Distributed traffic control method and system based on multi-agent dynamic structure recombination

Abstract

The invention provides a distributed traffic control method and system based on multi-agent dynamic structure recombination. The method comprises the following steps: initializing a traffic light controller of each intersection of a road network into an agent according to acquired road network data; the intelligent agent exchanges information with a neighbor intelligent agent; the intelligent agents independently carry out calculation according to the road condition information obtained by the intelligent agents, and the association degree value of each neighbor intelligent agent is obtained; the intelligent agent makes a decision according to the association degree value to obtain a corresponding decision result; and executing the decision result, and redistributing the control right of the intelligent agent to the intersection. According to the invention, multiple agents are introduced into the traffic control system, so that the problem of dimension explosion caused by a centralized traffic control system can be solved, and the congestion condition of a traffic intersection can be effectively relieved.

Description

Technical Field

[0001] The present invention relates to the technical field of traffic control, and in particular to a distributed traffic control method and system based on multi-agent dynamic structural reorganization. Background Art

[0002] With the continuous development of social economy, the number of private cars has continued to increase, and traffic congestion has become a serious social problem in big cities. In order to alleviate traffic congestion, on the one hand, it is necessary to improve the construction of transportation infrastructure, and on the other hand, it is necessary to build a reasonable traffic control system, the core of which is to effectively control the traffic lights to ensure smooth travel under different traffic volumes. However, the urban traffic network is a large-scale and complex system. The method of controlling traffic lights at a certain intersection is not feasible. It is usually necessary to divide the entire road network into multiple sub-areas, and then control the traffic lights in each sub-area. In previous methods, a centralized traffic control scheme is often used, that is, to collect traffic data from the entire road network, centrally calculate and divide the sub-areas, and then control the traffic lights in each sub-area. This scheme has high computational complexity, poor robustness, and is not flexible enough. Therefore, it is necessary to adopt a distributed method for large-scale traffic control systems, so multi-agents are introduced.

[0003] Intelligent agents can think independently and interact with the environment. Specifically, each intelligent agent has the ability to communicate with neighboring intelligent agents and also has the ability to make self-decisions, which can meet the real-time requirements of traffic control in large-scale road networks.

[0004] In summary, the market needs a distributed traffic control method and system based on multi-agent dynamic structural reorganization that can adopt a distributed approach to large-scale traffic control systems. Summary of the invention

[0005] In view of the defects in the prior art, the object of the present invention is to provide a distributed traffic control method and system based on multi-agent dynamic structural reorganization.

[0006] A distributed traffic control method based on multi-agent dynamic structure reorganization provided by the present invention includes:

[0007] Initialization step: Initialize the traffic light controller at each intersection of the road network as an intelligent agent based on the acquired road network data;

[0008] Interaction step: the agent exchanges information with neighboring agents;

[0009] Calculation steps: The agent independently calculates based on the road condition information it obtains to obtain the correlation value with each neighboring agent;

[0010] Decision-making step: the agent makes a decision according to the correlation degree value to obtain a corresponding decision result;

[0011] Reconstruction step: execute the decision result and reallocate the control of the intelligent agent over the intersection.

[0012] Preferably, the intelligent agent includes traffic data, working mode, calculation module, communication module and decision module;

[0013] The working mode includes an active mode and a passive mode;

[0014] The active mode represents that the agent is controlling one or more intersections;

[0015] The passive mode means that the agent loses control of the current intersection.

[0016] Preferably, the factors affecting the degree of association between the agent and the neighboring agents include the road section length L, the number of lanes N, and the traffic flow F, and the calculation formulas are as follows:

[0017]

[0018] Where n represents the number of road segments connecting the boundaries of two agents, L A→B 、N A→B 、F A→B They represent the length of the road section L, the number of lanes N, and the traffic volume F between agents A and B respectively;

[0019] The calculation formula of the degree of association between agents is as follows:

[0020]

[0021] Among them, α represents the degree of association between agents.

[0022] Preferably, the section length, number of lanes and traffic flow between the agent and its neighboring agents are respectively the average values ​​of the section length, number of lanes and traffic flow of multiple sections connecting the boundaries of the agent and its neighboring agents.

[0023] Preferably, the decision making step comprises:

[0024] Step S4.1: Select the association threshold t between two agents A and B;

[0025] Step S4.2: According to the association threshold t, determine whether the two agents A and B simultaneously satisfy the following formula:

[0026] α A→B >t

[0027] α B→A >t

[0028] If yes, then execute step S4.3; if no, then the decision result is split, where α A→B , α B→A They represent the degree of association between agent A and agent B, and the degree of association between agent B and agent A respectively;

[0029] Step S4.3: Further determine whether the two agents A and B satisfy the following formula:

[0030]

[0031] If yes, the decision result is merge; if no, the decision result is no operation. A→B is the traffic flow from agent A to agent B, F B→A is the traffic flow from agent B to agent A.

[0032] Preferably, the selection of the correlation threshold t is related to the interval time of decision-making. The time of decision-making is adjusted according to actual needs. Under different decision-making times, the corresponding correlation threshold t is different.

[0033] Preferably, agents that need to be merged are placed in merge candidates, and agents that need to be split are placed in split candidates.

[0034] Preferably, the agent merging is a two-way process, which requires the merging parties to communicate and jointly decide;

[0035] The agent splitting is a one-way process. If a neighboring agent is among the splitting candidates, it will be split directly.

[0036] Preferably, the agent merging is to concentrate the control rights of multiple intersections on an agent at a certain intersection, the agent at the certain intersection is a core agent, the core agent is set to an active working mode, and the remaining agents are set to a passive working mode and obey the control of the core agent, thereby achieving coordinated control of the entire sub-area;

[0037] The core agent is selected on a first-come, first-served basis.

[0038] A distributed traffic control system based on multi-agent dynamic structural reorganization provided by the present invention includes:

[0039] Initialization module: Initialize the traffic light controller at each intersection of the road network into an intelligent agent based on the acquired road network data;

[0040] Interaction module: the agent exchanges information with neighboring agents;

[0041] Calculation module: The agent independently calculates the road condition information it obtains and obtains the correlation value with each neighboring agent;

[0042] Decision-making module: the agent makes a decision according to the correlation degree value to obtain a corresponding decision result;

[0043] Reconstruction module: execute the decision result and reallocate the control of the intelligent agent over the intersection.

[0044] Compared with the prior art, the present invention has the following beneficial effects:

[0045] 1. The present invention introduces multi-agents into the traffic control system, which can solve the dimensionality explosion problem caused by the centralized traffic control system and effectively alleviate the congestion at traffic intersections.

[0046] 2. The present invention proposes corresponding intelligent agent merging and splitting algorithms that can realize distributed control and management of large-scale traffic control systems with higher efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Other features, objects and advantages of the present invention will become more apparent from the detailed description of non-limiting embodiments made with reference to the following drawings:

[0048] Figure 1 It is a schematic diagram of the working method of the present invention;

[0049] Figure 2 is a diagram of the intelligent agent structure of the present invention;

[0050] Figure 3 It is a schematic diagram of the intelligent agent merging of the present invention;

[0051] Figure 4 A process for processing merge requests between intelligent agents of the present invention to achieve communication consensus;

[0052] Figure 5 Schematic diagram of the intelligent agent splitting of the present invention. DETAILED DESCRIPTION

[0053] The present invention is described in detail below in conjunction with specific embodiments. The following embodiments will help those skilled in the art to further understand the present invention, but are not intended to limit the present invention in any form. It should be noted that, for those of ordinary skill in the art, several changes and improvements can also be made without departing from the concept of the present invention. These all belong to the protection scope of the present invention.

[0054] The present invention introduces multiple agents into the traffic control system, each agent corresponds to a traffic light controller at one or more intersections, and then controls the traffic lights. As the traffic flow changes, the agents will communicate with each other and make decisions based on the acquired data, thereby achieving merging and splitting, and then continuously changing the signal light control.

[0055] Embodiment 1

[0056] According to the present invention, a distributed traffic control method based on multi-agent dynamic structural reorganization is provided, which includes first initializing each intersection of the input road network, initializing the traffic light controller of each intersection as an agent, exchanging data between adjacent agents, and the obtained data will be used to calculate the degree of association between the agents. According to the calculated degree of association, the agents will decide whether to merge or split. Finally, the agents will reorganize the structure, that is, redistribute the control rights of the intersections. Specifically, if Figure 1 As shown, including:

[0057] Initialization step: Based on the acquired road network data, the traffic light controller at each intersection of the road network is initialized as an agent, that is, each agent initially has control over an intersection. The structure of the initialized agent is as follows: Figure 2 As shown. The intelligent agents include traffic flow data, working mode, calculation module, communication module and decision module. The intelligent agent will store the traffic flow data of the current intersection. The working mode of each intelligent agent includes active mode and passive mode. The active mode means that the intelligent agent is controlling one or more intersections, and the passive mode means that the intelligent agent has lost control of the current intersection. At the same time, each intelligent agent also has the ability of calculation, communication and decision-making.

[0058] Interaction step: The agent exchanges information with neighboring agents. The information includes traffic flow information. After the agent is initialized, it will continuously exchange traffic flow information with neighboring agents. The acquired traffic flow information will be used in the decision step to determine whether the two agents can be merged.

[0059] Calculation steps: The agent independently calculates based on the road condition information it obtains, and obtains the degree of association with each neighboring agent. As the traffic flow changes, the traffic flow data maintained by the agent will also change continuously, and the agent will also calculate the degree of association α with the neighboring agent in real time. Factors that affect the degree of association between two agents include the length of the section L, the number of lanes N, and the traffic flow F. Since each agent may control multiple intersections, the section length, number of lanes, and traffic flow between the two agents are the average values ​​of the section length, number of lanes, and traffic flow of multiple sections connected by the boundaries of the two agents. The calculation formulas for the section length L, number of lanes N, and traffic flow F between agents A and B are as follows

[0060]

[0061] Where n is the number of road segments connecting the boundaries of the two agents.

[0062] When the road section is longer, vehicles entering from the control range of the previous agent will gradually disperse as the driving distance and the number of lanes increase, and arrive at the control range of the next agent in an increasingly unpredictable manner. At this time, the degree of association between agents decreases significantly. On the contrary, when the road section is very short, vehicles will arrive at the next intersection relatively continuously, and the degree of association between agents will increase, that is,

[0063]

[0064] As a dynamic influencing factor between the two agents, the traffic flow F is significantly proportional to the degree of association between the agents. When the traffic flow between the agents is large, congestion is likely to occur between the two agents, and more coordinated control is needed. On the contrary, when the traffic flow is small and in a dispersed state, the two agents do not need to coordinate, that is,

[0065] α∝f

[0066] In summary, we can get the formula for calculating the degree of association between agents:

[0067]

[0068] Decision-making step: The agent makes a decision based on the correlation value to obtain the corresponding decision result. First, select a correlation threshold t. The selection of the correlation threshold is related to the interval time of decision-making. The decision time can be flexibly adjusted according to actual needs. When the traffic flow changes greatly, a quick decision can be made. When the traffic conditions are relatively stable, a decision can be made at a longer interval. Under different decision times, the corresponding correlation threshold t is different. For the correlation degree α between two agents A and B A→B , α B→A :

[0069] Step S4.1: Select the association threshold T between two agents A and B.

[0070] Step S4.2: According to the association threshold t, determine whether the two agents A and B simultaneously satisfy the following formula:

[0071] α A→B >t

[0072] α B→A >t

[0073] If yes, then execute step S4.3; if no, then the decision result is split, where α A→B , αB→A They represent the degree of association between agent A and agent B, and the degree of association between agent B and agent A respectively.

[0074] Step S4.3: Further determine whether the two agents A and B satisfy the following formula:

[0075]

[0076] If yes, the decision result is merge; if no, the decision result is no operation. A→B is the traffic flow from agent A to agent B, F B→A It is the traffic flow from agent B to agent A, that is, the traffic flow between the two agents in both directions will not differ too much.

[0077] After the above steps S4.1 to S4.3, three decision results will be generated: merge, split, and no operation. At the same time, the intelligent agents that need to be merged will be placed in the merge candidates, and the intelligent agents that need to be split will be placed in the split candidates.

[0078] Reconstruction step: execute the decision result and reallocate the control of the agent over the intersection. The decision results include agent merging, agent splitting, and no operation. The agent merging is a two-way process, which requires communication between the two parties to make a joint decision; while agent splitting is a one-way process. If the neighboring agent is a candidate for splitting, it will be split directly.

[0079] like Figure 3 As shown in the figure, the merging of agents is to concentrate the control of several intersections on the agent at a certain intersection. The agent at this intersection is called the core agent. This core agent is set to the active working mode, and the other agents are set to the passive working mode and obey the control of the core agent, so as to achieve coordinated control of the entire sub-area. The selection of the core agent adopts a first-come, first-served method, that is, the agent that first issues a merging application will become the core agent.

[0080] Before the merger, Figure 4 As shown in , each agent will send a merge request to the neighboring agents that meet the conditions. However, since the scheduling order of each agent is different, the agent that has not received the merge request will send the request first, and the agent that has received the request needs to process the merge request. If the sender is in its own merge candidate, the merge operation is performed; otherwise, the sender needs to be notified, and the sender can delete the receiver from its own merge candidate to avoid continuously sending merge requests. In this way, a decision consensus can be reached in the end, such as Figure 4 shown.

[0081] Agent splitting is to reactivate some agents at intersections and let them regain control of their own intersections, such as Figure 5 As shown in the figure, the split candidates of the middle core agent are the left and bottom agents, which will only retain the merger with the right agent. So a large agent will be split into three small agents, and the core agent of each small agent is determined by the core agent of the original large agent.

[0082] The purpose of the present invention is to provide a distributed traffic control system based on multi-agent dynamic structural reorganization, which is used as an optimization method for centralized traffic control systems to alleviate traffic congestion.

[0083] Embodiment 2

[0084] The present invention also provides a distributed traffic control system based on multi-agent dynamic structural reorganization. The distributed traffic control system based on multi-agent dynamic structural reorganization can be implemented by executing the process steps of the distributed traffic control method based on multi-agent dynamic structural reorganization, that is, technical personnel in this field can understand the distributed traffic control method based on multi-agent dynamic structural reorganization as a preferred implementation of the distributed traffic control system based on multi-agent dynamic structural reorganization.

[0085] A distributed traffic control system based on multi-agent dynamic structural reorganization provided by the present invention includes:

[0086] Initialization module: Initialize the traffic light controller at each intersection of the road network into an intelligent agent according to the acquired road network data. The intelligent agent includes traffic flow data, working mode, calculation module, communication module and decision module. The working mode includes active mode and passive mode. The active mode means that the intelligent agent is controlling one or more intersections. The passive mode means that the intelligent agent loses control of the current intersection.

[0087] Interaction module: The agent exchanges information with neighboring agents.

[0088] Calculation module: The agent independently calculates based on the road condition information it obtains to obtain the correlation value with each neighboring agent. The factors affecting the correlation between the agent and the neighboring agents include the road section length L, the number of lanes N, and the traffic flow F. The calculation formulas are as follows:

[0089]

[0090]

[0091] Where n represents the number of road segments connecting the boundaries of two agents, L A→B 、N A→B 、F A→BThey represent the length L of the road section between agents A and B, the number of lanes N, and the traffic flow F respectively.

[0092] The calculation formula of the degree of association between agents is as follows:

[0093]

[0094] Wherein, α represents the degree of association between agents. The length of the road section, number of lanes, and traffic volume between the agent and its neighboring agents are the average lengths of the road sections, number of lanes, and traffic volume of multiple road sections connecting the boundaries of the agent and its neighboring agents.

[0095] Decision-making module: The agent makes a decision based on the correlation degree value to obtain a corresponding decision result. The decision-making module includes:

[0096] Module M4.1: Select the association threshold t between two agents A and B. The selection of the association threshold t is related to the decision interval time. The decision time is adjusted according to actual needs. Under different decision times, the corresponding association threshold t is different.

[0097] Module M4.2: According to the association threshold t, determine whether the two agents A and B simultaneously satisfy the following formula:

[0098] α A→B >t

[0099] α B→A >t

[0100] If yes, then module M4.3 is triggered. If no, then the decision result is split, where α A→B , α B→A They represent the degree of association between agent A and agent B, and the degree of association between agent B and agent A respectively.

[0101] Module M4.3: Further determine whether the two agents A and B satisfy the following formula:

[0102]

[0103] If yes, the decision result is merge. If no, the decision result is no operation. A→B is the traffic flow from agent A to agent B, F B→A is the traffic flow from agent B to agent A.

[0104] Reconstruction module: trigger the decision result and reallocate the control rights of the intelligent agents over the intersections. The intelligent agents that need to be merged are placed in the merge candidates, and the intelligent agents that need to be split are placed in the split candidates. The intelligent agent merging is a two-way process, which requires the merging parties to communicate and jointly decide. The intelligent agent splitting is a one-way process. If the neighboring intelligent agent is in the split candidate, it will be directly split. The intelligent agent merging is to concentrate the control rights of multiple intersections on the intelligent agent of a certain intersection. The intelligent agent of a certain intersection is the core intelligent agent. The core intelligent agent is set to the active working mode, and the remaining intelligent agents are set to the passive working mode and obey the control of the core intelligent agent, thereby realizing the coordinated control of the entire sub-area. The core intelligent agent is selected on a first-come, first-served basis.

[0105] Those skilled in the art know that, in addition to realizing the system and its various devices, modules, and units provided by the present invention in a purely computer-readable program code, it is entirely possible to realize the same functions in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers by logically programming the method steps. Therefore, the system and its various devices, modules, and units provided by the present invention can be considered as a hardware component, and the devices, modules, and units included therein for realizing various functions can also be regarded as structures within the hardware component; the devices, modules, and units for realizing various functions can also be regarded as both software modules for realizing the method and structures within the hardware component.

[0106] The above describes the specific embodiments of the present invention. It should be understood that the present invention is not limited to the above specific embodiments, and those skilled in the art can make various changes or modifications within the scope of the claims, which does not affect the essence of the present invention. In the absence of conflict, the embodiments of the present application and the features in the embodiments can be combined with each other arbitrarily.

Claims

1. A distributed traffic control method based on multi-agent dynamic structure reorganization, characterized in that: include: Initialization step: Initialize the traffic light controller at each intersection of the road network as an intelligent agent based on the acquired road network data; Interaction step: the agent exchanges information with neighboring agents; Calculation steps: The agent independently calculates based on the road condition information it obtains to obtain the correlation value with each neighboring agent; Decision-making step: the agent makes a decision according to the correlation degree value to obtain a corresponding decision result; Reconstruction step: execute the decision result and reallocate the control of the intelligent agent over the intersection.

2. The distributed traffic control method based on multi-agent dynamic structure reorganization according to claim 1 is characterized in that: The intelligent agents include traffic data, working mode, calculation module, communication module and decision module; The working mode includes an active mode and a passive mode; The active mode represents that the agent is controlling one or more intersections; The passive mode means that the agent loses control of the current intersection.

3. The distributed traffic control method based on multi-agent dynamic structure reorganization according to claim 1 is characterized in that: The factors that affect the degree of association between the agent and its neighboring agents include the length of the road section L, the number of lanes N, and the traffic flow F, and the calculation formulas are as follows: Where n represents the number of road segments connecting the boundaries of two agents, L A→B 、N A→B 、F A→B They represent the length of the road section L, the number of lanes N, and the traffic volume F between agents A and B respectively; The calculation formula of the degree of association between agents is as follows: Among them, α represents the degree of association between agents.

4. The distributed traffic control method based on multi-agent dynamic structure reorganization according to claim 3 is characterized in that: The length of the road section, the number of lanes, and the traffic flow between the agent and the neighboring agents are respectively the average values ​​of the length of the road section, the number of lanes, and the traffic flow of multiple road sections connecting the boundaries of the agent and the neighboring agents.

5. The distributed traffic control method based on multi-agent dynamic structure reorganization according to claim 1 is characterized in that: The decision-making steps include: Step S4.1: Select the association threshold t between two agents A and B; Step S4.2: According to the association threshold t, determine whether the two agents A and B simultaneously satisfy the following formula: α a→B >t α B→A >t If yes, then execute step S4.3; if no, then the decision result is split, where α A→B , α B→A They represent the degree of association between agent A and agent B, and the degree of association between agent B and agent A respectively; Step S4.3: Further determine whether the two agents A and B satisfy the following formula: If yes, the decision result is merge; if no, the decision result is no operation. a→B is the traffic flow from agent A to agent B, F B→A is the traffic flow from agent B to agent A.

6. The distributed traffic control method based on multi-agent dynamic structure reorganization according to claim 1 is characterized in that: The selection of the correlation threshold t is related to the interval time of decision-making. The time of decision-making is adjusted according to actual needs. Under different decision-making times, the corresponding correlation threshold t is different.

7. The distributed traffic control method based on multi-agent dynamic structure reorganization according to claim 5 is characterized in that: Agents that need to be merged are placed in the merge candidates, and agents that need to be split are placed in the split candidates.

8. The distributed traffic control method based on multi-agent dynamic structure reorganization according to claim 1 is characterized in that: The agent merging is a two-way process, which requires communication between the two parties and a joint decision; The agent splitting is a one-way process. If a neighboring agent is among the splitting candidates, it will be split directly.

9. The distributed traffic control method based on multi-agent dynamic structure reorganization according to claim 8 is characterized in that: The agent merging is to concentrate the control rights of multiple intersections on the agent at a certain intersection. The agent at the certain intersection is the core agent. The core agent is set to an active working mode, and the other agents are set to a passive working mode and obey the control of the core agent, thereby realizing coordinated control of the entire sub-area. The core agent is selected on a first-come, first-served basis.

10. A distributed traffic control system based on multi-agent dynamic structure reorganization, characterized in that: include: Initialization module: Initialize the traffic light controller at each intersection of the road network into an intelligent agent based on the acquired road network data; Interaction module: the agent exchanges information with neighboring agents; Calculation module: The agent independently calculates the road condition information it obtains and obtains the correlation value with each neighboring agent; Decision-making module: the agent makes a decision according to the correlation degree value to obtain a corresponding decision result; Reconstruction module: execute the decision result and reallocate the control of the intelligent agent over the intersection.

Images