Travel time prediction method and travel time prediction program

Abstract

Quickly determine the travel time for the predicted interval. [Solution] The travel time prediction method involves generating a travel time prediction model using travel history data, which includes location information of the departure point, location information of the arrival point, and travel time information between the departure point and the arrival point, as training data. The computer then inputs the location information of the departure point and the location information of the arrival point of the section to be predicted into the travel time prediction model to predict the travel time of the section to be predicted.

Description

Technical Field

[0001] The present invention relates to a travel time prediction method and a travel time prediction program.

Background Art

[0002] In local governments and the like, the number of emergency cases has been increasing year by year due to the increasing demand for emergency transportation accompanying the increase in the elderly population, the increase in heatstroke patients, and so on. Local governments and the like may review the team deployment including increasing the number of teams in order to shorten the time until the emergency team (emergency medical technicians and firefighters) arrives at the scene.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] When simulating the team deployment in a local government, when sending any team to the scene, it is necessary to find out how much time is required for the movement to the scene for each combination of the team and the scene. Conventionally, a technique for predicting the arrival time when a moving object arrives at a destination is known (see, for example, Patent Document 1 and the like).

[0005] However, when it is necessary to find out a large number of travel times as in the above simulation, it is necessary to find out each travel time in a short time.

[0006] In one aspect, an object of the present invention is to provide a travel time prediction method and a travel time prediction program capable of finding out the travel time of a prediction target section in a short time.

Means for Solving the Problems

[0007] In one embodiment, the travel time prediction method is a method in which a computer performs the following process: generating a travel time prediction model using travel history data, which includes location information of a departure point, location information of an arrival point, and information on the travel time between the departure point and the arrival point, as training data; and predicting the travel time of a target section by inputting the location information of the departure point and location information of the arrival point of the target section into the travel time prediction model.

[0008] In another embodiment, the travel time prediction method is a method in which a computer performs the following processing: acquiring location information of the starting point and the destination point of a prediction target section; predicting the travel time between the starting point and the destination point of the prediction target section based on the location information of the starting point and the destination point of the prediction target section and a travel time prediction model that predicts the travel time between the starting point and the destination point based on the location information of the starting point and the destination point; and outputting the predicted travel time between the starting point and the destination point of the prediction target section. [Effects of the Invention]

[0009] The travel time for the predicted interval can be determined in a short amount of time. [Brief explanation of the drawing]

[0010] [Figure 1] Figure 1 is a schematic diagram showing the configuration of a troop deployment decision support system according to one embodiment. [Figure 2] Figure 2(a) shows an example of the hardware configuration of a user terminal, and Figure 2(b) shows an example of the hardware configuration of a server. [Figure 3] Figure 3 is a functional block diagram of the server. [Figure 4] Figure 4 shows the data structure of the dispatch history database. [Figure 5] Figure 5 shows an overview of the travel time prediction model generation by the learning unit. [Figure 6] Figures 6(a) and 6(b) illustrate the effects of the travel time prediction model. [Figure 7] Figure 7 shows information on potential deployment locations. [Figure 8] Figure 8 is a map that shows the deployment information for each of the deployment candidates in Figure 7 at a specific point in time. [Figure 9] Figure 9 is a flowchart showing the processing performed by the server during the learning phase. [Figure 10] Figure 10 is a flowchart showing the simulation process performed by the server. [Figure 11] Figure 11 is a flowchart showing the specific processing steps in step S36 of Figure 10. [Figure 12] Figures 12(a) and 12(b) are diagrams illustrating the process shown in Figure 11. [Figure 13] Figure 13 is a diagram illustrating the information output by the output unit. [Figure 14] Figure 14 is a diagram illustrating a modified example. [Modes for carrying out the invention]

[0011] Below, one embodiment of the troop deployment decision support system will be described in detail based on Figures 1 to 14.

[0012] Figure 1 schematically shows the configuration of a troop deployment decision support system 100 according to one embodiment. The troop deployment decision support system 100 of this embodiment is a system that provides the necessary information and supports consideration when a local government considers how many troops (ambulance teams and fire teams) should be deployed to each location within the local government (existing fire stations, new fire stations, etc.). In this embodiment, the troops to be deployed to each location are ambulance teams, and each ambulance team will dispatch the same type of vehicle (ambulance).

[0013] As shown in Figure 1, the troop deployment decision support system 100 comprises a user terminal 70 and a server 10. The user terminal 70 and the server 10 are connected to a network 80 such as the internet or a mobile phone network.

[0014] The user terminal 70 is a PC (Personal Computer), a tablet terminal, a smartphone, etc., and is a terminal operated by local government officials or the like. At the user terminal 70, local government officials or the like input information on deployment candidates (information specifying how many teams should be deployed at which locations). Also, the user terminal 70 collects information on past dispatch histories from a database or the like that manages the dispatch histories. The user terminal 70 transmits the input information on deployment candidates and the collected information on past dispatch histories to the server 10. Further, the user terminal 70 acquires and displays information for supporting the consideration of team deployment transmitted from the server 10.

[0015] Fig. 2(a) shows the hardware configuration of the user terminal 70. As shown in Fig. 2(a), the user terminal 70 includes a CPU (Central Processing Unit) 190, a ROM (Read Only Memory) 192, a RAM (Random Access Memory) 194, a storage (such as an SSD (Solid State Drive), an HDD (Hard Disk Drive)) 196, a network interface 197, a display unit 193, an input unit 195, a drive 199 for a portable storage medium, etc. Each of these constituent parts of the user terminal 70 is connected to a bus 198.

[0016] Returning to Fig. 1, based on the information (information on deployment candidates and past dispatch histories) acquired from the user terminal 70, the server 10 generates information for supporting the consideration of team deployment and transmits it to the user terminal 70.

[0017] Figure 2(b) shows the hardware configuration of server 10. As shown in Figure 2(b), server 10 includes a CPU 90, ROM 92, RAM 94, storage 96, network interface 97, and a portable storage medium drive 99, etc. Each of these components of server 10 is connected to bus 98. In server 10, the CPU 90 executes programs (including travel time prediction programs) stored in ROM 92 or storage 96, or programs read from the portable storage medium 91 by the portable storage medium drive 99, thereby realizing the functions of each component shown in Figure 3. Note that the functions of each component in Figure 3 may be realized by integrated circuits such as ASICs (Application Specific Integrated Circuits) or FPGAs (Field Programmable Gate Arrays).

[0018] Figure 3 shows a functional block diagram of server 10. In server 10, the CPU 90 executes programs to realize the functions of dispatch history acquisition unit 20, learning unit 22, deployment candidate acquisition unit 24, simulation unit 26, and output unit 28. Figure 3 also illustrates the dispatch history DB 40, trained model DB 42, deployment candidate DB 44, and simulation result DB 46 stored in storage 96, etc.

[0019] The dispatch history acquisition unit 20 acquires dispatch history information collected by the user terminal 70 and stores it in the dispatch history DB 40. The dispatch history DB 40 stores dispatch history information as shown in Figure 4. Dispatch history information is information obtained when an ambulance (ambulance crew) is actually dispatched to a scene (dispatch location). Specifically, as shown in Figure 4, the dispatch history information has data for the following items: "Dispatch ID", "Dispatch location latitude", "Dispatch location longitude", "Date", "Day of the week", "Fire station latitude", "Fire station longitude", "Dispatch time", "Arrival time", and "Departure time". "Dispatch ID" is identification information assigned to each dispatch history piece in order of dispatch date and time. "Dispatch location latitude" and "Dispatch location longitude" are the latitude and longitude information of the dispatch location (such as the location of the patient using the ambulance). "Date" and "Day of the week" are the date and day of the week when the dispatch request was made. "Fire station latitude" and "fire station longitude" are the location information (latitude and longitude) of the fire station where the dispatched team was actually stationed. "Dispatch time" is the time when the ambulance actually started its dispatch after the dispatch request was received, "arrival time" is the time when the ambulance actually arrived at the location, and "return time" is the time when the dispatched team returned to the fire station.

[0020] Returning to Figure 3, the learning unit 22 generates a travel time prediction model using the information stored in the dispatch history DB 40 and stores it in the trained model DB 42. Figure 5 shows an overview of the travel time prediction model generation by the learning unit 22.

[0021] As shown in the upper part of Figure 5, the learning unit 22 generates a trained model (travel time prediction model) by performing machine learning using a large amount of training data obtained from the dispatch history DB 40. Here, each of the large amount of training data is a combination of data obtained from each dispatch history, including the date, day of the week, time of day, location information of the dispatch destination (latitude and longitude), location information of the fire station (latitude and longitude), and travel time (= arrival time - dispatch time). As shown in the lower part of Figure 5, the travel time prediction model obtained in this way takes the date, day of the week, time of day, and information on the target section to be predicted (data on location information of the dispatch destination (latitude and longitude) and location information of the fire station (latitude and longitude)) as input data, and outputs the travel time by ambulance for the target section as a prediction result.

[0022] By using this travel time prediction model, travel time can be predicted in a shorter time compared to general route planning methods. Furthermore, even when a travel time prediction model is generated by machine learning using routes 1 to 7 connecting stations A to C and destinations x, y, and z, as shown by solid lines in Figure 6(a), the travel time for routes 8 (between station A and destination z) and 9 (between station C and destination x), as shown by dashed lines in Figure 6(b), can also be predicted accurately and quickly.

[0023] Returning to Figure 3, the deployment candidate acquisition unit 24 acquires deployment candidate information for teams created by local government officials on user terminals 70 and stores it in the deployment candidate DB 44. Deployment candidate information for teams is, for example, the information shown in Figure 7. Specifically, for each deployment candidate 1, 2...M, it is defined which team operates at which fire station from what time to what time, and as shown in Figure 7, it includes information such as "team ID," "fire station," "fire station latitude," "fire station longitude," and "operating hours." "Team ID" is identification information assigned to each team, "fire station" is the name of the fire station, and "fire station latitude" and "fire station longitude" are the location information of the fire station. "Operating hours" is the operating hours for each team. When the deployment information for each deployment candidate at a certain point in time is represented on a map, it looks like Figure 8. The circled numbers on the map in Figure 8 indicate the locations of fire stations, and the values ​​of the circled numbers indicate the number of teams deployed at each fire station. For example, local government officials use the current deployment information (information on deployments actually in operation) as a basis to increase or decrease the number of units, add or add fire stations, etc., in the table in Figure 7, set the number of units for each fire station, and create multiple (M) candidate unit deployments. Note that the candidate unit deployments may be automatically generated (for example, randomly under predetermined conditions) by the user terminal 70 or server 10.

[0024] The simulation unit 26 simulates the ambulance travel time to each scene (dispatch destination) assuming that a dispatch request similar to the dispatch history stored in the dispatch history DB 40 occurs when each of the team deployment candidates is adopted. Specific examples of the simulation will be described later. The simulation unit 26 stores the simulation results for each of the team deployment candidates in the simulation results DB 46. If the simulation yields an average travel time to each scene for each deployment candidate, the simulation results DB 46 will store the average travel time linked to the deployment candidate.

[0025] The output unit 28 reads the simulation results from the simulation results DB 46 and outputs them to the user terminal 70. For example, if the simulation result is the average travel time, the output unit 28 outputs the average travel time for each deployment candidate to the user terminal 70. As a result, the display unit 193 of the user terminal 70 displays the average travel time for each deployment candidate (see Figure 13). This allows local government officials to refer to the average travel time for each deployment candidate and consider which deployment candidate should be adopted.

[0026] (Regarding the processing on Server 10) Next, the processing of server 10 will be explained in detail, following the flowcharts in Figures 9 to 11, and referring to other diagrams as appropriate. Server 10 executes the learning phase processing (Figure 9) and the simulation processing (Figures 10 and 11).

[0027] (Processing during the learning phase) Figure 9 shows a flowchart illustrating the processing performed by server 10 during the learning phase.

[0028] When the process shown in Figure 9 begins, first, in step S10, the dispatch history acquisition unit 20 waits until dispatch history information is input from the user terminal 70. Once dispatch history information is input from the user terminal 70, the process moves to step S12, where the dispatch history acquisition unit 20 acquires the dispatch history information and stores it in the dispatch history DB 40 (Figure 4).

[0029] Next, in step S14, the learning unit 22 extracts and deletes abnormal and missing data from the dispatch history information stored in the dispatch history DB 40 (cleansing process). The learning unit 22 extracts abnormal and missing data based on predetermined rules. Abnormal data includes, for example, data with extremely short travel times, and missing data includes data where some items are missing. Note that the process in step S14 may be performed by a human.

[0030] Next, in step S16, the learning unit 22 randomly divides the dispatch history information stored in the dispatch history DB 40 into learning data and evaluation data. For example, the learning unit 22 randomly divides the dispatch history information into learning data and evaluation data at a predetermined ratio.

[0031] Next, in step S18, the learning unit 22 generates a travel time prediction model by performing machine learning using the training data (see upper part of Figure 5).

[0032] Next, in step S20, the learning unit 22 evaluates the travel time prediction model using the evaluation data. Specifically, the learning unit 22 inputs the date, day of the week, time of day, destination location information, and fire station location information extracted from each of the evaluation data into the travel time prediction model to obtain a predicted value for the travel time of each of the evaluation data. Then, for each of the evaluation data, it calculates the difference between the predicted value of the travel time and the actual travel time, and checks whether the aggregated result of the difference (e.g., the average value) falls within a predetermined range.

[0033] Next, in step S22, the learning unit 22 determines whether the prediction accuracy of the travel time prediction model was above a predetermined level based on the evaluation in step S20. That is, the learning unit 22 determines whether the aggregated result of the differences described above falls within a predetermined range. If the determination in step S22 is affirmative, the process proceeds to step S24, where the learning unit 22 stores the travel time prediction model in the trained model DB 42. After that, the entire process shown in Figure 9 is completed.

[0034] On the other hand, if the judgment in step S22 is rejected, that is, if the evaluation results show that the prediction accuracy of the travel time prediction model is below a predetermined level, the learning unit 22 returns to step S16 and restarts the process of splitting the data into training data and evaluation data (step S16). Subsequently, if the judgment in step S22 is affirmed as a result of the processing and judgment in steps S16 to S22, the entire process shown in Figure 9 is completed after the processing in step S24.

[0035] (Simulation process) Next, the simulation process performed by server 10 will be explained in detail, following the flowcharts in Figures 10 and 11, and referring to other diagrams as appropriate.

[0036] When the process shown in Figure 10 begins, first, in step S30, the deployment candidate acquisition unit 24 waits until information on M deployment candidates is entered. Once information on M deployment candidates is entered from the user terminal 70, the process moves to step S32, where the deployment candidate acquisition unit 24 acquires information on the M deployment candidates and stores it in the deployment candidate DB 44. The deployment candidate DB 44 stores deployment candidates 1, 2, ..., M as shown in Figure 7.

[0037] Next, in step S34, the simulation unit 26 selects an unselected deployment candidate. For example, the simulation unit 26 selects deployment candidate 1.

[0038] Next, in step S36, the simulation unit 26 performs a simulation using the dispatch history and the selected deployment candidates. In step S36, the simulation unit 26 performs the process shown in Figure 11.

[0039] (Processing in step S36 (Figure 11)) When the process shown in Figure 11 begins, first, in step S50, the simulation unit 26 retrieves the first dispatch history from the dispatch history DB 40. The simulation unit 26 retrieves the dispatch history with dispatch ID=1 from the dispatch history DB 40 shown in Figure 4.

[0040] Next, in step S52, the simulation unit 26 updates the unit deployment information according to the acquired dispatch history. Specifically, the simulation unit 26 refers to the dispatch time of the acquired dispatch history and adjusts the unit deployment to match that time. For example, suppose the simulation unit 26 deploys the unit based on the dispatch time of dispatch history for dispatch ID=1, taking into account the operating time in Figure 7, and as a result, the deployment information shown in Figure 12(a) is obtained.

[0041] Next, in step S54, the simulation unit 26 selects one unselected team from among the teams that have not yet been dispatched. The simulation unit 26 selects one team from each fire station, and does not select more than one team from a single fire station.

[0042] Next, in step S56, the simulation unit 26 calculates the travel time required for the selected team to arrive at the destination of the dispatch history (dispatch ID=1) acquired in step S50. In this case, the simulation unit 26 calculates the travel time using a travel time prediction model (Figure 5). Specifically, the simulation unit 26 inputs the date, day of the week, and time of day included in the dispatch history of dispatch ID=1, as well as the location information of the dispatch destination (latitude and longitude) and the departure location information (latitude and longitude) of the fire station where the selected team is stationed, as information for the prediction target section, into the travel time prediction model in Figure 5. Then, the simulation unit 26 uses the travel time prediction model to calculate the travel time required for the selected team to travel to the dispatch destination (travel time for the prediction target section).

[0043] Next, in step S58, the simulation unit 26 determines whether or not it has finished selecting all the units that have not yet been deployed. If the determination in step S58 is negative, the process returns to step S54. After that, the processing and determination in steps S54, S56, and S58 are repeated to determine the time required for each unit that has not yet been deployed to travel to its destination (travel time). If the determination in step S58 is positive, the simulation unit 26 proceeds to step S60.

[0044] When the program moves to step S60, the simulation unit 26 resets the squad selection. That is, the simulation unit 26 sets all squads to "unselected".

[0045] Next, in step S62, the simulation unit 26 determines which team to dispatch. For example, if the travel times to the destination for each fire station's team are as shown in Figure 12(b), the simulation unit 26 determines the team with the shortest travel time (in Figure 12(b), the team with a travel time of 3 minutes) to be dispatched. However, the simulation unit 26 is not limited to this, and may, for example, determine teams within a predetermined time from the shortest travel time (3 minutes in Figure 12(b)) as dispatch candidates. The simulation unit 26 may then determine which team to dispatch from the fire station with the largest number of teams waiting at the same fire station. Alternatively, the simulation unit 26 may determine which team to dispatch by taking other criteria and travel time into consideration.

[0046] Next, in step S64, the simulation unit 26 determines the withdrawal time of the dispatched unit. In this case, the simulation unit 26 may use the difference between the withdrawal time and the arrival time included in the currently acquired dispatch history as the withdrawal time, or it may use a predetermined time that is set in advance as the withdrawal time.

[0047] Next, in step S66, the simulation unit 26 determines whether or not all dispatch history included in the dispatch history DB 40 has been acquired. If the determination in step S66 is negative, the process proceeds to step S68, where the simulation unit 26 acquires the next dispatch history. After that, the process returns to step S52.

[0048] Returning to step S52, the simulation unit 26 updates the unit deployment information according to the acquired dispatch history. In this case, units that have already been dispatched at the time of dispatch in the acquired dispatch history and whose return time has not yet elapsed are treated as not having returned to the fire station.

[0049] Subsequently, the simulation unit 26 repeatedly executes steps S52 to S68. That is, for each deployment history, the simulation unit 26 calculates the travel time required for each unit to move to the deployment destination and decides which unit to deploy. When processing for all deployment histories is completed, the decision in step S66 is affirmed, and the system proceeds to step S70. At the stage of proceeding to step S70, the travel times for all deployment histories have been calculated assuming that the deployment candidate selected in step S34 of Figure 10 is adopted.

[0050] Next, in step S70, the simulation unit 26 performs aggregation processing. In this case, as an example, the simulation unit 26 calculates the average travel time for all dispatch history (average travel time for ambulances). The simulation unit 26 stores the average travel time in the simulation results DB 46, linked to the deployment candidate number.

[0051] Therefore, once the process in Figure 11 is completed, the process proceeds to step S38 in Figure 10.

[0052] When the program proceeds to step S38 in Figure 10, the simulation unit 26 determines whether all deployment candidates have been selected. If the determination in step S38 is negative, the program returns to step S34. If the program returns to step S34, the simulation unit 26 selects any unselected deployment candidates (S34) and executes a simulation for the selected deployment candidates to calculate the average travel time (S36).

[0053] On the other hand, if the simulations for all deployment candidates are completed and the judgment in step S38 is affirmed, the process proceeds to step S40.

[0054] When the system moves to step S40, the output unit 28 outputs optimal deployment information. Here, as an example, the output unit 28 reads the average travel time for each deployment candidate from the simulation results DB 46 and outputs the deployment candidate with the shortest average travel time to the user terminal 70. For example, as shown in Figure 13, if the average travel time for deployment candidate 2 is the shortest after calculating the average travel time for each deployment candidate, deployment candidate 2 is output. In this case, the display unit 193 of the user terminal 70 displays information indicating that deployment candidate 2 is the optimal deployment.

[0055] In step S40, the output unit 28 may output a screen to the user terminal 70 displaying the deployment candidates in order of shortest average travel time. In this case, the local government officials may actually adopt the deployment candidate with the shortest average travel time, or they may select a deployment candidate to actually adopt by comprehensively considering the average travel time and the costs and effort required to change from the current troop deployment.

[0056] As described in detail above, according to this embodiment, the learning unit 22 generates a travel time prediction model using dispatch history, which includes location information of the departure point (fire station), location information of the arrival point (dispatch destination), and travel time information between the departure point and the arrival point, as learning data (S18 in Figure 9). Then, the simulation unit 26 inputs the location information of the departure point (fire station) and the location information of the arrival point (dispatch destination) for the prediction target section (the section that the selected team travels to the dispatch destination) into the travel time prediction model to predict the travel time for the prediction target section (S56 in Figure 11). This makes it possible to predict travel time in a shorter time compared to general route search methods. In this case, by using the travel time prediction model, it is possible to predict travel time even for routes that were not included in the learning data, as explained using Figures 6(a) and 6(b). Furthermore, as in this embodiment, when it is necessary to perform a large number of travel time predictions in the simulation, the simulation time can be significantly reduced.

[0057] Furthermore, in this embodiment, each training data set includes information about the date and time of travel (year, month, day, day of the week, and time of day). When the simulation unit 26 predicts travel time, it also inputs information about the date and time to be predicted (year, month, day, day of the week, and time of day) into the travel time prediction model. This makes it possible to predict travel time according to the year, month, day, day of the week, and time of day, that is, to predict travel time that takes into account congestion trends, etc. Note that the information about the date and time does not have to include all of the year, month, day, day of the week, and time of day, and may also include other information about the date and time (information about whether it is a public holiday, information about whether there is an event, etc.).

[0058] Furthermore, in this embodiment, the learning unit 22 uses the dispatch history (travel history) of emergency vehicles (e.g., ambulances) as training data, so it can generate a travel time prediction model specifically for emergency vehicles. This makes it possible to accurately predict the travel time of emergency vehicles, which differ from that of ordinary vehicles.

[0059] In the above embodiment, the output unit 28 was described as outputting information on the optimal deployment candidate based on the average travel time of each deployment candidate, or outputting a screen that sorts and displays each deployment candidate based on the average travel time, but it is not limited to this. For example, the output unit 28 may divide the local government into regions, calculate the average travel time to the deployment destinations included in each region, and output a screen that color-codes each region based on the calculated average value, as shown in Figure 14. Local government officials can check the average travel time for each region for each deployment candidate, as shown in Figure 14, to see if there are any disparities in average travel time between regions. This allows them to select a more appropriate deployment candidate.

[0060] In the above embodiment, the unit deployment decision support system 100 shown in Figure 1 was used when deciding on the deployment of ambulances and fire engines at a fire station, but this is not the only option. For example, when deploying people or units to multiple locations, such as a security company, taxi company, or delivery company, the unit deployment decision support system 100 shown in Figure 1 may also be used.

[0061] In the above embodiment, the case in which the server 10 executes the processes shown in Figures 9 to 11 was described, but the invention is not limited to this, and each user terminal 70 may also execute the processes shown in Figures 9 to 11. Furthermore, the devices that execute each process may differ, such as the server 10 executing the process in Figure 9 and the user terminals 70 executing the processes in Figures 10 and 11.

[0062] The above processing functions can be implemented by a computer. In this case, a program describing the processing content of the functions that the processing unit should have is provided. By executing this program on a computer, the above processing functions are implemented on the computer. The program describing the processing content can be recorded on a computer-readable recording medium (excluding carrier waves).

[0063] When distributing a program, it may be sold in the form of a portable recording medium such as a DVD (Digital Versatile Disc) or CD-ROM (Compact Disc Read Only Memory) on which the program is recorded. Alternatively, the program can be stored in the storage device of a server computer and transferred from the server computer to other computers via a network.

[0064] A computer executing a program stores the program, for example, on a portable storage medium or transferred from a server computer, in its own memory. The computer then reads the program from its memory and executes the processing according to the program. Alternatively, the computer can directly read the program from the portable storage medium and execute the processing according to that program. Furthermore, the computer can sequentially execute the processing according to the program received each time it is transferred from a server computer.

[0065] The embodiments described above are preferred examples of the present invention. However, the invention is not limited thereto, and various modifications are possible without departing from the spirit of the invention.

[0066] Furthermore, the following additional information is disclosed regarding the above description of the embodiments. (Note 1) A travel time prediction model is generated using travel history data, which includes location information of the departure point, location information of the arrival point, and travel time information between the departure point and the arrival point, as training data. By inputting the location information of the starting point and the location information of the destination section into the travel time prediction model, the travel time of the destination section is predicted. A method for predicting travel time, characterized in that the processing is performed by a computer. (Note 2) Each of the aforementioned movement history data includes information regarding the date and time the movement took place. In the aforementioned prediction process, information regarding the date and time of travel within the prediction target section, and location information of the departure and arrival points of the prediction target section are input to the travel time prediction model to predict the travel time within the prediction target section. A method for predicting travel time as described in Appendix 1, characterized by the features described herein. (Note 3) The travel time prediction method according to Note 2, characterized in that the information relating to the date and time includes at least one of the month, day of the week, and time of day. (Note 4) The method for predicting travel time according to any one of Notes 1 to 3, characterized in that the travel history data is travel history data of an emergency vehicle. (Note 5) Obtain the location information of the starting point and the location information of the destination point of the prediction target section. Based on the location information of the departure point and arrival point of the aforementioned prediction target section, and a travel time prediction model that predicts the travel time between the departure point and the arrival point based on the location information of the departure point and the arrival point, the travel time between the departure point and the arrival point of the aforementioned prediction target section is predicted. Outputs the travel time between the departure point and the arrival point of the predicted target section. A method for predicting travel time, characterized in that the processing is performed by a computer. (Note 6) A travel time prediction model is generated using travel history data, which includes location information of the departure point, location information of the arrival point, and travel time information between the departure point and the arrival point, as training data. By inputting the location information of the starting point and the location information of the destination section into the travel time prediction model, the travel time of the destination section is predicted. A travel time prediction program characterized by having a computer perform the processing. (Note 7) Each of the aforementioned movement history data includes information regarding the date and time the movement took place. In the aforementioned prediction process, information regarding the date and time of travel within the prediction target section, and location information of the departure and arrival points of the prediction target section are input to the travel time prediction model to predict the travel time within the prediction target section. A travel time prediction program as described in Appendix 6, characterized by the above. (Note 8) The travel time prediction program according to Note 7, characterized in that the information relating to the date and time includes at least one of the month, day of the week, and time of day. (Note 9) The travel time prediction program according to any one of Notes 6 to 8, characterized in that the travel history data is travel history data of an emergency vehicle. (Note 10) Obtain the location information of the starting point and the location information of the destination point of the prediction target section. Based on the location information of the departure point and arrival point of the aforementioned prediction target section, and a travel time prediction model that predicts the travel time between the departure point and the arrival point based on the location information of the departure point and the arrival point, the travel time between the departure point and the arrival point of the aforementioned prediction target section is predicted. Outputs the travel time between the departure point and the arrival point of the predicted target section. A travel time prediction program characterized by having a computer perform the processing. [Explanation of Symbols]

[0067] 10 servers 20 Dispatch History Acquisition Unit 22 Learning Department 24 Deployment candidate acquisition unit 26 Simulation Department 28 Output section 40 Dispatch History Database 42 Pre-trained model database 44 Deployment candidate DB 46 Simulation Results Database 70 User terminals 100 Unit Deployment Decision Support System

Claims

1. A travel time prediction model is generated using travel history data, which includes location information of the starting point, location information of the destination point, and travel time information between the starting point and the destination point, as training data. By inputting the location information of the starting point and the location information of the destination section into the travel time prediction model, the travel time of the destination section is predicted. A method for predicting travel time, characterized in that the processing is performed by a computer.

2. Each of the aforementioned movement history data includes information regarding the date and time the movement took place. In the aforementioned prediction process, information regarding the date and time of travel within the prediction target section, and location information of the departure and arrival points of the prediction target section are input to the travel time prediction model to predict the travel time within the prediction target section. The travel time prediction method according to feature 1.

3. The travel time prediction method according to claim 2, characterized in that the information relating to the date and time includes at least one of the month, day of the week, and time of day.

4. The method for predicting travel time according to claim 1, characterized in that the aforementioned travel history data is travel history data of an emergency vehicle.

5. Obtain the location information of the starting point and the location information of the destination point for the prediction section. Based on the location information of the departure point and arrival point of the aforementioned prediction target section, and a travel time prediction model that predicts the travel time between the departure point and the arrival point based on the location information of the departure point and the arrival point, the travel time between the departure point and the arrival point of the aforementioned prediction target section is predicted. Outputs the travel time between the departure point and the arrival point of the predicted target section. A method for predicting travel time, characterized in that the processing is performed by a computer.

6. A travel time prediction model is generated using travel history data, which includes location information of the starting point, location information of the destination point, and travel time information between the starting point and the destination point, as training data. By inputting the location information of the starting point and the location information of the destination section into the travel time prediction model, the travel time of the destination section is predicted. A travel time prediction program characterized by having a computer perform the processing.

7. Obtain the location information of the starting point and the location information of the destination point for the prediction section. Based on the location information of the departure point and arrival point of the aforementioned prediction target section, and a travel time prediction model that predicts the travel time between the departure point and the arrival point based on the location information of the departure point and the arrival point, the travel time between the departure point and the arrival point of the aforementioned prediction target section is predicted. Outputs the travel time between the departure point and the arrival point of the predicted target section. A travel time prediction program characterized by having a computer perform the processing.

Images