A Simulation-Based Method for Optimizing Parking Strategies in Automated Valet Parking Lots

Abstract

This invention discloses a simulation-based method for optimizing parking strategies in automated valet parking lots. The method includes: setting the simulation duration, parking demand, expected vehicle arrival time interval, and average vehicle parking time for the automated valet parking lot; randomly generating vehicle arrival and departure time series based on the set parameters; constructing a test simulation model using the time series and given parking space allocation, conflict resolution, vehicle departure, and relocation strategies to simulate the operation of the automated valet parking lot and collecting evaluation parameters; constructing an evaluation system for the parking strategies of the automated valet parking lot based on the evaluation parameters using the analytic hierarchy process (AHP); calculating the utility value of each parking strategy and comparing them to determine the optimal parking strategy for the automated valet parking lot. This invention comprehensively considers the time efficiency, space efficiency, and user experience of automated valet parking lots, quantitatively evaluating their parking strategies.

Description

Technical Field

[0001] This invention discloses a simulation-based method for optimizing parking strategies in automated valet parking lots, which relates to the technology of evaluating parking facilities in urban traffic planning and management, and belongs to the technical field of calculation, estimation or counting. Background Technology

[0002] Automated valet parking refers to a technology where autonomous vehicles enter a parking lot, park themselves, and automatically leave the parking lot upon instruction from the user. Furthermore, because these autonomous vehicles are driverless and interconnected, automated valet parking lots can select different parking strategies to achieve better overall parking lot utilization efficiency.

[0003] To improve the space utilization of automated valet parking lots, existing research has proposed a layout where autonomous vehicles are parked end-to-end. However, this layout inevitably leads to some vehicles being blocked by other parked vehicles during exit, reducing the parking lot's time efficiency and user experience. To balance space utilization with time efficiency, existing research has analyzed the factors influencing spatiotemporal efficiency, finding that autonomous vehicle parking strategies and parking lot layouts have a certain impact on overall efficiency. However, some existing studies focus on comparing autonomous vehicle parking strategies under a fixed layout, while others focus on parking lot layout design under a fixed parking strategy. The two studies use different evaluation indicators for the overall parking lot efficiency, and these indicators are correlated, leading to differences in the optimal parking strategy for autonomous vehicles under different parking lot layouts. Therefore, existing research cannot provide a comprehensive and objective evaluation of the spatiotemporal efficiency of different autonomous vehicle parking strategies under different parking lot layouts.

[0004] In conclusion, while automated valet parking systems can be designed with compact layouts to improve space utilization, this can reduce parking time efficiency and user experience. Existing research provides a comprehensive and objective evaluation of different autonomous vehicle parking strategies under various layouts. Therefore, a comprehensive efficiency evaluation system for automated valet parking systems is urgently needed. Summary of the Invention

[0005] Technical issues:

[0006] The purpose of this invention is to provide a simulation-based method for optimizing parking strategies in automated valet parking lots. This method comprehensively considers the time efficiency, space efficiency, and user experience of vehicle parking in automated valet parking lots, proposing an optimal method for vehicle parking strategies. This provides important reference for the layout optimization and strategy improvement of automated valet parking lots, and solves the problem that existing research struggles to comprehensively and objectively evaluate the time and space efficiency of different autonomous vehicle parking strategies under different parking lot layouts.

[0007] Technical solution: A simulation-based method for optimizing parking strategies in automated valet parking lots, comprising the following steps:

[0008] Step 1: Initialize design parameters, which include: number of parking spaces in the automated valet parking lot, simulation duration of the automated valet parking lot, parking demand, expected vehicle arrival time interval, and average vehicle parking time.

[0009] Step 2: Based on the initial values ​​of simulation duration, parking demand, expected vehicle arrival time interval, and average vehicle parking time, generate the time series of vehicles arriving at the parking lot and the time series of vehicles leaving the parking lot.

[0010] Step 3: Based on the time sequence of vehicles arriving at the parking lot and the time sequence of vehicles leaving the parking lot, a K-stack high-density parking lot layout with unidirectional lane flow is adopted to simulate the automatic valet parking strategy.

[0011] Step 4: Collect the following evaluation parameters generated during the simulation of the parking strategy of the automated valet parking lot: average parking lot occupancy rate D1, average vehicle search time D2, average vehicle departure time delay D3, cumulative number of conflicts D4, and cumulative relocation distance D5.

[0012] Step 5: Based on the evaluation parameters generated during the simulation of the parking strategy of the automated valet parking lot, establish a hierarchical structure model based on the analytic hierarchy process to obtain the weight matrix of each evaluation parameter.

[0013] Step 6: Calculate the utility values ​​of different parking strategies obtained from the simulation, and use them as the evaluation index of the parking strategy. The smaller the evaluation index value, the better the parking strategy.

[0014] Furthermore, in step 2, the time series of vehicles arriving at the parking lot is randomly generated based on a negative exponential distribution according to the initial values ​​of simulation duration, parking demand, and expected vehicle arrival time interval.

[0015] Furthermore, in step 2, the time sequence of the vehicles leaving the parking lot is generated based on the simulation duration, parking demand, initial values ​​of average vehicle parking duration, and the time sequence of vehicles arriving at the parking lot.

[0016] Furthermore, in step 3, a K-stack high-density parking lot layout is adopted, setting vehicles to travel in one direction on the road and in two directions between parking spaces and between parking spaces and roads. The length, width, flow direction, entrance and exit positions, length-to-width ratio of parking spaces, and width of parking islands of each road are fixed to conduct dynamic simulation of the automated valet parking lot.

[0017] Furthermore, step 3 specifically involves:

[0018] Construct a Car list describing parked vehicles, wherein the Car list contains a vehicle class for each parked vehicle;

[0019] Construct a two-dimensional array Layout to describe the layout of the parking lot. The two-dimensional array Layout describes the occupancy status of each parking space and each lane in the parking island.

[0020] Iterate through the time sequence of vehicles arriving at the parking lot and the time sequence of vehicles leaving the parking lot, and determine whether the current vehicle is an arriving vehicle or a departing vehicle;

[0021] If the current vehicle is an arriving vehicle, then assign a parking space to the current vehicle and update the Layout 2D array and the time sequence of the vehicle's arrival at the parking lot;

[0022] If the current vehicle is a departing vehicle, then a departure simulation is performed on the current vehicle, and the two-dimensional array describing the Layout and the time series of the vehicle leaving the parking lot are updated. If there is an obstructing vehicle in the departure direction of the current vehicle, then the obstructing vehicle is repositioned, that is, a new parking space is assigned to the obstructing vehicle, so that the obstructing vehicle can drive to the new parking space to park, freeing up departure space for the current vehicle, and the two-dimensional array of the Layout is updated.

[0023] Five evaluation parameters representing parking lot time efficiency (B1), space efficiency (B2), and user experience (B3) were collected during the simulation of different parking space allocation strategies in automated valet parking lots. The evaluation parameters include: average parking lot occupancy rate (D1), average vehicle search time (D2), average vehicle departure time delay (D3), cumulative number of conflicts (D4), and cumulative relocation distance (D5).

[0024] Furthermore, the hierarchical model in step 5 is divided into a target layer, a criterion layer, and an indicator layer. The target layer is to select the optimal parking strategy; the criterion layer includes time efficiency B1, space efficiency B2, and user experience B3; and the indicator layer includes average parking lot occupancy rate D1, average vehicle search time D2, average vehicle departure time delay D3, cumulative number of conflicts D4, and cumulative relocation distance D5.

[0025] Furthermore, the calculation method for each evaluation parameter obtained in step 4 is as follows:

[0026] Parking lot occupancy rate D1:

[0027]

[0028]

[0029] Where: T is the simulation duration;

[0030] Occupancy t The parking lot occupancy rate at simulation second t. i ;

[0031] Number of parked vehicles t Let the number of all parked vehicles be Number_of_parked_vehicles after the simulation ends in second t. t ;

[0032] Total_berths is the total number of parking spaces in the parking lot;

[0033] Average vehicle search time D2:

[0034] Search id =park id -arrive id ,

[0035]

[0036] Where: N is the total number of parking spaces;

[0037] Search id The parking search time for the idth parking vehicle;

[0038] park id The simulation time for the idth parking vehicle to arrive at the parking space;

[0039] arrive id The simulation time for the idth parking vehicle to enter the parking lot;

[0040] Average vehicle departure time delay D3:

[0041] Delay id′ =real_deoart id′ -deoart id′ ,

[0042]

[0043] Among them: Delay id′ The departure time of vehicle id′ was delayed;

[0044] real_depart id′ This represents the actual departure time of vehicle id′.

[0045] depart id′ Let id′ be the simulation time for the departing vehicle to leave the parking space;

[0046] N′ represents the total number of departing vehicles;

[0047] Total number of conflicts D4:

[0048]

[0049] Among them: conflicts t Let t be the number of collisions that occur in simulation seconds;

[0050] Cumulative relocation distance D5:

[0051]

[0052] Where: N″ is the total number of repositioning vehicles;

[0053] reloction t,id″ Let be the relocation distance of the id″-th repositioning vehicle in simulation seconds t, that is, the distance that the id″-th repositioning vehicle travels to the new parking space in simulation seconds t.

[0054] Furthermore, in step 6, the utility value of the parking strategy is P = w1D1 + w2D2 + w3D3 + w4D4 + w5D5, where w1, w2, w3, w4, and w5 are the weights of D1, D2, D3, D4, and D5, respectively.

[0055] The present invention also provides a simulation-based automatic valet parking lot parking strategy optimization device, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps of the above-described simulation-based automatic valet parking lot parking strategy optimization method.

[0056] The present invention also provides a computer-readable storage medium having a computer program stored thereon, characterized in that, when the computer program is executed by a processor, it implements the steps of the above-described simulation-based automatic valet parking lot parking strategy optimization method.

[0057] This invention, employing the aforementioned technical solution, offers the following advantages: It comprehensively considers the time efficiency, space efficiency, and user experience of automated valet parking lots, and, based on the characteristics of automated valet parking lots, proposes new evaluation indicators and provides a method for quantitatively evaluating parking strategies in automated valet parking lots. This invention utilizes numerical simulation of parking lots to model the entire operation process of automated valet parking lots under different parking strategies, thereby obtaining evaluation index values ​​for the time efficiency, space efficiency, and user experience of automated valet parking lots. This invention is applicable to the quantitative evaluation of different parking strategies under different parking lot layouts and varying occupancy rates, providing a basis for optimizing the layout and improving the strategies of automated valet parking lots. Attached Figure Description

[0058] Figure 1 This is a flowchart of the present invention;

[0059] Figure 2 This is a layout diagram of a K-stack automated valet parking lot with a unidirectional lane flow, used in the simulation.

[0060] Figure 3 This is a schematic diagram of the A1 parking strategy;

[0061] Figure 4 This is a schematic diagram of the A2 parking strategy;

[0062] Figure 5 It is a hierarchical structure model. Detailed Implementation

[0063] The present invention will be further explained below with reference to the accompanying drawings and specific embodiments. It should be understood that these examples are only for the purposes of this invention and are not intended to limit the scope of the invention. After reading this invention, any modifications of the invention in various equivalent forms by those skilled in the art will fall within the scope defined by the appended claims.

[0064] like Figure 1 As shown, a simulation-based method for optimizing parking strategies in an automated valet parking lot includes the following steps:

[0065] Step 1: Initialize the design parameters of the automated valet parking lot. The design parameters include: number of parking spaces in the automated valet parking lot, simulation duration of the automated valet parking lot, parking demand, expected vehicle arrival time interval, and average vehicle parking time.

[0066] Step 2: Based on the initial values ​​of simulation duration, parking demand, expected vehicle arrival time interval, and average vehicle parking time, generate the time series of vehicles arriving at the parking lot and the time series of vehicles leaving the parking lot.

[0067] Step 3: Based on the vehicle arrival time series and vehicle departure time series generated in Step 2, a K-stack high-density parking lot layout with unidirectional lane flow is adopted to construct an operation simulation model of the automated valet parking lot and simulate the entire parking process of the automated valet parking lot; collect the evaluation parameters representing the parking lot time efficiency B1, space efficiency B2 and user experience B3 generated during the simulation of different parking strategies of the automated valet parking lot: average parking lot occupancy rate D1, average vehicle search time D2, average vehicle departure time delay D3, cumulative number of conflicts D4, and cumulative relocation distance D5.

[0068] Step 4: Based on the evaluation parameters generated during the simulation of the parking strategy of the automated valet parking lot, establish a hierarchical structure model using the analytic hierarchy process (AHP).

[0069] Step 5: Based on the hierarchical model established in Step 4, calculate the weight matrix of each evaluation parameter, calculate the utility value of different parking strategies obtained from the simulation, and use this as the evaluation index of the parking strategy. The smaller the evaluation index value, the better the parking strategy.

[0070] Furthermore, the method for generating the time series of vehicles arriving at the parking lot and the time series of vehicles leaving the parking lot in step 2 includes two parts:

[0071] The first part, the time series of vehicles arriving at the parking lot, is randomly generated based on a negative exponential distribution, according to the parking demand and expected arrival time interval of vehicles initialized in step 1.

[0072] Let the parking demand be Demand (unit: vehicles), and the expected arrival time interval of vehicles be t. e (Unit: seconds / vehicle), the average arrival rate of vehicles is 1 / t e (Unit: vehicles / second), the actual arrival time interval T' of vehicles follows the parameter 1 / t. e The negative exponential distribution, i.e. Generate Demand-1 random numbers t1, t2, ..., t' based on probability T'. Demand-1 If the arrival time of the first autonomous vehicle in the parking lot is set to 0, then the arrival time T of the i-th autonomous vehicle in the parking lot is... i for

[0073] The second part, the time series of vehicles leaving the parking lot, is generated based on the parking demand, average parking time of vehicles, and time series of vehicles arriving at the parking lot initialized in step 1.

[0074] Let the average parking time of vehicles be μ. e (Unit: seconds / vehicle), the average vehicle departure rate is 1 / μ e (Unit: vehicles / second), the actual parking time M of the vehicles follows a parameter of 1 / μ. e The negative exponential distribution, i.e. Generate Demand-1 random numbers μ1, μ2, ..., μ based on probability M. Demand-1 Then the time when the j-th autonomous vehicle leaves the parking lot is T. i +μ i .

[0075] Furthermore, the full-process simulation model of the automated valet parking lot in step 3 includes: a K-stack high-density parking lot layout with unidirectional lane flow, parking space allocation strategy, vehicle relocation strategy, vehicle departure strategy, and simulation process data update and collection.

[0076] (1) The basic layout of a K-stack high-density parking lot with unidirectional lane flow is a combination of the following elements: unidirectional traffic lanes, parking lot entrances and exits, parking islands, and parking stacks. A parking island refers to a rectangular area where autonomous vehicles are concentrated. Each parking island contains Y horizontally arranged parking stacks. Each parking stack can accommodate two-way parking, with a maximum of K vehicles stacked end-to-end in each direction. On the parking island, vehicles can only travel laterally in both directions and cannot travel vertically. Between adjacent parking islands and between the parking islands and the vertical edges of the parking lot, there are two traffic lanes with opposite flow directions. Simultaneously, at the parking lot entrances and exits, there are two horizontally arranged entrance / exit lanes with opposite flow directions. To simplify the parking space size, each parking space in the automated valet parking lot is 3 meters long and 3 meters wide, and the traffic lane width is 3 meters.

[0077] Based on the basic layout of a unidirectional K-stack high-density parking lot, a two-dimensional array `layout` can be constructed to describe the current occupancy status of each parking space and road in the parking lot. If there are no vehicles parked or driving in each parking space or lane, it is represented as 0; if there are vehicles parked or driving in each parking space or lane, it is represented as 1; if a parking space becomes the target parking space for a vehicle, it is represented as 2.

[0078] (2) The parking space allocation strategy of automated valet parking lots includes the following steps:

[0079] First, determine the parking space allocation strategies for several automated valet parking lots that need to be evaluated. Then, according to the determined parking space allocation strategies, sequentially check each parking island to see if there are any vacant spaces. If a vacant space exists that meets the criteria, it is allocated to the newly arriving vehicle; otherwise, entry is stopped. This process continues until all parking spaces in the parking lot are occupied.

[0080] The vehicle relocation and departure strategies include the following steps: Detecting the direction with the minimum sum of the departure distance of the departing vehicle and the relocation distance of the blocking vehicle, and using this direction as the vehicle's departure direction; relocating the blocking vehicle, treating it as a parking-seeking vehicle, and moving it to a new parking space along the shortest route; and having the vehicle to depart travel to the exit along the shortest route in the current departure direction to complete the departure.

[0081] (3) Based on the basic layout of the K-stack high-density parking lot with unidirectional lane flow and the parking space allocation strategy, vehicle relocation strategy and departure strategy of the automated valet parking lot, the whole process of automated valet parking is simulated, and the parameters representing the parking lot time efficiency, space efficiency and user experience evaluation generated during the simulation are collected: average parking lot occupancy rate D1, average vehicle search time D2, average vehicle departure time delay D3, cumulative number of conflicts D4, and cumulative relocation distance D5.

[0082] Construct the parking space allocation strategy function Parking_strategy, the vehicle entry and relocation function Function_Arrive, the vehicle departure strategy function Function_Depart, and the vehicle class describing the parked vehicle. The vehicle class contains member variables, and the functions and member variables are shown in Tables 1 and 2.

[0083] Table 1 Functions and Their Functions

[0084]

[0085] Table 2 Member variables of the vehicle class

[0086]

[0087] Where: T is the simulation duration in seconds (s); N is the total number of vehicles.

[0088] Construct a Car list describing the parked vehicles, where each Car list contains a Vehicle class for each parked vehicle. Construct a 2D layout array describing the parking lot layout.

[0089] In each simulation second, the following process is completed: traverse the Car list describing the parked vehicles, determine whether a vehicle has entered the target parking space, and if a vehicle has arrived at the parking space, change its state from "arriving" to "parking" and update the vehicle's position information in the layout.

[0090] Iterate through the Car list describing parked vehicles, determine if any vehicle is in the parking search phase, and if so, determine if there is a conflict between the parking search vehicle and the departing vehicle, and between the parking search vehicle and the parking search vehicle. If a conflict exists, according to the conflict resolution strategy, make one side stay for one round to avoid the conflict, and update the layout.

[0091] Iterate through the time sequence of vehicle arrivals at the parking lot, determine if any vehicles are entering, and if so, execute the `Function_Arrive` function to add the vehicle to the `vehicle` class, saving information such as the vehicle's sequence number (id), target parking space tuple (pos), current coordinate tuple (coordinate), speed (velocity), arrival time (arrive), stay time (stay), departure time (depart), actual departure time (real depart), route list (route), and current state (state). Update the vehicle's position information in the layout.

[0092] Iterate through the time series of vehicles leaving the parking lot, checking if any vehicle's departure time equals the current simulation time. If such a vehicle exists, change its state from "parking" to "departing" and execute the `Function_Depart` function. Iterate through all vehicles, checking if any parked vehicles are blocking the current vehicle's departure. If such a vehicle exists, treat it as a parking-seeking vehicle, change its state from "parking" to "arriving," and execute the `Function_Arrive` function. Update the vehicle position information in the layout.

[0093] Iterate through the list of cars describing parked vehicles, checking if any car is on the departure route. If a car is found, check if there is a conflict between departing cars. If a conflict exists, use the conflict resolution strategy to make one car stay for one round to avoid further conflict. Update the position information of the departing cars and update it in the layout.

[0094] At the end of the simulation process, the following evaluation parameters representing the parking lot's time efficiency B1, space efficiency B2, and user experience B3 generated during the simulation of the automated valet parking lot are collected and calculated: average parking lot occupancy rate D1, average vehicle search time D2, average vehicle departure time delay D3, cumulative number of conflicts D4, and vehicle relocation distance D5.

[0095] Specifically, the process of relocating, moving, and obtaining the number of vehicle conflicts involves the following steps:

[0096] Relocation: Assigning a new parking space to a vehicle that is blocking the exit.

[0097] Relocation: The process by which a vehicle being repositioned travels from its current parking space to its newly assigned parking space, following a specific path.

[0098] Conflict: If two vehicles reach the same position in the next simulation second, it is considered that a conflict has occurred between the two vehicles.

[0099] Specifically, different arrival rate scenarios were used to conduct multiple simulations for different strategies. The average value of each result was taken to obtain five evaluation indicators: average parking lot occupancy rate D1, average vehicle parking search time D2, average vehicle departure time delay D3, cumulative number of conflicts D4, and cumulative relocation distance D5.

[0100] Furthermore, in step 4, based on the evaluation parameters representing parking lot time efficiency B1, space efficiency B2, and user experience B3 generated during the simulation of different parking space allocation strategies in the automated valet parking lot—namely, the average parking lot occupancy rate D1, average vehicle parking search time D2, average vehicle departure time delay D3, cumulative number of conflicts D4, and cumulative relocation distance D5—a hierarchical structure model is established using the analytic hierarchy process.

[0101] The hierarchical model is divided into three layers: the target layer, the criterion layer, and the indicator layer. The target layer focuses on selecting the optimal parking strategy; the criterion layer considers time efficiency (B1), space efficiency (B2), and user experience (B3); and the indicator layer measures average parking lot occupancy (D1), average vehicle search time (D2), average vehicle departure time delay (D3), cumulative conflict count (D4), and cumulative relocation distance (D5).

[0102] Construct pairwise comparison matrices between factors at each level to assess the importance of each indicator in the parking lot. In the matrix, the importance of each factor relative to the previous level is quantified numerically, and can be represented by the following pairwise comparison matrix:

[0103]

[0104] Wherein: g i′j′ Let be the comparison scale between elements i′ and j′ at the same level; G is the pairwise comparison matrix at this level; n is the number of elements in this level, and also the order of matrix G; where g describes the magnitude of the mutual influence of factors. i′j′ The values ​​are quantified according to the table. This matrix represents the importance of each factor at this level to a factor at the previous level. For example, for the indicator layer elements D1 and D2, which has a greater influence on B1 than D2 on B1, then g12 represents its scale.

[0105] Each factor in the criteria layer has a pairwise comparison matrix with the target layer, and each factor in the indicator layer has three pairwise comparison matrices with the three factors in the criteria layer. A pairwise comparison matrix represents the importance of each factor in the next lower level to a factor or indicator in the previous level, as shown in Table 3.

[0106] Table 3 Comparison Scale

[0107]

[0108] Further, in step 5, based on the established hierarchical model, the weight matrix of each evaluation parameter is calculated to obtain the comprehensive utility evaluation formula. The utility values ​​of different parking space allocation strategies are then calculated, and the optimal parking strategy for the automated valet parking lot is determined through comparison. The calculation method is as follows:

[0109] (1) Perform hierarchical single sorting and consistency check. By performing a consistency check on matrix G, subjective inconsistencies are reduced.

[0110] Test the consistency of matrix G and calculate the consistency index CI.

[0111]

[0112] Where: CI is the consistency index; λ max is the largest eigenvalue of the matrix; n is the order of the matrix;

[0113] Find the random consistency index RI, as shown in Table 4.

[0114] Table 4 Random Consistency Index Values

[0115]

[0116] Calculate the random consistency ratio CR

[0117]

[0118] When CR < 0.10, the inconsistency of matrix G is considered to be within the acceptable range and meets the consistency requirement. Therefore, the eigenvector corresponding to its largest eigenvalue can be used for normalization and as the weight vector. Otherwise, matrix G needs to be adjusted.

[0119] (2) Hierarchical overall ranking and its consistency test

[0120] The process of calculating the relative importance ranking weights of all factors at the same level to the overall goal is called hierarchical overall ranking. This process is carried out layer by layer from the highest level to the lowest level. Criterion layer B contains three factors: time efficiency B1, space efficiency B2, and user experience B3, with hierarchical overall ranking weights of b1, b2, and b3 respectively. Indicator layer D contains five factors: average parking lot occupancy rate D1, average vehicle parking time D2, average vehicle departure time delay D3, cumulative conflict count D4, and cumulative relocation distance D5, with their respective overall ranking weights for Bj as d1, b2, and b3. 1j″ ,d 2j″, d 3j″ ,d 4j″ ,d 5j″ .

[0121] Formula for calculating the total ranking weight of the indicator layer:

[0122]

[0123] Where: w i″ The total ranking weights of the five factors D1, D2, D3, D4, and D5 in the indicator layer D are taken in the range [0,1].

[0124] b j″ The total ranking weights of the three factors B1, B2, and B3 in the criterion layer B are defined, with values ​​ranging from [0,1].

[0125] d i″j″ The total ranking weight of the five factors D1, D2, D3, D4, and D5 of the indicator layer D relative to the three factors B1, B2, and B3 of the criterion layer B is given, with a value range of [0,1].

[0126] The total ranking weight table for a single layer is shown in Table 5.

[0127] Table 5 Overall Ranking Table for Single Layer

[0128]

[0129]

[0130] The overall hierarchical ranking also requires a consistency check from the highest to the lowest level. The random consistency ratio (CR) of the overall ranking at level B is:

[0131]

[0132] Where: CR″ is the overall random consistency ratio of the criterion layer B;

[0133] CI j″ The consistency index of all factors in indicator layer D with factor Bj in criterion layer B.

[0134] RI j″ The random consistency index of all factors in indicator layer D with respect to factor Bj in criterion layer B.

[0135] When CR < 0.10, the overall hierarchical ranking result is considered to have satisfactory consistency.

[0136] (3) Calculate the weight of each evaluation indicator.

[0137] Once the consistency check passes, the eigenvector corresponding to the largest eigenvalue of the matrix becomes the weight vector. The weight vectors of each layer are normalized, and the resulting weighted values ​​are the weights of each indicator in indicator layer D, denoted as w1, w2, w3, w4, and w5. Therefore, the formula for the overall utility of different parking strategies is:

[0138] P = w1D1 + w2D2 + w3D3 + w4D4 + w5D5

[0139] Where P is the overall utility value of a parking strategy.

[0140] (4) Processing evaluation indicators

[0141] In the evaluation index system for parking strategies in automated valet parking lots, different indicators have varying effects on the evaluation results. Among the five evaluation indicators, the values ​​of different indicators differ across different strategies, their orders of magnitude differ, and their impact on the results differ. Therefore, it is necessary to process the different evaluation indicators. The processing method is as follows:

[0142] The Softmax function is used to normalize the evaluation parameters of different parking space allocation strategies, resulting in the evaluation parameter matrix of different parking space allocation strategies.

[0143] Softmax function:

[0144]

[0145] Where: z is an evaluation parameter for different strategies; z p and z q Let be one of the values.

[0146] (5) Calculate the comprehensive utility value of different parking strategies, and analyze and compare them to find the optimal parking strategy.

[0147] Example

[0148] The specific implementation steps of a simulation-based method for optimizing parking strategies in automated valet parking lots are as follows:

[0149] Step 1: Initialize the design parameters of the automated valet parking lot, including the number of parking spaces, the basic layout of the automated valet parking lot, the simulation duration, parking demand, expected vehicle arrival time interval, and average vehicle parking time. In this example, the number of parking spaces is set to 84, the basic layout is a K-stack automated valet parking lot with unidirectional lane flow, the length and width of each parking space and the road width are both 3 meters, the simulation duration is 36,000 simulation seconds, the parking demand is 120 vehicles, the expected vehicle arrival time intervals are 5 minutes, 1 minute and 30 seconds respectively, and the average vehicle parking time is 10 hours.

[0150] Step 2: Based on the initial values ​​of simulation duration, parking demand, expected vehicle arrival time interval, and average vehicle parking duration, generate the time series of vehicles arriving at the parking lot and the time series of vehicles leaving the parking lot based on Poisson distribution and negative exponential distribution.

[0151] Step 3: Based on the time series of vehicles arriving at the parking lot and the time series of vehicles leaving the parking lot generated in Step 2, use the following method... Figure 2The lane direction shown represents a unidirectional K-stack high-density parking lot layout. An operational simulation model of an automated valet parking lot is constructed to simulate the entire parking process. This includes: a parking space allocation strategy function (Parking_strategy), a vehicle entry and relocation function (Function_Arrive), a vehicle departure strategy function (Function_Depart), and a vehicle class that records vehicle information. This class contains member variables such as vehicle ID (id), target parking space tuple (pos), current coordinate tuple (coordinate), speed (velocity), entry time (arrive), stay time (stay), departure time (depart), actual departure time (real depart), route list (route), and current state (state).

[0152] Construct a Car list describing the parked vehicles, where each Car list contains a Vehicle class for each parked vehicle. Construct a 2D layout array describing the parking lot layout.

[0153] In each simulation second, the following process is completed: traverse the Car list describing the parked vehicles, determine whether a vehicle has entered the target parking space, and if a vehicle has arrived at the parking space, change its state from "arriving" to "parking" and update the vehicle's position information in the layout.

[0154] Iterate through the Car list describing parked vehicles, determine if any vehicle is in the parking search phase, and if so, determine if there is a conflict between the parking search vehicle and the departing vehicle, and between the parking search vehicle and the parking search vehicle. If a conflict exists, according to the conflict resolution strategy, make one side stay for one round to avoid the conflict, and update the layout.

[0155] Iterate through the time sequence of vehicle arrivals at the parking lot, determine if any vehicles are entering, and if so, execute the `Functon_Arrive` function to add the vehicle to the `vehicle` class, saving information such as the vehicle's sequence number (id), target parking space tuple (pos), current coordinate tuple (coordinate), speed (velocity), arrival time (arrive), stay time (stay), departure time (depart), actual departure time (real depart), route list (route), and current state (state). Update the vehicle's position information in the layout.

[0156] Iterate through the time series of vehicles leaving the parking lot, checking if any vehicle's departure time equals the current simulation time. If such a vehicle exists, change its state from "parking" to "departing" and execute the `Function_Depart` function. Iterate through all vehicles, checking if any parked vehicles are blocking the current vehicle's departure. If such a vehicle exists, treat it as a parking-seeking vehicle, change its state from "parking" to "arriving," and execute the `Function_Arrive` function. Update the vehicle position information in the layout.

[0157] Iterate through the list of Cars describing parked vehicles, checking if any vehicle is on the departure route. If a vehicle is found, check if there is a conflict between departing vehicles. If a conflict exists, use the conflict resolution strategy to make one vehicle stay for one round to avoid further conflict. Update the position information of the departing vehicles and update it in the layout.

[0158] At the end of the simulation process, the following evaluation parameters representing the parking lot's time efficiency B1, space efficiency B2, and user experience B3 generated during the simulation of the automated valet parking lot are collected and calculated: average parking lot occupancy rate D1, average vehicle search time D2, average vehicle departure time delay D3, cumulative number of conflicts D4, and vehicle relocation distance D5.

[0159] A1 strategy under three arrival rate scenarios (e.g.) Figure 3 As shown), A2 strategy (such as) Figure 4 As shown, each simulation was performed 3 times, and the average value of each result was taken to obtain five evaluation indicators: average parking lot occupancy rate D1, average vehicle parking time D2, average vehicle departure time delay D3, cumulative number of conflicts D4, and cumulative relocation distance D5.

[0160] Step 4: Based on the evaluation parameters generated during the simulation of different parking space allocation strategies in the automated valet parking lot, representing the parking lot time efficiency B1, space efficiency B2, and user experience B3, and considering the following parameters: average parking lot occupancy rate D1, average vehicle search time D2, average vehicle departure time delay D3, cumulative number of conflicts D4, and cumulative relocation distance D5, establish a hierarchical structure model using the analytic hierarchy process.

[0161] Build as Figure 5 The hierarchical model shown uses a nine-level scaling method to represent its relative importance, constructing a contrast matrix between the target layer A and the criterion layer B:

[0162]

[0163] Construct contrast matrices between the criterion-layer time efficiency B1, space efficiency B2, and user experience B3 and the indicator-layer D, respectively:

[0164]

[0165]

[0166]

[0167] Step 5: Based on the hierarchical model established in Step 4, calculate the weight matrix of each evaluation parameter to obtain the comprehensive utility evaluation formula. Further calculate the utility values ​​of different parking space allocation strategies and compare them to obtain the optimal parking strategy for the automated valet parking lot.

[0168] The values ​​in the contrast matrix represent the relative importance of the three criteria, and the largest eigenvalue λ of matrix G is calculated. max =3.0055, the corresponding eigenvector is normalized, and the resulting weight matrix is:

[0169] W = (0.5954 0.2764 0.1282) τ

[0170] A consistency test was performed on the matrix, and the calculated values ​​were CI = 0.00275, RI = 0.58, and CR = 0.00474 < 0.1. Therefore, the matrix passed the consistency test and has satisfactory consistency. Similarly, the calculation process for the criterion layer and the indicator layer is the same.

[0171] Consistency checks were performed on matrices G1, G2, and G3 respectively. The calculated values ​​were CI1 = 0.0062, CI2 = 0.010275, CI3 = 0.0038, RI = 1.12, CR1 = 0.00554 < 0.1, CR2 = 0.00917 < 0.1, and CR3 = 0.00339 < 0.1. Therefore, the matrices passed the consistency check and have satisfactory consistency.

[0172] The eigenvectors corresponding to the largest eigenvalues ​​are normalized to obtain the following weight matrices:

[0173] W1=(0.0650 0.3839 0.2231 0.1179 0.2101) τ ,

[0174] W2=(0.2580 0.1007 0.0519 0.1242 0.4652) τ ,

[0175] W3=(0.0722 0.1459 0.3753 0.0755 0.3311)τ .

[0176] A consistency check was performed on the overall hierarchical ranking, and the calculated CR” = 0.00627 < 0.1, indicating that the result satisfies the consistency requirement. After weighting the index layer, the weights are shown in Table 6.

[0177] Table 6 Weight values ​​of each indicator

[0178]

[0179] The comprehensive evaluation formula for parking strategies can be obtained as follows:

[0180] P=0.1193D1+0.2751D2+0.1953D3+0.1142D4+0.2961D5.

[0181] The proportional method was used to normalize the various indicators, and the results are shown in Table 7. The comprehensive evaluation formula for parking strategies was used to calculate the parking effects of strategies A1 and A2 under different conditions, and the results are shown in Tables 8 and 9.

[0182] Table 7 shows the index values ​​of strategies A1 and A2 and their normalized values ​​under three scenarios.

[0183]

[0184]

[0185] Table 8 shows the advantages of the A2 strategy in terms of indicator values ​​compared to the A1 strategy in three scenarios.

[0186]

[0187]

[0188] Table 9 shows the combined benefit values ​​of strategies A1 and A2 under three scenarios.

[0189]

[0190] The results show that, compared to strategy A1, strategy A2 significantly reduces the number of vehicle conflicts and vehicle relocation distance in automated valet parking lots under the three arrival rate scenarios, outperforming strategy A1 by 22.817% and 18.623%, respectively. Regarding average parking time and departure time delay, strategy A2's advantage is not significant, outperforming strategy A1 by 7.072% and 0.573%, respectively. However, strategy A1 has a greater advantage in terms of average parking lot occupancy, possibly because strategy A1 involves more relocations and more vehicles on the road, resulting in a lower occupancy rate. Furthermore, the advantages of strategy A2 over strategy A1 differ significantly under low and high arrival rate scenarios. This may be because vehicle arrivals and departures are more dispersed under low arrival rates, while they are more concentrated under high arrival rates, and the differences between the strategies contribute to this result. On the other hand, since vehicles travel in one direction in the parking lot, there is little difference between strategies A1 and A2 in terms of the average search time from the parking lot entrance to the parking space and the average departure time from the parking space to the parking lot exit.

[0191] Overall, under low, medium, and high arrival rates, strategy A2 outperforms strategy A1. The combined benefits of time efficiency, space efficiency, and user satisfaction of vehicle parking are 6.7%, 20.7%, and 6.4% higher than strategy A1, respectively, with an average of 11.3%.

[0192] In one embodiment, a vehicle lane-changing trajectory deviation calculation device based on traffic simulation is also provided, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps of the above-described simulation-based automatic valet parking lot parking strategy optimization method.

[0193] In one embodiment, a computer-readable storage medium is also provided, on which a computer program is stored, which, when executed by a processor, implements the steps of the above-described simulation-based automatic valet parking lot parking strategy optimization method.

[0194] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0195] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A simulation-based method for optimizing parking strategies in automated valet parking lots, characterized in that, Step 1: Initialize the design parameters, which include: number of parking spaces in the automated valet parking lot, simulation duration of the automated valet parking lot, parking demand, expected vehicle arrival time interval, and average vehicle parking time. Step 2: Based on the initial values ​​of simulation duration, parking demand, expected vehicle arrival time interval, and average vehicle parking time, generate the time series of vehicles arriving at the parking lot and the time series of vehicles leaving the parking lot. Step 3: Based on the time sequence of vehicles arriving at the parking lot and the time sequence of vehicles leaving the parking lot, a K-stack high-density parking lot layout with unidirectional lane flow is adopted to simulate the automatic valet parking strategy. Step 4: Collect the evaluation parameters generated during the simulation of the parking strategy of the automated valet parking lot, including: average parking lot occupancy rate D1, average vehicle search time D2, average vehicle departure time delay D3, cumulative number of conflicts D4, and cumulative relocation distance D5. Step 5: Based on the evaluation parameters generated during the simulation of the parking strategy of the automated valet parking lot, establish a hierarchical structure model based on the analytic hierarchy process to obtain the weight matrix of each evaluation parameter. Step 6: Calculate the utility values ​​of different parking strategies obtained from the simulation, and use them as the evaluation index of the parking strategy. The smaller the evaluation index value, the better the parking strategy. Step 3 specifically involves: Construct a Car list describing parked vehicles, wherein the Car list contains a vehicle class for each parked vehicle; Construct a two-dimensional array Layout to describe the layout of the parking lot. The two-dimensional array Layout describes the occupancy status of each parking space and each lane in the parking island. Iterate through the time sequence of vehicles arriving at the parking lot and the time sequence of vehicles leaving the parking lot, and determine whether the current vehicle is an arriving vehicle or a departing vehicle; If the current vehicle is an arriving vehicle, then assign a parking space to the current vehicle and update the Layout 2D array and the time sequence of the vehicle's arrival at the parking lot; If the current vehicle is a departing vehicle, then a departure simulation is performed on the current vehicle, and the Layout 2D array and the time sequence of the vehicle leaving the parking lot are updated. If there is an obstructing vehicle in the departure direction of the current vehicle, then the obstructing vehicle is repositioned, that is, a new parking space is assigned to the obstructing vehicle, so that the obstructing vehicle can drive to the new parking space to park, freeing up departure space for the current vehicle, and the Layout 2D array is updated. Five evaluation parameters representing parking lot time efficiency (B1), space efficiency (B2), and user experience (B3) were collected during the simulation of different parking space allocation strategies in automated valet parking lots. The evaluation parameters include: average parking lot occupancy rate (D1), average vehicle search time (D2), average vehicle departure time delay (D3), cumulative number of conflicts (D4), and cumulative relocation distance (D5).

2. The simulation-based method for optimizing parking strategies in automated valet parking lots according to claim 1, characterized in that, In step 2, the time series of vehicles arriving at the parking lot is randomly generated based on the initial values ​​of simulation duration, parking demand, and expected vehicle arrival time interval.

3. The simulation-based method for optimizing parking strategies in automated valet parking lots according to claim 1, characterized in that, In step 2, the time sequence of vehicles leaving the parking lot is generated based on the simulation duration, parking demand, initial values ​​of average vehicle parking duration, and the time sequence of vehicles arriving at the parking lot.

4. The simulation-based method for optimizing parking strategies in automated valet parking lots according to claim 1, characterized in that, In step 3, a K-stack high-density parking lot layout is adopted, setting vehicles to travel in one direction on the road and in two directions between parking spaces and between parking spaces and roads. The length, width, flow direction, entrance and exit positions, length-to-width ratio of parking spaces, and width of parking islands of each road are fixed to conduct dynamic simulation of the automated valet parking lot.

5. The simulation-based method for optimizing parking strategies in automated valet parking lots according to claim 1, characterized in that, In step 5, the hierarchical model is divided into a target layer, a criterion layer, and an indicator layer. The target layer is to select the optimal parking strategy; the criterion layer includes time efficiency B1, space efficiency B2, and user experience B3; and the indicator layer includes average parking lot occupancy rate D1, average vehicle search time D2, average vehicle departure time delay D3, cumulative number of conflicts D4, and cumulative relocation distance D5.

6. The simulation-based method for optimizing parking strategies in automated valet parking lots according to claim 1, characterized in that, The calculation method for obtaining each evaluation parameter in step 4 is as follows: Average parking lot occupancy : ,in: This refers to the simulation duration. The parking lot occupancy rate at simulation second t. ; Let t be the number of all parked vehicles in the parking space after the simulation ends in t seconds. ; This represents the total number of parking spaces in the parking lot; Average vehicle parking time : ,in: The total number of parking spaces; The parking search time for the idth parking vehicle; The simulation time for the idth parking vehicle to arrive at the parking space; The simulation time for the idth parking vehicle to enter the parking lot; Average vehicle departure time delay : ,in: For the first The departure time of one vehicle was delayed; For the first The actual departure time of each departing vehicle; For the first Simulated time for each departing vehicle to leave its parking space; This represents the total number of vehicles leaving the site. Total number of conflicts : ,in: for The number of collisions occurring per second in the simulation; Cumulative relocation distance : ,in: This represents the total number of repositioned vehicles; For the t-th simulation second The relocation distance of the repositioning vehicle, i.e., the first The distance a repositioning vehicle travels to a new parking space in t seconds during simulation.

7. The simulation-based method for optimizing parking strategies in automated valet parking lots according to claim 1, characterized in that, The utility value of the parking strategy in step 6 , , , , , The weights are D1, D2, D3, D4, and D5, respectively.

8. A simulation-based automatic valet parking lot parking strategy optimization device, comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the simulation-based automatic valet parking lot parking strategy optimization method as described in any one of claims 1 to 7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the simulation-based automatic valet parking lot parking strategy optimization method as described in any one of claims 1 to 7.

Images