Tidal lane dynamic control method and system based on traffic flow prediction and medium

Abstract

The invention relates to a reversible lane dynamic control method and system based on traffic flow prediction and a medium. The method comprises the following steps: constructing a reversible lane dynamic control model; constructing a road section topological graph and outputting a first intermediate feature based on the topological graph; constructing a reversible lane topological graph and outputting a second intermediate feature based on the topological graph; generating a condition sample from the actual road network information and the traffic flow data, and processing the condition sample to obtain a plurality of sample fragments; judging whether an abnormal event exists or not, and generating an abnormal event feature if the abnormal event exists; if no abnormal event exists, obtaining an abnormal event-free feature and directly inputting the abnormal event-free feature into the road segment similar graph convolutional neural network; outputting a third intermediate feature by using the road segment similarity graph convolutional neural network; predicting the traffic flow through a road section upstream and downstream map convolutional neural network; the reversible lane dynamic control is realized based on the predicted traffic flow. The system and the medium are used for realizing the method. Compared with the prior art, the reversible lane dynamic control method can timely and efficiently realize reversible lane dynamic control.

Description

Technical Field

[0001] The present invention relates to the field of intelligent transportation, and in particular to a tidal lane dynamic control method, system and medium based on traffic flow prediction. Background Art

[0002] The tidal phenomenon in urban traffic refers to the phenomenon that due to the over-concentration of urban residential and work areas, a large number of residential communities are built outside the city center, and many people choose to work in the city center and live in the suburbs, resulting in a one-way asymmetric congestion in which a large number of vehicles flow from the residential area to the work area in the early morning, and the traffic flow returns from the work area to the residential area in the evening. This phenomenon is particularly obvious during peak hours in the morning and evening, making the traffic flow on certain sections of the road show significant differences and imbalances in time. In response to the above phenomenon, a strategy for tidal lane control is proposed. Tidal lanes, also known as variable lanes, refer to a traffic organization method that changes the driving direction or type of vehicles on certain lanes at different times. By adjusting the lane direction to adapt to traffic needs in different time periods, traffic congestion can be alleviated and road capacity and utilization can be improved. Existing tidal lane control includes the following two methods: 1) Based on real-time detected traffic flow data, this method analyzes the current data to generate a tidal lane control plan for the corresponding moment, but the road conditions change in this process, so although this method achieves tidal lane control, it has a certain lag; 2) Generate a tidal lane control method based on a predicted traffic phenomenon such as queuing vehicles or lane saturation. Although this method has enhanced timeliness, queuing vehicles or lane saturation cannot fully feedback the lane traffic status, and cannot guarantee the operating efficiency of the generated tidal lane control method.

[0003] Therefore, it is a problem that needs to be solved to provide a tidal lane control method that is timely and efficient. Summary of the invention

[0004] The purpose of the present invention is to overcome the defects of the above-mentioned prior art and to provide a tidal lane dynamic control method, system and medium based on traffic flow prediction.

[0005] The purpose of the present invention can be achieved by the following technical solutions:

[0006] According to a first aspect of the present invention, a tidal lane dynamic control method based on traffic flow prediction is provided, the method comprising:

[0007] Constructing a tidal lane dynamic control model, the tidal lane dynamic control model includes a traffic flow prediction module and a tidal lane dynamic control module, wherein the traffic flow prediction module includes a road section graph convolutional neural network, a tidal lane graph convolutional neural network, an abnormal event graph convolutional neural network, a road section similarity graph convolutional neural network and a road section upstream and downstream graph convolutional neural network;

[0008] Construct a road section topology map based on actual road network information and traffic flow data, and use a road section map convolutional neural network to output a first intermediate feature, wherein the first intermediate feature includes a road section weight;

[0009] Construct a tidal lane topology map based on actual road network information and traffic flow data, and use a tidal lane map convolutional neural network to output a second intermediate feature, wherein the second intermediate feature includes a tidal weight;

[0010] Generating condition samples from actual road network information and traffic flow data, and processing the condition samples using a sliding window to obtain a plurality of sample segments, each of which includes a plurality of traffic flow data;

[0011] By using the attention mechanism, the first intermediate feature, the second intermediate feature and the sample fragment are combined to determine whether there is an abnormal event. If there is an abnormal event, the first intermediate feature, the second intermediate feature and the sample fragment are input into the abnormal event graph convolutional neural network to calculate the abnormal event feature, and the abnormal event feature is input into the road section similarity graph convolutional neural network; if there is no abnormal event, the first intermediate feature, the second intermediate feature and the sample fragment corresponding to the non-abnormal event are directly input into the road section similarity graph convolutional neural network as the non-abnormal event feature;

[0012] Using a road section similarity graph convolutional neural network to process abnormal event features or non-abnormal event features, and output a third intermediate feature, wherein the third intermediate feature includes a road section similarity weight;

[0013] The upstream and downstream graph convolutional neural network of the road section predicts the traffic flow of the target road section at the next moment based on the first intermediate feature, the second intermediate feature, the abnormal event feature and the third intermediate feature;

[0014] The dynamic control of tidal lanes is achieved by using the tidal lane dynamic control module based on the predicted road section traffic flow.

[0015] As a preferred technical solution, the method for obtaining the road section weight is:

[0016] Acquire information of adjacent road sections and non-adjacent road sections based on the road section topology map and acquire vehicle quantity data of the road section based on the corresponding traffic flow data;

[0017] The adjacent road section weight matrix is ​​calculated based on the vehicle data information of the adjacent road sections, and the non-adjacent road section weight matrix is ​​calculated based on the vehicle quantity data of the non-adjacent road sections.

[0018] As a preferred technical solution, the method for obtaining the tidal weight is:

[0019] Obtain the road section density and congestion from the traffic flow data and calculate the tidal weight value, which is expressed as:

[0020]

[0021] Among them, w t represents the tidal weight value, i and j represent the road section, t represents the time, k j represents the congestion density, and ρ represents the road segment density.

[0022] As a preferred technical solution, the method for determining whether there is an abnormal event is: dividing the road section status into fast, slow, slow and stopped based on the road section density, wherein the road section status of slow and stopped corresponds to abnormal events, and the road section status of fast and slow corresponds to no abnormal events.

[0023] As a preferred technical solution, the abnormal event feature includes an abnormal event weight, and the method for obtaining the abnormal event weight is:

[0024] The queuing probability is calculated based on traffic flow data, and its expression is: Where P(X=k) represents the queue probability, k represents the number of observed vehicles, and λ represents the average number of vehicles per unit time and space;

[0025] Update the vehicle position, vehicle speed and vehicle density, wherein the vehicle speed is related to the vehicle density, and its expression is: v(t+△t)=f(v(t),ρ(t)), wherein v(t) represents the speed, ρ(t) represents the vehicle density, and f(·) represents the speed update function; the vehicle density is related to the vehicle speed, and its calculation formula is: ρ(t+△t)=g(ρ(t),v(t)), wherein g(·) represents the density update function;

[0026] The abnormal event weight is calculated based on the queuing probability and the updated vehicle position, vehicle speed and vehicle density, and its expression is:

[0027] w a (i,j)=P(X=k)×[x(t+△t)+v(t+△t)+ρ(t+△t)], where, w a represents the weight value of the abnormal event, and i and j represent road sections.

[0028] As a preferred technical solution, the method for obtaining the road section similarity weight is:

[0029] Construct a similarity topology graph based on abnormal event features or non - abnormal event features, obtain information on any two road segments in the similarity topology graph obtained, and calculate the distance between the two using the longest common subsequence method;

[0030] Calculate the similarity weight value of any two road segments based on the distance, and its expression is:

[0031] w h (i,j)=e -lcss×ε×o(nm) ,

[0032] where, w h represents the similarity weight value, i and j both represent road segments, lcss represents the distance, ε represents the considered threshold, o(nm) represents the time complexity, and n and m respectively represent the lengths of the two road segments.

[0033] As a preferred technical solution, the method for predicting the traffic flow of the target road segment at the next moment is:

[0034] Calculate the weight data of the target road segment and the upstream and downstream of the target road segment respectively based on the first intermediate feature, the second intermediate feature, the abnormal event feature, and the third intermediate feature;

[0035] Use the weight data of the target road segment and the upstream and downstream of the target road segment to calculate the time - related weight of the target road segment at the next moment, and its expression is:

[0036] W r (t + 1)=W r (t)+∑downW r (t)-∑upW r (t),

[0037] where, W r (t) represents the weight data of the target road segment, ∑downW r (t) represents the weight data of the downstream road segment, and ∑upW r (t) represents the weight data of the upstream road segment;

[0038] Generate a traffic flow prediction for the next moment based on the time - related weight.

[0039] As a preferred technical solution, the weight data of the target road segment and the upstream and downstream of the target road segment are obtained by weighted summation based on the first intermediate feature, the second intermediate feature, the abnormal event feature, and the third intermediate feature of the corresponding road segments.

[0040] According to the second aspect of the present invention, a tidal lane dynamic control system based on traffic flow prediction is provided, and the system is used to implement the above - mentioned method.

[0041] According to a third aspect of the present invention, there is provided a computer-readable storage medium having a computer program stored thereon, wherein the program implements the method described above when executed by a processor.

[0042] Compared with the prior art, the present invention realizes the control of tidal lanes based on the predicted traffic flow of the target road section, thereby ensuring the timeliness of the control, and fully considering the possible impact of abnormal events on traffic flow in the process of predicting traffic flow. The prediction expansion capability of the model is enhanced by calculating the similarity weights of road sections, so that the traffic flow analysis results of one road can be expanded to multiple similar roads, so that the traffic flow can be predicted more comprehensively and accurately, thereby generating a more accurate tidal lane control strategy. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 is a flow chart of the method of the present invention;

[0044] Figure 2 It is the adjacent road section topology map of the present invention;

[0045] Figure 3 It is a non-adjacent road section topology map of the present invention;

[0046] Figure 4 A schematic diagram of a tidal lane according to the present invention;

[0047] Figure 5 A simplified topological diagram of a tidal lane of the present invention;

[0048] Figure 6 It is the target road section topology map of the present invention. DETAILED DESCRIPTION

[0049] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0050] Unless otherwise defined, the technical terms or scientific terms involved in this application should be understood by people with ordinary skills in the technical field to which this application belongs. The words "one", "a", "a", "the" and the like involved in this application do not indicate a quantitative limitation, and may represent the singular or plural. The terms "include", "comprise", "have" and any of their variations involved in this application are intended to cover non-exclusive inclusions; for example, a process, method, system, product or device that includes a series of steps or modules (units) is not limited to the listed steps or units, but may also include steps or units that are not listed, or may also include other steps or units inherent to these processes, methods, products or devices. The words "connect", "connected", "coupled" and the like involved in this application are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. The "multiple" involved in this application refers to two or more. "And / or" describes the association relationship of associated objects, indicating that there may be three relationships, for example, "A and / or B" can represent: A exists alone, A and B exist at the same time, and B exists alone. The character " / " generally indicates that the objects before and after are in an "or" relationship. The terms "first", "second", "third", etc. involved in this application are only used to distinguish similar objects and do not represent a specific ordering of the objects.

[0051] This embodiment provides a method for dynamic control of tidal lanes based on traffic flow prediction, that is, first predicting the traffic flow at the next moment based on the traffic flow data at the current moment and then realizing dynamic control of the tidal lanes. Because the attention mechanism shows higher accuracy in traffic flow prediction tasks, it can better capture the dynamic changes of traffic flow, and the traffic road network subgraph construction method of the attention mechanism constructs a more accurate and efficient subgraph model by intelligently selecting and weighting the nodes and edges in the traffic network, which provides a new idea for traffic flow prediction and management. Based on the above reasons, this embodiment selects the attention mechanism and graph convolutional neural network to realize traffic flow prediction.

[0052] Specifically, the flowchart of the tidal lane dynamic control method provided in this embodiment is as follows: Figure 1 As shown, the detailed steps include:

[0053] S1. Model construction:

[0054] A tidal lane dynamic control model is constructed, which includes a traffic flow prediction module and a tidal lane dynamic control module. The traffic flow prediction module includes a road section graph convolutional neural network, a tidal lane graph convolutional neural network, an abnormal event graph convolutional neural network, a road section similarity graph convolutional neural network and a road section upstream and downstream graph convolutional neural network.

[0055] S2. Preliminary data processing:

[0056] S21, constructing a road section topology map based on actual road network information and traffic flow data, using the road section map convolutional neural network to output a first intermediate feature, and the first intermediate feature includes a road section weight. The detailed steps for obtaining the road section weight are:

[0057] S211, based on the road section topology map, obtain information about adjacent road sections and non-adjacent road sections, wherein the topology map of the adjacent road sections is as follows: Figure 2 As shown in the figure, the topological diagram of non-adjacent road sections is as follows Figure 3 As shown; the vehicle quantity data of the road section is obtained according to the traffic flow data of the corresponding road section.

[0058] S212, if the number of vehicles on section A is 100, 50 vehicles flow from section A to section B, 20 vehicles flow from section A to section C, and 10 vehicles flow from section A to section D, then the weight matrix of adjacent sections is shown in Table 1, and its expression is: W 1,c =∑W 1,ij , where W 1,c represents the weight of adjacent road segments, W 1,ij Represents the weight of two adjacent road segments i relative to j.

[0059] Table 1 Adjacent road segment weight matrix

[0060] A B C D A 1 / 5 1 / 2 1 / 5 1 / 10

[0061] On the basis of adjacent sections, add non-adjacent section E, and assume that the number of vehicles on section E is 30, and there are 20 vehicles from section E to section A. The weight matrix of non-adjacent sections is shown in Table 2, and its expression is: W 2,c =∑∏(w 2,i ,w 2,j ), where W 2,c represents the weight of non-adjacent road segments, w 2,i represents the weight of road section i, w 2,j represents the weight of road segment j.

[0062] Table 2 Weight matrix of non-adjacent road sections

[0063] A B C D E A 1 / 5 1 / 2 1 / 5 1 / 10 0 E 2 / 3 0 0 0 1 / 3

[0064] The adjacent link weight matrix and the non-adjacent link weight matrix are collectively referred to as link weights.

[0065] S22, sections containing tidal lanes such as Figure 4 As shown in the figure, the tidal lane topology is constructed based on the actual road network information and traffic flow data. Figure 5 As shown, lane 2 is a tidal lane. Figure 5The tidal weight matrices of the tidal lanes are shown in Tables 3 and 4. The tidal lane graph convolutional neural network is used to output the second intermediate feature, and the second intermediate feature includes the tidal weight. The method for obtaining the tidal weight is:

[0066] Obtain the road section density and congestion from the traffic flow data and calculate the tidal weight value, which is expressed as:

[0067]

[0068] Among them, w t represents the tidal weight value, i and j represent the road section, t represents the time, k j represents the congestion density, and ρ represents the road segment density.

[0069] Table 3 Tidal weight matrix before switching

[0070] A B C A 1 1 0 B 0 1 1 C 0 1 1

[0071] Table 4 Tidal weight matrix after reversal

[0072] A B C A 1 0 0 B 1 1 1 C 0 1 1

[0073] S23, generating condition samples S from actual road network information and traffic flow data t (In t ,Out t ), In t is an N×W matrix, where N represents the number of roads and W represents traffic flow data; a sliding window is used to map the condition sample S t (In t ,Out t ) to obtain multiple sample fragments, set the size of the delimited window to w, the step length to d, and t It is divided into K sample segments, and each sample segment is an N×w matrix including multiple traffic flow data. And because the time series data has a compact feature, the data in adjacent time slots have strong local interactions, so the sliding window w>1.

[0074] S3. Abnormal event judgment:

[0075] The attention mechanism is used to combine the first intermediate feature, the second intermediate feature and the sample fragment to determine whether there is an abnormal event based on the road section density. When the road section density ρ>100% or 80%≤ρ≤90%, that is, the state of the road section is stopped or slow, it will be marked as an abnormal event; if the road section density is ρ≤50% or 50≤ρ≤80%, that is, the state of the road section is fast and slow, it will be marked as no abnormal event.

[0076] S4. Subsequent data processing:

[0077] S41, if it is an abnormal event section, the first intermediate feature, the second intermediate feature and the sample segment of the section are input into the abnormal event graph convolutional neural network to calculate the abnormal event feature; if there is no abnormal event, step S42 is executed to input the abnormal event feature into the section similarity graph convolutional neural network, and the detailed steps are as follows:

[0078] S411. The probability of vehicles queuing is regarded as Poisson distribution, and the queuing probability is calculated based on traffic flow data. The expression is: Where P(X=k) represents the queue probability, k represents the number of observed vehicles, and λ represents the average number of vehicles per unit time and space;

[0079] S412, update the vehicle position, the expression is: x(t+△t)=x(t)+v(t)×△t, △t represents the event step, x(t) represents the vehicle position, and v(t) represents the speed.

[0080] S413. Update vehicle speed, which is expressed as: v(t+△t)=f(v(t),ρ(t)), where v(t) represents speed, ρ(t) represents vehicle density, and f(·) represents speed update function.

[0081] S414, updating vehicle density, which is calculated as follows: ρ(t+△t)=g(ρ(t),v(t)), where g(·) represents a density update function;

[0082] S415. Calculate the abnormal event weight based on the queuing probability and the updated vehicle position, vehicle speed and vehicle density. The expression is:

[0083] w a (i,j)=P(X=k)×[x(t+△t)+v(t+△t)+ρ(t+△t)], where, w a represents the weight value of the abnormal event, and i and j represent road sections.

[0084] S42, if there is no abnormal event, directly inputting the first intermediate feature, the second intermediate feature and the sample segment corresponding to the abnormal event into the road segment similarity graph convolutional neural network as the abnormal event feature;

[0085] S43, using the road section similarity graph convolutional neural network to process abnormal event features or non-abnormal event features, and output a third intermediate feature, wherein the third intermediate feature includes a road section similarity weight, and a method for calculating the road section similarity weight is:

[0086] S431, constructing a similarity topology map according to the abnormal event characteristics or the non-abnormal event characteristics, obtaining information of any two road sections in the similarity topology map, and calculating the distance between the two using the longest common subsequence method;

[0087] S432. Calculate the similarity weight value between any two road segments based on the aforementioned distance, and its expression is:

[0088] w h (i,j) = e -lcss×ε×o(nm) ,

[0089] where w h represents the similarity weight value, i and j both represent road segments, lcss represents the distance, ε represents the considered threshold, which is set according to the specific application scenario and data characteristics, o(nm) represents the time complexity, and n and m respectively represent the lengths of the two road segments.

[0090] S44. The upstream and downstream graph convolutional neural network of the road segment predicts the traffic flow of the target road segment at the next moment based on the first intermediate feature, the second intermediate feature, the abnormal event feature, and the third intermediate feature. The detailed steps include:

[0091] S441. Calculate the weight data of the target road segment and the upstream and downstream of the target road segment respectively based on the first intermediate feature, the second intermediate feature, the abnormal event feature, and the third intermediate feature. The positions of the target road segment and its upstream and downstream road segments are as Figure 6 shown, and the expression of the weight data of the target road segment is:

[0092] W r (t) = w c + w t + w a + w h ,

[0093] where W r (t) represents the weight of the target road segment, w c represents the road segment weight of the target road segment, w t represents the tidal weight of the target road segment, w a represents the abnormal event weight of the target road segment, w h represents the similarity weight of the target road segment.

[0094] The expression of the upstream weight data is: ∑up W r (t) = w' c + w' t + w' a + w' h , and the expression of the downstream weight data is: ∑down W r (t) = w'' c + w'' t + w'' a + w'' h .

[0095] S442. Calculate the time - related weight of the target road section at the next moment using the weight data of the target road section and its upstream and downstream sections. The expression is as follows:

[0096] W r (t + 1)=W r (t)+∑W_down r (t)-∑W_up r (t),

[0097] where, W r (t) represents the weight data of the target road section, ∑W_down r (t) represents the weight data of the downstream road section, ∑W_up r (t) represents the weight data of the upstream road section;

[0098] S443. Generate a traffic flow prediction for the next moment based on the time - related weight.

[0099] S5. Lane control:

[0100] Based on the predicted road section traffic flow, use the tidal lane dynamic control module to achieve dynamic control of the tidal lane. Specifically, if Figure 4 lane 3 is congested, then lane 2 has the same direction as lane 3; if lane 1 is congested, then lane 2 has the same direction as lane 1.

[0101] This embodiment also provides a tidal lane dynamic control system based on traffic flow prediction for implementing the above - mentioned method. Those skilled in the art can clearly understand that for the sake of convenience and brevity of description, the specific working process described can refer to the corresponding process in the foregoing method embodiment and will not be elaborated herein.

[0102] The system of the present invention includes a central processing unit (CPU), which can execute various appropriate actions and processes according to computer program instructions stored in a read - only memory (ROM) or computer program instructions loaded from a storage unit into a random - access memory (RAM). In the RAM, various programs and data required for device operation can also be stored. The CPU, ROM, and RAM are connected to each other through a bus. An input / output (I / O) interface is also connected to the bus.

[0103] Multiple components in the device are connected to the I / O interface, including: an input unit, such as a keyboard, a mouse, etc.; an output unit, such as various types of displays, speakers, etc.; a storage unit, such as a disk, an optical disc, etc.; and a communication unit, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit allows the device to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0104] The processing unit performs the various methods and processes described above, such as methods S1 to S5. For example, in some embodiments, methods S1 to S5 may be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as a storage unit. In some embodiments, part or all of the computer program may be loaded and / or installed on the device via a ROM and / or a communication unit. When the computer program is loaded into the RAM and executed by the CPU, one or more steps of methods S1 to S5 described above may be performed. Alternatively, in other embodiments, the CPU may be configured to execute methods S1 to S5 in any other appropriate manner (e.g., by means of firmware).

[0105] The functions described above herein may be performed at least in part by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic components that may be used include: field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), and the like.

[0106] The program code for implementing the method of the present invention can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer or other programmable data processing device, so that the program code, when executed by the processor or controller, enables the functions / operations specified in the flow chart and / or block diagram to be implemented. The program code can be executed entirely on the machine, partially on the machine, partially on the machine as a stand-alone software package and partially on a remote machine, or entirely on a remote machine or server.

[0107] In the context of the present invention, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0108] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed by the present invention, and these modifications or replacements should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A tidal lane dynamic control method based on traffic flow prediction, characterized in that: The method includes: Constructing a tidal lane dynamic control model, the tidal lane dynamic control model includes a traffic flow prediction module and a tidal lane dynamic control module, wherein the traffic flow prediction module includes a road section graph convolutional neural network, a tidal lane graph convolutional neural network, an abnormal event graph convolutional neural network, a road section similarity graph convolutional neural network and a road section upstream and downstream graph convolutional neural network; Construct a road section topology map based on actual road network information and traffic flow data, and use a road section map convolutional neural network to output a first intermediate feature, wherein the first intermediate feature includes a road section weight; Construct a tidal lane topology map based on actual road network information and traffic flow data, and use a tidal lane map convolutional neural network to output a second intermediate feature, wherein the second intermediate feature includes a tidal weight; Generating condition samples from actual road network information and traffic flow data, and processing the condition samples using a sliding window to obtain a plurality of sample segments, each of which includes a plurality of traffic flow data; By using the attention mechanism, the first intermediate feature, the second intermediate feature and the sample fragment are combined to determine whether there is an abnormal event. If there is an abnormal event, the first intermediate feature, the second intermediate feature and the sample fragment are input into the abnormal event graph convolutional neural network to calculate the abnormal event feature, and the abnormal event feature is input into the road section similarity graph convolutional neural network; if there is no abnormal event, the first intermediate feature, the second intermediate feature and the sample fragment corresponding to the non-abnormal event are directly input into the road section similarity graph convolutional neural network as the non-abnormal event feature; Using a road section similarity graph convolutional neural network to process abnormal event features or non-abnormal event features, and output a third intermediate feature, wherein the third intermediate feature includes a road section similarity weight; The upstream and downstream graph convolutional neural network of the road section predicts the traffic flow of the target road section at the next moment based on the first intermediate feature, the second intermediate feature, the abnormal event feature and the third intermediate feature; The dynamic control of tidal lanes is achieved by using the tidal lane dynamic control module based on the predicted road section traffic flow.

2. According to claim 1, a tidal lane dynamic control method based on traffic flow prediction is characterized in that: The method for obtaining the road section weight is: Acquire information of adjacent road sections and non-adjacent road sections based on the road section topology map and acquire vehicle quantity data of the road section based on the corresponding traffic flow data; The adjacent road section weight matrix is ​​calculated based on the vehicle data information of the adjacent road sections, and the non-adjacent road section weight matrix is ​​calculated based on the vehicle quantity data of the non-adjacent road sections.

3. According to claim 1, a tidal lane dynamic control method based on traffic flow prediction is characterized in that: The method for obtaining the tidal weight is: Obtain the road section density and congestion from the traffic flow data and calculate the tidal weight value, which is expressed as: Among them, w t represents the tidal weight value, i and j represent the road section, t represents the time, k j represents the congestion density, and ρ represents the road segment density.

4. The method for dynamic control of tidal lanes based on traffic flow prediction according to claim 1 is characterized in that: The method for determining whether there is an abnormal event is: dividing the road section status into fast, slow, slow and stopped based on the road section density, wherein the road section status of slow and stopped corresponds to abnormal events, and the road section status of fast and slow corresponds to no abnormal events.

5. The method for dynamic control of tidal lanes based on traffic flow prediction according to claim 1 is characterized in that: The abnormal event feature includes an abnormal event weight. The method for obtaining the abnormal event weight is: The queuing probability is calculated based on traffic flow data, and its expression is: Where P(X=k) represents the queue probability, k represents the number of observed vehicles, and λ represents the average number of vehicles per unit time and space; Update the vehicle position, vehicle speed and vehicle density, wherein the vehicle speed is related to the vehicle density, and its expression is: v(t+△t)=f(v(t),ρ(t)), wherein v(t) represents the speed, ρ(t) represents the vehicle density, and f(·) represents the speed update function; the vehicle density is related to the vehicle speed, and its calculation formula is: ρ(t+△t)=g(ρ(t),v(t)), wherein g(·) represents the density update function; The abnormal event weight is calculated based on the queuing probability and the updated vehicle position, vehicle speed and vehicle density, and its expression is: w a (i,j)=P(X=k)×[x(t+△t)+v(t+△t)+ρ(t+△t)], where, w a represents the weight value of the abnormal event, and i and j represent road sections.

6. The method for dynamic control of tidal lanes based on traffic flow prediction according to claim 1 is characterized in that: The method for obtaining the road section similarity weight is: Construct a similarity topology map based on abnormal event features or non-abnormal event features, obtain information on any two road sections in the similarity topology map, and calculate the distance between the two using the longest common subsequence method; The similarity weight value of any two road segments is calculated based on the distance, and the expression is: w h (i,j)=e -lcss×ε×o(nm) , Among them, w h represents the similarity weight value, i and j both represent road segments, lcss represents distance, ε represents the consideration threshold, o(nm) represents time complexity, and n and m represent the lengths of two road segments respectively.

7. The method for dynamic control of tidal lanes based on traffic flow prediction according to claim 1 is characterized in that: The method for predicting the traffic flow of the target road section at the next moment is: Calculate the weight data of the target road section and the upstream and downstream of the target road section respectively based on the first intermediate feature, the second intermediate feature, the abnormal event feature and the third intermediate feature; The time-related weight of the target section at the next moment is calculated by using the weight data of the target section and the upstream and downstream of the target section. The expression is: W r (t+1)=W r (t)+∑ 下 W r (t)-∑ 上 W r (t), Among them, W r (t) represents the target road section weight data, ∑ 下 W r (t) represents the weight data of the downstream section, ∑ 上 W r (t) represents the weight data of the upstream section; A traffic flow prediction for the next moment is generated based on the time-related weights.

8. The method for dynamic control of tidal lanes based on traffic flow prediction according to claim 7 is characterized in that: The weight data of the target road section and the upstream and downstream of the target road section are obtained based on the weighted sum of the first intermediate feature, the second intermediate feature, the abnormal event feature and the third intermediate feature of the corresponding road section.

9. A tidal lane dynamic control system based on traffic flow prediction, characterized in that: The system is used to implement the method according to any one of claims 1 to 8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method according to any one of claims 1 to 8 is implemented.

Images