Traffic information prediction device and method

Abstract

This disclosure relates to a traffic information prediction device and method. The traffic information prediction device includes: a communication unit configured to receive real-time traffic information data and past traffic information data; and a processor configured to set up a link group including multiple links, predict the link group travel time (ETA) using a first artificial intelligence model based on the real-time traffic information data and past traffic information data, and predict link-level traffic information using a second artificial intelligence model based on the predicted link group travel time (ETA).

Description

Technical Field

[0001] This disclosure relates to a traffic information prediction device and method that collects real-time data and then predicts traffic information based on estimated time of arrival (ETA). Background Technology

[0002] Typically, navigation devices installed in vehicles provide map information and guide passengers to their destination based on the map.

[0003] This navigation device determines the vehicle's current location based on location information received from Global Positioning System (GPS) satellites, and then reads data related to the current location from an internal road database (DB) or a received road DB to display the read data along with the vehicle's location on the screen.

[0004] Recently, navigation devices can not only set routes and provide guidance along those routes, but also calculate the estimated arrival time of a vehicle to its destination based on traffic conditions and provide that estimated arrival time to the user.

[0005] In this scenario, the navigation device can not only request route searches from the server for route exploration, but also receive data from devices that manage and analyze traffic conditions, and calculate estimated arrival times based on the received data.

[0006] However, the estimated arrival time prediction uses the vehicle's entry and exit times based on link measurements, and the problem is that the accuracy of the predicted travel time for each link is low due to the variability of link-level data. Specifically, link-level data limits the ability to derive patterns for predicting the required time. Summary of the Invention

[0007] The purpose of this disclosure is to provide a traffic information prediction device and method that calculates the travel time of each link group unit by setting up a link group including multiple links, and predicts link-based traffic information based on the travel time.

[0008] A traffic information prediction device according to one aspect of the present disclosure includes: a communication unit configured to receive real-time traffic information data and past traffic information data; and a processor configured to set up a link group including multiple links, predict the travel time (ETA) of the link group using a first artificial intelligence model based on the real-time traffic information data and past traffic information data, and predict link-level traffic information using a second artificial intelligence model based on the predicted travel time (ETA) of the link group.

[0009] The processor can be configured to input traffic information data, including speed, traffic flow, accidents, and location information, into a first artificial intelligence model and a second artificial intelligence model, and input the link-group level predicted travel time (ETA) calculated from the first artificial intelligence model into the second artificial intelligence model, and then output link-level predicted traffic information.

[0010] The processor can calculate the travel time (ETA) of a link group by summing the travel times (ETA) of multiple links at the same nth time point via a first artificial intelligence model.

[0011] The processor can calculate the predicted speed for a link group by converting the travel time (ETA) of the link group into speed based on a first artificial intelligence model.

[0012] The processor can predict link-level traffic information based on a second artificial intelligence model, using traffic information heatmaps of multiple links contained in a link group.

[0013] The processor can input link attribute information, including spatial information, link information, traffic information, and accident information, as well as link group attribute information, including length information, traffic information, and time information, into each of the first artificial intelligence model and the second artificial intelligence model.

[0014] The first and second artificial intelligence models can be implemented by modeling traffic information data based on embedding and transformation modules.

[0015] The converter module can learn the importance of each link in a group of links based on the degree of impact of traffic condition predictions.

[0016] According to another aspect of the traffic information prediction method disclosed herein, the method includes: a processor using a first artificial intelligence model to predict the travel time (ETA) of a link group based on real-time traffic information data and past traffic information data; and a processor using a second artificial intelligence model to predict link-level traffic information based on the travel time (ETA) of the link group.

[0017] In the process of predicting the travel time (ETA) of a link group, the processor can calculate the travel time (ETA) of the link group by summing the travel times (ETA) of each link at the same nth time point for multiple links included in the link group through a first artificial intelligence model.

[0018] In the process of predicting the travel time (ETA) of a link group, the processor can calculate the predicted speed of the link group by converting the travel time (ETA) into speed based on the first artificial intelligence model.

[0019] In the process of predicting link-level traffic information, the processor can use a second artificial intelligence model to predict link-level traffic information using traffic information heatmaps of multiple links contained in the link group.

[0020] Traffic information prediction methods may also include having a processor input link attribute information, including spatial information, link information, traffic information, and accident information, as well as link group attribute information, including length information, traffic information, and time information, into each of the first artificial intelligence model and the second artificial intelligence model.

[0021] During the input process, the processor can be configured to input traffic information data, including speed, traffic flow, accidents, and location information, into a first artificial intelligence model and a second artificial intelligence model, and input the link group-level predicted ETA calculated from the first artificial intelligence model into the second artificial intelligence model.

[0022] The first and second artificial intelligence models can be implemented by modeling traffic information data based on embedding and transformation modules.

[0023] The converter module can learn the importance of each link in a group of links based on the degree of impact of traffic condition predictions.

[0024] According to this disclosure, the traffic information prediction device and method according to the embodiments of this disclosure can predict travel time by analyzing the entry and exit points of vehicles based on link groups, and can convert the predicted travel time into link-level data, thereby improving prediction accuracy.

[0025] The traffic information prediction device and method according to this disclosure can easily derive traffic condition patterns by analyzing real-time data based on link groups and easily predict link-based traffic information.

[0026] The traffic information prediction device and method disclosed herein can provide accurate traffic information, thereby improving the ease and efficiency of route planning. Attached Figure Description

[0027] Figure 1 This is a diagram that briefly illustrates the configuration of a system including a traffic information prediction device according to an embodiment of the present disclosure.

[0028] Figure 2 This is a block diagram that briefly illustrates the control configuration of a traffic information prediction device according to an embodiment of the present disclosure.

[0029] Figure 3 This is a diagram illustrating traffic information prediction using an artificial intelligence model of a traffic information prediction device according to embodiments of the present disclosure.

[0030] Figure 4 This is a diagram illustrating the process of predicting traffic information based on link groups in a traffic information prediction device according to an embodiment of the present disclosure.

[0031] Figure 5 This is a diagram illustrating the process of converting link-group-based traffic information into link-level traffic information in a traffic information prediction device according to an embodiment of the present disclosure.

[0032] Figure 6 This is an example diagram illustrating an artificial intelligence model of a traffic information prediction device according to an embodiment of the present disclosure.

[0033] Figure 7 This is a flowchart illustrating a traffic information prediction method using a traffic information prediction device according to an embodiment of the present disclosure. Detailed Implementation

[0034] The present disclosure will be described below with reference to the accompanying drawings.

[0035] In this process, for clarity and convenience of description, the thickness of the lines or the dimensions of the components shown in the accompanying drawings may be enlarged. Furthermore, the following terms are defined with reference to the functions in this disclosure and may vary according to the intentions or practices of the user or operator. Therefore, these terms should be defined based on the entire contents of this specification.

[0036] Figure 1 This is a diagram that briefly illustrates the configuration of a system including a traffic information prediction device according to an embodiment of the present disclosure.

[0037] refer to Figure 1 The system including traffic information prediction equipment according to this disclosure may include a vehicle 10, a route search server 20, a traffic server 30, and a prediction device 100.

[0038] Vehicle 10 can request route search server 20 to search for routes through navigation terminal.

[0039] The route search server 20 can request data from the traffic server 30 in response to a route search request received from the vehicle 10, and can receive real-time traffic information and predicted future traffic information.

[0040] Here, traffic server 30 can store data about road information, road infrastructure installed on the road, and links in a database (DB). Traffic server 30 can store real-time traffic information received from the position sensors of vehicle 10 or road infrastructure, and can provide traffic information in response to requests from route search server 20. For example, road infrastructure is configured to collect information about road and traffic conditions and includes closed-circuit television (CCTV), road weather information system (RWIS), vehicle detection system (VDS), variable message signs (VMS), slope metering system (RMS), lane control system (LCS), etc.

[0041] Traffic server 30 can provide data on road and traffic conditions to vehicle 10 and route search server 20. Here, traffic server 30 may include a database (DB) for storing the data.

[0042] The prediction device 100 can be connected to the traffic server 30, which can set up multiple links on each road into a link group, and can predict traffic information at future points in time by analyzing real-time traffic information based on the link group using real-time traffic information.

[0043] The prediction device 100 can predict the travel time at future points in time (also known as "ETA: Estimated Time of Arrival") based on link groups, convert the predicted travel time into link-level information, and generate link-level predicted traffic information. The prediction device 100 provides the link-level predicted traffic information to the traffic server 30. In some cases, the prediction device 100 may be included in the traffic server 30.

[0044] Therefore, traffic server 30 can provide link-level predicted traffic information to route search server 20.

[0045] The route search server 20 can set up multiple routes based on the predicted traffic information data received from the traffic server 30, calculate the estimated arrival time for each route, and provide the calculation results to the navigation terminal of the vehicle 10.

[0046] Therefore, vehicle 10 can display each received driving route and estimated arrival time on the screen of the navigation terminal.

[0047] Figure 2 This is a block diagram that briefly illustrates the control configuration of a traffic information prediction device according to an embodiment of the present disclosure.

[0048] refer to Figure 2The prediction device 100 may include a memory 120, a communication unit 130, and a processor 110. Furthermore, the prediction device 100 may also include an artificial intelligence model and an independent database (DB) 140 for training the artificial intelligence model.

[0049] The memory 120 can store data transmitted / received through the communication unit 130, artificial intelligence model data, and analysis data from the processor 110. Furthermore, the memory 120 can temporarily store training data for the artificial intelligence model and real-time traffic information received from the traffic server 30. The memory 120 may include storage devices, including volatile memory (such as random access memory (RAM)) and non-volatile memory (such as read-only memory (ROM), electrically erasable programmable ROM (EEPROM), or flash memory).

[0050] Database 140 can store large amounts of traffic information used to predict artificial intelligence models. For example, database 140 can store training data for artificial intelligence models, real-time traffic information data and past traffic information data received from traffic server 30, link group-level travel time data and link-level predicted traffic information data predicted by artificial intelligence models.

[0051] The communication unit 130 can receive data stored in the traffic server 30 while communicating with the traffic server 30, and can send data calculated by the processor 110 to the traffic server 30. The communication unit 130 can communicate with previously registered terminals or vehicles 10.

[0052] The communication unit 130 can communicate with the communication server 30 via a wired or wireless communication module, and in some cases can use the communication bus of the communication server 30 to send or receive data within the communication server 30. For example, the communication unit 130 can communicate via at least one of short-range communication such as Wi-Fi or Bluetooth, mobile communication, wireless access in a vehicular environment (WAVE) for communication with vehicles and road infrastructure, and vehicle-to-everything (V2X) for vehicle-to-object communication, and can also send or receive data via controller area network (CAN) communication or local interconnection network (LIN) communication.

[0053] Processor 110 may include at least one microprocessor. Processor 110 may operate based on data and algorithm data stored in memory.

[0054] The processor 110 can learn past and real-time traffic information based on an artificial intelligence model and can predict future traffic information. The processor 110 can store data in the database 140 when using the artificial intelligence model to predict traffic information, load data into the memory 120, process data in real time, and store the predicted traffic information in the database (DB) 140.

[0055] The processor 110 can set up a link group comprising multiple links based on links (road segments) set at multiple locations on the road. The processor 110 can set up a single link group containing links contained in a specific area unit or related area. Furthermore, for road segments set up based on areas, etc., when establishing the route of vehicle 10, the processor 110 can set up a link group containing multiple links in the road segment.

[0056] The processor 110 can classify traffic information data collected from each link and can predict link group-level travel time (ETA) by inputting past traffic information data and real-time traffic information data of each link group into a first artificial intelligence model, which is an artificial intelligence model for link groups.

[0057] Here, the processor 110 can predict the travel time (ETA) based on a first artificial intelligence model by deriving patterns of link group-level traffic information. The predicted travel time (ETA) can include information about the travel time (ETA) of a vehicle traveling through the corresponding link group at a future point in time.

[0058] The processor 110 can input the link attribute information and link group attribute information as input data into the first artificial intelligence model, and calculate the travel time (ETA) of each link at a certain time interval relative to the predicted time point. It can also calculate the predicted travel time (ETA) of each link group by summing the travel times (ETA) of each link at the future time point.

[0059] Furthermore, the processor 110 predicts the time of travel (ETA) at the link group level and then inputs the link group-level ETA into a second artificial intelligence model to calculate link-level predicted traffic information from the link group-level ETA. The processor 110 can calculate predicted traffic information for each link at a future time point for multiple links contained in a link group.

[0060] The processor 110 can retrain the first and second artificial intelligence models based on predicted traffic information.

[0061] The processor 110 can store the link-group level travel time (ETA), the link-group level predicted travel time (ETA), and the link-level predicted traffic information pattern data in the database 140. Furthermore, the processor 110 transmits the stored link-level predicted traffic information to the traffic server 30 via the communication unit 130.

[0062] Figure 3 This is a diagram illustrating traffic information prediction using an artificial intelligence model of a traffic information prediction device according to embodiments of the present disclosure.

[0063] refer to Figure 3 The prediction device 100 can be configured to include a link group with multiple links, and can calculate the predicted traffic information of the link group.

[0064] The processor 110 can reference national highways or expressways and, depending on the route settings, set up link groups based on at least one of specific road segments, areas, and intersections. For example, when setting up a route from Seoul to Busan, the processor 110 can set up at least one link group, which includes multiple links around multiple reference points including Seoul Toll Station, Dongtan, Anseong, Cheonan, Nami, Hoedok, Pyeongyong, Yeongdong, Gimcheon, Dongdaegu, and Yeongcheon.

[0065] Processor 110 can use a first artificial intelligence model to predict traffic information for future timeframes for a link group. In step S10, processor 110 can predict the link group-level travel time (ETA) using the first artificial intelligence model, convert the predicted value into a speed value, and output the speed value in step S20. For example, processor 110 can predict traffic information for future timeframes by predicting the travel time (ETA) over a 5-hour period at 5-minute intervals.

[0066] Here, the processor 110 can use a first artificial intelligence model to calculate the travel time (ETA) of multiple links included in the link group at the nth time point corresponding to a future time point, and can calculate the travel time (ETA) of the link group by summing the travel times of each link at the nth time point. That is, the processor 110 calculates the travel time of the link group by summing the travel time of link A at the first time point and the travel time of link B, rather than summing the travel time of link A at the first time point and the travel time of link B at the second time point.

[0067] In step S30, the processor 110 can convert the predicted travel time (ETA) of the link group into speed and input the speed into the second artificial intelligence model. In step S40, the processor 110 can convert the link group level into the link level and calculate the link-level predicted traffic information.

[0068] The processor 110 can use a second artificial intelligence model to generate link-level predicted speeds for multiple links contained in a link group, and can calculate the link-level predicted speeds as link-level predicted traffic information.

[0069] The prediction device 100 uses two artificial intelligence models to compare past data with real-time data and derives patterns based on link groups to predict the estimated travel time (ETA) required to traverse a link group. Here, the prediction device 100 can calculate the link-based travel time (ETA) based on past traffic information data for multiple links contained in each link group.

[0070] Here, the prediction device 100 can predict link-based traffic information by reflecting the deceleration segment B in the link of the road segment A where the speed is reduced due to traffic congestion, etc.

[0071] Figure 4 This is a diagram illustrating the process of predicting traffic information based on link groups in a traffic information prediction device according to an embodiment of the present disclosure.

[0072] refer to Figure 4 The processor 110 can calculate the travel time (ETA) of a link group based on a first artificial intelligence model. The processor 110 can collect data in step S110, input the data into the first artificial intelligence model in step S120, and calculate the predicted travel time (ETA) for each link group in step S130 by summing the travel times of each link at specific time intervals corresponding to future time points.

[0073] Processor 110 can input link-level traffic information as input data into the first artificial intelligence model. For example, traffic information may include speed, traffic flow, accidents, and location information.

[0074] Because future congestion may vary based on changes in traffic flow over the past, processor 110 can more accurately predict traffic information by inputting data, including past traffic flow information and real-time traffic flow information, into the first artificial intelligence model.

[0075] Here, the processor 110 can use the link attribute information and link group attribute information of traffic information as input data for the first artificial intelligence model.

[0076] Link attribute information can include spatial information, link information, traffic information, and accident information. Spatial information can be static or dynamic location information for each link; link information can include links of corresponding road type and link length; traffic information can include real-time speed, past speed data, and traffic flow; and accident information can include accident information (accident alerts, construction information, and event or assembly information) for links included in the route.

[0077] Link group attribute information can include length information, traffic information, and time information. Length information can include route length and straight-line distance between the origin and destination (origin-destination: OD), traffic information can include real-time speed, past speed, and traffic flow, and time information can include weekday information and the start time of travel (one-minute intervals).

[0078] The first AI model can be implemented by modeling traffic information based on a linear transformer, and can be generated by modeling attention to specific parts of the information. The first AI model can convert input data into vectors, calculate similarity, and compute attention weights. The first AI model can then compute a weighted sum by multiplying the attention weights by the vectors.

[0079] Therefore, the first artificial intelligence model predicts the travel time of each link within a 5-hour time period for each link group at 5-minute intervals (e.g., 5 minutes, 10 minutes, and 15 minutes). The processor 110 can, based on the output prediction results, sum the travel times of multiple links in the link group at the nth time, which is a specific future time point, by summing the travel times of the corresponding links at the first time point as a future time point and summing the travel times of the corresponding links at the second time point as a future time point, thereby outputting the link group-level predicted travel time (ETA). That is, the link group-level travel time (ETA) can be calculated by summing the travel times of the corresponding links at the same nth time point.

[0080] For example, when links A, B to Z are included in a link group, the processor 110 can predict the travel time (ETA) of the link group at a 5-minute future time point by using a first artificial intelligence model to sum the travel time (ETA) of links A, B, and Z at a 5-minute future time point.

[0081] In this way, processor 110 can calculate traffic information for link groups over a 5-hour time period at 5-minute intervals for future times. Because processor 110 uses real-time data as input, traffic congestion can be reflected in the prediction results.

[0082] Figure 5 This is a diagram illustrating the process of converting link-group-based traffic information into link-level traffic information in a traffic information prediction device according to an embodiment of the present disclosure.

[0083] refer to Figure 5 The processor 110 can output the link group-level predicted travel time (ETA) by using a first artificial intelligence model to calculate the link group-level travel time (ETA) for each future time point, and then calculate the link-level predicted traffic information by inputting the link group-level travel time (ETA) into a second artificial intelligence model.

[0084] The processor 110 can convert the link group-level travel time at each future time point into speed, and can convert the speed into traffic information or extraction patterns for each link contained in the link group.

[0085] Here, the processor 110 can input link attribute information and link group attribute information as input data into the second artificial intelligence model, and can additionally input predicted travel time (ETA) into the second artificial intelligence model. The predicted travel time (ETA) is the predicted link group-level travel time for each future time point, and is the link group-level predicted traffic information previously calculated by the first artificial intelligence model.

[0086] Similar to the first AI model, the second AI model can be implemented by modeling traffic information based on a linear transformer and can be generated through attention modeling that processes information by focusing on specific parts. Here, the second AI model can predict link-level traffic information based on heatmaps of traffic information from multiple links contained in a link group.

[0087] The second AI model can be generated by learning data about multiple links contained in a link group. This second AI model can not only learn traffic information based on links, but also use data generated from learning performed on link groups to calculate more accurate predictions.

[0088] The second artificial intelligence model can consider the link contribution information and link group attribute information of the link group-level predicted travel time (i.e., the travel time at the first time point, the travel time at the second time point) to the travel time at the m-th time point, in order to predict the link-level travel time.

[0089] Therefore, the second AI model can predict traffic information based on links, including travel time at each future point in time. For example, link-level traffic information prediction can include predictions (travel time or speed) for a maximum 5-hour period obtained by predicting traffic information at 5-minute intervals as link-level future points in time.

[0090] Figure 6 This is an example diagram illustrating an artificial intelligence model of a traffic information prediction device according to an embodiment of the present disclosure.

[0091] refer to Figure 6 The processor 110 can use two artificial intelligence models to predict travel time (ETA) based on link groups and predict traffic information again based on links, thereby accurately predicting future traffic information.

[0092] Each AI model receives link attribute information, speed attribute information, accident attribute information, and link group attribute information in step S210, analyzes the data through the embedding module and transformation module in step S220, and outputs the predicted value in step S230.

[0093] Artificial intelligence models can be divided into embedded modules and linear transformer (attention) modules.

[0094] The embedding module is configured to extract feature vectors from various features of traffic information and concatenate multiple strings after extracting the feature vectors of link attribute features. Furthermore, the embedding module can extract feature vectors from various features of link group attributes and can generate two-dimensional (2D) data (X) of link attributes and link group attributes. emb ).

[0095] The embedding module applies geospatial and traffic variables to link counts in link attribute information from multiple input datasets and calculates time variables for link group attribute information, thereby generating 2D data (X) about links and time. emb For example, 2D data (X) emb (This can include Link-1 X) emb1 Link-1 X emb3 Link-1 X embN Time X embN+1 wait.

[0096] The linear transformation module employs linear attention, which has a time complexity of O(n) and requires relatively low memory usage and computational cost. This is because the linear transformation module is based on the 2D data (X) generated by the link generation embedding module. emb Therefore, the importance of learning which links have a greater impact on future traffic condition prediction (linear link attention) is not the importance of learning features.

[0097] That is, the transformer module can process 2D data (X) emb Attention weights are applied to learn links that significantly influence traffic condition predictions. In this case, the linear transformation module can convert the weights (A) (where the weights A) into the weights (A'). ij∈[0.1]) applied to i th X emb And sum the results to output the function f(X). emb The function f(X) emb ) can include (AV)1+X emb1 (AV)2+X emb2 wait.

[0098] Therefore, the first AI model can predict travel time (ETA) based on link groups, and the second AI model can predict link-level traffic information based on predicted ETA.

[0099] Figure 7 This is a flowchart illustrating a traffic information prediction method using a traffic information prediction device according to an embodiment of the present disclosure.

[0100] refer to Figure 7 The prediction device 100 is connected to the traffic server 30 and receives traffic information prediction data from the traffic server 30.

[0101] In step S310, processor 110 collects past traffic information data and real-time traffic information data from traffic server 30. In step S320, processor 110 stores the past traffic information data and real-time traffic information data in database 140.

[0102] The processor 110 inputs the link attribute information and link group attribute information from the collected traffic information into the first artificial intelligence model, which serves as the link group-level ETA prediction model, and then analyzes the data in real time in step S330.

[0103] Here, as the first artificial intelligence model, a transformer model (attention) can be used. In step S340, the first artificial intelligence model can predict the travel time (ETA) on a link group basis based on the input traffic information data (i.e., speed, traffic flow, accidents, and location information). When calculating the link group-level predicted travel time (ETA) from the first artificial intelligence model, the processor 110 can convert the link group-level predicted travel time (ETA) into speed data and output the speed data.

[0104] In step S350, the processor 110 inputs the link attribute information and link group attribute information into the second artificial intelligence model, and inputs the predicted travel time (ETA) output from the first artificial intelligence model into the second artificial intelligence model to analyze the input data.

[0105] In step S360, the second artificial intelligence model can predict link-level traffic information from link group-level predicted travel time (ETA) based on the input data. Here, the second artificial intelligence model can predict link-level traffic information based on traffic information heatmaps of multiple links contained in a link group.

[0106] In step S370, the processor 110 sends the prediction data (i.e., link-level predicted traffic information) to the traffic server 30 via the communication unit 130. The traffic server 30 can store the received predicted traffic information and can send the predicted traffic information to the route search server 20 in response to a request.

[0107] Furthermore, in step S380, processor 110 can store real-time data and predicted data for each future time point. Processor 110 can use the calculated data to retrain the artificial intelligence model.

[0108] Therefore, the traffic information prediction device and method according to this disclosure can be configured to include a link group comprising multiple links, analyze the entry and exit points of vehicles based on the link group to predict traffic information, and convert the predicted traffic information into link-level information, thereby improving the accuracy of the prediction.

[0109] Furthermore, the traffic information prediction device and method according to this disclosure can easily derive patterns by analyzing real-time data based on link groups, can easily predict traffic information, and can provide accurate traffic information, thereby improving the ease and efficiency of route setup.

[0110] While this disclosure has been described with respect to specific embodiments shown in the accompanying drawings, these embodiments are for illustrative purposes only, and it will be apparent to those skilled in the art that various modifications and equivalent implementations may be made without departing from the scope of the art to which this disclosure pertains. Therefore, the scope of this disclosure should be defined by the appended claims.

Claims

1. A traffic information prediction device, comprising: The communication unit is configured to receive real-time traffic information data and past traffic information data; as well as The processor is configured to set up a link group comprising multiple links, use a first artificial intelligence model to predict the travel time of the link group based on the real-time traffic information data and the past traffic information data, and use a second artificial intelligence model to predict link-level traffic information based on the predicted travel time of the link group.

2. The traffic information prediction device according to claim 1, wherein, The processor is configured to: Traffic information data, including speed, traffic flow, accidents, and location information, is input into the first artificial intelligence model and the second artificial intelligence model, and The link-group level predicted travel time calculated from the first artificial intelligence model is input into the second artificial intelligence model, and then the link-level predicted traffic information is output.

3. The traffic information prediction device according to claim 1, wherein, The processor calculates the travel time of the link group by summing the travel times of the multiple links at the same nth time point via the first artificial intelligence model.

4. The traffic information prediction device according to claim 1, wherein, The processor calculates the predicted speed for the link group by converting the travel time of the link group into speed based on the first artificial intelligence model.

5. The traffic information prediction device according to claim 1, wherein, The processor, based on the second artificial intelligence model, uses traffic information heatmaps of the multiple links included in the link group to predict the link-level traffic information.

6. The traffic information prediction device according to claim 1, wherein, The processor inputs link attribute information, including spatial information, link information, traffic information, and accident information, as well as link group attribute information, including length information, traffic information, and time information, into each of the first artificial intelligence model and the second artificial intelligence model.

7. The traffic information prediction device according to claim 1, wherein, The first artificial intelligence model and the second artificial intelligence model are implemented by modeling traffic information data based on an embedding module and a converter module.

8. The traffic information prediction device according to claim 7, wherein, The converter module learns the importance of each of the multiple links included in the link group based on the degree of impact of traffic condition prediction.

9. A traffic information prediction method, comprising: The processor uses a first artificial intelligence model to predict the travel time of the link group based on real-time traffic information data and past traffic information data; and The processor uses a second artificial intelligence model to predict link-level traffic information based on the travel time of the link group.

10. The traffic information prediction method according to claim 9, wherein, In predicting the travel time of the link group, the processor calculates the travel time of the link group by summing the travel times of each link at the same nth time point for multiple links included in the link group via the first artificial intelligence model.

11. The traffic information prediction method according to claim 9, wherein, In predicting the travel time of the link group, the processor calculates the predicted speed of the link group by converting the travel time of the link group into speed based on the first artificial intelligence model.

12. The traffic information prediction method according to claim 9, wherein, In the process of predicting the link-level traffic information The processor, based on the second artificial intelligence model, uses traffic information heatmaps of multiple links included in the link group to predict the link-level traffic information.

13. The traffic information prediction method according to claim 9, further comprising: The processor inputs link attribute information, including spatial information, link information, traffic information, and accident information, as well as link group attribute information, including length information, traffic information, and time information, into each of the first artificial intelligence model and the second artificial intelligence model.

14. The traffic information prediction method according to claim 13, wherein, During the input process, the processor is configured to: Traffic information data, including speed, traffic flow, accidents, and location information, is input into the first artificial intelligence model and the second artificial intelligence model, and The link group-level predicted travel time calculated from the first artificial intelligence model is input into the second artificial intelligence model.

15. The traffic information prediction method according to claim 9, wherein, The first artificial intelligence model and the second artificial intelligence model are implemented by modeling traffic information data based on an embedding module and a converter module.

16. The traffic information prediction method according to claim 15, wherein, The converter module learns the importance of each link in the multiple links included in the link group based on the degree of impact of traffic condition prediction.

Images