A method for matching shared parking supply and demand under AVP conditions

Abstract

The present invention relates to a method for matching shared parking supply and demand under AVP conditions, including: using AVP to move vehicles at the beginning and end of the parking demand time and the parking space supply time to reduce the number of vehicle moves, and dividing the shared parking supply and demand time with the above time as the breakpoint; on the basis of time segmentation, taking the minimum number of AVP vehicle moves to maximize the shared parking demand as the optimization goal, and establishing a corresponding mathematical model; considering the pairing scheme as an ant foraging trajectory, and the time period pairing as a trajectory point, combining the two strategies of random exploration and deep exploration to determine the specific pairing corresponding to each trajectory point on the ant foraging route; considering the number of vehicle moves of the pairing scheme as the length of the ant foraging trajectory, and combining the improved ant colony algorithm to solve and obtain the parking scheme. Compared with the prior art, the present invention can maximize the shared parking demand with the minimum number of AVP vehicle moves, while improving the parking space utilization rate, reducing the vehicle cost and the accident rate.

Description

Technical Field

[0001] The present invention relates to the technical field of shared parking, and in particular to a shared parking supply and demand matching method under AVP conditions. Background Art

[0002] With the current increase in the number of cars, the problem of parking difficulties is becoming more and more serious. Shared parking is a way to solve the problem of parking difficulties by using existing parking facilities. Compared with the construction of new parking facilities, it not only has low renovation costs and short cycles, but also has more significant effects. In shared parking, parking space owners earn income by renting out parking spaces during idle periods, and vehicle owners use shared parking spaces to park their vehicles. The higher the parking space utilization rate, the greater the income obtained by both parties and the less waste of resources. However, how to reasonably match these shared parking demands and supplies with time and space differences is an urgent problem to be solved in order to realize shared parking.

[0003] The Autonomous Valet Robot (AVP) achieves vehicle relocation by moving the vehicle and parking it in different parking spaces in two consecutive time periods. If it is applied to shared parking and parking is done during small hours when the parking spaces are idle, it can solve the problem of low parking space utilization and unsatisfied shared parking demand in traditional shared parking, which is caused by the fact that vehicles can only be parked in one parking space during their shared parking demand time. While maximizing the parking demand, it further improves the parking space utilization rate of shared parking. However, inappropriate AVP moving rules or unreasonable parking plans may also lead to frequent vehicle relocation, resulting in increased vehicle use costs and increased accident rates. Summary of the invention

[0004] The purpose of the present invention is to overcome the defects of the above-mentioned prior art and provide a shared parking supply and demand matching method under AVP conditions, so as to simultaneously achieve the minimum AVP car moving times and maximize the satisfaction of shared parking needs.

[0005] The purpose of the present invention can be achieved by the following technical solution: A method for matching supply and demand of shared parking under AVP conditions, comprising the following steps:

[0006] S1. In order to reduce the number of unnecessary vehicle moves during shared parking, it is stipulated that autonomous parking can only be used to move vehicles at the beginning and end of the parking demand time and the parking space supply time, and the shared parking supply and demand time is divided with the above time as the breakpoint to describe the vehicle displacement during shared parking;

[0007] S2. Based on the time segmentation, define e v，p，i is a pairing of vehicle v parked at parking space p in time period i, e 0，p，i Indicates that there is no vehicle parked in parking space p during time period i, Ei =(e1, e2, ..., e j ) is the time period pairing formed by all reasonable combinations of pairs in time period i; M = (E 1 , E 2 , ..., E n ) represents the pairing scheme composed of reasonable combinations of all time period pairings, n is the number of split time periods, and each pairing scheme has a specific number of car moves;

[0008] The optimization goal is to maximize the shared parking demand with the least number of AVP car moves, and establish a corresponding mathematical model;

[0009] S3, consider the pairing scheme as the foraging trajectory of ants, and pair the time periods E i Consider it as the i-th trajectory point of the ant, and combine the two strategies of random exploration and deep exploration to determine the specific pairings corresponding to each trajectory point on the ant foraging route;

[0010] S4. The number of car moves in the pairing scheme is regarded as the length of the ant foraging trajectory. Combined with the improved ant colony algorithm, the specific parking scheme that maximizes the shared parking demand with the minimum number of AVP car moves is solved.

[0011] Furthermore, the mathematical model established in step S2 includes an objective function and a first constraint condition, a second constraint condition, and a third constraint condition, wherein the objective function is specifically that the distance that the AVP moves the vehicle each time is the smallest and the number of times the AVP moves the vehicle is the smallest;

[0012] The first constraint condition specifically means that the parking demand in any time period must be met;

[0013] The second constraint condition is specifically a restriction on the supply of shared parking in any period of time;

[0014] The third constraint condition is specifically a range of decision variables for a vehicle to be parked in a berth during a certain period of time.

[0015] Furthermore, the specific process of step S2 is as follows:

[0016] In order to meet all shared parking needs, assuming that the parking demand at any time is less than or equal to the supply, Indicates whether there is a supply of parking space p in time period i; if there is a supply, its value is 1, otherwise it is 0;

[0017] Indicates whether vehicle v has parking demand in time period i; if there is parking demand, its value is 1, otherwise it is 0;

[0018] l p，qrepresents the AVP moving distance from berth p to berth q, W represents the cost of one shift, and the corresponding mathematical model is established as follows:

[0019] Objective function:

[0020]

[0021] The first constraint is:

[0022]

[0023] Second constraint:

[0024]

[0025] The third constraint is:

[0026]

[0027] Among them, the first sum term in the objective function makes the distance that AVP moves each vehicle the smallest, the second sum term makes the number of vehicle moves of AVP the smallest, and in the third constraint condition, when vehicle v is parked in parking space p during time period i, the decision variable Otherwise 0.

[0028] Furthermore, the random exploration strategy in step S3 specifically randomly generates a parking plan for a period of time as the trajectory point of the ant;

[0029] The deep exploration strategy specifically constructs a classic transportation problem from the pheromone matrix of the trajectory points, and selects pairing solutions with more pheromones as candidate values ​​of the trajectory points.

[0030] Furthermore, the step S3 specifically includes the following steps:

[0031] S31, set the random exploration probability γ∈(0,1) to determine the next trajectory point of the ant;

[0032] S32, generate a random number ε∈(0,1), if γ>ε, then randomly generate a time period pair E on time period i i , as the i-th trajectory point of the ant, otherwise execute step S33;

[0033] S33. Construct pheromone matrix: All shared parking demands in time period i constitute a set V i 、All open berths constitute the set P i , with the set V i and set P i The inner elements are row and column elements, forming the pheromone matrix corresponding to the ith trajectory point of the ant represents the pheromone on the pairing e(v, p, i);

[0034] S34, constructing a transportation problem based on the pheromone matrix and solving it to obtain candidate values ​​of the ant i trajectory points;

[0035] S35, solving heuristic information of candidate values ​​of trajectory points;

[0036] S36, calculating the probability corresponding to the alternative values ​​of the trajectory point;

[0037] S37. Select one of the three trajectory point candidate values ​​as the i-th trajectory point of the ant using the roulette method.

[0038] Furthermore, the specific process of constructing the transportation problem in step S34 is as follows: the parking demand in time period i is the production volume of the shipping place, the supply of berths is the demand volume of the sales place, and the pheromone matrix C i For the cost matrix, construct a classic transportation problem.

[0039] Furthermore, the specific process of solving the candidate values ​​of the ant i trajectory points in step S34 is: using the greedy algorithm, the Vogel method, and the Hungarian algorithm to solve the constructed transportation problem, and obtain the three time period pairs on time period i It is used as the candidate value of the ant's i-th trajectory point, and the corresponding pheromones are Among them, GA, VA, and HA correspond to the greedy algorithm, Vogel's method, and Hungarian algorithm, respectively.

[0040] Furthermore, the specific process of step S35 is: find Respectively with E i-1 The cost of moving between The corresponding heuristic information is constructed in this way:

[0041]

[0042]

[0043]

[0044] Furthermore, the probability calculation formula corresponding to the candidate values ​​of the trajectory points in step S36 is specifically:

[0045]

[0046] in, for As the probability of the trajectory point, α is the pheromone control parameter, β is the heuristic information control parameter, and Y is the algorithm set composed of GA, VA, and HA. A set of candidate values ​​for trajectory points;

[0047] The same can be said and are respectively used as the probability of trajectory points.

[0048] Furthermore, the step S4 specifically includes the following steps:

[0049] S41, initialization parameters: pheromone control parameter α, heuristic information control parameter β, pheromone volatility coefficient ρ, coefficient Q of trajectory length converted to pheromone concentration, number of ants m, random exploration probability γ, maximum number of iterations Max, current number of iterations is τ = 1;

[0050] S42. Randomly generate m ants:

[0051] First, randomly generate time period pairs for all time periods, and pair the obtained n time periods E i , i∈{1, 2, ..., n} as ant m a The track points on the foraging track, reasonably combine these time periods to form a pairing scheme M a , generates an ant m a ;

[0052] Then, repeat the above steps to generate m ants;

[0053] S43. Calculate the pheromone on the ant track:

[0054] First, according to the mathematical model established in step S2, the length z(x) of the ant foraging trail is calculated;

[0055] Then, using the formula:

[0056]

[0057] Pheromones were calculated from the trajectory length;

[0058] S44, pheromone update: when ants m a After matching v，p，i hour, otherwise, Update the pheromone matrix C using the following formula i , pheromone in i∈{1, 2, ..., n} Then update the pheromone matrix of all trajectory points:

[0059]

[0060] S45, generate m new ants: use the random exploration strategy and the deep exploration strategy to determine all the trajectory points of an ant, that is, all the time period pairings included in a pairing scheme E 1 , E2 , ..., E n , generate a complete foraging trajectory of an ant;

[0061] Repeat the above operation to generate new m ants;

[0062] S46, check the termination condition: if the current number of iterations τ = Max, terminate the algorithm and output the optimal result;

[0063] Otherwise, the number of iterations τ:=τ+1 and return to execute step S43.

[0064] Compared with the prior art, the present invention introduces AVP into shared parking to move vehicles, and can use the idle time of parking spaces for parking, thereby improving the utilization rate of parking spaces while maximizing the satisfaction of parking needs. In order to achieve the minimum number of vehicle moves while maximizing the satisfaction of shared parking needs, the present invention takes into account the impact of AVP on shared parking, focuses on the matching relationship between parking needs and parking spaces during shared parking, regards time period pairing as trajectory points on the foraging trajectory of ants, and selects trajectory points using two strategies, random exploration strategy and deep exploration strategy, to reduce the alternative values ​​of ant trajectory points, thereby being able to solve a specific parking plan that maximizes the satisfaction of shared parking needs with the minimum number of AVP vehicle moves.

[0065] On the basis of time segmentation, the present invention takes the minimum AVP car moving times to maximize the shared parking demand as the optimization goal, combines the two strategies of random exploration and deep exploration to determine the specific pairing corresponding to the trajectory points on the ant foraging trajectory, and proposes a targeted improved ant colony algorithm. First, the supply and demand time of shared parking is divided into smaller time periods, and AVP is used to move cars between consecutive time periods; secondly, a mathematical model with the minimum number of car moving times to maximize the shared parking demand is established as the objective function, and the number of car moving times of the pairing scheme is solved; then the pairing scheme is regarded as the ant foraging trajectory, and the time period pairing is regarded as the trajectory point, and the alternative values ​​of the trajectory point are determined by combining the two strategies; finally, the number of car moving times of the pairing scheme is regarded as the trajectory length, and a targeted improved ant colony algorithm is designed. The present invention introduces automatic parking robots into shared parking. By optimizing the number of AVP car moving times in the shared parking process, while maximizing the shared parking demand and improving the utilization rate of parking spaces, it can greatly reduce resource waste, reduce vehicle use costs, and reduce the incidence of traffic accidents. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] Figure 1 It is a schematic diagram of the method flow of the present invention;

[0067] Figure 2 It is a schematic diagram of the application process of the embodiment;

[0068] Figure 3 It is a time segmentation diagram in the embodiment. DETAILED DESCRIPTION

[0069] The present invention is described in detail below with reference to the accompanying drawings and specific embodiments.

[0070] Example

[0071] like Figure 1 As shown, a method for matching shared parking supply and demand under AVP conditions includes the following steps:

[0072] S1. In order to reduce the number of unnecessary vehicle moves during shared parking, it is stipulated that autonomous parking can only be used to move vehicles at the beginning and end of the parking demand time and the parking space supply time, and the shared parking supply and demand time is divided with the above time as the breakpoint to describe the vehicle displacement during shared parking;

[0073] S2. Based on the time segmentation, define e v，p，i is a pairing of vehicle v parked at parking space p in time period i, e 0，p，i Indicates that there is no vehicle parked in parking space p during time period i, E i =(e1, e2, ..., e j ) is the time period pairing formed by all reasonable combinations of pairs in time period i; M = (E 1 , E 2 , ..., E n ) represents the pairing scheme composed of reasonable combinations of all time period pairings, n is the number of split time periods, and each pairing scheme has a specific number of car moves;

[0074] The optimization goal is to maximize the shared parking demand with the least number of AVP car moves, and establish a corresponding mathematical model;

[0075] S3, consider the pairing scheme as the foraging trajectory of ants, and pair the time periods E i Consider it as the i-th trajectory point of the ant, and combine the two strategies of random exploration and deep exploration to determine the specific pairings corresponding to each trajectory point on the ant foraging route;

[0076] S4. The number of car moves in the pairing scheme is regarded as the length of the ant foraging trajectory. Combined with the improved ant colony algorithm, the specific parking scheme that maximizes the shared parking demand with the minimum number of AVP car moves is solved.

[0077] This technical solution takes into account the current shared parking. A vehicle can only be parked in a parking space when the parking demand time is less than or equal to the time when the parking space is open. However, with the introduction of AVP shared parking, a vehicle can be parked as long as the shared parking supply and demand times intersect, thereby improving the utilization rate of parking spaces while maximizing the satisfaction of parking needs. However, frequent AVP moves can also cause waste of resources and accidents. In order to solve the problem of maximizing the satisfaction of shared parking needs with the least number of AVP moves, the present invention is based on time period segmentation and uses the ant colony algorithm as a framework. It combines the two strategies of random exploration and deep exploration to determine the trajectory points of ants, and designs a targeted improved ant colony algorithm, such as Figure 2 As shown, the steps are as follows:

[0078] Step 1: To reduce the number of unnecessary AVP moves during shared parking, it is stipulated that vehicles can only be moved at the beginning and end of the parking demand time and the parking space supply time. The shared parking demand time and the parking space supply time are divided by the above time to describe the displacement of vehicles during shared parking.

[0079] Step 2: Based on the time segmentation, define e v，p，i is a pairing of vehicle v parked at parking space p in time period i, e 0，p，i Indicates that there is no vehicle parked in parking space p during time period i. i =(e1, e2, ..., e j ) is the time period pairing composed of all reasonable combinations of pairs in time period i. 1 , E 2 , ..., E n ) represents the pairing scheme composed of reasonable combinations of all time periods, n is the number of divided time periods, and each pairing scheme has a specific number of car moves. The optimization goal is to maximize the shared parking demand with the minimum number of AVP car moves, and establish the corresponding mathematical model.

[0080] Step 3: Consider the pairing scheme as the foraging trajectory of ants and pair the time periods E i Considered as the ith trajectory point of the ant. Combining the two strategies of random exploration and deep exploration, the specific pairings corresponding to each trajectory point on the ant foraging route are selected.

[0081] Step 4: Consider the number of car moves in the pairing scheme as the length of the ant foraging trajectory, and design a targeted improved ant colony algorithm to solve the specific parking plan that meets the shared parking demand with the least number of AVP car moves.

[0082] In step 2, in order to ensure that all parking demands can be met, it is assumed that the parking demand in any period is less than or equal to the supply. Indicates whether there is a supply of parking space p in time period i; if there is a supply, its value is 1, otherwise it is 0. Indicates whether vehicle v has parking demand in time period i; if there is parking demand, its value is 1, otherwise it is 0. p，q represents the AVP moving distance from berth p to berth q, and W represents the cost of one shift. The following model (1-4) is established to solve the number of shifts of the matching solution.

[0083]

[0084]

[0085]

[0086]

[0087] In the above models (1-4), equation (1) is the objective function, and equations (2-4) are constraints. The first sum term in the objective function (1) minimizes the distance that the AVP moves each time, and the second sum term minimizes the number of times the AVP moves. Constraint (2) indicates that the parking demand at any time period must be met. Constraint (3) indicates the shared parking supply limit at any time period. In constraint (4), when vehicle v is parked at parking space p in time period i, the decision variable Otherwise 0.

[0088] In step 3, the random exploration strategy: randomly generate a parking plan for a period of time as the trajectory point of the ant; the deep exploration strategy constructs the classic transportation problem from the pheromone matrix of the trajectory point, and selects the pairing plan with more pheromones as the candidate value of the trajectory point. Assuming the random exploration probability γ∈(0,1), the steps to determine the next trajectory point of the ant are as follows:

[0089] Step 3.1: Generate a random number ε∈(0,1). If γ>ε, then randomly generate a time period pair E on time period i i , as the ith trajectory point of the ant. Otherwise, perform the following operations.

[0090] Step 3.2: Construct the pheromone matrix. All shared parking demands in time period i constitute the set V i 、All open berths constitute the set P i . Take the set V i and the set P i The inner elements are row and column elements, which constitute the pheromone matrix corresponding to the ith trajectory point of the ant. represents the pheromone on the pairing e(v, p, i).

[0091] Step 3.3: Construct and solve the transportation problem based on the pheromone matrix. The parking demand in time period i is the production volume of the shipping place, and the supply berth is the demand volume of the sales place. The pheromone matrix Ci For the cost matrix, construct a classic transportation problem. Use the greedy algorithm (GA), Vogel's method (VA), and Hungarian algorithm (HA) to solve and get the three time period pairs on time period i It is used as the candidate value of the ant's i-th trajectory point. The corresponding pheromones are

[0092] Step 3.4: Obtain heuristic information for alternative values ​​of trajectory points. With E i-1 The cost of moving between Constructing heuristic information

[0093] Step 3.5: Calculate the probability of alternative values. Use formula (5) to calculate the choice (or ) as the probability of trajectory points.

[0094]

[0095] In formula (5), α and β are the algorithm parameters corresponding to pheromone and heuristic information. or The probability of being the ith trajectory point of the ant or The calculation formula is similar to formula (5), and the variables can be changed.

[0096] Step 3.6: Use the roulette method to select one of the three candidate values ​​as the i-th trajectory point of the ant. In step 4, the steps of the ant colony algorithm are as follows.

[0097] Step 4.1: Initialize parameters. Pheromone control parameter α, heuristic information control parameter β, pheromone volatility coefficient ρ, coefficient Q of converting trajectory length into pheromone concentration, number of ants m, random exploration probability γ, maximum number of iterations Max, and the current number of iterations is τ = 1.

[0098] Step 4.2: Randomly generate m ants. First, randomly generate time period pairs for all time periods, and pair the obtained n time period pairs E i , i∈{1, 2, ..., n} as ant m a The track points on the foraging track, reasonably combine these time periods to form a pairing scheme M a , generates an ant m a Then, repeat the above steps to generate m ants.

[0099] Step 4.3: Calculate the pheromone on the ant's trail. First, use formulas (1 to 4) to calculate the length z(x) of the ant's foraging trail. Then, use formula Pheromones were calculated from the trajectory lengths.

[0100] Step 4.4: Pheromone update. When the ant m a After matching v，p，i hour, Write, Update the pheromone matrix C using formula (6) i , pheromone in i∈{1, 2, ..., n} Then update the pheromone matrix of all trajectory points.

[0101]

[0102] Step 4.5: Generate m new ants. Use the random exploration strategy and the deep exploration strategy to determine all the trajectory points of an ant, that is, all the time period pairings included in a pairing scheme E 1 , E 2 , ..., E n , generate a complete foraging track of an ant. Repeat the above operation to generate new m ants.

[0103] Step 4.6: Check the termination condition. If the current number of iterations τ = Max, then terminate the algorithm and output the optimal result. Otherwise, the number of iterations τ: = τ + 1 and go to step 4.3.

[0104] The specific process of applying the above solution in this embodiment is as follows:

[0105] 1. To reduce unnecessary car moving during shared parking, it is stipulated that AVP car moving only occurs at the beginning and end of parking demand time and parking space supply time, respectively denoted by t v，s ,t v，e ,t p，s ,t p，e In order to describe the moving situation of shared parking during the process, the shared parking supply and demand time is divided into smaller time periods with the above time as the breakpoint.

[0106] 1.1、The start and end time of shared parking supply and demand t v，s ,t v，e ,t p，s ,t p，e Put it into set T.

[0107] 1.2. Eliminate duplicate elements in set T, arrange the above elements in ascending order, and obtain an ordered set of moments.

[0108] 1.3. Use the elements in set T as breakpoints to divide the shared parking demand time and supply time. In this embodiment, the opening hours of 8 shared parking spaces in a shared parking area are p1 (7:30-12:00, 13:00-16:30), p2 (6:00-11:30), p3 (8:30-14:15, 16:00-20:00), p4 (7:15-11:45, 13:30-18:15), p5 (9:30-16:00, 16:45-19:30), p6 (6:45-12:45, 14:45-18:15), p7 (9:00-13:45, 15:00-18:30), and p8 (8:00-15:15, 6:00-18:45). The visual display is as follows: Figure 3 shown. Figure 3 In the example, a day is divided into 24 parts with 1 hour as the time scale. The shaded part indicates that the berth does not support sharing, and the blank part indicates that the berth is open during this period. The start and end times of the shared time of these 8 berths are used as breakpoints to divide the berth opening time. After the division, berth p1 changes from the original two shared time periods to 15 time periods after the division. For simplicity, this example only considers the start and end times of the shared berth opening time.

[0109] Second, in order to ensure that all parking needs can be met, it is assumed that the parking demand in any period is less than or equal to the supply. Based on the period segmentation, e(v, p, i) represents a pairing of vehicle v parked in parking space p in period i, and e(0, p, i) represents that there is no vehicle parked in parking space p in period i. i =(e1, e2, ..., e j ) represents the time period pairing formed by all reasonable combinations of pairs in time period i. 1 , E 2 , ..., E n ) represents the pairing scheme composed of reasonable combinations of all time periods, and n is the number of divided time periods. The optimization goal is to maximize the shared parking demand with the least number of AVP car moves, and the corresponding mathematical model is established.

[0110] use Indicates whether there is a supply of parking space p in time period i; if there is a supply, its value is 1, otherwise it is 0. Indicates whether vehicle v has parking demand in time period i; if there is parking demand, its value is 1, otherwise it is 0. p，q represents the AVP moving distance from berth p to berth q, and W represents the cost of one shift. The following model is established to solve the number of shifts of the pairing solution.

[0111]

[0112]

[0113]

[0114]

[0115] The first formula in the above model is the objective function, and the second to fourth formulas are three constraints. The first sum term in the objective function minimizes the distance that AVP moves each vehicle, and the second sum term minimizes the number of vehicle moves by AVP. The first constraint indicates that the parking demand at any time period must be met. The second constraint indicates the supply limit of shared parking at any time period. In the third constraint, when vehicle v is parked at parking space p in time period i, the decision variable Otherwise 0.

[0116] 3. Consider the pairing scheme as the foraging trajectory of ants and pair them in time periods. i The random exploration and deep exploration strategies are combined to select the trajectory points on the ant foraging route, and then a targeted improved ant colony algorithm is designed to solve the parking plan that meets the shared parking demand with the least number of AVP moves.

[0117] 3.1. Initialization parameters: pheromone control parameter α, heuristic information control parameter β, pheromone volatility coefficient ρ, coefficient Q of converting trajectory length into pheromone concentration, number of ants m, random exploration probability γ, maximum number of iterations Max, and the current number of iterations is τ = 1.

[0118] 3.2. Randomly generate the initial m ants.

[0119] 3.2.1. Randomly generate an ant. Randomly generate time period pairs for all time periods, and pair the obtained n time periods E i , i∈{1, 2, ..., n} as ant m a The track points on the foraging track, reasonably combine these time periods to form a pairing scheme M a , generates an ant m a .

[0120] 3.2.2. Repeat the above steps to generate m ants and obtain the foraging trajectories of the m ants.

[0121] 3.3. Calculate the pheromones of ant trajectories.

[0122] 3.3.1. Calculate the length of the ant's foraging trail. a Number of car moves z(x(m a )) is considered as ant m a The length of the foraging trajectory is calculated using the established mathematical model.

[0123] 3.3.2. Using the formula By Ant M a The trajectory length z(x(m a ))Calculate pheromones

[0124] 3.4. Update the pheromone matrix of trajectory points.

[0125] 3.4.1. Construct the pheromone matrix corresponding to the trajectory point. When the parking demand and parking space supply in a period are huge, there are many time period pairs, and it is difficult to determine the specific pairing corresponding to each ant trajectory point. To solve this problem, construct the pheromone matrix C corresponding to the ith trajectory point of the ant i , i∈{1, 2, ..., n}.

[0126] 3.4.2. Update the pheromones in the pheromone matrix. The pheromone representing the pairing e(v, p, i); is a 0-1 variable. When ant m a When e(v, p, i) is matched, its value is 1; otherwise, it is 0. The pheromone update formula is as shown in formula (5). Use this formula to update the pheromone matrix C i .

[0127]

[0128] 3.5. Combining random exploration strategy and deep exploration strategy to select ant trajectory points. Let the random exploration probability γ∈(0,1).

[0129] 3.5.1. Generate a random number ε∈(0,1). If γ>ε, then randomly generate a time period pair E on time period i i , as the ant's i-th trajectory point. Otherwise, perform the following operations and use the deep exploration strategy to determine the ant's next trajectory point.

[0130] 3.5.2. Construct and solve the transportation problem based on the pheromone matrix to obtain the candidate values ​​of the trajectory points and the pheromone and heuristic information of each candidate value. First, the parking demand in time period i is the production volume of the shipping place, and the supply berth is the demand volume of the sales place. The pheromone matrix C i For the cost matrix, a classic transportation problem is constructed. Secondly, the greedy algorithm (GA), Vogel's method (VA), and Hungarian algorithm (HA) are used to solve the three time period pairs on time period i. It is used as the candidate value of the ant's i-th trajectory point. The corresponding pheromones are Finally, find and The cost of moving between Constructing heuristic information

[0131] 3.5.3. Calculate the probability of alternative values. Use the following calculation (or ) is the probability of the ith trajectory point.

[0132]

[0133] 3.5.4. Use the roulette wheel method to select one of the three candidate values ​​as the i-th trajectory point of the ant.

[0134] 3.6. Determine whether an ant has been generated. If n trajectory points of an ant have been generated, then reasonably combine these trajectory points to generate an ant and proceed to the next step. Otherwise, return to step 3.5.1.

[0135] 3.7. Determine whether m ants have been generated. If m ants have been generated for the current iteration, proceed to the next step. Otherwise, return to step 3.5.1.

[0136] 3.8. Check the termination condition. If the current number of iterations τ = Max, terminate the algorithm and output the optimal result. Otherwise, the number of iterations τ: = τ + 1 and go to step 3.3.

[0137] Applying this technical solution in practice can optimize the number of AVP car moves during shared parking, while maximizing the satisfaction of shared parking needs and improving parking space utilization.

Claims

1. A method for matching shared parking supply and demand under AVP conditions, characterized in that: The following steps are involved: S1. In order to reduce the number of unnecessary vehicle moves during shared parking, it is stipulated that autonomous parking can only be used to move vehicles at the beginning and end of the parking demand time and the parking space supply time, and the shared parking supply and demand time is divided with the above time as the breakpoint to describe the vehicle displacement during shared parking; S2. Based on the time segmentation, define e v,p,i is a pairing of vehicle v parked at parking space p in time period i, e 0,p,i Indicates that there is no vehicle parked in parking space p during time period i, E i =(e1,e2,…,e j ) is the time period pairing formed by all reasonable combinations of pairs in time period i; M = (E 1 ,E 2 ,…,E n ) represents the pairing scheme composed of reasonable combinations of all time period pairings, n is the number of split time periods, and each pairing scheme has a specific number of car moves; The optimization goal is to maximize the shared parking demand with the least number of AVP car moves, and establish a corresponding mathematical model; S3, consider the pairing scheme as the foraging trajectory of ants, and pair the time periods E i Consider it as the i-th trajectory point of the ant, and combine the two strategies of random exploration and deep exploration to determine the specific pairings corresponding to each trajectory point on the ant foraging route; S4. Considering the number of car moves in the pairing scheme as the length of the ant foraging trajectory, combined with the improved ant colony algorithm, a specific parking scheme is obtained that maximizes the shared parking demand with the minimum number of AVP car moves; The specific process of step S2 is: In order to meet all shared parking needs, assuming that the parking demand at any time is less than or equal to the supply, Indicates whether there is a supply of parking space p in time period i; if there is a supply, its value is 1, otherwise it is 0; Indicates whether vehicle v has parking demand in time period i; if there is parking demand, its value is 1, otherwise it is 0; l p,q represents the AVP moving distance from berth p to berth q, W represents the cost of one shift, and the corresponding mathematical model is established as follows: Objective function: The first constraint is: Second constraint: The third constraint is: Among them, the first sum term in the objective function makes the distance that AVP moves each vehicle the smallest, the second sum term makes the number of vehicle moves of AVP the smallest, and in the third constraint condition, when vehicle v is parked in parking space p during time period i, the decision variable Otherwise, 0; The random exploration strategy in step S3 specifically randomly generates a parking plan for a period of time as the trajectory point of the ant; The deep exploration strategy specifically constructs a classic transportation problem from the pheromone matrix of the trajectory points, and selects the pairing scheme with more pheromones as the candidate value of the trajectory point; Step S3 specifically includes the following steps: S31, set the random exploration probability γ∈(0,1) to determine the next trajectory point of the ant; S32. Generate a random number ε∈(0,1). If γ>ε, then randomly generate a time period pair E on time period i. i , as the i-th trajectory point of the ant, otherwise execute step S33; S33. Construct pheromone matrix: All shared parking demands in time period i constitute a set V i 、All open berths constitute the set P i , with the set V i and the set P i The inner elements are row and column elements, forming the pheromone matrix corresponding to the ith trajectory point of the ant represents the pheromone on the pairing e(v,p,i); S34, constructing a transportation problem based on the pheromone matrix and solving it to obtain candidate values ​​of the ant i trajectory points; S35, solving heuristic information of candidate values ​​of trajectory points; S36, calculating the probability corresponding to the alternative values ​​of the trajectory point; S37, using a roulette method to select one of the three trajectory point candidate values ​​as the i-th trajectory point of the ant; The specific process of constructing the transportation problem in step S34 is as follows: the parking demand in time period i is the production volume of the shipping place, the supply berth is the demand volume of the sales place, and the pheromone matrix C i For the cost matrix, construct a classic transportation problem; The specific process of solving the candidate values ​​of the ant i trajectory points in step S34 is: using the greedy algorithm, the Vogel method, and the Hungarian algorithm to solve the constructed transportation problem, and obtain the three time period pairs on time period i It is used as the candidate value of the ant's i-th trajectory point, and the corresponding pheromones are Among them, GA, VA, and HA correspond to the greedy algorithm, Vogel's method, and Hungarian algorithm, respectively.

2. A method for matching shared parking supply and demand under AVP conditions according to claim 1, characterized in that: The mathematical model established in step S2 includes an objective function and a first constraint condition, a second constraint condition, and a third constraint condition. The objective function is specifically that the distance that the AVP moves the vehicle each time is the smallest and the number of times the AVP moves the vehicle is the smallest; The first constraint condition specifically means that the parking demand in any time period must be met; The second constraint condition is specifically a restriction on the supply of shared parking in any period of time; The third constraint condition is specifically a range of decision variables for a vehicle to be parked in a berth during a certain period of time.

3. The method for matching shared parking supply and demand under AVP conditions according to claim 1, characterized in that: The specific process of step S35 is: Respectively with E i-1 The cost of moving between The corresponding heuristic information is constructed in this way: in, They are The corresponding heuristic information.

4. A method for matching shared parking supply and demand under AVP conditions according to claim 3, characterized in that: The probability calculation formula corresponding to the candidate values ​​of the trajectory points in step S36 is specifically: in, for As the probability of the trajectory point, α is the pheromone control parameter, β is the heuristic information control parameter, and Y is the algorithm set composed of GA, VA, and HA. A set of candidate values ​​for trajectory points; The same can be said and are respectively used as the probability of trajectory points.

5. A method for matching shared parking supply and demand under AVP conditions according to claim 4, characterized in that: The step S4 specifically comprises the following steps: S41, initialization parameters: pheromone control parameter α, heuristic information control parameter β, pheromone volatility coefficient ρ, coefficient Q of trajectory length converted to pheromone concentration, number of ants m, random exploration probability γ, maximum number of iterations Max, current number of iterations is τ = 1; S42. Randomly generate m ants: First, randomly generate time period pairs for all time periods, and pair the obtained n time periods E i ,i∈{1,2,…,n} as ant m a The track points on the foraging track, reasonably combine these time periods to form a pairing scheme M a , generates an ant m a ; Then, repeat the above steps to generate m ants; S43. Calculate pheromones on ant tracks: First, according to the mathematical model established in step S2, the length z(x) of the ant foraging trail is calculated; Then, using the formula: Pheromones were calculated from the trajectory length; S44, pheromone update: when ants m a After matching v,p,i hour, otherwise, Update the pheromone matrix C using the following formula i , pheromone in i∈{1,2,…,n} Then update the pheromone matrix of all trajectory points: S45, generate m new ants: use the random exploration strategy and the deep exploration strategy to determine all the trajectory points of an ant, that is, all the time period pairings included in a pairing scheme E 1 ,E 2 ,…,E n , generate a complete foraging trajectory of an ant; Repeat the above operation to generate m new ants; S46, check the termination condition: if the current number of iterations τ = Max, terminate the algorithm and output the optimal result; Otherwise, the number of iterations τ:=τ+1 and the process returns to execute step S43.

Images