A large parking facility optimization design method and system based on multi-agent simulation

Abstract

The application provides a large parking facility optimization design method and system based on multi-agent simulation, and belongs to the technical field of parking facility optimization design. The parking facility optimization design method is based on a large parking facility model built by multi-agent simulation. All feasible parking facility optimization schemes are input into the simulation model, including parking facility feasible optimization scheme design, parking facility multi-agent simulation model establishment, and parking facility feasible scheme simulation deduction. The system includes a data acquisition module, an optimization strategy generation module, and a multi-agent simulation module that cooperate with each other to filter the optimal parking facility optimization scheme by comparing the simulation results of all feasible optimization schemes. The application can give the optimal parking facility optimization scheme for different large parking facilities and different traffic demand scenarios, and is suitable for various scenarios. The application provides reliable decision-making data basis for effectively improving parking resource utilization and parking efficiency.

Description

Technical Field

[0001] This invention belongs to the field of parking facility optimization design technology, and particularly relates to a method and system for optimizing the design of large-scale parking facilities based on multi-agent simulation. Background Technology

[0002] With rapid economic development, the number of motor vehicles in large cities is increasing year by year, and traffic problems have become a common issue faced by cities worldwide. As a crucial node in urban transportation systems, the scientific and effective internal traffic organization schemes of large parking facilities are essential for alleviating traffic congestion and improving traffic efficiency. Traditional parking facility traffic organization designs often fail to meet the ever-increasing parking demand, especially in large urban environments such as commercial centers, residential areas, and tourist attractions. Parking lots often face problems such as peak traffic flow and inefficient space utilization, leading to challenges such as traffic congestion, wasted parking spaces, and poor user experience.

[0003] To address these issues, traditional traffic organization design methods for parking facilities are often based on experience and rules, lacking scientific quantitative analysis and optimization. Simulation modeling provides a highly flexible, controllable, and repeatable simulation environment, helping traffic planners, researchers, and policymakers better understand and evaluate the impact of different traffic organization schemes on urban traffic systems. It has been widely applied in traffic organization scheme analysis and optimization. Therefore, simulation modeling can be used for the optimized design of large parking facilities.

[0004] Existing technologies have yielded results in modeling and optimizing large parking facilities, but these are mostly techniques for micro-level traffic modeling and comparing two optimization schemes, with few techniques addressing the design of all feasible optimization schemes for parking facilities. Simulating the heterogeneity between different individuals in a traffic system is extremely difficult, and multi-agent simulation technology offers a novel approach. Unlike existing micro-level traffic modeling techniques, multi-agent simulation technology can simulate the behavior and interactions of various agents (such as vehicles, drivers, and pedestrians) in the real world, simulating the choices and decision-making behavior of independent individuals, thereby helping to understand the operational patterns of large-scale traffic systems. Furthermore, the number of traffic organization optimization schemes for large parking facilities is relatively small, making it difficult to deduce the optimal parking facility optimization scheme.

[0005] To address the aforementioned issues, it is necessary to invent a method and system for optimizing the design of large-scale parking facilities based on multi-agent simulation, to establish a more realistic multi-agent simulation model and to design and compare all feasible optimization schemes for large-scale parking facilities. Summary of the Invention

[0006] The purpose of this invention is to provide a method for optimizing the design of large-scale parking facilities based on multi-agent simulation. The method is characterized by using a large-scale parking facility model built through multi-agent simulation, inputting all feasible optimization schemes into the simulation model, and obtaining simulation results for different optimization schemes. The method includes the following steps:

[0007] S1: Collect various vehicle information based on data acquisition equipment within large parking facilities;

[0008] S2: Design all feasible optimization solutions for the parking facilities, including the following steps:

[0009] S21: Abstract the road network structure of large parking facilities into a directed graph;

[0010] S22: Calculate the connectivity of a directed graph and filter all connected graphs;

[0011] S23: Convert all connected graphs into traffic organization schemes for parking facilities to obtain all feasible optimization schemes;

[0012] S3: Establish a multi-agent simulation model of parking facilities based on all feasible optimization solutions, including the following steps;

[0013] S31: Establish intelligent agents for vehicles, roads, intersections, and parking lots;

[0014] S32: Based on vehicle trajectory data within parking facilities and stochastic utility theory, a utility function for parking space selection is established, and drivers park according to the calculation results of the utility function for parking space selection.

[0015] S33: Based on the road network structure of large parking facilities, use the elements of vehicles, roads, intersections and parking lots in multi-agent simulation software to model the environment and complete the multi-agent simulation model of parking facilities;

[0016] S4: Based on the multi-agent simulation model of parking facilities, simulate and deduce feasible solutions for parking facilities, select the optimal solution after deduction, and complete the optimization design of large-scale parking facilities based on multi-agent simulation.

[0017] Furthermore, in S2, S21 specifically refers to: based on the road network structure of large parking facilities, with intersections and entrances / exits as nodes of a directed graph, and the directed edges between nodes determined according to the road connection of the parking facilities, the large parking facilities are abstracted into a directed graph, and the directed graph is represented by an adjacency matrix. Then, the directed edges between two nodes include three possible cases.

[0018] In S22, a feasible path between any two nodes is searched using a depth-first search algorithm, and the connectivity of the graph is calculated to filter all connected graphs.

[0019] In S23, based on the correspondence between the road network structure of large parking facilities and the directed graph, all connected graphs are transformed into traffic organization schemes for parking facilities.

[0020] Furthermore, in S22, the depth-first search algorithm starts from a node and continuously searches along the directed edges between nodes, i.e., intersections and entrances / exits, until all feasible paths are found or a certain depth limit is reached. Specifically, the depth-first search algorithm starts from a vertex in the graph and continuously searches for adjacent, unvisited vertices. If multiple vertices that meet the conditions are found, one vertex is selected and the algorithm continues from that vertex; if no vertices that meet the conditions are found, the algorithm backtracks to previously visited vertices and searches for unvisited vertices. When there are no unvisited vertices in the graph, the depth-first search algorithm terminates.

[0021] Furthermore, in S31, the vehicle agent is used to simulate the driver's operation in the road network, the decision-making at each decision point, and the process and result of cruise parking; the vehicle agent continuously searches for parking spaces during driving and decides whether to select a parking space based on the practicality and availability of the parking space; the vehicle agent stores the time of the vehicle entering and leaving the parking lot and the time of entering and leaving the parking space, and uses it to calculate the time used to search for an available parking space and the time used to leave the parking lot.

[0022] The road agent is used to describe information such as the coordinates of the start and end points of a road, the number of vehicles on the road, the number of parking spaces on the road, and the intersections connecting the start and end points.

[0023] The intersection agent is used to describe information such as roads entering and exiting the intersection, the direction of roads connecting to the intersection, and the number of vehicles in the intersection.

[0024] The parking space agent is used to describe information such as parking space coordinates, the road it is attached to, parking space occupancy status, and parking space occupancy rate.

[0025] Furthermore, S32 specifically involves: after the vehicle enters the parking facility, it drives towards its destination and searches for available parking spaces. The utility of the parking space is calculated based on the utility function of the selected parking space. If the utility of all available parking spaces is less than the minimum acceptable utility, the driver continues to search for parking spaces. If the utility of some of the found available parking spaces is greater than the minimum acceptable utility, the driver selects the available parking space with the highest utility and parks the vehicle in that parking space.

[0026] Furthermore, in S32, the establishment of the utility function for parking space selection includes the following steps:

[0027] S321: Extract the vehicle coordinates, destination coordinates, coordinates of all available parking spaces, and final parking space selection result from the vehicle trajectory data when the vehicle selects a parking space. After processing, obtain the distance between the vehicle and the available parking space, the distance from the parking space to the destination, and the parking space selection result. The parking space selection result is represented as follows: add a label to each available parking space. When the parking space is selected, add label=1; when the parking space is not selected, add label=0.

[0028] S322: Use Spearman correlation coefficient to test the correlation between the distance from the vehicle to the parking space, the distance from the parking space to the destination, and the parking space selection result;

[0029] S323: Set the extracted data of each vehicle as a group. For each group of N data points, substitute the parameter estimates and data into the utility function for parking space selection to calculate the utility of selecting each parking space.

[0030] S324: If the driver selects the parking space with the highest utility, then add a label=1 to the parking space with the highest utility and add a label=0 to other parking spaces;

[0031] S325: Compare with the actual parking space selection results, calculate the accuracy rate, and select the set of parameters with the highest accuracy rate as the calibration result.

[0032] Furthermore, in S32, the utility function for parking space selection is expressed as:

[0033] U i =-(αTD) i +βWD i )+ε i

[0034] Among them, U i The utility of selecting parking space i for the driver; TD i WD is the driving distance from the driver to parking space i. i ε is the distance the driver walks from parking space i to their destination; i The random error term for the utility of the decision-maker in choosing parking space i; α and β are TD i and WD i The parameters.

[0035] Furthermore, in S1, the simulation and deduction of feasible parking facility solutions includes the following steps:

[0036] S41: Multi-scheme simulation and deduction: Based on all feasible optimization schemes output by S23, each scheme is substituted into the multi-agent simulation model for operation, different traffic demand scenarios are loaded, and the evaluation index of each scheme is output with the average vehicle travel time and average delay time as evaluation indexes.

[0037] S42: Multi-scheme comparative analysis: Based on the evaluation indicators of each scheme output by S41, the optimal optimization scheme under different traffic demand scenarios is selected through comparison.

[0038] A system for optimizing the design of large parking facilities based on multi-agent simulation is provided. The system implements the optimization design method for large parking facilities based on multi-agent simulation. It is characterized by including a data acquisition module, an optimization strategy generation module, and a multi-agent simulation module that cooperate with each other. The system compares and analyzes the evaluation indicators of different optimization schemes based on multiple modules to select the optimal optimization scheme for large parking facilities.

[0039] Furthermore, the data acquisition module collects data within the parking facility, including data acquisition devices, such as sensors and cameras. The data acquisition module collects traffic data information such as vehicle flow, parking duration, and driver behavior through sensors and cameras, and transmits the traffic data information to the system for processing.

[0040] The optimization strategy generation module is used to generate all feasible optimization schemes based on the road network structure and abstraction of the map structure of large parking facilities.

[0041] The multi-agent simulation module is used to build a multi-agent simulation model. First, based on the collected traffic data and optimization strategies, it generates all feasible optimization schemes output by the module. Then, it simulates traffic flow, driver behavior, parking process and other traffic operation conditions under different traffic demand scenarios. Finally, it loads all feasible optimization schemes one by one and outputs the evaluation index of each feasible optimization scheme.

[0042] Compared with the prior art, the beneficial effects of the present invention are mainly reflected in:

[0043] 1. This invention uses multi-agent simulation technology to more accurately simulate vehicle flow and driver behavior in large parking scenarios, thereby providing data analysis and solution optimization that are closer to the real situation, and improving the real-time performance and accuracy of the solution.

[0044] 2. The system of the present invention is more flexible in optimization design and can generate customized optimization strategies according to different scenarios and needs, making it more suitable for different types of parking facilities.

[0045] 3. The system of the present invention provides a comprehensive design scheme evaluation module, which can evaluate the effects of different schemes through multiple indicators and parameters, such as parking efficiency and traffic congestion, so as to more comprehensively measure the merits of the schemes. Attached Figure Description

[0046] Figure 1 This is a flowchart of the large-scale parking facility optimization design method based on multi-agent simulation of the present invention;

[0047] Figure 2 A flowchart illustrating the design of a feasible optimization scheme for a large parking facility according to the present invention;

[0048] Figure 3 A flowchart for establishing a multi-agent simulation model of a large parking facility according to the present invention;

[0049] Figure 4 This is a flowchart illustrating the simulation of a feasible solution for a large-scale parking facility according to the present invention.

[0050] Figure 5 This is a schematic diagram of the optimized design system structure for large parking facilities according to the present invention. Detailed Implementation

[0051] The following will describe in more detail, with reference to the schematic diagrams, a method and system for optimizing the design of large parking facilities based on multi-agent simulation, according to the present invention. The diagrams illustrate preferred embodiments of the present invention. It should be understood that those skilled in the art can modify the present invention described herein while still achieving the beneficial effects of the present invention. Therefore, the following description should be understood as being of general knowledge to those skilled in the art and is not intended to limit the present invention.

[0052] As one aspect of the present invention, a method for optimizing the design of large-scale parking facilities based on multi-agent simulation is provided. This method optimizes the traffic organization scheme of parking facilities based on an optimization design approach that includes designing feasible optimization schemes for parking facilities, establishing multi-agent simulation models of parking facilities, and simulating and deducing feasible schemes for parking facilities.

[0053] Figure 1 To optimize the flowchart of the design methodology, the steps of the optimization design methodology include:

[0054] Step 1: Collect information including vehicle flow, parking duration, and driver behavior using data acquisition equipment within large parking facilities;

[0055] Step 2: Design all feasible optimization schemes based on the road network structure and graph structure of large parking facilities;

[0056] Step 3: Input all feasible solutions into the multi-agent simulation model to obtain the corresponding evaluation indicators;

[0057] Step 4: Compare and analyze all feasible solutions and select the optimal solution.

[0058] Figure 2 The flowchart for the design of a feasible optimization scheme for a large parking facility provided by this invention includes the following steps: abstracting the road network structure into a graph, filtering the connected graph, and outputting the feasible optimization scheme. The specific steps are as follows:

[0059] 1): The road network structure is abstracted as a diagram:

[0060] Taking the B1 level parking lot of Plot 15 in Jinding Tiandi, Jinqiao, Shanghai as an example, the intersections and entrances / exits in the road network structure are treated as nodes in the graph, resulting in a total of 8 nodes. The directed edges between nodes are determined based on the actual road connections, resulting in a total of 9 directed edges. There are three possible cases for each pair of nodes. Taking the directed edge between nodes 1 and 2 as an example, these include a one-way edge from node 1 to node 2, a one-way edge from node 2 to node 1, and a two-way edge from node 1 to node 2.

[0061] 2): Filtering connected graphs:

[0062] There are three possible outcomes for every directed edge between any two nodes. Therefore, the directed graph output by S1 has 39 possible outcomes, or 19683 different directed graphs. A depth-first search (DFS) algorithm is used to search for feasible paths between any two nodes. A directed graph is considered connected only if every two nodes have a feasible path. The final output is all connected graphs.

[0063] 3): Output of feasible optimization solutions:

[0064] Based on the correspondence between the road network structure of large parking facilities and directed graphs, directed edges in the connected graph are mapped to corresponding roads. Taking the connection between the directed edge and the actual road between nodes 1 and 2 as an example, if the directed edge is a one-way edge from node 1 to node 2, the actual road connection is a one-way road from node 1 to node 2; if the directed edge is a one-way edge from node 2 to node 1, the actual road connection is a one-way road from node 2 to node 1; if the directed edge is a two-way edge from node 1 to node 2, the actual road connection is a two-way road from node 1 to node 2. By transforming all connected graphs into traffic organization schemes for parking facilities, a total of 128 feasible optimization schemes are obtained.

[0065] Figure 3 This is a flowchart illustrating the establishment of a multi-agent simulation model for a large parking facility provided in an embodiment of the present invention. The simulation modeling steps include: multi-agent attribute design, parking search model establishment, and simulation environment modeling. The specific steps are as follows:

[0066] (1): Multi-agent attribute design:

[0067] Use the multi-agent simulation software Anylogic to create agents for vehicles, roads, intersections, and parking lots in the simulation model.

[0068] The vehicle agent's initial speed is set to 5 km / h, its maximum speed to 15 km / h, and its maximum acceleration and deceleration to 2 m² / s. The vehicle's driving behavior is controlled by controls such as CarSource, CarMoveTo, CarDispose, SelectOutput5, and Delay, along with corresponding code, simulating the driver's operations in the road network, decisions at various decision points, and the process and outcome of cruising and parking. A parking search model is integrated into the vehicle agent, continuously searching for parking spaces during the journey and deciding whether to select a space based on its usability and availability. The timeMeasureStart and timeMeasureEnd controls are used to record the vehicle agent's entry and exit times from the parking lot and parking spaces. Subtracting the timeMeasureStart from the timeMeasureEnd time yields the time taken to search for an available parking space and the time taken to leave the parking lot.

[0069] Create a road agent based on the actual number of roads, and store information such as the coordinates of the start and end points of the corresponding roads, the number of lanes in both directions, the number of vehicles in each lane, and parking spaces.

[0070] Create intersection agents based on the actual number of intersections, and store information such as the roads entering and exiting the corresponding intersections, and the number of vehicles in the corresponding intersections.

[0071] A parking space agent is created based on the actual number of parking spaces, storing the coordinates of the corresponding parking space, the attached road, the parking space occupancy status, and the parking space occupancy rate. The size of each parking space is set to 5.3 meters × 2.4 meters.

[0072] (2): Parking search model establishment:

[0073] Using parking lot vehicle trajectory data from the Dragon Lake Parking (DLP) dataset, Berkeley's first high-definition dataset for parking scenarios, we extracted the vehicle coordinates, destination coordinates, coordinates of all available parking spaces, and the final parking space selection result when a vehicle selects a parking space. After processing, we obtained the distance between the vehicle and the available parking space, the distance from the parking space to the destination, and the parking space selection result. The parking space selection result is to add a label to each available parking space. If the parking space is selected, the label is added as label=1; if it is not selected, the label is added as label=0.

[0074] The Spearman correlation coefficient was used to test the correlation between the distance from the vehicle to the parking space, the distance from the parking space to the destination, and the parking space selection result. The significance test values ​​of the distance from the vehicle to the parking space, the distance from the parking space to the destination, and the parking space selection result were all less than 0.01, indicating a significant correlation, and the correlation was negative.

[0075] A utility function for parking space selection is established based on stochastic utility theory, and the utility function is calibrated through parameter optimization experiments using Anylogic.

[0076] The data of each extracted vehicle is grouped together. For each group of N data points, the parameter estimates and data are substituted into the utility function for parking space selection to calculate the utility of selecting each parking space.

[0077] Assuming the driver chooses the parking space with the highest utility, label the parking space with the highest utility as label=1, and label the other parking spaces as label=0;

[0078] The accuracy rate is calculated by comparing the results with the actual parking space selection, and the set of parameters with the highest accuracy rate is selected as the calibration result.

[0079] The final output utility function for parking space selection is

[0080] U i =-(αTD) i +βWD i )+ε i

[0081] Among them, U i TD's utility in selecting parking space i for the driver i WD is the driving distance from the driver to parking space i. i ε is the distance the driver walks from parking space i to their destination. i The random error term for the utility of the decision-maker in choosing option i.

[0082] (3): Simulation environment modeling:

[0083] Based on the B1 level parking lot drawings of Jin Ding Tiandi Plot 15 in Jinqiao, Shanghai, environmental modeling was performed using elements such as roads, parking spaces, and intersections in the road library of the multi-agent simulation software Anylogic.

[0084] Figure 4 This is a flowchart illustrating the simulation of a feasible solution for a large parking facility provided in an embodiment of the present invention. The simulation steps include: multi-scheme simulation, parking search model establishment, and simulation environment modeling. The specific steps are as follows:

[0085] 1. Simulation and deduction of multiple scenarios:

[0086] Load different traffic demand scenarios, such as morning rush hour and evening rush hour, and substitute each of the 128 feasible optimization schemes output by S3 into the multi-agent simulation model in Anylogic for simulation. The simulation time is 3600s and the preloading time is 300s. Output the average vehicle travel time and average delay time for different feasible optimization schemes under different traffic demand scenarios.

[0087] 2: Comparative Analysis of Multiple Options:

[0088] Based on the evaluation indicators of each scheme output by S7, the optimal optimization scheme under different traffic demand scenarios is selected by comparison.

[0089] This invention also provides a schematic diagram of a large-scale parking facility optimization design system, as shown in the schematic diagram below. Figure 5 As shown in the diagram. First, the data acquisition module collects various data within the parking facility in real time using equipped sensors, cameras, and other devices, such as vehicle flow, parking duration, and driver behavior. Second, the optimization strategy generation module generates all feasible optimization schemes based on the road network structure of the large parking facility and an abstract map structure. The collected parking facility traffic data and feasible optimization schemes are input into a multi-agent simulation module to simulate traffic flow, driver behavior, and parking processes under different scenarios, outputting evaluation metrics for different optimization schemes. By comparing and analyzing the evaluation metrics of different optimization schemes, the optimal optimization scheme for the large parking facility is selected.

[0090] The above are merely preferred embodiments of the present invention and do not constitute any limitation on the present invention. Any equivalent substitutions or modifications made by those skilled in the art to the technical solutions and content disclosed in the present invention without departing from the scope of the present invention shall be deemed to have remained within the protection scope of the present invention.

Claims

1. A method for optimizing the design of large-scale parking facilities based on multi-agent simulation, characterized in that, The parking facility optimization design method is based on a large-scale parking facility model built using multi-agent simulation. All feasible parking facility optimization schemes are input into the simulation model to obtain simulation results for different optimization schemes. The method includes the following steps: S1: Collect various vehicle information based on data acquisition equipment within large parking facilities; S2: Design all feasible optimization solutions for the parking facilities, including the following steps: S21: Abstract the road network structure of large parking facilities into a directed graph; S22: Calculate the connectivity of a directed graph and filter all connected graphs; S23: Convert all connected graphs into traffic organization schemes for parking facilities to obtain all feasible optimization schemes; S3: Establish a multi-agent simulation model of parking facilities based on all feasible optimization solutions, including the following steps; S31: Establish intelligent agents for vehicles, roads, intersections, and parking lots; S32: Based on vehicle trajectory data within parking facilities and stochastic utility theory, a utility function for parking space selection is established, and drivers park according to the calculation results of the utility function for parking space selection. In step S32, establishing the utility function for parking space selection includes the following steps: S321: Extract the vehicle coordinates, destination coordinates, coordinates of all available parking spaces, and final parking space selection result from the vehicle trajectory data when the vehicle selects a parking space. After processing, obtain the distance between the vehicle and the available parking space, the distance from the parking space to the destination, and the parking space selection result. The parking space selection result is expressed as follows: add a label to each available parking space. When the parking space is selected, add label=1; when the parking space is not selected, add label=0. S322: Use Spearman correlation coefficient to test the correlation between the distance from the vehicle to the parking space, the distance from the parking space to the destination, and the parking space selection result; S323: Set the extracted data of each vehicle as a group. For each group of N data points, substitute the parameter estimates and data into the utility function for parking space selection to calculate the utility of selecting each parking space. S324: If the driver selects the parking space with the highest utility, then add a label=1 to the parking space with the highest utility and add a label=0 to other parking spaces; S325: Compare with the actual parking space selection results, calculate the accuracy, and select the set of parameters with the highest accuracy as the calibration result; S33: Based on the road network structure of large parking facilities, use the elements of vehicles, roads, intersections and parking lots in multi-agent simulation software to model the environment and complete the multi-agent simulation model of parking facilities; S4: Based on the multi-agent simulation model of parking facilities, simulate and deduce feasible solutions for parking facilities, select the optimal solution after deduction, and complete the optimization design of large-scale parking facilities based on multi-agent simulation.

2. The method for optimizing the design of large-scale parking facilities based on multi-agent simulation according to claim 1, characterized in that, In S2, S21 specifically refers to: based on the road network structure of large parking facilities, taking intersections and entrances / exits as nodes of a directed graph, determining the directed edges between nodes according to the road connection of the parking facilities, and abstracting the large parking facilities into a directed graph. The directed graph is represented by an adjacency matrix, and the directed edges between two nodes include three possible cases. In step S22, a feasible path between any two nodes is searched using a depth-first search algorithm, and the connectivity of the graph is calculated to filter all connected graphs. In step S23, based on the correspondence between the road network structure of large parking facilities and the directed graph, all connected graphs are transformed into traffic organization schemes for parking facilities.

3. The method for optimizing the design of large-scale parking facilities based on multi-agent simulation according to claim 2, characterized in that, In step S22, the depth-first search algorithm starts from a node and continuously searches along the directed edges between nodes, i.e., intersections and entrances / exits, until all feasible paths are found.

4. The method for optimizing the design of large-scale parking facilities based on multi-agent simulation according to claim 1, characterized in that, In step S31, the vehicle agent is used to simulate the driver's operation in the road network, the decision-making at each decision point, and the process and result of cruise parking; the vehicle agent continuously searches for parking spaces during driving and decides whether to select a parking space based on the practicality and availability of the parking space; the vehicle agent stores the time of entering and exiting the parking lot and the time of entering and exiting the parking space, and is used to calculate the time used to search for an available parking space and the time used to leave the parking lot. The road agent is used to describe the coordinates of the road's start and end points, the number of vehicles on the road, the number of parking spaces on the road, and information about the intersections connecting the start and end points. The intersection agent is used to describe information such as roads entering and exiting the intersection, the direction of roads connecting to the intersection, and the number of vehicles in the intersection. The parking space agent is used to describe information such as the coordinates of the parking space, the road it is attached to, the occupancy status of the parking space, and the parking space occupancy rate.

5. The method for optimizing the design of large-scale parking facilities based on multi-agent simulation according to claim 1, characterized in that, S32 specifically involves: after the vehicle enters the parking facility, it drives towards its destination and searches for available parking spaces. The utility of the parking space is calculated based on the utility function of the selected parking space. If the utility of all available parking spaces is less than the minimum acceptable utility, the driver continues to search for parking spaces. If the utility of some of the found available parking spaces is greater than the minimum acceptable utility, the driver selects the available parking space with the highest utility and parks the vehicle in that parking space.

6. The method for optimizing the design of large-scale parking facilities based on multi-agent simulation according to claim 1, characterized in that, In step S32, the utility function for selecting a parking space is expressed as follows: ； in, Selecting a parking space for the driver The utility; For the driver to the parking space The driving distance; For the driver in the parking space The walking distance from the parking lot to the destination; Selecting parking spaces for decision-makers The random error term of utility; and for and The parameters.

7. The method for optimizing the design of large-scale parking facilities based on multi-agent simulation according to claim 1, characterized in that, In step S4, establishing a simulation of the feasible solution for the parking facility includes the following steps: S41: Multi-scheme simulation and deduction: Based on all feasible optimization schemes output by S23, each scheme is substituted into the multi-agent simulation model for operation, different traffic demand scenarios are loaded, and the evaluation index of each scheme is output with the average vehicle travel time and average delay time as evaluation indexes. S42: Multi-scheme comparative analysis: Based on the evaluation indicators of each scheme output by S41, the optimal optimization scheme under different traffic demand scenarios is selected through comparison.

8. A system for optimizing the design of large parking facilities based on multi-agent simulation, implementing the optimization design method for large parking facilities based on multi-agent simulation as described in any one of claims 1-7, characterized in that, The system includes a data acquisition module, an optimization strategy generation module, and a multi-agent simulation module that work together. Based on the comparison and analysis of evaluation indicators of different optimization schemes by multiple modules, the system selects the optimal optimization scheme for large parking facilities.

9. The system of the large-scale parking facility optimization design method based on multi-agent simulation according to claim 8, characterized in that, The data acquisition module collects data within the parking facility, including a data acquisition device. The data acquisition module collects traffic data information through the data acquisition device and transmits the traffic data information to the system for processing. The optimization strategy generation module is used to generate all feasible optimization schemes based on the road network structure and abstract map structure of large parking facilities. The multi-agent simulation module is used to construct a multi-agent simulation model. First, based on the collected traffic data and optimization strategies, it generates all feasible optimization schemes output by the module. Then, it simulates traffic operation under different traffic demand scenarios. Finally, it loads all feasible optimization schemes one by one and outputs the evaluation index of each feasible optimization scheme.

Images