A genetic programming hyper-heuristic traffic flow assignment method based on knowledge transfer

Abstract

The present invention provides a genetic programming hyper-heuristic traffic flow allocation method based on knowledge transfer, comprising the following steps: obtaining road network information; abstracting the road network into a weighted directed graph; constructing a traffic simulation model based on the road network information and determining an optimization target; evaluating each routing strategy in the population using the traffic simulation model; extracting knowledge based on the source domain; and when the problem scenario changes, iteratively optimizing the target domain based on the knowledge learned from the source domain, and applying the optimal individual as the final routing strategy to traffic flow allocation. The present invention effectively solves the problem of the effectiveness of routing strategies for traffic flow allocation when the road network structure changes, while improving the efficiency of retraining in new scenarios. By applying transfer learning to traffic flow optimization, the general knowledge learned in the source domain is transferred to the target domain, which can significantly improve the efficiency of retraining under the new road network structure, is applicable to traffic flow allocation strategies of different types of road network structures, and improves the generalization ability of routing strategies.

Description

Technical Field

[0001] The present invention relates to the technical field of traffic optimization, in particular to a genetic programming hyper-heuristic traffic flow allocation method based on knowledge migration. Background Art

[0002] With the accelerating pace of urbanization and the continued growth of car ownership among urban residents, the contradiction between urban transportation demand and supply has become increasingly prominent. With increasingly complex urban road networks and increasing traffic volume, travel route selection, as an important component of transportation decision-making, is the most frequent and common problem faced by drivers.

[0003] Traffic flow allocation has become a critical issue in the field of intelligent transportation. With the rapid development of artificial intelligence (AI), the interconnectedness of vehicle and road information within the road network and the travel demands across the entire network have formed a complex swarm intelligence system. Traditional static traffic flow allocation schemes are no longer effective and struggle to cope with complex traffic scenarios. For example, it is difficult to achieve efficient traffic flow routing and allocation in environments with dynamic traffic demand and a changing road network environment.

[0004] Therefore, most of the current research on traffic flow allocation methods is based on intelligent optimization algorithms. In addition to traditional meta-heuristic algorithms, hyper-heuristic algorithms provide new ideas for solving traffic flow allocation problems and show effectiveness in automatically evolving routing strategies to cope with uncertain environments.

[0005] While existing hyper-heuristic methods achieve traffic flow allocation by optimizing the decision-making rules of individuals facing multiple roads, improving the overall traffic capacity of the road network to a certain extent, problems still exist. For example, when the routing strategy generated by genetic programming changes the road network structure, such as when it switches from a grid to a radial network, the previous routing strategy may no longer be effective and a new strategy must be retrained. Traditional hyper-heuristic algorithms have difficulty evolving traffic flow allocation strategies that are universally applicable to different types of road network structures. Summary of the Invention

[0006] In response to the shortcomings of the existing technology, the present invention provides a genetic programming hyper-heuristic traffic flow allocation method based on knowledge transfer. The present invention effectively solves the problem of the effectiveness of routing strategies for traffic flow allocation when the road network structure changes, while improving the efficiency of retraining in new scenarios.

[0007] The technical solution of the present invention is: a genetic programming hyper-heuristic traffic flow allocation method based on knowledge transfer, comprising the following steps:

[0008] S1), obtain road network information;

[0009] S2) Abstract the road network into a weighted directed graph;

[0010] S3) Constructing a traffic simulation model based on road network information and determining optimization objectives;

[0011] S4), using the constructed traffic simulation model to evaluate each routing strategy in the population;

[0012] S5), extract knowledge based on the source domain;

[0013] S6) When the problem scenario changes, the target domain is iteratively optimized based on the knowledge learned from the source domain, and the optimal individual is used as the final routing strategy for traffic flow distribution.

[0014] Preferably, in step S2), the road network is abstracted into a weighted directed graph, specifically:

[0015] S21) Obtain a node set V = {v1, v2, ..., v N}, N is the total number of intersections in the road network;

[0016] S22) Obtain edge set E according to the roads in the road network, and for element e ij ∈E represents the connection node v in the road network i and node v j An edge of the road, the length of the road is used as the attribute of the edge;

[0017] S23) According to the node set V and the edge set E, a directed graph G = (V, E) is constructed.

[0018] Preferably, in step S3), a traffic simulation model is constructed based on the road network information, specifically:

[0019] A traffic simulation model is constructed for the traffic flow distribution problem based on the location information of the intersection in the road network, whether there are traffic lights at the intersection, the length of each road in the road network, the starting and ending point information of the vehicle, and the departure time.

[0020] Preferably, in step S3), the optimization goal is to minimize the average travel time of all vehicles, that is:

[0021]

[0022] Where f is the average travel time of all vehicles; N vehicle is the total number of vehicles entering the road network during the simulation time; is the travel time of vehicle i from entering the road network until it reaches its destination and leaves the road network.

[0023] Preferably, in step S4), the constructed traffic simulation model evaluates each routing strategy in the population, specifically including:

[0024] S41), generating a vehicle at a specific time point and entering the system based on the constructed traffic simulation model and traffic flow information;

[0025] S42), based on the real-time traffic flow information in the road network, calculating the heuristic value of each optional road at the intersection ahead of the vehicle through the routing strategy, and adding the road with the highest heuristic value to the vehicle's driving route;

[0026] S43), each vehicle cyclically performs the above operations until it reaches the destination within the simulation time, and when the simulation ends, calculates the average travel time to evaluate the fitness value of each individual in the population;

[0027] S44), and use the genetic operator to iterate the population. When the maximum number of iterations is reached, the individual with the largest fitness value in the population is output, otherwise the genetic operator operation is continued on the population.

[0028] Preferably, step S42) specifically includes the following steps:

[0029] S421) Pre-processing the roads available for the vehicle under the constraint that the vehicle's driving route cannot have loops;

[0030] S422) Calculate the candidate road set S through the routing strategy road The heuristic value corresponding to each road in is selected, and the road with the largest heuristic value is added to the current vehicle's route. This step is repeated for all vehicles until the simulation ends.

[0031] Preferably, in step S5), a genetic programming-based hyper-heuristic algorithm is used to solve the source domain to perform knowledge extraction.

[0032] Preferably, in step S6), when the problem scenario changes, the probability distribution of the GP tree with excellent performance in the final population is learned from the source domain as knowledge; a good initial population is created in the target domain based on the knowledge learned from the source domain, and then the genetic operator is iterated on the population until the stopping condition is met, and the optimal individual in the population is output as the routing strategy.

[0033] The beneficial effects of the present invention are:

[0034] 1. This invention effectively solves the problem of the effectiveness of routing strategies for traffic flow distribution when the road network structure changes, while improving the efficiency of retraining in new scenarios;

[0035] 2. This invention applies transfer learning to traffic flow optimization, transferring general knowledge learned in the source domain to the target domain. This can significantly improve the efficiency of retraining under new road network structures, evolve traffic flow allocation strategies that are applicable to different types of road network structures, and improve the generalization ability of routing strategies.

[0036] 3. The present invention first solves the source problem domain. When the road network in the problem scenario changes, such as from a grid shape to a radial shape, the previously trained routing strategy may no longer be effective, and a new strategy needs to be retrained. Extracting knowledge from the solution of the source problem and migrating it to the solution of the current problem can effectively improve the solution efficiency and accuracy in the current scenario. The trained routing strategy can more effectively optimize the distribution of traffic flow under the current road network structure. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 Schematic diagram of the flow of traffic flow allocation method of the present invention;

[0038] Figure 2 Flowchart of retraining in the target domain based on source domain knowledge in an embodiment of the present invention. DETAILED DESCRIPTION

[0039] The specific embodiments of the present invention will be further described below with reference to the accompanying drawings:

[0040] The embodiments described below with reference to the accompanying drawings are exemplary and intended only to explain the present invention, and are not to be construed as limiting the present invention. The step numbers in the following embodiments are provided solely for ease of illustration and do not limit the order of the steps. The order of execution of the steps in the embodiments may be adjusted adaptively based on the understanding of those skilled in the art.

[0041] In the description of the present invention, it should be understood that descriptions involving orientations, such as up, down, front, back, left, right, etc., indicating orientations or positional relationships, are based on the orientations or positional relationships shown in the accompanying drawings. They are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation. Therefore, they cannot be understood as limitations on the present invention.

[0042] In the description of the present invention, "several" means one or more, "many" means more than two, "greater than," "less than," and "exceed" are understood to exclude the number itself, while "above," "below," and "within" are understood to include the number itself. The use of "first" and "second" in the description is solely for the purpose of distinguishing technical features and should not be construed as indicating or implying relative importance, implicitly specifying the number of the indicated technical features, or implicitly specifying the order of the indicated technical features.

[0043] In the description of the present invention, unless otherwise clearly defined, terms such as setting, installing, and connecting should be understood in a broad sense, and technicians in the relevant technical field can reasonably determine the specific meanings of the above terms in the present invention based on the specific content of the technical solution.

[0044] Example 1

[0045] like Figure 1 As shown, this embodiment provides a genetic programming hyper-heuristic traffic flow allocation method based on knowledge transfer. This embodiment is mainly used for traffic flow allocation in dynamic open scenarios when the road network structure changes. Specifically, it includes the following steps:

[0046] S1) Obtain road network information.

[0047] S2) Abstract the road network into a weighted directed graph; specifically:

[0048] S21) Obtain a node set V = {v1, v2, ..., v N}, N is the total number of intersections in the road network;

[0049] S22) Obtain edge set E according to the roads in the road network, and for element e ij ∈E represents the connection node v in the road network i and node v j An edge of the road, the length of the road is used as the attribute of the edge;

[0050] S23) According to the node set V and the edge set E, a directed graph G = (V, E) is constructed.

[0051] S3) Construct a traffic simulation model based on the road network information and determine the optimization target; specifically:

[0052] A traffic simulation model is constructed for the traffic flow distribution problem based on the location information of the intersection in the road network, whether there are traffic lights at the intersection, the length of each road in the road network, the starting and ending point information of the vehicle, and the departure time.

[0053] The optimization goal is to minimize the average travel time of all vehicles, that is:

[0054]

[0055] Where f is the average travel time of all vehicles; N vehicle is the total number of vehicles entering the road network during the simulation time; is the travel time of vehicle i from entering the road network until it reaches its destination and leaves the road network.

[0056] S4), using the constructed traffic simulation model to evaluate each routing strategy in the population;

[0057] S41) Start simulation based on the road network information and traffic flow information. For each vehicle in the road network, i When a vehicle comes to a new road, its optional roads are preprocessed to filter out the next optional road set S for the vehicle. road ;

[0058] S42), using the routing strategy to calculate the heuristic value for each road in the optional road set, selecting the road with the largest heuristic value to add to the current vehicle's driving route, and repeating this step until the destination is reached;

[0059] S43), if the simulation reaches the set maximum time, calculate the average travel time of all vehicles as the fitness value of the routing strategy;

[0060] S44) Iteratively evolve individuals in the population, including operations such as selection, crossover, and mutation, while ensuring that the population size remains unchanged;

[0061] S45) If the evolution reaches the maximum number of iterations, execute step S46) otherwise return to step S41) to start the next round of evolution;

[0062] S46) Output the optimal individual in the population and the execution ends.

[0063] S5) Extract knowledge from the source domain; when the problem scenario changes, iteratively optimize the target domain based on the knowledge learned from the source domain, and use the optimal individual as the final routing strategy for traffic flow allocation, such as Figure 2 As shown, specifically:

[0064] S51), applying a hyper-heuristic algorithm based on genetic programming on the source domain, and sampling a subset of excellent individuals in the source domain from the final population after the iteration using tournament selection;

[0065] S52), based on the sampled subset, learning the probability of each function node or terminal node in the excellent individual appearing at a given position in the heuristic rule;

[0066] S53) Generate new individuals based on the probability distribution of function nodes or terminal nodes learned from the source domain and use them as part of the initial population of the target domain. Another part of the individuals in the population are randomly initialized to obtain a good initial population, thereby improving the quality of the initial state in the target domain.

[0067] S54), return to step S4), run the hyper-heuristic algorithm based on genetic programming on the target domain for iterative solution, and after the iteration, output the optimal individual as the final routing strategy for large-scale traffic flow distribution under the current road network structure.

[0068] The above embodiments and descriptions are only for explaining the principles and best embodiments of the present invention. Without departing from the spirit and scope of the present invention, the present invention may be subject to various changes and improvements, which shall fall within the scope of the invention to be protected.

Claims

1. A genetic programming hyper-heuristic traffic flow allocation method based on knowledge transfer, characterized by: The following steps are involved: S1), obtain road network information; S2) Abstract the road network into a weighted directed graph; S3) Construct a traffic simulation model based on the road network information and determine the optimization target; specifically: A traffic simulation model is constructed for the traffic flow distribution problem based on the location information of the intersection in the road network, whether there are traffic lights at the intersection, the length of each road in the road network, the starting and ending points of the vehicles, and the departure time; The optimization goal is to minimize the average travel time of all vehicles, that is: Where f is the average travel time of all vehicles; N vehicle is the total number of vehicles entering the road network during the simulation time; is the travel time of vehicle i from entering the road network until it reaches its destination and leaves the road network; S4) Evaluate each routing strategy in the population using the constructed traffic simulation model; specifically including: S41), generating a vehicle at a specific time point and entering the system based on the constructed traffic simulation model and traffic flow information; S42), based on the real-time traffic flow information in the road network, calculating the heuristic value of each optional road at the intersection ahead of the vehicle through the routing strategy, and adding the road with the highest heuristic value to the vehicle's driving route; specifically including: S422) Calculate the candidate road set S through the routing strategy road The heuristic value corresponding to each road in is selected, and the road with the largest heuristic value is added to the current vehicle's route. This step is repeated for all vehicles until the simulation ends. S43), each vehicle cyclically performs the above operations until it reaches the destination within the simulation time, and when the simulation ends, calculates the average travel time to evaluate the fitness value of each individual in the population; S44), and using the genetic operator to iterate the population, when the maximum number of iterations is reached, the individual with the largest fitness value in the population is output, otherwise the genetic operator operation is continued on the population; S5), extract knowledge based on the source domain; Use a genetic programming-based hyper-heuristic algorithm to solve the source domain for knowledge extraction; S6) When the problem scenario changes, the target domain is iteratively optimized based on the knowledge learned from the source domain, and the optimal individual is used as the final routing strategy for traffic flow allocation; specifically: When the problem scenario changes, the probability distribution of the GP trees that perform well in the final population is learned from the source domain as knowledge; A good initial population is created in the target domain based on the knowledge learned from the source domain. Then, the genetic operator is iteratively applied to the population until the stopping condition is met, and the optimal individual in the population is output as the routing strategy.

2. The genetic programming hyper-heuristic traffic flow allocation method based on knowledge transfer according to claim 1 is characterized by: In step S2), the road network is abstracted into a weighted directed graph, specifically: S21) Obtain a node set V = {v1, v2, ..., v N }, N is the total number of intersections in the road network; S22) Obtain edge set E according to the roads in the road network, and for element e ij ∈E represents the connection node v in the road network i and node v j An edge of the road, the length of the road is used as the attribute of the edge; S23) According to the node set V and the edge set E, a directed graph G = (V, E) is constructed.

3. The genetic programming hyper-heuristic traffic flow allocation method based on knowledge transfer according to claim 1 is characterized by: Step S42) also includes: S421) Pre-processing the roads available for the vehicle under the constraint that the vehicle's driving route cannot have loops.

Images