Time-space fusion perception transformation device and method based on traffic flow prediction

Abstract

The invention relates to the technical field of traffic, in particular to a time-space fusion perception transformation device and method based on traffic flow prediction.The device comprises a historical period perception module, a traffic data processing module, an information embedding module, a dynamic flow interaction graph module and a space and time self-attention module, and the steps are as follows: S1, firstly, historical traffic data are processed by using a K-POP clustering algorithm; obtaining a representative short-term traffic mode set containing the key historical information; s2, embedding past periodic feature information into traffic data by using a self-attention mechanism through a historical period sensing module, and expanding traffic features of a single node to the outside of a single moment; according to the space-time fusion perception transformation device and method based on traffic flow prediction, challenges caused by traffic data dynamics, periodicity and delay can be effectively solved through the method, and the space-time fusion perception transformation device and method can be used as a key model of an RNDT system algorithm layer to accurately predict future traffic flow and provide decision support.

Description

Technical Field

[0001] The present invention relates to the field of traffic technology, and in particular to a time-space fusion perception transformation device and method based on traffic flow prediction. Background Art

[0002] Traffic flow prediction is considered to be one of the most critical tasks in RNDT. Graph structures can describe the relationships between different entities and are very suitable for modeling traffic data and road network structures. Methods based on graph neural networks (GNNs) have also been widely explored in the field of traffic prediction. In the latest technological developments, researchers combine GNNs with sequence information processing mechanisms such as recurrent neural networks (RNNs) or attention mechanisms. This integration aims to capture spatial and temporal relationships simultaneously and promote accurate prediction of traffic flow information. This integrated network is often called spatiotemporal graph neural network (STGNN), which currently exhibits state-of-the-art performance in the field of traffic prediction.

[0003] Although GNN models are very effective, they still face significant limitations: traffic nodes at different times often show similar traffic flows, but their traffic interaction patterns may not be equal. For example, commercial nodes may experience similar traffic peaks during meal times on different dates, but the traffic flow after the peak varies between weekdays and weekends. Relying only on the time embedding of a single moment as input will cause the model to fail to fully capture the interactive characteristics; in reality, traffic transfers between different nodes are not instantaneous, and there may be unavoidable delays due to congestion. In addition, the historical traffic conditions of the nodes will also affect the current traffic conditions and the surrounding environment. Using the time embedding of a single moment as input will ignore important historical and relevant interactive information; urban traffic networks are constantly evolving, which means that when building spatial dependency models, pre-defining static physical road networks may lead to incomplete transmission of traffic information, limiting the ability to learn dynamic patterns of urban traffic. Summary of the invention

[0004] The purpose of the present invention is to provide a spatiotemporal fusion perception transformation device and method based on traffic flow prediction, so as to solve the problem that traffic nodes in different periods often show similar traffic flows, but their traffic interaction patterns may not be equal, and traffic transfers between different nodes are not instantaneous. Due to congestion, there may be inevitable delays. In addition, the historical traffic conditions of the nodes will also affect the current traffic conditions and the surrounding environment. The predefined static physical road network may lead to incomplete transmission of traffic information, which limits the ability to learn the dynamic patterns of urban traffic. To achieve the above purpose, the present invention provides the following technical solutions: a spatiotemporal fusion perception transformation device and method based on traffic flow prediction, including a historical period perception module, traffic data processing, information embedding, a dynamic flow interaction graph, and a spatial and temporal self-attention module, and the steps are as follows: S1: First, the historical traffic data is processed using the K-POP clustering algorithm to obtain a representative short-term traffic pattern set containing key historical information; S2: The historical cycle perception module uses the self-attention mechanism to embed past periodic feature information into traffic data, extending the traffic characteristics of a single node beyond a single moment; S3: Through the dynamic traffic interaction graph generation algorithm, the impact of historical status on the current status is taken into account, so as to distinguish the interaction relationship between different periods with similar traffic flow status; S4: The self-attention module is used to capture spatial and temporal dependencies to further optimize the graph; S5: The final output graph will serve as the basis for traffic flow prediction.

[0005] The historical cycle perception module is used to capture historical information and introduce multi-scale time correlation to distinguish traffic cycles and accurately perceive time delays.

[0006] The dynamic traffic interaction graph is generated by a novel dynamic traffic interaction graph generation algorithm based on semantic similarity between features embedded in historical cycle information, and the generated graph contains semantic hierarchical interaction relationships between different traffic nodes.

[0007] The spatial and temporal self-attention module includes spatial self-attention and temporal self-attention. The spatial self-attention uses the graph attention mechanism to simultaneously capture physical-level spatial dependencies and semantic-level spatial dependencies, and the temporal self-attention captures the complex dynamic temporal dependencies in traffic data.

[0008] A spatiotemporal fusion perception transformation device based on traffic flow prediction includes a shell, the shell is used to be loaded on a vehicle in use and is used to protect internal components, a historical cycle perception module is fixedly connected to the bottom inner wall of the shell, a space and time self-attention module is fixedly connected to the bottom inner wall of the shell, a storage module for information storage is fixedly connected to the bottom inner wall of the shell, a PCB board is fixedly connected to the bottom inner wall of the shell and the PCB board is electrically connected to the historical cycle perception module, the space and time self-attention module and the storage module, a processing module for information processing and calculation is fixedly connected to the top of the PCB board, a positioning module for vehicle positioning is fixedly connected to the top of the processing module, and a wireless communication module for network communication is fixedly connected to the top of the PCB board.

[0009] Compared with the prior art, the present invention has the following beneficial effects: In the present invention, the challenges brought by the dynamics, periodicity and delay of traffic data can be effectively solved by this method. It can be used as a key model of the algorithm layer of the RNDT system to accurately predict future traffic flows and provide decision support.

[0010] In the present invention, a historical cycle awareness (HCA) module is used to embed historical information, clearly perceive traffic cycles and accurately perceive time delays, and a dynamic flow interaction graph (DFIG) generation algorithm is used to simulate the spatial dependency of the semantic level and distinguish interactions in different periods with similar traffic flow states. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Figure 1 It is a flowchart of the present invention; Figure 2 It is a schematic diagram of the three-dimensional structure of the device of the present invention; Figure 3 It is a schematic diagram of the cross-sectional structure of the device of the present invention.

[0012] In the figure: 1. Shell; 2. Historical cycle perception module; 3. Spatial and temporal self-attention module; 4. Storage module; 5. PCB board; 6. Processing module; 7. Positioning module; 8. Wireless communication module. DETAILED DESCRIPTION

[0013] 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 only 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 technical personnel in this field without creative work are within the scope of protection of the present invention.

[0014] See also Figure 1-Figure 3 The present invention provides a technical solution: a spatiotemporal fusion perception transformation device and method based on traffic flow prediction, including a historical cycle perception module 2, traffic data processing, information embedding, a dynamic traffic interaction diagram, and a spatial and temporal self-attention module 3, the steps are as follows: S1: First, the historical traffic data is processed using the K-POP clustering algorithm to obtain a representative short-term traffic pattern set containing key historical information; S2: Historical cycle perception module 2 uses the self-attention mechanism to embed past periodic feature information into traffic data, extending the traffic characteristics of a single node beyond a single moment; S3: Through the dynamic traffic interaction graph generation algorithm, the impact of historical status on the current status is taken into account, so as to distinguish the interaction relationship between different periods with similar traffic flow status; S4: spatial and temporal self-attention module 3 to capture spatial and temporal dependencies for further graph optimization; S5: The final output graph will serve as the basis for traffic flow prediction.

[0015] The historical cycle perception module 2 is used to capture historical information and introduce multi-scale time correlation to distinguish traffic cycles and accurately perceive time delays.

[0016] The dynamic traffic interaction graph is generated by a novel dynamic traffic interaction graph generation algorithm based on the semantic similarity between features embedded in historical cycle information. The generated graph contains the semantic hierarchical interaction relationships between different traffic nodes.

[0017] The spatial and temporal self-attention module 3 includes spatial self-attention and temporal self-attention. The spatial self-attention uses the graph attention mechanism to simultaneously capture the physical-level spatial dependencies and the semantic-level spatial dependencies, and the temporal self-attention captures the complex dynamic temporal dependencies in traffic data.

[0018] A spatiotemporal fusion perception transformation device based on traffic flow prediction includes a shell 1, which is used to be loaded on a vehicle in use and to protect internal components. A historical cycle perception module 2 is fixedly connected to the bottom inner wall of the shell 1, a space and time self-attention module 3 is fixedly connected to the bottom inner wall of the shell 1, a storage module 4 for information storage is fixedly connected to the bottom inner wall of the shell 1, a PCB board 5 is fixedly connected to the bottom inner wall of the shell 1, and the PCB board 5 is electrically connected to the historical cycle perception module 2, the space and time self-attention module 3 and the storage module 4, a processing module 6 for information processing and calculation is fixedly connected to the top of the PCB board 5, a positioning module 7 for vehicle positioning is fixedly connected to the top of the processing module 6, and a wireless communication module 8 for network communication is fixedly connected to the top of the PCB board 5.

[0019] The above shows and describes the basic principles, main features and advantages of the present invention. Technical personnel in this industry should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions are only preferred examples of the present invention and are not used to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A device and method for temporal and spatial fusion perception transformation based on traffic flow prediction, characterized in that: It includes a historical cycle perception module (2), traffic data processing, information embedding, dynamic traffic interaction graph, and spatial and temporal self-attention module (3). The steps are as follows: S1: First, the historical traffic data is processed using the K-POP clustering algorithm to obtain a representative short-term traffic pattern set containing key historical information; S2: Historical cycle perception module (2) uses the self-attention mechanism to embed past periodic feature information into traffic data, extending the traffic characteristics of a single node beyond a single moment; S3: Through the dynamic traffic interaction graph generation algorithm, the impact of historical status on the current status is taken into account, so as to distinguish the interaction relationship between different periods with similar traffic flow status; S4: consists of spatial and temporal self-attention modules (3) to capture spatial and temporal dependencies to further optimize the graph; S5: The final output graph will serve as the basis for traffic flow prediction.

2. The device and method for time-space fusion perception transformation based on traffic flow prediction according to claim 1 is characterized in that: The historical cycle perception module (2) is used to capture historical information and introduce multi-scale time correlation, distinguish traffic cycles and accurately perceive time delays.

3. The device and method for time-space fusion perception transformation based on traffic flow prediction according to claim 1 is characterized in that: The dynamic traffic interaction graph is generated by a novel dynamic traffic interaction graph generation algorithm based on semantic similarity between features embedded in historical cycle information, and the generated graph contains semantic hierarchical interaction relationships between different traffic nodes.

4. The device and method for time-space fusion perception transformation based on traffic flow prediction according to claim 1 is characterized in that: The spatial and temporal self-attention module (3) includes spatial self-attention and temporal self-attention, wherein the spatial self-attention utilizes a graph attention mechanism to simultaneously capture physical-level spatial dependencies and semantic-level spatial dependencies, and the temporal self-attention captures complex dynamic temporal dependencies in traffic data.

5. A spatiotemporal fusion perception transformation device based on traffic flow prediction, characterized in that: The invention comprises a shell (1), wherein the shell (1) is used to be loaded on a vehicle for use and to protect internal components, a historical cycle perception module (2) is fixedly connected to the inner wall at the bottom of the shell (1), a spatial and temporal self-attention module (3) is fixedly connected to the inner wall at the bottom of the shell (1), a storage module (4) for information storage is fixedly connected to the inner wall at the bottom of the shell (1), a PCB board (5) is fixedly connected to the inner wall at the bottom of the shell (1), and the PCB board (5) is electrically connected to the historical cycle perception module (2), the spatial and temporal self-attention module (3) and the storage module (4), a processing module (6) for information processing and calculation is fixedly connected to the top of the PCB board (5), a positioning module (7) for positioning the vehicle position is fixedly connected to the top of the processing module (6), and a wireless communication module (8) for network communication is fixedly connected to the top of the PCB board (5).

Images