.

[0005] The technical solution of the present invention is as follows: A method for motivating electric vehicle users for new energy power transmission, comprising the following steps:

[0006] Step 1: The platform publishes the power transmission task information of the new energy power station;

[0007] Step 2: The platform collects information on electric vehicle users who sign up to participate in the transmission of new energy electricity;

[0008] Step 3: Construct a social optimization electric vehicle user selection problem with the goal of maximizing social utility, and use an electric vehicle user selection incentive algorithm to obtain a set of electric vehicle user winners and the rewards paid by the platform to the winners;

[0009] The electric vehicle user selection incentive algorithm includes a winner selection stage and a reward determination stage;

[0010] The winner selection stage includes: for each new energy power station, for each unselected electric vehicle user, using a deep reinforcement learning algorithm to solve a first electric vehicle sub-path from the starting point of the electric vehicle user to the current new energy power station and satisfying energy constraints and deadline constraints; using a deep reinforcement learning algorithm to solve a second electric vehicle sub-path from the current new energy power station to the designated destination and satisfying energy constraints and deadline constraints; the first electric vehicle sub-path and the second electric vehicle sub-path are integrated into an electric vehicle path from the starting point of the electric vehicle to the designated destination; the electric vehicle user with the lowest driving energy consumption among the electric vehicle paths of all unselected electric vehicle users is returned as the winner;

[0011] The reward determination phase includes: for each winner, executing the winner selection phase in the electric vehicle user set that does not include the winner to obtain a new winner user set, and then calculating the maximum reward of the current electric vehicle user that will not be replaced by the electric vehicle users in the new winner user set.

[0012] Furthermore, in the winner selection stage, after solving the first electric vehicle sub-path, if the available capacity of the current electric vehicle user when arriving at the current new energy power station is not greater than the amount of electricity to be delivered by the current new energy power station, the electric vehicle path calculation of the next unselected electric vehicle user is started; if the available capacity of the current electric vehicle user when arriving at the current new energy power station is greater than the amount of electricity to be delivered by the current new energy power station, the second electric vehicle sub-path is solved.

[0013] Furthermore, the following constraints are satisfied when executing step 3:

[0014]

[0015]

[0016]

[0017]

[0018]

[0019]

[0020]

[0021] Among them, w is the winner set of electric vehicle users; b e represents the bid price of the electric vehicle user e for the delivered unit electricity; Assemble for new energy power station; is a binary variable, indicating whether electric vehicle user e transmits electricity from new energy power station f. It means that electric car user e is supplying electricity from new energy power station f, otherwise D f represents the electric energy stored in the renewable energy power station f; q e represents the path of the electric vehicle; |a| represents the length of section a; γ e is the energy consumption per unit distance of electric vehicle user e; t e represents the deadline of electric vehicle user e; the electric vehicle of electric vehicle user e passes through the road section a∈q e The remaining energy after is the battery capacity of the electric vehicle user e; It indicates the remaining power of user e's electric car after passing through the new energy power station f.

[0022] Further,

[0023]

[0024] in, represents the initial power of electric vehicle user e; represents the electric vehicle path q e The subpath before the segment a, Indicates the path Length.

[0025] Furthermore, when solving the first electric vehicle sub-path and the second electric vehicle sub-path, the electric vehicle is used as an intelligent agent to formalize the Markov decision process. The elements of the Markov decision process include:

[0026] For each time step t, the state vector is defined as: a e represents the road section where the electric car of electric car user e is located, represents the remaining power of electric vehicle user e, represents the remaining deadline of electric vehicle user e,

[0027]

[0028] Among them, the binary variable y e Indicates whether the electric vehicle of electric vehicle user e is discharged at the designated destination L. If so, y e =1; otherwise y e =0; α r Represents the charging power of new energy power stations for electric vehicles; α d represents the discharge power of the electric vehicle at the specified destination L;

[0029] For each time step t, the action vector is defined as: Binary variables Indicates whether the electric car of user e has passed through section a. If so, otherwise

[0030] The reward is defined as: r e (S e (t),A e (t))=-γ e |a|.

[0031] Furthermore, the maximum reward is in The sum of the costs of each electric vehicle user in the new winner user set, is the sum of the costs of each electric vehicle user in the winner user set, b e is the bid price of the current winner user for the transmission unit of electricity, D f Represents the electric energy stored in the renewable energy power station f.

[0032] Further,

[0033] in, is a binary variable, indicating whether electric vehicle user e transmits electricity from new energy power station f. It means that electric car user e is supplying electricity from new energy power station f, otherwise γ e is the energy consumption per unit distance of electric vehicle user e, |q e | represents the electric vehicle path q e Length, Represents the electric vehicle users in the new set of winner users.

[0034] Another technical solution of the present invention is a computer storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the aforementioned electric vehicle user incentive method for new energy power transmission is implemented.

[0035] Another technical solution of the present invention is a program product, including a computer program or a computer executable instruction, which, when executed by a processor, implements the aforementioned electric vehicle user incentive method for new energy power transmission.

[0036] Compared with the prior art, the present invention has the following beneficial effects:

[0037] The present invention is aimed at the new energy power transmission scenario based on electric vehicles, formalizes the social optimization electric vehicle user selection problem, calls the electric vehicle user selection incentive algorithm, and obtains the electric vehicle user winner set and their rewards, which are improved in terms of social utility, task completion rate and reward-cost ratio compared with the existing technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 The present invention is a flowchart of the method.

[0039] Figure 2 This is an example diagram of a traffic network for a simulation experiment to verify the performance of the method of the present invention.

[0040] Figure 3 This is a performance comparison diagram of the social utility of the method of the present invention in a small-scale experiment as the number of electric vehicle users changes.

[0041] Figure 4 This is a performance comparison diagram of the task completion rate of the method of the present invention in a small-scale experiment as the number of electric vehicle users changes.

[0042] Figure 5 This is a performance comparison diagram of the reward-to-cost ratio of the method of the present invention in a small-scale experiment as the number of electric vehicle users changes.

[0043] Figure 6 This is an empirical cumulative distribution function analysis diagram of the social utility of the method of the present invention in a large-scale experiment.

[0044] Figure 7 This is an analysis diagram showing the utility of the winner 9 in a large-scale experiment of the method of the present invention as the price changes.

[0045] Figure 8 This is an analysis diagram showing the utility of loser 47 in a large-scale experiment using the method of the present invention as the price changes.

[0046] Fig. 9 This is a performance comparison diagram of the social utility of the method of the present invention in a large-scale experiment as the number of electric vehicle users changes.

[0047] Fig.10 This is a performance comparison diagram of the task completion rate of the method of the present invention in a large-scale experiment as the number of electric vehicle users changes.

[0048] Fig.11 This is a performance comparison diagram of the social utility of the method of the present invention in a large-scale experiment as the number of renewable energy power stations changes.

[0049] Fig.12 This is a performance comparison chart of the task completion rate of the method of the present invention in a large-scale experiment as the number of renewable energy power stations changes. DETAILED DESCRIPTION

[0050] In order to more clearly illustrate the technical solution of the present invention, the present invention will be further described below; obviously, the following description is only a part of the embodiments, and for ordinary technicians in this field, without paying creative work, the technical solution of the present invention can also be applied to other similar scenarios based on these; in order to more clearly illustrate the technical solution of the present invention, the technical solution of the present invention is further described in detail below with reference to the accompanying drawings.

[0051] The present invention provides an electric vehicle user incentive method based on deep reinforcement learning for renewable energy power transmission, which can dispatch a socially optimized set of electric vehicle users to maximize social utility. Consider a group of renewable energy power stations located in the suburbs, and a city charging station connected to the city's main power grid and capable of receiving electricity from the renewable energy power station. The city charging station is set as the designated destination for electric vehicles to transmit renewable energy power. Consider a group of electric vehicle users who travel from their respective starting points to the designated destination. Electric vehicles transmit electricity from renewable energy power stations to the main power grid through vehicle-to-grid technology. The transportation network is formalized as a completely wireless graph. in, represents a node set consisting of traffic intersections, electric vehicle starting points, new energy power stations, and charging stations in the city; ε represents the connection The edge set between any two points in . Consider a new energy power transmission application platform and a community of electric vehicle users. The application platform considers a set of F new energy power stations with power transmission needs. The limited amount of electricity delivered by each renewable energy power station is recorded as The set of electric vehicle users interested in power delivery is denoted as The application platform A community of electric vehicle users who publish power delivery tasks.

[0052] Every electric car user Submit a quote to the application platform e , which is the expected reward for electric vehicle user e for delivering a unit of electric energy. e It represents the real cost of electric vehicle user e delivering a unit of electricity. e Indicates the starting point of electric vehicle user e. e Indicates the cut-off time for electric vehicle user e. Represents the battery capacity of the electric vehicle of electric vehicle user e. Represents the initial charge of the electric vehicle of electric vehicle user e. γ e It represents the energy consumption per unit distance of electric vehicle of electric vehicle user e. represents the driving speed of the electric vehicle of electric vehicle user e on any road section a∈ε. Electric vehicle user e submits a six-tuple to the application platform

[0053] Application Platform Select an Electric Vehicle User Winner Collection The winner will be informed of the reward. The electric vehicle user delivers all the electricity from the new energy power station to the designated destination, i.e. the charging station L in the city. The reward received by each electric vehicle user e is recorded as p e The utility of an electric car user e is the difference between the reward received and the actual cost:

[0054]

[0055] in, Winners collection for electric vehicle users; is a binary variable, indicating whether the electric vehicle user e transmits the electricity of the new energy power station f. Indicates that electric vehicle user e delivers electricity to new energy power station f, otherwise

[0056] Obviously, the loser’s utility is 0. This is because the loser’s electricity transmission cost is 0 and the reward is 0.

[0057] Please note that e May be different from c e This is because electric vehicle users are selfish, and the real cost is personal privacy information, which is unknown to other users and application platforms. Electric vehicle users may take certain strategic behaviors to maximize their own utility.

[0058] Since the renewable energy power stations are deployed in the suburbs and are relatively scattered, any electric vehicle can visit at most one renewable energy power station during its driving. e Indicates a line from the starting point s e To the specified destination L satisfies t e Constrained path, user e's electric car passes through the road segment a∈q e The remaining energy after It can be calculated by the following formula:

[0059]

[0060] in, represents the electric vehicle path q e The subpath before the segment a, Indicates the path The length of |a| represents the length of section a; a(f) represents the section whose end point is the new energy power station f; It indicates the remaining power of user e’s electric car after passing through the new energy power station f.

[0061] If and only if the available capacity of the battery of user e's electric vehicle is not less than the amount of electricity waiting to be delivered by the new energy power station f, user e's electric vehicle visits the power station, that is, Among them, a(f) represents the road section whose end point is the new energy power station f; It indicates the remaining power of user e’s electric car after passing through the new energy power station f.

[0062] Benefits of application platform Among them, λ s represents the unit price of electricity sold by the main grid. The utility u0 of the application platform is the difference between the revenue and the payment, that is,

[0063] Consider incentives Output Electric Vehicle User Winners Collection and the rewards paid to the winners When the incentive method is true, it can be seen that Social utility u:

[0064]

[0065] because is a known number published by the platform, so is a constant. The goal of the application platform is to maximize social utility and deliver electricity from renewable energy power stations to the main grid while meeting energy constraints and deadline constraints. This goal is equivalent to minimizing social costs. The above problem is named: social optimization electric vehicle user selection problem.

[0066] The social optimization electric vehicle user selection problem is an NP-hard problem: Consider a special case of the social optimization electric vehicle user selection problem, where each user's electric vehicle does not visit any renewable energy power station, that is, remove constraints (1-b), (1-d) and (1-e). Further, assume that each user's electric vehicle has enough power, that is, remove constraint (1-c). At this time, the social optimization electric vehicle user selection problem is simplified to: solving a shortest path from the starting point to the specified destination that satisfies the deadline constraint. This problem is essentially a restricted shortest path problem that is known to be NP-hard (R. Hassin, Approximation schemes for the restricted shortest path problem, Math. Oper. Res., vol. 17, no. 1, pp. 36–42, Feb. 1992.). Therefore, the social optimization electric vehicle user selection problem is NP-hard.

[0067] The essence of the social optimization electric vehicle user selection problem is to solve the scheduling path of a group of electric vehicles, maximize the total remaining energy, and ensure that the selected electric vehicles can reach the destination before the deadline, and their remaining power is always non-negative during the driving process. In the scheduling path, each electric vehicle passes through a new energy power station and then reaches the destination. Therefore, the method of the present invention calculates a shortest path from the starting point of the electric vehicle to the new energy power station, and then from the new energy power station to the designated destination. Then, the above two shortest paths are integrated into one path.

[0068] The electric vehicle scheduling process is formalized as a Markov decision process. The electric vehicle is defined as an intelligent agent, and the key elements of the Markov decision process are defined as follows:

[0069] (1) Status: The status reflects the road section a where the electric car of user e is located. e , remaining power and remaining deadline The remaining deadline for the current segment Among them, the binary variable y e Indicates whether the electric vehicle of user e is discharged at the designated destination L. If discharged, y e =1; otherwise y e =0. r Represents the charging power of new energy power stations for electric vehicles. d represents the discharge power of the electric vehicle at the specified destination L. Therefore, the state vector of each time step t∈[1,T] is defined as:

[0070] (2) Action: At each time step t, the agent needs to select a road segment. Binary variables Indicates whether the electric car of user e has passed through section a. If so, otherwise Then, the action vector for each time step t∈[1,T] is defined as:

[0071] (3) Reward: Reward is the feedback after the agent performs a certain action. The method of the present invention defines the driving cost as the reward of the agent at time step t, which can be calculated by the following formula:

[0072] r e (S e (t),A e (t))=-γ e |a|

[0073] The workflow of the method of the present invention is as follows Figure 1 shown.

[0074] Step 1: The application platform publishes the power transmission task information of the new energy power station.

[0075] The application platform releases F renewable energy power stations with power transmission needs, denoted as The amount of electricity that each new energy power station needs to deliver is limited, recorded as

[0076] Step 2: The application platform collects information on electric vehicle users who sign up to participate in the transmission of new energy electricity.

[0077] The set of electric vehicle users interested in power delivery is denoted as Application platform The electric vehicle user community publishes the power delivery task. Submit a quote to the application platform e , which is the expected reward for electric vehicle user e for delivering a unit of electric energy. e It represents the real cost of electric vehicle user e delivering a unit of electricity. e Indicates the starting point of electric vehicle user e. e Indicates the cut-off time for electric vehicle user e. Indicates the battery capacity of the user's electric car. Represents the initial power of user e's electric car. γ e It indicates the energy consumption per unit distance of user's electric vehicle. represents the driving speed of the electric car of user e on any road section a∈ε. Electric car user e submits a six-tuple to the application platform

[0078] Step 3: Construct a social optimization electric vehicle user selection problem with the goal of maximizing social utility, and use the electric vehicle user selection incentive algorithm to obtain the electric vehicle user winner set and the rewards paid by the platform to the winners. Specifically, establish constraints (1-a) to ensure that each electric vehicle user can reach the specified destination L before the deadline; establish constraints (1-b) to ensure that each user The electric vehicle visits at most one renewable energy power station during driving; establish constraints (1-c) to ensure that each user The remaining power of the electric vehicle during driving is always non-negative; the constraint (1-d) is established to ensure that the electric vehicle of user e visits the renewable energy power station if and only if the available capacity of the battery of the electric vehicle of user e is not less than the power waiting to be delivered by the renewable energy power station f; the constraint (1-e) is established to ensure that all the power of the renewable energy power station is delivered to the main power grid. The formalized problem constructed is as follows:

[0079]

[0080]

[0081]

[0082]

[0083]

[0084]

[0085]

[0086] 3.1 Initialize the electric vehicle user set Electric vehicle user collection for the application platform Electric Vehicle User Winners Roundup Initialized to empty, the reward set for electric car user winners Initialized to empty, for For each renewable energy power station, execute step 3.2;

[0087] 3.2 For each renewable energy power station Initialize an empty electric vehicle user set W and execute step 3.3;

[0088] 3.3 For each electric vehicle user Execute steps 3.4 to 3.21 to calculate a line from the starting point s e Passing through the renewable energy power station f to reach the designated destination L, satisfying the energy constraints and deadline t e The full path of the constraint q e ;

[0089] 3.4 Section a e The end point is initialized to the starting point s of the electric car of user e e , d e Initialized as a new energy power station f, Initialized to the initial power of user e's electric car Initialized as the deadline t of user e’s electric car e , execute steps 3.6 to 3.20 to calculate a line from the starting point s e To meet the energy constraints and deadline t of the renewable energy power station f e Constrained shortest path Assigned to

[0090] 3.5 If Then section a e The end point is initialized to the new energy power station f, d e Initialize to the specified destination L; Initialized as the remaining power of user e's electric car after passing through the new energy power station f Initialize to Execute steps 3.6 to 3.20 to calculate a route from the renewable energy power station f to the specified destination L that meets the energy constraints and deadlines. Constrained shortest path Assigned to

[0091] 3.6 Initialize neural network parameters θ; Initialize neural network parameters is θ; based on the neural network parameters θ and Initialize the evaluation neural network and the target neural network respectively; initialize a constant ∈ between 0 and 1; initialize an empty set M; initialize an integer Initialize the number of episodes K of deep reinforcement learning and the number of time steps T of an episode; perform steps 3.7 to 3.20 for each episode;

[0092] 3.7 Initializing an empty collection Starting state S e (1) Initialized as:

[0093] 3.8 For each time step t∈[1,T], execute steps 3.9-3.20;

[0094] 3.9 Generate a random number ζ between 0 and 1; if ζ≤∈, randomly select a segment from the set ε as the current action a; otherwise, find Q(S e (t),a e The road segment with the largest (t);θ) value is the current action a;

[0095] 3.10a Join the collection Calculate the reward value of the current action: r e (S e (t),A e (t)) is -γ e |a|;

[0096] 3.11 Calculate the remaining power of the electric vehicle for Calculate the remaining deadline for the current electric car user for

[0097] 3.12 Next state S′ e (t) is initialized to:

[0098] 3.13 If or but Assign the value to empty and terminate the algorithm;

[0099] 3.14 If the end point of the current path is a r Is the designated destination e , then terminate the algorithm;

[0100] 3.15 {S e (t),A e (t),r e (S e (t),A e (t)),S′ e (t)} join the set M;

[0101] 3.16 Randomly select m sample data from M {S e (j),A e (j),r e (S e (j),A e (j)),S′ e (j)};

[0102] 3.17 Estimate the Q value Y(t) and assign it to

[0103] 3.18 The gradient descent method is used to update the parameters θ(t) of the evaluation neural network:

[0104] 3.19 If t is The modulus is 0, then the target neural network parameter Assign value to θ(t);

[0105] 3.20 Update current status S e (t) is S′ e (t);

[0106] 3.21 If and are not empty, electric car user e is incorporated into the set W, and Integrate into a line from the starting point s e A full path q that passes through the renewable energy power station f to reach the designated destination L e ;

[0107] 3.22 Calculate the energy consumption γ of the electric vehicle for each user e∈W e |q e |, return the electric car user e′ with the lowest energy consumption and add it to the set from Remove e′ from Assign a value of 1;

[0108] 3.23 Initialize each electric vehicle user Payment e is 0;

[0109] 3.24 Calculate the cost for each user e∈W Sum and assign to

[0110] 3.25 for each electric vehicle user Execute steps 3.26 to 3.28 to calculate the reward p paid by the platform to e e ;

[0111] 3.26 Based on the set excluding electric vehicle users e Execute step 3.1 to calculate the winner set of electric vehicle users

[0112] 3.27 Calculate each user Cost Sum and assign to

[0113] 3.28 Take p e and The maximum value is assigned to p e ;p e Add to collection

[0114] 3.29 Return to Electric Vehicle User Winners Collection Electric Vehicle User Winner Rewards Collection

[0115] The electric vehicle user incentive method of the present invention is individual rational: According to the definition of electric vehicle user utility, u e =p e -c e D f Based on the reward determination rule in step 3.28, we know Because user e's offer is real, that is, b e =c e Therefore, we know

[0116]

[0117] Obviously, That is, based on the reward calculation rule in step 3.28, the unit cost of electricity transmission for the electric vehicle user set including user e is not greater than the unit cost of electricity transmission for the electric vehicle user set not including user e. e ≥0, that is, the incentive method for electric vehicle users is individually rational.

[0118] The electric vehicle user incentive method of the present invention is true: an auction mechanism is true if and only if: (1) the selection rule is monotonic: if user e bids b e Win the auction, then other conditions satisfy b′ e ≤ b e The bid of user e can also win; (2) The reward each winner receives is the key price: if the bid of user e is greater than the key price, he cannot win the auction (R.Myerson, Optimal auction design, Math.Oper.Res., vol.6, no.1, pp.58–73, Feb.1981). It can be seen that the electric vehicle user incentive algorithm is monotonic, and the reward p e is the key price. The monotonicity of the selection rule is obvious, because the electric car user e with a smaller bid cannot be pushed back in the ranking. Next, we prove that p e is the key price for electric vehicle user e, that is, the bid price is higher than p e This could prevent electric car users from winning the auction.

[0119] If electric car user e is selected and submits a false bid, Therefore, it can be known that Then, we know Because, the platform will select a set of electric vehicle users that does not include user e. Therefore, electric vehicle user e will not win the auction.

[0120] It should be noted that the specific methods of the above embodiments can form a computer program product. Therefore, the computer program product implemented in this application can be stored on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.).

[0121] A simulation experiment was conducted on the implementation method of the present invention to verify the performance of the method of the present invention when the number of electric vehicle users and the number of new energy power stations are different.

[0122] The coordinate data of the new energy power station in the simulation experiment comes from the coordinates of the wind power stations in Gusu District and Xiangcheng District of Suzhou City (https: / / map.baidu.com / ). In order to compare and analyze the performance of the method of the present invention with the benchmark algorithm, the following performance indicators are selected: (1) social utility; (2) algorithm running time; (3) task completion rate, that is, the ratio of the number of electric vehicle user winners to the number of new energy power stations; (4) reward cost ratio, that is, the ratio of the sum of the rewards of the electric vehicle user winners to the sum of their costs. The cut-off time and the number of passengers of each electric vehicle user follow the normal distribution of μ1=2.1,σ1=0.1 respectively. Figure 2 A specific example is provided, which includes 15 starting points of electric vehicle users, 5 new energy power stations and 1 designated destination, where the dots represent the starting points of electric vehicle users, the square blocks represent the new energy power stations, and the cylinder represents the designated destination L. The distance between any two nodes is calculated through Baidu Maps. The default parameter settings of the simulation experiments are shown in Table 1. All simulation experiments are run on a computer with an 11th Gen Intel(R) Core(TM) i7-1185G7@3.00GHz and 16G memory. Each measurement value is averaged over 100 instances.

[0123] Table 1 Parameter settings

[0124]

[0125]

[0126] The method of the present invention is compared with the following three benchmark algorithms:

[0127] 1. The optimal solution to the social optimization electric vehicle user selection problem. Enumerate all feasible electric vehicle user winner sets for the social optimization electric vehicle user selection problem and select the winner set with the largest social utility.

[0128] 2. Greedy time user selection algorithm. For each new energy power station, first, for each user's electric car, a path with the shortest driving time from the starting point through the new energy power station to the designated destination is solved; then, the user with the lowest social cost among all the unselected electric vehicles on the above path is selected. When the power transmission tasks of all new energy power stations are completed, the iteration ends.

[0129] 3. Greedy energy user selection algorithm. For each new energy power station, first, for each user's electric car, a shortest driving path from the starting point through the new energy power station to the designated destination is solved; then, the user with the lowest social cost among all the unselected electric vehicles on the above path is selected. When the power transmission tasks of all new energy power stations are completed, the iteration ends.

[0130] In order to compare with the optimal solution of the social optimization electric vehicle user selection problem, a small-scale simulation experiment was first conducted. Three new energy power stations and three traffic nodes were set up. The number of electric vehicle users was increased from 5 to 1, and the social utility, task completion rate, reward-cost ratio and running time were measured. Figure 3 and 4 It shows that the social utility and task completion rate of the method of the present invention are 55.39% and 52.83% of the optimal solution on average, respectively. Figure 5 It shows that the reward-cost ratio of the method of the present invention is no more than 7.8, and the reward-cost ratio of the optimal solution is no more than 10. Table 2 shows that the running time of the method of the present invention and the optimal solution increases with the increase in the number of electric vehicle users. When the number of electric vehicle users is 11, the method of the present invention can be completed in 35.63 seconds, which is much less than the running time of the optimal solution.

[0131] Table 2 The running time of the optimal solution of the method of the present invention and the social optimization electric vehicle user selection problem varies with the number of electric vehicle users

[0132]

[0133] Then, a large-scale simulation experiment was conducted to evaluate the scalability of the method of the present invention. The number of electric vehicle users increased from 20 to 100, and the number of new energy power stations increased from 10 to 20. First, the individual rationality and authenticity of the method of the present invention were verified. For individual rationality, the empirical cumulative distribution function of the utility of all electric vehicle users was analyzed. Figure 6 As shown, electric car users have no negative utility, indicating that the method of the present invention has individual rationality. Next, a winner (ID = 9) and a loser (ID = 47) are randomly selected to verify the authenticity of the method of the present invention. The above two electric car users are allowed to submit different quotes of their actual costs to the platform, such as Figure 7 and 8 As shown in Figure 1, when the winner 9 submits the true cost price b9 = 0.55, he always obtains the maximum utility. Correspondingly, when the loser 47 submits the true cost price b 47 =c 47 =1.40, he always obtains non-negative utility.

[0134] further, Fig. 9 and 10It shows that the social utility and task completion of the greedy energy user selection algorithm, the greedy time user selection algorithm and the method of the present invention first increase with the increase in the number of electric vehicle users. This is because when the number of electric vehicle users increases, all algorithms can find electric vehicle users with better social utility for the new energy power station, and more electric vehicle users can complete more power transmission tasks of the new energy power station. Then, the social utility and task completion rate of all algorithms remain stable.

[0135] The solution of the method of the present invention is always better than the solution of the user selection algorithm of greedy energy and the user selection algorithm of greedy time. On average, the social utility of the method of the present invention is increased by 514.93% and 539.06% respectively, and the task completion rate is increased by 249.62% and 244.39% respectively compared with the user selection algorithm of greedy energy and the user selection algorithm of greedy time. Therefore, the method of the present invention is significantly better than the baseline algorithm. This is because the user selection algorithm of greedy energy always selects the electric vehicle user with the largest remaining power, while ignoring the driving time. Similarly, the user selection algorithm of greedy time always selects the electric vehicle user with the smallest driving time, while ignoring the energy constraint.

[0136] Fig.11 and 12 It shows that the social utility and task completion of the greedy energy user selection algorithm, the greedy time user selection algorithm and the method of the present invention decrease first with the increase of the number of renewable energy power stations. On average, the social utility and task completion rate of the method of the present invention are increased by 467.71% and 470.92% respectively, and the task completion rate is increased by 203.38% and 194.83% respectively compared with the greedy energy user selection algorithm and the greedy time user selection algorithm.

[0137] In summary, Figure 3-12 It shows that the method of the present invention can obtain better performance in terms of social utility and task completion rate compared with the user selection algorithm based on greedy energy and the user selection algorithm based on greedy time. It shows that the method of the present invention has good scalability.

Claims

1. A method for motivating electric vehicle users for new energy power transmission, characterized in that: The following steps are involved: Step 1: The platform publishes the power transmission task information of the new energy power station; Step 2: The platform collects information on electric vehicle users who sign up to participate in the transmission of new energy electricity; Step 3: Construct a social optimization electric vehicle user selection problem with the goal of maximizing social utility, and use an electric vehicle user selection incentive algorithm to obtain a set of electric vehicle user winners and the rewards paid by the platform to the winners; The electric vehicle user selection incentive algorithm includes a winner selection stage and a reward determination stage; The winner selection stage includes: for each new energy power station, for each unselected electric vehicle user, using a deep reinforcement learning algorithm to solve a first electric vehicle sub-path from the starting point of the electric vehicle user to the current new energy power station and satisfying energy constraints and deadline constraints; using a deep reinforcement learning algorithm to solve a second electric vehicle sub-path from the current new energy power station to the designated destination and satisfying energy constraints and deadline constraints; the first electric vehicle sub-path and the second electric vehicle sub-path are integrated into an electric vehicle path from the starting point of the electric vehicle to the designated destination; the electric vehicle user with the lowest driving energy consumption among the electric vehicle paths of all unselected electric vehicle users is returned as the winner; The reward determination phase includes: for each winner, executing the winner selection phase in the electric vehicle user set that does not include the winner to obtain a new winner user set, and then calculating the maximum reward of the current electric vehicle user that will not be replaced by the electric vehicle users in the new winner user set.

2. The electric vehicle user incentive method for new energy power transmission according to claim 1 is characterized in that: In the winner selection stage, after solving the first electric vehicle sub-path, if the available capacity of the current electric vehicle user when arriving at the current new energy power station is not greater than the amount of electricity to be delivered by the current new energy power station, the electric vehicle path calculation of the next unselected electric vehicle user is started; if the available capacity of the current electric vehicle user when arriving at the current new energy power station is greater than the amount of electricity to be delivered by the current new energy power station, the second electric vehicle sub-path is solved.

3. The electric vehicle user incentive method for new energy power transmission according to claim 1 is characterized in that: The following constraints are met when executing step 3: st in, Winners collection for electric car users; b e represents the bid price of the electric vehicle user e for the delivered unit electricity; Assemble for new energy power station; is a binary variable, indicating whether electric vehicle user e transmits electricity from new energy power station f. It means that electric car user e is supplying electricity from new energy power station f, otherwise D f represents the electric energy stored in the renewable energy power station f; q e represents the path of the electric vehicle; |a| represents the length of section a; γ e is the energy consumption per unit distance of electric vehicle user e; t e represents the deadline of electric vehicle user e; the electric vehicle of electric vehicle user e passes through the road section a∈q e The remaining energy after is the battery capacity of the electric vehicle user e; It indicates the remaining power of user e's electric car after passing through the new energy power station f.

4. The electric vehicle user incentive method for new energy power transmission according to claim 3 is characterized in that: in, represents the initial power of electric vehicle user e; represents the electric vehicle path q e The subpath before the segment a, Indicates the path Length.

5. The electric vehicle user incentive method for new energy power transmission according to claim 3 is characterized in that: When solving the first electric vehicle sub-path and the second electric vehicle sub-path, the electric vehicle is used as an agent to formalize the Markov decision process. The elements of the Markov decision process include: For each time step t, the state vector is defined as: a e represents the road section where the electric car of electric car user e is located, represents the remaining power of electric vehicle user e, represents the remaining deadline of electric vehicle user e, Among them, the binary variable y e Indicates whether the electric vehicle of electric vehicle user e is discharged at the designated destination L. If so, y e =1; otherwise y e =0; α r Represents the charging power of new energy power stations for electric vehicles; α d represents the discharge power of the electric vehicle at the specified destination L; For each time step t, the action vector is defined as: Binary variables Indicates whether the electric car of user e has passed through section a. If so, otherwise The reward is defined as: r e (S e (t),A e (t))=-γ e |a|.

6. The electric vehicle user incentive method for new energy power transmission according to claim 1 is characterized in that: The maximum reward is in The sum of the costs of each electric vehicle user in the new winner user set, is the sum of the costs of each electric vehicle user in the winner user set, b e is the bid price of the current winner user for the transmission unit of electricity, D f Represents the electric energy stored in the renewable energy power station f.

7. The electric vehicle user incentive method for new energy power transmission according to claim 6 is characterized in that: in, is a binary variable, indicating whether electric vehicle user e transmits electricity from new energy power station f. It means that electric car user e is supplying electricity from new energy power station f, otherwise γ e is the energy consumption per unit distance of electric vehicle user e, |q e | represents the electric vehicle path q e Length, Represents the electric vehicle users in the new set of winner users.

8. A computer storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, the electric vehicle user incentive method for new energy power transmission described in any one of claims 1 to 7 is implemented.

9. A program product comprising a computer program or computer executable instructions, characterized in that: When the computer program or computer executable instructions are executed by a processor, the electric vehicle user incentive method for new energy power transmission described in any one of claims 1 to 7 is implemented.

Images