Heuristic vehicle path planning method applied to microscopic traffic simulation engine

Abstract

The invention discloses a heuristic vehicle path planning method applied to a microscopic traffic simulation engine. The heuristic vehicle path planning method comprises the following steps: constructing a road network road weight model; generating a vehicle in the road network, planning an initial path for the vehicle according to the starting point and the ending point, and presetting a simulation step length; a vehicle runs according to an initial path, dynamic path planning is carried out every time when the time reaches a simulation step length, and if the path needs to be planned again, a heuristic vehicle path planning method is adopted. The invention provides a new heuristic re-routing triggering mechanism which is combined with factors such as an OD matrix, an intersection congestion index, a road congestion index, a road congestion index threshold, an intersection congestion index threshold and the like to solve the problem of vehicle path re-planning in the urban region management process. The flexibility, authenticity and effectiveness of the vehicle in the microscopic simulation process are improved, a new hybrid heuristic function is provided, and the suitability of path planning of the vehicle under the multi-source environment condition in the simulation process is improved.

Description

Technical Field

[0001] The invention relates to a vehicle path planning method, and in particular to a heuristic vehicle path planning method applied to a microscopic traffic simulation engine. Background Art

[0002] In the process of urban traffic area management, when vehicles are congested, vehicle paths will be replanned to manage urban area traffic flow. There are many methods for vehicle path planning, such as graph search algorithms, artificial potential field algorithms and intelligent optimization algorithms. Graph search algorithms include Dijkstra algorithm, Floyd algorithm, A* algorithm, etc.; intelligent optimization algorithms include genetic algorithms, ant colony algorithms, particle swarm algorithms, etc.

[0003] Dijkstra's algorithm is an efficient graph theory algorithm that is specifically designed to compute the shortest path from a single source to all other vertices in a weighted directed graph. The algorithm is based on a greedy strategy and gradually builds a shortest path tree through an iterative process. Dijkstra's algorithm is suitable for graphs without negative weight edges because it relies on the cumulative property of edge weights to ensure monotonically increasing path lengths.

[0004] As an algorithm based on the principle of dynamic programming, the core goal of Floyd's algorithm is to solve the shortest path problem between all pairs of vertices in a weighted graph. The core strategy of this algorithm is to optimize and update the existing shortest path estimate by iteratively introducing intermediate vertices. Floyd's algorithm is applicable to directed and undirected graphs, and can handle cases with negative weight edges, but the premise is that there is no negative weight cycle in the graph.

[0005] The A* algorithm (A star algorithm) is an efficient heuristic search strategy that combines the completeness of breadth-first search and the efficiency of greedy best-first search. The core of the A* algorithm lies in its evaluation function, which consists of two parts: actual cost and heuristic estimate. Among them, represents the actual path length from the starting point to the current node, and is a heuristic function used to estimate the shortest path length from the node to the end point. This heuristic function is usually designed based on problem domain knowledge. For example, in path planning, it can be Euclidean distance or Manhattan distance. The A* algorithm searches for paths by preferentially expanding those nodes with the lowest values, thereby significantly reducing the search space and improving search efficiency while ensuring that the shortest path is found. When the heuristic function meets conditions such as being non-negative and admissible, the A* algorithm can guarantee to find the shortest path. Heuristic function As a manifestation of the A* algorithm heuristic, it enables the algorithm to find the optimal path with less cost, where the estimated distance Commonly used are Euclidean distance, Manhattan distance, diagonal distance and Chebyshev distance, such as the following four methods: , , , , in, , is the x-axis coordinate and y-axis coordinate of the current node, , are the x-axis and y-axis coordinates of the end point.

[0006] The application of artificial potential field algorithms in the field of transportation is mainly reflected in the simulation, optimization and control of traffic flow. The algorithm simulates and predicts the dynamic behavior of traffic flow by simulating the interaction forces between vehicles, thereby providing decision support for traffic management and planning. In traffic flow simulation, each vehicle can be regarded as a particle, and its movement on the road is affected by the vehicle in front (repulsion) and the destination (gravity). Repulsion ensures that the vehicle maintains a safe distance and avoids collision, while gravity guides the vehicle towards the destination. In addition, artificial potential field algorithms can also be applied to ITS for the control and optimization of dynamic traffic flow. For example, in traffic signal control, the algorithm can be used to adjust the timing of traffic lights to reduce the waiting time at the intersection and improve the traffic capacity of the road. In the path planning of emergency vehicles, the algorithm can help emergency vehicles find the fastest path to the destination while minimizing the impact on other vehicles.

[0007] Genetic algorithm is an optimization technique inspired by biological evolution theory. Its application in the field of traffic system optimization has shown its potential as an efficient search strategy, especially in solving complex traffic flow distribution and traffic signal control problems. By encoding the flow of vehicles in the traffic network to form "chromosomes" representing different traffic configurations, genetic algorithms iteratively optimize traffic flow by simulating the mechanisms of natural selection, such as elite strategies, crossover and mutation. The fitness function usually evaluates the performance of each solution based on indicators such as traffic flow, delay or energy consumption.

[0008] The ant colony algorithm is a meta-heuristic optimization algorithm inspired by the foraging behavior of ants in nature. In the field of transportation system optimization, the ant colony algorithm is used to solve problems such as traffic flow distribution, route selection, and traffic signal control. The algorithm simulates the positive feedback mechanism of ants in the process of finding food: ants leave pheromones on the path, and other ants tend to walk along the path with higher pheromone concentration, thus gradually forming the optimal path. In the traffic network, the pheromone concentration of each path is positively correlated with the number of vehicles passing through the path, while the travel time of the path is negatively correlated with it. Through an iterative process, the algorithm can learn efficient paths that reduce congestion and travel time.

[0009] Particle swarm optimization is an optimization technology based on swarm intelligence. In the field of traffic system optimization, particle swarm optimization is more effective in solving problems such as traffic flow distribution, traffic signal control and path planning. The algorithm simulates the search behavior of a group of particles in the solution space, where each particle represents a potential solution and updates its flight direction and speed by tracking the historical best position of individuals and groups. In the traffic network, the position of the particles encodes the allocation of vehicles or signal timing, while the update of speed reflects the optimization process based on fitness functions such as traffic flow, delay or energy consumption.

[0010] However, these algorithms have some problems, such as: Dijkstra algorithm is not suitable for finding the shortest path between all vertex pairs, nor is it suitable for handling negative weight edges. Artificial potential field algorithms may fall into local minima, that is, the robot may stop moving at a point that is not the global optimal, especially in complex environments. Genetic algorithms may require a large number of iterations to find the optimal solution or the approximate optimal solution, which makes it less fast than some traditional optimization algorithms in terms of convergence speed. The performance of ant colony algorithms is highly dependent on parameter settings, such as crossover rate, mutation rate, and population size. The selection of these parameters has a significant impact on the effectiveness of the algorithm. Summary of the invention

[0011] The purpose of the present invention is to provide a heuristic vehicle path planning method applied to a microscopic traffic simulation engine to solve the problem that the above-mentioned path planning algorithms have poor effects in the process of microscopic traffic simulation.

[0012] In order to achieve the above-mentioned object, the technical solution adopted by the present invention is as follows: a heuristic vehicle path planning method applied to a microscopic traffic simulation engine, applied to a road network traffic simulation, wherein the road network includes a plurality of roads and a plurality of intersections, comprises the following steps; S1, constructing a road network road weight model to generate a road congestion index for each road and an intersection congestion index for each intersection, and presetting a road congestion index threshold ThR and an intersection congestion index threshold ThC; Construct a road congestion index array and an intersection congestion index array, wherein the road congestion index array is composed of the road congestion indexes of all roads, and the intersection congestion index array is composed of the congestion indexes of all intersections in the road network; S2, generating vehicles in the road network, planning the initial path for the vehicles according to the starting point and the end point, and presetting the simulation step length; S3, the vehicle travels along the initial path, and performs dynamic path planning every time the time reaches a simulation step. The dynamic path planning in the t-th simulation step includes steps S31 to S36; S31, obtaining the current position of the vehicle, the road congestion index array at the current moment, the intersection congestion index array and the OD matrix; S32, determining whether the current OD matrix has changed compared to the OD matrix in the t-1th dynamic path planning, and if so, replanning the path to generate the tth simulation path, and taking the next intersection of the current position as the relevant intersection, and all roads from the current position to the end point as the relevant roads; S33, comparing the intersection congestion index of the relevant intersection in the t-th simulation path and the t-1th simulation path to see if it has changed, if so, proceeding to step S34, otherwise, taking the t-1th simulation path as the t-th simulation path; S34, determining whether the intersection congestion index of the relevant intersection exceeds the intersection congestion index threshold ThC, if so, proceeding to step S35, otherwise, taking the t-1th simulation path as the tth simulation path; S35, determining whether the road congestion index of the relevant roads has changed, if at least one relevant road has changed, proceeding to step S36, otherwise, taking the t-1th simulation path as the tth simulation path; S36, determining whether the road congestion index of the relevant road exceeds the road congestion index threshold ThR; If at least one relevant road exceeds, the path is replanned, otherwise the t-1th simulation path is used as the tth simulation path.

[0013] As a preferred embodiment: In S1, each road and each intersection in the road network is numbered, where the road congestion index of the i-th road is , intersection congestion index of the jth intersection According to the following formula: , In the formula, The actual travel time, is the free flow time; For the the length of the road; For the Real-time traffic speed of roads, is the free flow speed; , The j-th intersection corresponds to n entrance roads, and the k-th entrance road is , the actual road speed is , 1≤k≤n.

[0014] Preferably, in S32 and S36, a heuristic vehicle path planning method is used to replan the path, and the heuristic vehicle path planning method includes the following steps: a1, select the A-star algorithm, the heuristic function of the A-star algorithm is h(n); a2, construct a mixed heuristic function ; , In the formula, , is the x-axis coordinate and y-axis coordinate of the current node, , is the x-axis coordinate and y-axis coordinate of the parent node of the current node. , are the x-axis and y-axis coordinates of the end point, T is the distance from the starting point to the end point, is the road congestion index of the next road at the current location in the t-1th simulation path, is the intersection congestion index of the next intersection of the current position in the t-1th simulation path; a3, using a mixed heuristic function Replace the heuristic function in the A-star algorithm with h(n) to obtain a heuristic vehicle path planning method.

[0015] About OD matrix: OD matrix is ​​a table of the number of travel exchanges between all origins (Origin) and destinations (Destination) in the transportation network, reflecting the basic needs of users for the transportation network and the spatial distribution of traffic in the road network. The dynamic OD matrix reflects the time-varying traffic demand between each OD pair (corresponding to each specific origin-destination pair in the transportation network) within a specific period of time. In the intelligent transportation system (ITS), the dynamic OD matrix is ​​important data in the traveler information system (ATIS) and the traffic management system (ATMS). It is also the basic data of the dynamic traffic allocation model and some micro traffic simulators. It directly affects the real-time effectiveness of the ITS. There are several estimation methods for the OD matrix, such as the static matrix estimation method and the dynamic matrix estimation method. The present invention does not improve them and only uses the matrix.

[0016] Compared with the prior art, the advantages of the present invention are: (1) A new heuristic rerouting trigger mechanism is proposed, which combines factors such as OD matrix, intersection congestion index, road congestion index, road congestion index threshold ThR, and intersection congestion index threshold ThC to solve the problem of vehicle path replanning in the process of urban area governance. It improves the flexibility, authenticity, and effectiveness of vehicles in the micro-simulation process.

[0017] (2) The road congestion index and the intersection congestion index are combined with the heuristic function of the A-star algorithm to obtain a new hybrid heuristic function ,This method improves the adaptability of the vehicle path planning under ,multi-source environment conditions during the simulation process. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a flow chart of the present invention; Figure 2 The road network of the map used in Example 2; Figure 3 For Figure 2 Traffic flow statistics of the area shown in the figure after 30 minutes of short-term traffic simulation; Figure 4 For Figure 2 Traffic flow statistics of the area shown in the figure after 60 minutes of short-term traffic simulation; Figure 5 For Figure 2 The results of the checkpoint congestion coefficient for a 30-minute short-term traffic simulation in the area shown; Figure 6 For Figure 2 The congestion coefficient results of the checkpoint in the shown area after a 60-minute short-term traffic simulation. DETAILED DESCRIPTION

[0019] The present invention will be further described below in conjunction with the accompanying drawings.

[0020] Example 1: See Figure 1 , a heuristic vehicle path planning method applied to a microscopic traffic simulation engine, applied to a road network traffic simulation, the road network comprising a plurality of roads and a plurality of intersections, comprising the following steps; S1, constructing a road network road weight model to generate a road congestion index for each road and an intersection congestion index for each intersection, and presetting a road congestion index threshold ThR and an intersection congestion index threshold ThC; Construct a road congestion index array and an intersection congestion index array, wherein the road congestion index array is composed of the road congestion indexes of all roads, and the intersection congestion index array is composed of the congestion indexes of all intersections in the road network; S2, generating vehicles in the road network, planning the initial path for the vehicles according to the starting point and the end point, and presetting the simulation step length; S3, the vehicle travels along the initial path, and performs dynamic path planning every time the time reaches a simulation step. The dynamic path planning in the t-th simulation step includes steps S31 to S36; S31, obtaining the current position of the vehicle, the road congestion index array at the current moment, the intersection congestion index array and the OD matrix; S32, determining whether the current OD matrix has changed compared to the OD matrix in the t-1th dynamic path planning, and if so, replanning the path to generate the tth simulation path, and taking the next intersection of the current position as the relevant intersection, and all roads from the current position to the end point as the relevant roads; S33, comparing the intersection congestion index of the relevant intersection in the t-th simulation path and the t-1th simulation path to see if it has changed, if so, proceeding to step S34, otherwise, taking the t-1th simulation path as the t-th simulation path; S34, determining whether the intersection congestion index of the relevant intersection exceeds the intersection congestion index threshold ThC, if so, proceeding to step S35, otherwise, taking the t-1th simulation path as the tth simulation path; S35, determining whether the road congestion index of the relevant roads has changed, if at least one relevant road has changed, proceeding to step S36, otherwise, taking the t-1th simulation path as the tth simulation path; S36, determining whether the road congestion index of the relevant road exceeds the road congestion index threshold ThR; If at least one relevant road exceeds, the path is replanned, otherwise the t-1th simulation path is used as the tth simulation path.

[0021] In S1, each road and each intersection in the road network is labeled, where the road congestion index of the i-th road is , intersection congestion index of the jth intersection According to the following formula: , In the formula, The actual travel time, is the free flow time; For the the length of the road; For the Real-time traffic speed of roads, is the free flow speed; , The j-th intersection corresponds to n entrance roads, and the k-th entrance road is , the actual road speed is , 1≤k≤n.

[0022] In S32 and S36, a heuristic vehicle path planning method is used to replan the path, and the heuristic vehicle path planning method includes the following steps; a1, select the A-star algorithm, the heuristic function of the A-star algorithm is h(n); a2, construct a mixed heuristic function ; , In the formula, , is the x-axis coordinate and y-axis coordinate of the current node, , is the x-axis coordinate and y-axis coordinate of the parent node of the current node. , are the x-axis and y-axis coordinates of the end point, T is the distance from the starting point to the end point, is the road congestion index of the next road at the current location in the t-1th simulation path, is the intersection congestion index of the next intersection of the current position in the t-1th simulation path; a3, using a mixed heuristic function Replace the heuristic function in the A-star algorithm with h(n) to obtain a heuristic vehicle path planning method.

[0023] Example 2: See Figures 2 to 6 In order to illustrate the effect of the present invention, the Anhui Xuancheng open data set is used to provide the experiment with checkpoint data, signal point data, road section travel time, and intersection lane traffic flow data. This experiment will use the traffic road network formed by Zhongshan Road, Zhuangyuan Road, Lingxi Road, Jincheng Road, and Diezhang Road combined with the heuristic vehicle path planning algorithm to verify the optimization of urban area road network management. The figure includes 6 intersections, namely HK-101, HK-103, HK-104, HK-96, HK-95, and HK-94. In this embodiment, the intersection is also called a checkpoint.

[0024] This experiment uses real data to compare with three sets of microscopic traffic simulation results, which are derived from the following three methods.

[0025] Method 1: The original microscopic traffic simulation results are called Original Simulation in English, abbreviated as OrS.

[0026] Method 2: Micro-traffic simulation results optimized by digital twin technology, called Optimized Simulation, abbreviated as OpS.

[0027] Method 3: The method of the present invention is an optimized microscopic traffic simulation result combined with a heuristic vehicle path planning algorithm, which is called Heuristic Simulation in English, abbreviated as HS.

[0028] right Figure 2 The traffic flow statistics of the urban area shown in the figure are as follows: Figure 3As shown in the figure, the initial OD matrix is ​​used in OpS to monitor the traffic flow data of 6 checkpoints in the simulated road network. Real Data is real data, showing the real data of traffic flow at 6 checkpoints in Xuancheng, Anhui at 8:30. Comparing the checkpoint traffic flow data of Real Data, OrS and OpS, it is shown that the authenticity of the micro-traffic simulation engine after optimization with digital twin technology is significantly improved, and the default simulation path planning algorithm can more accurately describe the real traffic flow under the simulated road network.

[0029] right Figure 2 The results of a 60-minute long-term traffic simulation in the urban area shown in the figure are as follows: Figure 4 As shown in the figure, the predicted OD matrix is ​​used in OpS. It can be seen that there is little difference between OpS and Real Data data, whether it is short-term or long-term traffic simulation. After using HS, the traffic volume of the six checkpoints has decreased, indicating that the total number of vehicles in the simulated road network has decreased. Compared with OpS, HS can better plan the vehicle's driving path and enable vehicles to leave the simulated road network in the urban area faster at the same time of simulation.

[0030] right Figure 2 The results of the checkpoint congestion coefficient for a 30-minute short-term traffic simulation in the area shown are as follows: Figure 5 As shown in the figure, by comparing the congestion coefficients of Real Data and OrS, it can be seen that the OrS simulation process was severely congested after 30 minutes, and the congestion coefficients of the six checkpoints were very high, while Real Data showed that only HK-101 and HK-104 were obviously congested. It can be seen that when using the original simulation engine for traffic simulation, the simulation lacks authenticity. After simulation using OpS, the congestion coefficient of OpS is close to the congestion coefficient of Real Data, indicating that the authenticity of OpS simulation of urban areas has been significantly improved. From the HS simulation results, it can be seen that compared with Real Data, except for the HK-104 checkpoint, the congestion coefficient of the other five checkpoints has decreased significantly.

[0031] right Figure 2 The results of the checkpoint congestion coefficient for a 60-minute short-term traffic simulation in the area shown are as follows: Figure 6 As shown in the figure, it can be seen that the congestion coefficients of the six checkpoints in the HS simulation results have decreased significantly. The decrease in the intersection congestion coefficient indirectly shows that the road congestion coefficient has decreased. For the urban area road network, the total number of vehicles in the road network has decreased, and the overall road network has been optimized. The path planned by the heuristic vehicle path planning algorithm for the vehicle can effectively improve the global efficiency of the urban area road network, and the planned driving path can provide effective reference and strong support for the actual vehicle journey.

[0032] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A heuristic vehicle path planning method applied to a microscopic traffic simulation engine, applied to a road network traffic simulation, wherein the road network includes a plurality of roads and a plurality of intersections, characterized in that: The steps include: S1, constructing a road network road weight model to generate a road congestion index for each road and an intersection congestion index for each intersection, and presetting a road congestion index threshold ThR and an intersection congestion index threshold ThC; Construct a road congestion index array and an intersection congestion index array, wherein the road congestion index array is composed of the road congestion indexes of all roads, and the intersection congestion index array is composed of the congestion indexes of all intersections in the road network; S2, generating vehicles in the road network, planning the initial path for the vehicles according to the starting point and the end point, and presetting the simulation step length; S3, the vehicle travels along the initial path, and performs dynamic path planning every time the time reaches a simulation step. The dynamic path planning in the t-th simulation step includes steps S31 to S36; S31, obtaining the current position of the vehicle, the road congestion index array at the current moment, the intersection congestion index array and the OD matrix; S32, determining whether the current OD matrix has changed compared to the OD matrix in the t-1th dynamic path planning, and if so, replanning the path to generate the tth simulation path, and taking the next intersection of the current position as the relevant intersection, and all roads from the current position to the end point as the relevant roads; S33, comparing the intersection congestion index of the relevant intersection in the t-th simulation path and the t-1th simulation path to see if it has changed, if so, proceeding to step S34, otherwise, taking the t-1th simulation path as the t-th simulation path; S34, determining whether the intersection congestion index of the relevant intersection exceeds the intersection congestion index threshold ThC, if so, proceeding to step S35, otherwise, taking the t-1th simulation path as the tth simulation path; S35, determining whether the road congestion index of the relevant roads has changed, if at least one relevant road has changed, proceeding to step S36, otherwise, taking the t-1th simulation path as the tth simulation path; S36, determining whether the road congestion index of the relevant road exceeds the road congestion index threshold ThR; If at least one relevant road exceeds, the path is replanned, otherwise the t-1th simulation path is used as the tth simulation path.

2. The heuristic vehicle path planning method applied to a microscopic traffic simulation engine according to claim 1, characterized in that: In S1, each road and each intersection in the road network is labeled, where the road congestion index of the i-th road is , intersection congestion index of the jth intersection According to the following formula: ， In the formula, The actual travel time, is the free flow time; For the the length of the road; For the Real-time traffic speed of roads, is the free flow speed; ， The j-th intersection corresponds to n entrance roads, and the k-th entrance road is , the actual road speed is , 1≤k≤n.

3. The heuristic vehicle path planning method applied to a microscopic traffic simulation engine according to claim 1, characterized in that: In S32 and S36, a heuristic vehicle path planning method is used to replan the path, and the heuristic vehicle path planning method includes the following steps; a1, select the A-star algorithm, the heuristic function of the A-star algorithm is h(n); a2, construct a mixed heuristic function ; ， In the formula, , is the x-axis coordinate and y-axis coordinate of the current node, , is the x-axis coordinate and y-axis coordinate of the parent node of the current node. , are the x-axis and y-axis coordinates of the end point, T is the distance from the starting point to the end point, is the road congestion index of the next road at the current location in the t-1th simulation path, is the intersection congestion index of the next intersection of the current position in the t-1th simulation path; a3, using a mixed heuristic function Replace the heuristic function in the A-star algorithm with h(n) to obtain a heuristic vehicle path planning method.

Images