Vehicle path dynamic adaptive planning method, system, computer equipment and storage medium

Abstract

The present invention discloses a vehicle path dynamic adaptive planning method, system, computer equipment and storage medium. The method of the present invention includes the following steps: Step 1. Establishing a dynamic loading model of the traffic flow of a section in an urban traffic network to predict the traffic flow; establishing a vehicle travel time prediction model considering the traffic status to predict the future section travel time; Step 2. Using the vehicle travel time prediction model, based on the improved ant colony algorithm, solving the path planning problem with the goal of minimizing the vehicle travel time under dynamic traffic conditions, thereby obtaining the shortest path for vehicle travel time. The method of the present invention comprehensively analyzes the propagation of traffic congestion and the changes in the traffic conditions of the section, considers the impact of the congestion propagation process on the traffic flow, establishes a dynamic loading model of the road network traffic flow from the aspects of the spatial correlation of the section and the dynamic changes in the traffic conditions, and realizes the accurate prediction of the vehicle travel time under the dynamic traffic situation.

Description

Technical Field

[0001] The present invention relates to the technical field of vehicle path planning, and in particular to a vehicle path dynamic adaptive planning method, system, computer equipment and storage medium. Background Art

[0002] Intelligent transportation systems play an important role in solving urban traffic congestion and improving the efficiency of urban road network operations. Road traffic state propagation law analysis and path planning are important research contents of intelligent transportation systems.

[0003] The analysis of the propagation law of road traffic status mainly explores the spatiotemporal distribution characteristics of traffic congestion, the spatial correlation of congestion, analyzes the changes in traffic status of bottleneck sections during congestion, and the propagation law of congestion under different traffic networks.

[0004] Path planning mainly provides vehicles with real-time optimal driving routes by analyzing traffic conditions, ensuring the shortest vehicle travel time and improving the traffic capacity of the entire traffic network. Path planning is an important research content of traffic congestion control.

[0005] At present, the path planning problem based on traffic flow is still mainly based on qualitative analysis of fixed traffic scenarios. There is a lack of research on the loading changes of road network traffic with strong correlation of real dynamic traffic flow. Therefore, it is impossible to quickly and accurately predict the actual road traffic status and obtain the optimal path planning in a dynamic traffic environment. Summary of the invention

[0006] The purpose of the present invention is to propose a vehicle path dynamic adaptive planning method to achieve accurate prediction of vehicle travel time under dynamic traffic conditions, thereby obtaining the optimal path planning under dynamic traffic conditions.

[0007] In order to achieve the above object, the present invention adopts the following technical scheme:

[0008] A vehicle path dynamic adaptive planning method comprises the following steps:

[0009] Step 1. First, a dynamic loading model of road traffic in the urban traffic network is established to predict traffic flow; then a vehicle travel time prediction model considering traffic conditions is established to predict future road section travel time;

[0010] Step 2. Using the vehicle travel time prediction model, based on the improved ant colony algorithm, the path planning problem with the goal of minimizing the vehicle travel time under dynamic traffic conditions is solved to obtain the path with the shortest vehicle travel time.

[0011] In addition, based on the above-mentioned vehicle path dynamic adaptive planning method, the present invention also proposes a corresponding vehicle path dynamic adaptive planning system, and its technical solution is as follows:

[0012] A vehicle path dynamic adaptive planning system, comprising:

[0013] The vehicle travel time prediction model establishment module is used to establish a dynamic loading model of road section traffic in the urban traffic road network to predict traffic flow, and then establish a vehicle travel time prediction model considering traffic conditions to predict future road section travel time;

[0014] The vehicle travel time shortest path solving module is used to solve the path planning problem with the goal of minimizing vehicle travel time under dynamic traffic conditions based on the vehicle travel time prediction model and the improved ant colony algorithm to obtain the vehicle travel time shortest path.

[0015] In addition, based on the above-mentioned vehicle path dynamic adaptive planning method, the present invention also proposes a computer device for implementing the above-mentioned vehicle path dynamic adaptive planning method.

[0016] The computer device includes a memory and a processor. The memory stores executable codes. When the processor executes the executable codes, the steps of the vehicle path dynamic adaptive planning method mentioned above are implemented.

[0017] In addition, based on the above-mentioned vehicle path dynamic adaptive planning method, the present invention also proposes a computer-readable storage medium for implementing the above-mentioned vehicle path dynamic adaptive planning method.

[0018] The computer-readable storage medium stores a program thereon, and when the program is executed by a processor, it is used to implement the steps of the vehicle path dynamic adaptive planning method mentioned above.

[0019] The present invention has the following advantages:

[0020] As described above, the present invention proposes a vehicle path dynamic adaptive planning method for the vehicle driving path planning problem under dynamic traffic conditions. The method comprehensively analyzes the propagation of traffic congestion and the changes in the traffic conditions of the road sections. By considering the impact of the congestion propagation process on the traffic flow, a dynamic loading model of the road network traffic flow is established from the aspects of the spatial correlation of the road sections and the dynamic changes in the traffic conditions, and the accurate prediction of the vehicle travel time under the dynamic traffic situation is achieved. In addition, the present invention also improves the ant colony algorithm and applies the algorithm to the problem of solving the shortest path between two points, inducing the vehicle to generate the optimal path that is adaptive to the current traffic flow, reducing time delays, and finally verifying the effectiveness of the method through experiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 A schematic diagram of a process flow of a vehicle path dynamic adaptive planning method according to an embodiment of the present invention;

[0022] Figure 2 This is a simplified diagram of the traffic phase relationship of the urban road network;

[0023] Figure 3 It is a classification map of spatially related road sections;

[0024] Figure 4 It is the road section capacity change diagram;

[0025] Figure 5 This is a schematic diagram of the principle of angle selection;

[0026] Figure 6 A flowchart of solving the path planning problem based on the improved ant colony algorithm;

[0027] Figure 7 This is a diagram of buffer setting in the experiment of the present invention;

[0028] Figure 8 It is a schematic diagram of all positioning points in the experiment of the present invention;

[0029] Fig. 9 This is a schematic diagram of the positioning points screened according to the buffer zone in the experiment of the present invention;

[0030] Fig.10 This is the result diagram of association rule mining in the experiment of the present invention;

[0031] Fig.11 is the traffic flow diagram of the initial road section in the experiment of the present invention;

[0032] Fig.12 This is a comparison chart of algorithm performance under different traffic flow conditions in the experiment of the present invention;

[0033] Fig.13 This is a schematic diagram of the traditional algorithm path planning;

[0034] Fig.14 Schematic diagram of path planning based on the improved algorithm. DETAILED DESCRIPTION

[0035] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments:

[0036] like Figure 1 As shown, this embodiment describes a vehicle path dynamic adaptive planning method, which includes the following steps:

[0037] Step 1. First, a dynamic loading model of road traffic in the urban traffic network is established to predict traffic flow; then a vehicle travel time prediction model considering traffic conditions is established to predict future road section travel time.

[0038] The process of establishing a dynamic loading model for road traffic is as follows:

[0039] Step 1.1.1. Abstract the urban traffic road network into a traffic network composed of nodes and edges. Under the topological structure of this traffic network, abstract the vehicle flow into traffic flow; where nodes are intersections and edges are road sections.

[0040] Since the speed and density of vehicle traffic are constantly changing, the traffic conditions of each road section are also constantly changing and interacting with each other, such as Figure 2 This is a simplified schematic diagram of traffic phase relationships in an urban road network.

[0041] Should Figure 2 A simplified urban road network traffic flow diagram is given.

[0042] Arrows are used in the figure to represent traffic flow in the traffic network. Each arrow in the figure has been marked with a serial number. The direction of the arrow represents the direction of traffic flow, the size of the arrow represents the size of traffic flow, and the line type is used to distinguish the direction of traffic flow.

[0043] The present invention utilizes the FP-Growth algorithm to analyze and calculate the spatial correlation of traffic flow by mining historical traffic data.

[0044] Step 1.1.2. Calculate the association rules between the sections of the floating vehicle's driving trajectory based on the FP-Growth algorithm, use the confidence of the association rules as the correlation coefficient of the spatial correlation of the traffic flow, and construct the traffic transfer probability matrix.

[0045] The specific steps of step 1.1.2 are as follows:

[0046] Step I.1 quantifies the impact and degree of impact between related road sections by establishing a road section spatial correlation matrix.

[0047] The road section spatial correlation matrix is ​​used to represent the size of the interaction between traffic flows on each road section. The vehicle trajectory can reflect the path of the vehicle, and express the source, destination, and size of the interaction flow of the vehicle.

[0048] The frequent item sets obtained by FP-Growth data mining method are used to obtain the confidence of the association rules between the two road sections. The confidence is used as the element of the spatial correlation matrix. Then the road section spatial correlation matrix W ij As shown in formula (1).

[0049]

[0050] In formula (1), row numbers i1, i2, i3…i m and column numbers j1, j2, j3…j m Indicates the road section number, w ij represents the spatial correlation between road sections i and j, where w ij With w ji are not equal, that is, the spatial weight matrix is ​​an asymmetric matrix.

[0051] Step I.2. Construct an adjacency matrix A to describe the adjacency relationship of road segments at the spatial level, as shown in formula (2). The adjacency matrix A is used to describe the adjacency relationship of road segments in the traffic network.

[0052]

[0053] In formula (2), row numbers i1, i2, i3…i m and column numbers j1, j2, j3…j m Represents the road segment number. The elements in the adjacency matrix A are represented by 0 and 1, where element 0 represents non-adjacency and element 1 represents adjacency.

[0054] Step I.3. Construct a reachability matrix to describe the connectivity between road segments.

[0055] Since the road sections are interconnected, even if two road sections are not adjacent in space, vehicles can still reach each other through several road sections. The connectivity between road sections is reflected by constructing a reachable matrix R.

[0056] The operation of the reachability matrix R is obtained through Boolean operations on the adjacency matrix A and the identity matrix E.

[0057] The subsequent calculation of the probability of selecting a road section needs to consider which sections the current road section is connected to (i.e., reachable). Although the subsequent calculation of traffic only considers the adjacent sections, the use of the reachability matrix can be used for unified calculation, which is more convenient.

[0058] like Figure 3 As shown, lines of different line types represent adjacent road segments with gradually increasing orders. Through simple calculation, it can be known that the number of k-order road segments adjacent to the initial road segment is m=6*k-3.

[0059] Step I.4. Construct the traffic transfer probability matrix.

[0060] Through the link connectivity relationship, the matrix and link spatial correlation matrix can be reached. The spatial correlation of adjacent links of the same order in the link spatial weight matrix is ​​normalized, and then the error processing is performed to obtain the flow transfer probability matrix.

[0061] The specific steps of step I.4 are:

[0062] First, the probability of road section selection is obtained, and the calculation formula is formula (3):

[0063]

[0064] Among them, U ij is the probability of road segment selection, w ig Represents the spatial weight of the segment g with the same order as the segment j among all the segments relative to the segment i. Since the segment selection probability tends to be centralized, the error coefficient E is obtained by comparing the segment selection probability with the average selection probability. ij , the calculation formula is formula (4):

[0065]

[0066] Among them, k represents the reachable order, represents the average selection probability.

[0067] Because the more accessible levels there are, the more optional sections there will be. From a macro perspective, the more sections there are, the probability of each section being selected will gradually approach, so it is called the average selection probability.

[0068] That is, the actual road section selection probability R ij for:

[0069] R ij =U ij +E ij (5)

[0070] Thus, the flow transfer probability matrix p is obtained. ij As shown in formula (6):

[0071]

[0072] Among them, R ig Represents the spatial weight of link g with the same order as link j among all links relative to link i.

[0073] Step 1.1.3. Establish a dynamic traffic loading model for a road section, specifically including establishing a dynamic traffic loading model for a congested road section, establishing a traffic loading model for an upstream road section, and establishing a traffic loading model for a downstream road section.

[0074] To simplify the problem, it is assumed that the time it takes for a vehicle to pass through an intersection is constant. The present invention only considers the time it takes for a vehicle to travel on a road section. The time it takes to pass through all road sections is t a The calculation formula is shown in formula (7):

[0075]

[0076] In the formula, t i is the time spent passing through road segment i, and n is the total number of road segments passed during the vehicle journey.

[0077] In vehicle travel prediction, it is necessary to calculate the changes in traffic flow and make real-time judgment and analysis on the traffic conditions of the changed road sections, as follows:

[0078] a. When the traffic flow is less than the actual traffic capacity of the road section, vehicles can pass through the current road section quickly and smoothly, and the current road section is determined to be unblocked;

[0079] b. When the traffic flow of a road section is greater than the actual traffic capacity of the road section, vehicles cannot pass through the road section smoothly and quickly, and need to slow down and queue up, resulting in congestion. The current road section is determined to be in a congested state.

[0080] Step II.1. Establish a dynamic traffic loading model for congested road sections.

[0081] When a road section enters a congested state, a bottleneck point of traffic capacity will be formed, which can be used to divide the road section into three specific sections, namely, the upstream section L a 、Bottleneck section L b , Downstream section L c .

[0082] like Figure 3 As shown, it is assumed that congestion begins to occur at time T0.

[0083] For the upstream section L a , the input traffic flow is The road traffic volume is After the congestion occurs, the road section L a Actual traffic capacity From the initial state Becomes bottleneck point with traffic capacity D avg .

[0084] Actual traffic capacity That is, if the road section is not blocked in the initial state, the traffic capacity for smooth traffic can be calculated according to the speed of the traffic flow; when congestion occurs, the speed of vehicles on the road section will decrease, and the traffic capacity will decrease, that is, it will drop to the bottleneck point traffic capacity D avg The above two traffic capacities are measured by the traffic volume of the road section.

[0085] For the bottleneck section L b , the road flow is The actual traffic capacity of the road section is D avg .

[0086] For the downstream section L c , road traffic Lc Actual traffic capacity of the road section Still in the initial state

[0087] When the time t x After that, the traffic congestion spreads to the entire upstream section. At this time, the upstream section L a Acceptable input traffic flow Depend on Reduced to bottleneck capacity D avg .

[0088] Step II.2. Establish the flow loading model of the upstream section.

[0089] L upstream of the congested bottleneck section a The road sections are similar, and the initial input traffic flow of the current upstream section is q r , the traffic volume of the road section is Q r =q r The actual traffic capacity of the road section is When congestion occurs on the downstream section, the actual capacity of the upstream section is Change to D avg , traffic began to become congested and queued.

[0090] When the time t s When the traffic congestion spreads to the intersection, the flow rate of the upstream section is reduced to D avg The traffic flow change is calculated based on the flow transfer probability matrix, and the current actual traffic capacity of each upstream section is calculated as shown in formula (8).

[0091] D r =∑D ro (8)

[0092] Among them, D r Indicates the actual traffic capacity of the current upstream section, D ro Indicates the directional flow size received by all upstream segments.

[0093] Step II.3. Establish a downstream section flow prediction model.

[0094] The traffic flow of downstream sections mainly comes from the traffic flow of all upstream sections. Therefore, when the actual traffic capacity of congested sections is reduced, the traffic flow of related downstream sections will also be reduced accordingly.

[0095] When the traffic on the congested section does not change, the output traffic remains unchanged, so the traffic on the downstream section will not change either.

[0096] There are two situations in which the flow rate of the upstream section decreases:

[0097] The upstream section is the section where congestion occurs or the upstream section is the downstream section of the congested section; since vehicles on the upstream section may drive into multiple downstream sections, the reduction in flow on the downstream section also needs to be calculated according to the actual transfer probability of the vehicle.

[0098] Flow rate after the flow rate of the downstream section is reduced The calculation formula is shown in formula (9).

[0099]

[0100] Among them, q ij represents the initial flow of the downstream section, p ij represents the flow transfer probability matrix of the upstream section flow to the downstream section, q i represents the initial flow of the upstream section, Q i Indicates the flow value after the flow of the upstream section is reduced.

[0101] The process of building a vehicle travel time prediction model is as follows:

[0102] Step 1.2.1. Establish a vehicle travel time calculation model under unobstructed conditions.

[0103] The vehicle travel time is calculated under unobstructed conditions, and the traffic model impedance function related to the road section traffic flow and actual traffic capacity is used to solve it. The improved queuing model proposed by Davidson is an asymptotic impedance function that can better describe the relationship between travel time and traffic flow. The calculation formulas are shown in formulas (10) and (11).

[0104]

[0105]

[0106] Among them, parameter J is the service level parameter, which is related to factors such as road grade and road width.

[0107] In this embodiment, the value of parameter J is 0.5 (0.4-0.6 for urban roads).

[0108] t i Indicates the travel time on a smooth road section. is the smooth travel time of road section i; Q i is the traffic flow through road section i; D i is the actual traffic capacity of road section i; L i is the length of road section i; V f The free flow speed.

[0109] Step 1.2.2. Establish a vehicle travel time calculation model under congestion conditions.

[0110] A traffic wave model is constructed to describe the congestion change. Assuming that vehicles are first-in-first-out on a certain road section, the driving time of a vehicle is the time it takes for all vehicles to pass through the road section before it. i for:

[0111]

[0112] When vehicles are waiting in a queue, the traffic density and propagation of congestion conditions on the current road section need to be considered.

[0113] According to the traffic wave model, the congestion propagation speed v s is formula (13).

[0114]

[0115] Where D avg is the bottleneck section capacity, k p is the traffic density of the upstream section of the bottleneck section, k i It is the traffic density under the condition of free driving of vehicles on the current road section.

[0116] Traffic density k of the upstream section of the bottleneck section p The calculation formula of is shown in formula (14).

[0117]

[0118] Among them, D c represents the traffic capacity of the road section, L represents the distance between the bottleneck section and the intersection of the upstream section, and n represents the number of lanes; D c It is related to the number of vehicle lanes and the distance between the bottleneck section and the intersection of the upstream section. The calculation formula is shown in formula (15).

[0119]

[0120] Among them, l represents the standard length of the vehicle, which is 4m, and d represents the distance between vehicles, which is 1m.

[0121] The traffic density k under the condition of free driving of vehicles on the current road section i The calculation formula of is shown in formula (16).

[0122]

[0123] The time calculation model for the traffic congestion state to spread across the entire upstream section is shown in formula (17).

[0124]

[0125] Assuming that the road section is blocked at time T0, if the vehicle enters the road section at time T1 and there is no traffic congestion at the intersection, the time the vehicle encounters traffic congestion is Δt, as shown in formula (18).

[0126]

[0127] The time it takes for a vehicle to pass through the upstream section of a congested section is the sum of the normal travel time of the initial section and the time it takes to pass through the congested section caused by the spread. The calculation model is shown in formula (19).

[0128]

[0129] In the formula, t e Indicates the time it takes for a vehicle to pass through a bottleneck section.

[0130] Step 1.2.3. Establish the relevant vehicle travel time prediction model, as shown in formula (20).

[0131] t L =∑[t i ·h]+∑[t e ·h]+∑t g (20)

[0132] In the formula, t L represents the path travel time; t i Indicates the travel time on a smooth road section; t e represents the travel time on a blocked road section, h is a decision variable, which takes the value of 1 when the road section is passed, otherwise it takes the value of 0; t g Indicates the time it takes a vehicle to pass through an intersection.

[0133] To simplify the problem, assume that the time it takes for a vehicle to pass through an intersection is constant.

[0134] Step 2. Using the vehicle travel time prediction model, based on the improved ant colony algorithm, the path planning problem with the goal of minimizing the vehicle travel time under dynamic traffic conditions is solved to obtain the path with the shortest vehicle travel time.

[0135] The actual travel time of each road section calculated by the vehicle travel time prediction model is used as the weight ratio of each road section of the ant colony algorithm, and the objective function of the vehicle travel time prediction model is used as the solution target of the ant colony algorithm, which can effectively promote and improve the solution efficiency of the ant colony algorithm. Among them, the objective function refers to the minimum travel time, that is, t L Minimum.

[0136] Ant colony algorithm is an important optimization method for path planning. Italian scholar M.Dorigoric proposed an optimization algorithm that simulates the intelligent behavior of ant colonies by observing the behavior of ants in nature - ant colony algorithm.

[0137] In view of the shortcomings of the original ant colony algorithm's pheromone update strategy, Stutzle et al. proposed the "maximum minimum ant colony system" to improve the problem that the ant colony algorithm is prone to stagnation.

[0138] In the process of solving the path planning problem, ants only need to consider two factors when choosing the next city: the concentration of pheromones remaining on the road and the length of the road between the two points.

[0139] The probability of the cth ant moving from segment i to segment j As shown in formula (21);

[0140]

[0141] Among them, τ ij is the pheromone concentration between road section i and road section j, α is the pheromone heuristic factor, which is a measure of τ ij Parameters; η ij is the pheromone concentration between road section i and road section j, β is the expected heuristic factor, which is a measure of η ij Parameters.

[0142] allowed c Indicates the road section that has not been visited by the cth ant yet.

[0143] The pheromone update model is shown in formula (22);

[0144]

[0145] Among them, τ ij (t), τ ij (t+1) represents the pheromone concentrations at the tth and t+1th time between road section i and road section j, ρ is the pheromone volatility coefficient, 0<ρ<1, Δτ ij Represents the growth of pheromone on the path from city i to city j.

[0146] It represents the growth of pheromone on the path of the cth ant from city i to city j.

[0147] In solving the shortest path problem in a road network, since the starting point and the end point are fixed, a fixed direction can be selected. This direction selection strategy can improve the intelligent search characteristics of the ant colony algorithm.

[0148] Through intelligent thinking, we can know that the optimal solution of the shortest path will not deviate too much from the direction of the line connecting the starting point and the end point. Therefore, from the perspective of the current node, the choice of the next node should be as biased as possible towards the direction of the line connecting the current node and the end point. The angle between the two straight lines can be used to describe the deviation of the direction.

[0149] The value of the angle is ∈[0,π], and the schematic diagram is as follows Figure 5 shown.

[0150] Among them, point A is the starting point, point B is the end point, the current node is point C, and the next nodes are two nodes D and E. Define the angle between CD and CB as θ, and the angle between CE and CB as Both meet

[0151] When the pheromone concentrations and road lengths of CD and CE are equal, the ant's choice of the next node should be more inclined to point D, which has a smaller angle with the end point. That is, the probability of selecting the successor node is improved according to this tendency.

[0152] In the improved ant colony algorithm, the coefficient η of the formula for calculating the transition probability of the next node is ij Improvements have been made to increase the influence of the direction angle. The improved coefficient η ij The calculation formula is shown in formula (23).

[0153]

[0154] Among them, w ij is the road impedance of the road section ij; θ represents the smallest angle between the road section to be selected next time and the terminal road section, and λ is the amplification factor related to the road impedance.

[0155] It can be seen from formula (23) that the smaller the impedance of the road section is and the closer the road section is to the destination, the greater the probability of being selected as a passing road section. Therefore, the improvement can effectively improve the disadvantage of the slow convergence speed of the ant colony algorithm in the early stage.

[0156] The traditional ant colony algorithm updates the pheromone based on the path taken by each ant. In order to improve the convergence speed and quality of the algorithm and improve the performance of the algorithm, it is necessary to make certain improvements to the pheromone update mechanism.

[0157] The present invention adds a "value" attribute to the ant body, and only the ants that successfully reach the end point are valuable ants. When the ants ultimately fail to reach the end point, they will be judged as worthless ants.

[0158] When the pheromone is finally updated, the pheromone on the walking path of valuable ants will be increased, and the pheromone on the walking path of worthless ants will be reduced.

[0159] And valuable ants will be ranked according to the length of the path they take, and only ants with values ​​exceeding the average are eligible to participate in the pheromone update that guides the next generation of ants to choose paths.

[0160] In the pheromone update mechanism, the total amount of pheromone owned by each ant is different, and the total amount of pheromone will decrease in order of value ranking. In order to prevent the search from stagnating and falling into a local optimal solution, the value of the pheromone content on each road section (i.e., each edge) is also limited to a reasonable range [τ min ,τ max ]middle.

[0161] The pheromone update mechanism designed in this way can not only reduce the search of invalid paths, improve search efficiency, and accelerate algorithm convergence, but also effectively prevent the algorithm results from falling into local optimal solutions.

[0162] like Figure 6 As shown in the figure, the process of solving the path planning problem based on the improved ant colony algorithm is as follows:

[0163] III.1. Initialize various ant colony algorithm parameters, select the starting and end points for solving the shortest path time, and set all ants to be born on the initial section.

[0164] III.2. Start the iteration process. Each iteration dynamically loads the road section traffic, calculates and updates the weight of the edge in the algorithm, and each ant selects the next road section by comparing the probability of traffic transfer according to the roulette method, and records the selected road section in the taboo table; among them, the edge in the algorithm refers to the road section, and the weight of the edge in the algorithm refers to the passing time.

[0165] III.3. Determine whether the current section is a terminal section.

[0166] If yes, continue to step III.4; if no, jump to step III.2 to continue searching.

[0167] III.4. Calculate the total path time of all ants and compare it with the shortest path time to determine whether to update the shortest path time; use the improved pheromone update strategy to update the pheromone concentration on the path.

[0168] III.5. Determine whether the iteration is finished; if it is finished, output the shortest time of the entire process; otherwise, jump to III.2 to continue the iteration.

[0169] The established vehicle travel time prediction model is combined with the improved ant colony algorithm and compared with the traditional ant colony algorithm method for solving the shortest path to verify the effectiveness of the vehicle travel time prediction model established by the present invention.

[0170] The example background is a traffic road network in a certain area of ​​Jing'an District, Shanghai. The road network data comes from OpenStreetMap, and the coordinate system used in the dataset is WGS-84. The floating car data used comes from the Shanghai taxi driving dataset publicly available from the Hong Kong University of Science and Technology. There are about 12 million data collected from 0:00 to 24:00 on February 20, 2007, with a sampling interval of less than 30s.

[0171] Each item in the data set includes taxi ID, recording time, longitude, latitude, instantaneous speed of the taxi, etc.

[0172] First, we process the obtained road network file appropriately, import the traffic road network file into QGIS, and refine the road network file, delete interference factors such as buildings, and irrelevant branches such as community roads. Then, we number each road and set a 20-meter buffer zone. The results are as follows: Figure 7 The taxi driving data is cleaned up to remove abnormal data and filter the track positioning points according to the range of the buffer zone. The processing results are shown in Figure 8 , Fig. 9 shown.

[0173] The vehicle driving trajectories of all relevant vehicles are converted into road segment sequences, and the road segment sequences are regarded as data sets.

[0174] The FP-Growth algorithm is run visually through the python3 GUI to calculate the support of each data item set and mine association rules to obtain association rules that meet the support and confidence levels. The results are as follows: Fig.10 shown.

[0175] The traffic capacity of this road section is analyzed. The theoretical traffic capacity of lanes specified in the Urban Road Design Code is shown in Table 1.

[0176] At the same time, the traffic capacity is also affected by the road classification coefficient. Since most of the sections in the research network are trunk roads, the influence coefficient is 0.80. Therefore, the traffic capacity of the road is 1384pcu·h -1 .

[0177] The vehicle simulation software SUMO was used to load the road network, and the driving trajectories of 23,810 vehicles were simulated according to the length of the road network to obtain the initial flow of each section of the road network, such as Fig.11 shown.

[0178] Table 1 Lane theoretical capacity

[0179]

[0180] Assume that the average time a vehicle spends traveling through an intersection is 30 seconds.

[0181] Verification and comparison are carried out for different traffic flow densities, the traffic flow loading model is used to load the initial flow of each section in the road network, and the road traffic flow is predicted through the traffic flow transfer probability matrix.

[0182] Three algorithms (original ant colony algorithm, maximum-minimum ant system algorithm, and improved ant colony algorithm of the present invention) are used for path planning. The algorithm effects are compared. Fig.12 shown.

[0183] Depend on Fig.12 It can be seen that the planning path time obtained by the first two algorithms increases rapidly with the increase in the number of vehicles, while after the time prediction model proposed in the present invention is combined with the improved ant colony algorithm, the time growth rate is relatively slow and the improvement effect is more significant.

[0184] In addition, the present invention also verifies the remarkable effectiveness of the travel time prediction model by vehicle flow simulation.

[0185] Take the number of vehicles that can reasonably be included in the road network (23,810 vehicles) as an example:

[0186] According to the traditional path planning method, the optimal path for a vehicle to travel from the starting point to the end point is: 48-122-121-56-213-164-166-165-163-162-193-62-130-131-128, such as Fig.13 shown.

[0187] If the road is congested, the total driving time will increase dramatically. If you continue to drive along the original route, the actual driving time will become 1564 seconds.

[0188] The optimal path obtained by the travel time prediction model combined with the improved ant colony algorithm of the present invention is: 11-108-144-27-24-165-163-162-193-123-104-133-115. Fig.14 As shown, the total driving time is 632 seconds.

[0189] If the vehicle continues to travel along the route planned by the traditional algorithm, the entire journey will be delayed by 932 seconds (1564 seconds - 632 seconds) compared to the shortest foreseeable time. Therefore, by comparing and verifying the experimental results under the same vehicle flow conditions, the effect of the vehicle travel time prediction model established by the present invention can be highlighted.

[0190] In summary, the method of the present invention can obtain the optimal vehicle driving route under different traffic conditions, improve the traffic efficiency of the road network, and the improved ant colony algorithm has been effectively improved in terms of convergence speed and accuracy.

[0191] In addition, based on the same inventive concept, an embodiment of the present invention also provides a vehicle path dynamic adaptive planning system corresponding to the above-mentioned vehicle path dynamic adaptive planning method.

[0192] The vehicle path dynamic adaptive planning system includes the following functional modules:

[0193] The vehicle travel time prediction model establishment module is used to establish a dynamic loading model of road section traffic in the urban traffic road network to predict traffic flow, and then establish a vehicle travel time prediction model considering traffic conditions to predict future road section travel time;

[0194] The vehicle travel time shortest path solving module is used to solve the path planning problem with the goal of minimizing vehicle travel time under dynamic traffic conditions based on the vehicle travel time prediction model and the improved ant colony algorithm to obtain the vehicle travel time shortest path.

[0195] It should be noted that, in the above-mentioned vehicle path dynamic adaptive planning system, the implementation process of the functions and effects of each module is specifically described in the implementation process of the corresponding steps in the above-mentioned method, which will not be repeated here.

[0196] In addition, the present invention also proposes a computer device for implementing the above-mentioned vehicle path dynamic adaptive planning method. The computer device includes a memory and one or more processors. The memory stores an executable code, and when the processor executes the executable code, it is used to implement the steps of the above-mentioned vehicle path dynamic adaptive planning method.

[0197] In this embodiment, the computer device is any device or apparatus with data processing capability, which will not be described in detail here.

[0198] In addition, an embodiment of the present invention further provides a computer-readable storage medium on which a program is stored. When the program is executed by a processor, it is used to implement the steps of the above-mentioned vehicle path dynamic adaptive planning method.

[0199] The computer-readable storage medium can be an internal storage unit of any device or apparatus with data processing capabilities, such as a hard disk or memory, or an external storage device of any device with data processing capabilities, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), an SD card, a flash card, etc. equipped on the device.

[0200] Of course, the above description is only a preferred embodiment of the present invention, and the present invention is not limited to the above embodiments. It should be noted that all equivalent substitutions and obvious deformation forms made by any technician familiar with the field under the guidance of this specification fall within the essential scope of this specification and should be protected by the present invention.

Claims

1. A vehicle path dynamic adaptive planning method, characterized in that: The steps include: Step 1. First, a dynamic loading model of road traffic in the urban traffic network is established to predict traffic flow; then a vehicle travel time prediction model considering traffic conditions is established to predict future road section travel time; Step 2. Using the vehicle travel time prediction model, based on the improved ant colony algorithm, the path planning problem with the goal of minimizing the vehicle travel time under dynamic traffic conditions is solved to obtain the shortest path for vehicle travel time; In step 1, the process of establishing a dynamic loading model for road traffic is as follows: Step 1.1.

1. Abstract the urban traffic road network into a traffic network consisting of nodes and edges. Under the topological structure of the traffic network, abstract the vehicle flow into traffic flow; where nodes are intersections and edges are road sections; Step 1.1.

2. Calculate the association rules between the sections of the floating vehicle's driving trajectory based on the FP-Growth algorithm, use the confidence of the association rules as the correlation coefficient of the spatial correlation of the traffic flow, and construct the traffic transfer probability matrix; The specific steps of step 1.1.2 are as follows: Step I.1 Use the frequent item sets obtained by the FP-Growth data mining method to obtain the confidence of the association rules between the two road sections. The confidence is used as the element of the spatial correlation matrix. Then the road section spatial correlation matrix W ij As shown in formula (1); In formula (1), row numbers i1, i2, i3…i m and column numbers j1, j2, j3…j m Indicates the road section number, w ij represents the spatial correlation between road sections i and j, where w ij With w ji are not equal, that is, the spatial weight matrix is ​​an asymmetric matrix; Step I.

2. Construct the adjacency matrix A to describe the adjacency relationship at the spatial level of the road segment, as shown in formula (2); In formula (2), row numbers i1, i2, i3…i m and column numbers j1, j2, j3…j m Represents the road segment number. The elements in the adjacency matrix A are represented by 0 and 1. Element 0 represents non-adjacency, and element 1 represents adjacency. Step I.

3. Construct a reachability matrix to describe the connectivity between road segments; The operation of the reachability matrix R is obtained through Boolean operations on the adjacency matrix A and the identity matrix E; Step I.

4. Construct the traffic transfer probability matrix; Through the constructed reachability matrix and section spatial correlation matrix, the spatial correlation of adjacent sections of the same order in the section spatial weight matrix is ​​normalized, and then error processing is performed to obtain the flow transfer probability matrix; The specific steps of step I.4 are: First, the probability of road section selection is obtained, and the calculation formula is formula (3): Among them, U ij is the probability of road segment selection, w ig Represents the spatial weight of the segment g with the same order as the segment j among all the segments relative to the segment i. Since the segment selection probability tends to be centralized, the error coefficient E is obtained by comparing the segment selection probability with the average selection probability. ij , the calculation formula is formula (4): Where k represents the reachable order, represents the average selection probability; That is, the actual road section selection probability R ij for: R ij =U ij +E ij (5) Thus, the flow transfer probability matrix p is obtained. ij As shown in formula (6): Among them, R ig Represents the spatial weight of the segment g with the same order as the segment j among all the segments relative to the segment i; Step 1.1.

3. Establish a dynamic traffic loading model for a road section, specifically including establishing a dynamic traffic loading model for a congested road section, establishing a traffic loading model for an upstream road section, and establishing a traffic loading model for a downstream road section; Step II.

1. Establish a dynamic traffic loading model for congested road sections; When a road section enters a congested state, a bottleneck point of traffic capacity will be formed, which can be used to divide the road section into three specific sections, namely, the upstream section L a 、Bottleneck section L b , Downstream section L c ; Assume that congestion starts at time T0; For the upstream section L a , the input traffic flow is The road traffic volume is After the congestion occurs, the road section L a Actual traffic capacity From the initial state Becomes bottleneck section with traffic capacity D avg ; For the bottleneck section L b , the road flow is The actual traffic capacity of the road section is D avg ; For the downstream section L c , road traffic L c Actual traffic capacity of the road section Still in the initial state When the time t x After that, the traffic congestion spreads to the entire upstream section. At this time, the upstream section L a Acceptable input traffic flow Depend on Reduced to bottleneck section capacity D avg ; Step II.

2. Establish the flow loading model of the upstream section; The initial input traffic flow of the current upstream section is q r , the traffic volume of the road section is Q r =q r The actual traffic capacity of the road section is When congestion occurs on the downstream section, the actual capacity of the upstream section is Change to D avg , traffic began to become crowded and queued; When the time t s When the traffic congestion spreads to the intersection, the flow rate of the upstream section is reduced to D avg ; The traffic flow change is calculated based on the flow transfer probability matrix. The current actual traffic capacity of each upstream section is calculated as shown in formula (8); D r =∑D ro (8) Among them, D r Indicates the actual traffic capacity of the current upstream section, D ro Indicates the directional flow size received by all upstream segments; Step II.

3. Establish a downstream section flow prediction model; The traffic volume of the downstream sections mainly comes from the traffic volume of all upstream sections. Therefore, when the actual traffic capacity of the congested section is reduced, the traffic volume of the related downstream sections will also be reduced accordingly. When the traffic on the congested section does not change, the output traffic remains unchanged, so the traffic on the downstream section will not change either; There are two situations in which the flow rate of the upstream section decreases: The upstream section is the section where congestion occurs or the upstream section is the downstream section of the congested section; since vehicles on the upstream section may enter multiple downstream sections, the reduction in the flow of the downstream section is calculated according to the actual transfer probability of the vehicles; Flow rate after the flow rate of the downstream section is reduced The calculation formula is shown in formula (9); Among them, q ij represents the initial flow of the downstream section, p ij represents the flow transfer probability matrix of the upstream section flow to the downstream section, q i represents the initial flow of the upstream section, Q i Indicates the flow value after the flow of the upstream section is reduced.

2. The vehicle path dynamic adaptive planning method according to claim 1, characterized in that: In the step 1, a vehicle travel time prediction model is established; Step 1.2.

1. Establish a vehicle travel time calculation model under unobstructed conditions; The vehicle travel time is calculated under unobstructed conditions, and the traffic model impedance function related to the traffic flow and actual traffic capacity of the road section is used to solve it. The calculation formulas are shown in formulas (10) and (11); Among them, parameter J is the service level parameter; t i Indicates the travel time on a smooth road section. is the smooth travel time of road section i; Q i is the traffic flow through road section i; D i is the actual traffic capacity of road section i; L i is the length of road section i; V f The speed of free circulation; Step 1.2.

2. Establish a vehicle travel time calculation model under congestion conditions; A traffic wave model is constructed to describe the congestion change. Assuming that vehicles are first-in-first-out on a certain road section, the driving time of a vehicle is the time it takes for all vehicles to pass through the road section before it. The speed v of a vehicle under normal driving conditions is i for: When vehicles are waiting in a queue, the traffic density and congestion spread of the current road section need to be considered; According to the traffic wave model, the congestion propagation speed v s is formula (13); Where D avg is the bottleneck section traffic capacity, k p is the traffic density of the upstream section of the bottleneck section, k i is the traffic density of the current road section under the condition of free driving of vehicles; the traffic density k of the upstream section of the bottleneck section p The calculation formula is shown in formula (14); Among them, D c represents the traffic capacity of the road section, L represents the distance between the bottleneck section and the intersection of the upstream section, and n represents the number of lanes; D c It is related to the number of vehicle lanes and the distance between the bottleneck section and the intersection of the upstream section. The calculation formula is shown in formula (15); Where, l represents the standard length of the vehicle, and d represents the distance between vehicles; The traffic density k under the condition of free driving of vehicles on the current road section i The calculation formula is shown in formula (16); The time calculation model for the traffic congestion state to spread over the entire upstream section is shown in formula (17); Assuming that the road section is blocked at time T0, if the vehicle enters the road section at time T1 and there is no traffic congestion at the intersection, the time the vehicle encounters traffic congestion is Δt, as shown in formula (18); The time it takes for a vehicle to pass through the upstream section of a congested section is the sum of the normal travel time of the initial section and the time it takes to pass through the congested section caused by the spread. The calculation model is shown in formula (19); Where, t e Indicates the time it takes for a vehicle to pass through a bottleneck section; Step 1.2.

3. Establish a relevant vehicle travel time prediction model, as shown in formula (20); t L =∑[t i ·h]+∑[t e ·h]+∑t g (20) Where, t L represents the path travel time; t i Indicates the travel time on a smooth road section; t e represents the travel time on a blocked road section, h is a decision variable, which takes the value of 1 when the road section is passed, otherwise it takes the value of 0; t g Indicates the time spent at the intersection.

3. The vehicle path dynamic adaptive planning method according to claim 2 is characterized in that: The step 2 is specifically as follows: The actual travel time of each road section calculated by the vehicle travel time prediction model is used as the weight ratio of each road section of the ant colony algorithm, and the objective function of the vehicle travel time prediction model is used as the solution target of the ant colony algorithm; The objective function is to find the minimum travel time, that is, t L Minimum.

4. The vehicle path dynamic adaptive planning method according to claim 3 is characterized in that: In step 2, the process of solving the path planning problem based on the improved ant colony algorithm is as follows: III.

1. Initialize various ant colony algorithm parameters and select the starting and ending points for solving the shortest path time; Assume that all ants are born on the initial section; III.

2. Start the iteration process. Each iteration dynamically loads the traffic flow of the road section, calculates and updates the weight of the edge in the algorithm, and each ant selects the next road section by comparing the probability of traffic transfer according to the roulette method, and records the selected road section in the taboo table; the edge in the ant colony algorithm is the road section, and the weight of the edge in the ant colony algorithm refers to the passing time; III.

3. Determine whether the current section is a terminal section; If yes, continue to step III.4; if no, jump to step III.2 to continue searching; III.

4. Calculate the total path time of all ants and compare it with the shortest path time to determine whether to update the shortest path time and use the improved pheromone update strategy to update the pheromone concentration on the path; III.

5. Determine whether the iteration is finished. If it is finished, output the shortest time of the entire process; otherwise, jump to III.2 to continue the iteration.

5. The vehicle path dynamic adaptive planning method according to claim 4, characterized in that: The probability of the cth ant moving from segment i to segment j As shown in formula (21); Among them, τ ij is the pheromone concentration between road section i and road section j, α is the pheromone heuristic factor, which is a measure of τ ij Parameters; η ij is the pheromone concentration between road section i and road section j, β is the expected heuristic factor, which is a measure of η ij Parameters; allowed c Indicates the road section that has not been visited by the cth ant yet; The pheromone update model is shown in formula (22); Among them, τ ij (t), τ ij (t+1) represents the pheromone concentrations at the tth and t+1th time between road section i and road section j, ρ is the pheromone volatility coefficient, 0<ρ<1, Δτ ij represents the growth of pheromone on the path from city i to city j; represents the growth of pheromone on the path of the cth ant from city i to city j; The improved coefficient η ij The calculation formula is shown in formula (23); Among them, w ij is the road impedance of the road section ij, λ is the amplification factor related to the road impedance, and θ represents the smallest angle between the road section to be selected next time and the termination section.

6. A vehicle path dynamic adaptive planning system for implementing the vehicle path dynamic adaptive planning method according to any one of claims 1 to 5, characterized in that: The vehicle path dynamic adaptive planning system includes: The vehicle travel time prediction model establishment module is used to establish a dynamic loading model of road section traffic in the urban traffic road network to predict traffic flow, and then establish a vehicle travel time prediction model considering traffic conditions to predict future road section travel time; The vehicle travel time shortest path solving module is used to solve the path planning problem with the goal of minimizing vehicle travel time under dynamic traffic conditions based on the vehicle travel time prediction model and the improved ant colony algorithm to obtain the vehicle travel time shortest path.

7. A computer device comprising a memory and a processor, wherein the memory stores executable code, characterized in that: When the processor executes the executable code, Used to implement the steps of the vehicle path dynamic adaptive planning method described in any one of claims 1 to 5.

8. A computer-readable storage medium having a program stored thereon, characterized in that: When the program is executed by a processor, it is used to implement the steps of the vehicle path dynamic adaptive planning method described in any one of claims 1 to 5.

Images