Operation support method, operation support system, and server

Abstract

To present information that can be understood by a driver by highly accurately predicting an accident risk.SOLUTION: An operation support method for supporting the operation of a vehicle comprises: a first step of making a computer acquire biological information of a driver who is driving the vehicle; a second step of acquiring environmental information of the driver; a third step of acquiring business information of the driver; a fourth step of generating biological index data from the biological information; a fifth step of generating integrated information by integrating the biological index data, the environmental information and the business information such that the time series is adjusted; a sixth step of inputting the integrated information to a preset accident risk prediction model to calculate accident risk information after a prescribed time; and a seventh step of inputting the accident risk, the integrated information and preset accident determination information to a preset factor calculation model to calculate factor information of the accident risk.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to an operation support method, an operation support system, and a server that predict the risk of traffic accidents and support the operation of transportation vehicles.

Background Art

[0002] In recent years, in the transportation industry such as trucks and buses, quantitative evaluation of the biological state has been carried out to prevent accidents caused by the health of drivers. For example, in Patent Document 1, among the biological states, evaluation of the autonomic nerve function based on the measurement of R-R interval data by various forms of heartbeat sensors that are easy to measure is known.

[0003] Patent Document 1 describes estimating the mental state of a driver from biological data or the like, generating mental data regarding the driving of the driver, and estimating the suitability or unsuitability of the driver's state.

[0004] Also, Patent Document 2 describes estimating an autonomic nerve function index of a driver from R-R interval data of biological data and predicting the accident risk after a predetermined time based on the autonomic nerve function index.

Prior Art Documents

Patent Documents

[0005]

Patent Document 1

Patent Document 2

Summary of the Invention

Problems to be Solved by the Invention

[0006] To predict the risk of accidents involving drivers during work, it is necessary to predict the risk of accidents in the near future in real time. During work, the driver's physical condition changes constantly, and factors contributing to accident risk include not only biological information but also environmental information and work-related information. However, environmental information such as driving conditions and work-related information such as the content of the work differ from driver to driver.

[0007] However, the conventional system had the problem that it was difficult to accurately predict the accident risk of each individual driver by accurately reflecting environmental and work information in biometric data. In addition, the conventional system issued a warning when the accident risk was high, but simply presenting a warning was not always enough to gain the driver's understanding.

[0008] Therefore, the present invention has been made in view of the above problems, and aims to predict accident risks with high accuracy and to present information that drivers can understand. [Means for solving the problem]

[0009] The present invention relates to a vehicle operation support method in which a computer having a processor and memory assists the operation of a vehicle, comprising: a first step of the computer acquiring biometric information of the driver while the vehicle is being driven; a second step of the computer acquiring environmental information of the driver; a third step of the computer acquiring work information of the driver; a fourth step of the computer generating biometric indicator data from the biometric information; and the computer processing the biometric indicator data, environmental information, and work information in a time series. granularity The process includes a fifth step of aligning and combining the information to generate integrated information, a sixth step of the computer inputting the integrated information into a pre-configured accident risk prediction model to calculate accident risk information after a predetermined time, and a seventh step of the computer inputting the accident risk information, the integrated information, and pre-configured accident judgment information into a pre-configured factor calculation model to calculate factor information for the accident risk information. [Effects of the Invention]

[0010] Therefore, in addition to predicting accident risk information after a predetermined time, the present invention can output information on the factors contributing to accident risk, thereby presenting information that is not unnatural to the driver.

[0011] Details of at least one embodiment of the subject matter disclosed herein are described in the accompanying drawings and the following description. Other features, aspects and effects of the subject matter disclosed are made apparent by the following disclosures, drawings and claims. [Brief explanation of the drawing]

[0012] [Figure 1] This block diagram shows an embodiment of the present invention and an example of the configuration of a vehicle support system. [Figure 2] This flowchart illustrates an embodiment of the present invention and outlines the processing performed by the operation support system. [Figure 3] This figure shows an example of biological information, illustrating an embodiment of the present invention. [Figure 4] This figure shows an example of environmental information illustrating an embodiment of the present invention. [Figure 5] This figure illustrates an embodiment of the present invention and shows an example of attendance data that constitutes business information. [Figure 6] This figure illustrates an embodiment of the present invention and shows an example of delivery data that constitutes business information. [Figure 7] This flowchart illustrates an embodiment of the present invention and shows an example of processing performed by the time-series biological-environmental-business information generation unit. [Figure 8] This graph illustrates an embodiment of the present invention and shows an example of heart rate data. [Figure 9] This graph illustrates an example of the present invention and shows an example of heart rate variability. [Figure 10] This graph illustrates an example of the spectral power density of heart rate variability, illustrating an embodiment of the present invention. [Figure 11] This figure illustrates an example of data formatting performed by the time-series biological-environmental-business information generation unit, illustrating an embodiment of the present invention. [Figure 12]This is a diagram showing an example of time-series biological-environment-business information, which illustrates an embodiment of the present invention. [Figure 13] This is a flowchart showing an example of the processing performed in the accident risk prediction model training unit, which illustrates an embodiment of the present invention. [Figure 14] This is a diagram showing an example of accident determination information, which illustrates an embodiment of the present invention. [Figure 15] This is a flowchart showing an example of the processing performed in the accident risk prediction unit, which illustrates an embodiment of the present invention. [Figure 16] This is a diagram showing an example of accident risk information, which illustrates an embodiment of the present invention. [Figure 17] This is a flowchart showing an example of the processing performed in the accident risk factor calculation unit, which illustrates an embodiment of the present invention. [Figure 18] This is a diagram showing an example of accident risk factor information, which illustrates an embodiment of the present invention. [Figure 19] This is a flowchart showing an example of the processing performed in the prediction result presentation unit, which illustrates an embodiment of the present invention. [Figure 20] This is a diagram showing an example of outputting a prediction result, which illustrates an embodiment of the present invention. [Figure 21] This is a diagram showing an example of the screen of the prediction result display terminal, which illustrates an embodiment of the present invention. [Figure 22] This is a diagram showing an example of the screen of the prediction result display terminal, which illustrates an embodiment of the present invention. [Figure 23] This is a graph showing an example of the data registered in the accident determination information, which illustrates an embodiment of the present invention. [Figure 24] This is a diagram showing an example of the training of the accident risk prediction model, which illustrates an embodiment of the present invention.

Mode for Carrying Out the Invention

[0013] Hereinafter, embodiments of the present invention will be described based on the accompanying drawings.

[0014] <System Configuration> Figure 1 is a block diagram showing an embodiment of the present invention and an example of the configuration of a vehicle operation support system. The vehicle operation support system of this embodiment includes a vehicle operation support server 1 that collects driving information of one or more vehicles 8, biometric information of the drivers of the vehicles 8, environmental information of the vehicles 8 or drivers, and work information of the drivers via a network 19, predicts the risk of a traffic accident for the driver (hereinafter referred to as accident risk), and notifies the driver if the predicted value of the accident risk exceeds a threshold.

[0015] Network 19 is connected to a driving information collection device 10 that acquires driving information of vehicle 8, a biometric information collection device 60 that acquires biometric information of the driver, an environmental information collection device 70 that acquires information about the environment of the driver operating vehicle 8, a work information collection device 80 that acquires information about the driver's work, and a prediction result display terminal 90 that outputs notifications from the operation support server 1, and can communicate with the operation support server 1.

[0016] The driving information collection device 10 collects information from a GNSS (Global Navigation Satellite System) 11 that detects the position information of the vehicle 8, a distance sensor 12 that detects the distance to the preceding vehicle, a speed sensor 13 that detects the speed of the vehicle 8, an acceleration sensor 14 that detects the movement of the vehicle 8, and a camera 15 that photographs the area around the vehicle 8, and transmits it to the operation support server 1.

[0017] The driving information collection device 10 is not limited to the sensors described above, and can also use distance measuring sensors to detect objects and distances around the vehicle 8, steering angle sensors to detect driving operations, etc. In addition, the driving information collection device 10 may be equipped with a driver ID reader (not shown) that reads a medium on which a driver identifier is recorded in order to identify the driver.

[0018] The biometric information collection device 60 includes sensors for a heart rate monitor 61 that detects heart rate data, a thermometer 62 that detects the driver's body temperature, and a blood pressure monitor 63 that detects the driver's blood pressure. The biometric information collection device 60 can use wearable devices that can be worn by the driver, as well as sensing devices attached inside the vehicle 8, such as the steering wheel, seat, and seat belt, and an image recognition system that captures images of the driver's facial expressions and behavior and analyzes the images.

[0019] The sensors of the biometric information collection device 60 are not limited to those described above, and can employ sensors that detect sweat volume, body temperature, blinking, eye movements, or brain waves, etc. Furthermore, the biometric information collection device 60 can set an identifier to identify the driver and add this identifier to various sensing data.

[0020] The environmental information collection device 70 includes a thermometer 71 and a barometer 72. The environmental information collection device 70 may be mounted on the vehicle 8 or, like the biometric information collection device 60, may be attached to the driver.

[0021] The work information collection device 80 has a work content input unit 81 for inputting the driver's work details. The driver can input information such as the type of work, the start and end times of work, breaks, and the details of delivery work from the work content input unit 81. The work information collection device 80 may be mounted on the vehicle 8, may be a portable terminal carried by the driver, or may be a terminal installed in an office or other location that can be operated remotely.

[0022] The prediction result display terminal 90 includes a factor label input unit 91 and an output unit 92. The factor label input unit 91 receives factor labels selected or entered by the driver for the factors that triggered the accident risk warning and transmits them to the operation support server 1. The output unit 92 consists of a display device and a speaker and outputs the accident risk warnings and cautionary notices transmitted from the operation support server 1. In addition to text data, the factor label input unit 91 can also accept voice input.

[0023] The prediction result display terminal 90 may be a portable terminal carried by the driver, a car navigation system installed in the vehicle 8, or a computer installed in an office or the like.

[0024] The operation support server 1 is a computer that includes a processor 2, memory 3, auxiliary storage device 4, communication interface 5, input device 6, and output device 7. Memory 3 loads the following functional units as programs: a time-series biological-environmental-business information generation unit (or integrated information generation unit) 31, an accident risk prediction model training unit 32, an accident risk prediction unit 33, an accident risk factor calculation unit 34, and a prediction result presentation unit 35. Each program is executed by the processor 2. Details of each functional unit will be described later.

[0025] Processor 2 operates as a functional unit that provides predetermined functions by executing processing according to the programs of each functional unit. For example, processor 2 functions as an accident risk prediction unit 33 by executing an accident risk prediction program. The same applies to other programs. Furthermore, processor 2 also operates as a functional unit that provides the functions of each of the multiple processes executed by each program. A computer and a computer system are devices and systems that include these functional units.

[0026] The auxiliary storage device 4 stores the data used by each of the above-mentioned functional units. The auxiliary storage device 4 stores biometric information 41, environmental information 42, business information 43, driving information 44, accident risk information 45, accident risk factor information 46, time-series biometric-environmental-business information 47, accident risk prediction model 48, accident risk factor calculation model 49, accident risk prediction model training data 50, presentation content dictionary 53, and accident risk factor labels 54.

[0027] The accident risk prediction model training data 50 includes historical time-series biological-environmental-operational information 51 and historical accident judgment information 52. Details of each data will be described later.

[0028] The input device 6 includes a mouse, keyboard, or touch panel. The output device 7 includes a display or speaker. The communication interface 5 is connected to the network 19 and communicates with the vehicle 8, the biometric information collection device 60, the environmental information collection device 70, the business information collection device 80, and the prediction result display terminal 90.

[0029] <Software Configuration> The time-series biological-environmental-business information generation unit 31 acquires driving information from the vehicle 8 and stores it in driving information, acquires biological information from the biological information collection device 60 and stores it in biological information 41, acquires environmental information from the environmental information collection device 70 and stores it in environmental information 42, and acquires business information from the business information collection device 80 and stores it in business information 43.

[0030] The time-series biological-environmental-business information generation unit 31 calculates the interval between R waves (RRI = RR Interval) of the heart rate data from the heart rate monitor 61 data (hereinafter referred to as heart rate data) of the biological information 41, calculates an autonomic nervous function (ANF) index from the RRI data (or heart rate interval data), and stores it in the biological information 41.

[0031] Furthermore, the time-series biological-environmental-business information generation unit 31 combines the granularity (calculation interval or analysis window width) of the autonomic nervous system function index, the granularity (measurement interval) of the environmental information 42, the granularity (collection interval) of the business information 43, and the granularity (measurement interval) of the operation information 44 to match the granularity of the autonomic nervous system function index, as described later, to generate time-series biological-environmental-business information 47.

[0032] The accident risk prediction model training unit 32 uses the accident risk prediction model training data 50 that has been collected in advance to train the machine learning model accident risk prediction model 48 and generate or update the accident risk prediction model 48.

[0033] The accident risk prediction unit 33 uses a trained accident risk prediction model 48 to take time-series bio-environment-work information 47 collected from the vehicle 8 and the driver as input and outputs accident risk information 45 that indicates the probability of the driver having an accident after a predetermined time.

[0034] The accident risk factor calculation unit 34 uses a pre-configured machine learning model accident risk factor calculation model 49 to calculate highly relevant factors as accident risk factor information 46, taking time-series biological-environmental-work information 47 and accident risk information 45 as input.

[0035] If the calculated accident risk information 45 satisfies a predetermined condition (i.e., the probability of an accident occurring exceeds a pre-set threshold Th1), the prediction result presentation unit 35 acquires accident risk factor information 46 corresponding to the accident risk information 45, retrieves a message (or warning) corresponding to the accident risk factor information 46 from a pre-set presentation content dictionary 53, and transmits it to the prediction result display terminal 90.

[0036] As will be described later, the prediction result presentation unit 35 receives the label entered by the driver in the factor label input unit 91 in response to the transmitted warning and stores the label in the accident risk factor label 54.

[0037] In this embodiment, we illustrate the case where the biometric information measured from the driver of vehicle 8 is used as an example, but it is not limited to the driver operating vehicle 8. For example, it may also apply to people who operate other moving objects such as airplanes or trains.

[0038] <Process Overview> Figure 2 is a flowchart outlining the processes performed by the operation support system. First, the accident risk prediction model training unit 32 trains and updates the accident risk prediction model 48 using the pre-configured accident risk prediction model training data 50.

[0039] In the operation support system of this embodiment, a machine learning model that estimates accident risk from past driving information and past biometric information is generated in advance as an accident risk prediction model 48, and the machine learning model is trained by inputting accident risk prediction model training data 50 into this accident risk prediction model 48 (S1).

[0040] The generation or training of the accident risk prediction model 48 may be carried out in the same manner as described in Patent Document 2. For example, a definition model is generated that estimates accident risk from the driving state of vehicle 8, using past driving data and hazard occurrence data as input. Next, the driving data collected in the past is input to the definition model to generate estimated accident risk data, which represents the probability of an accident occurring. Then, biometric indicator data of the driver is calculated from past biometric information, and a machine learning model can be generated as an accident risk prediction model that outputs the accident risk (probability) after a predetermined time from the biometric indicator data of vehicle 8 while it is in motion, using the estimated accident risk data and biometric indicator data as input.

[0041] In this embodiment, biometric data such as power spectral density calculated from the driver's heart rate data (described later) or autonomic nervous system indicators based on NN intervals (differences in intervals between R waves) calculated from time-domain analysis (described later) can be used. Alternatively, results analyzed and calculated from autonomic nervous system indicators may also be used.

[0042] The accident risk prediction model training data 50 for training the accident risk prediction model 48 consists of past time-series bio-environment-operation information 51 and past accident judgment information 52. The past time-series bio-environment-operation information 51 is data that combines the biometric information of drivers collected in the past with the environmental information and operation information (driving information) at the time the biometric information was collected, with the granularity of the time series aligned as described later.

[0043] The accident risk prediction model training unit 32 can add newly collected time-series biological-environmental-operational information 47 to the time-series biological-environmental-operational information 51. Furthermore, if the accident risk prediction model training unit 32 updates the accident risk factor labels 54, it can provide feedback to the accident judgment information 52.

[0044] As will be described later, the accident judgment information 52 is data collected from past accidents and near misses, and is in the same time series as the past time-series biological-environmental-operational information 51.

[0045] After training the accident risk prediction model 48 with past accident risk prediction model training data 50, the time-series bio-environment-operation information generation unit 31 acquires bio-information 41, environmental information 42, operation information 43, and driving information 44 (S2).

[0046] The time-series biological-environmental-business information generation unit 31 generates time-series biological-environmental-business information 47 by performing predetermined preprocessing on the biological information 41 of the vehicle 8 driver, and environmental information 42, business information 43, and driving information 44 that correspond to the time series of the biological information 41 (S3).

[0047] The time-series biological-environmental-business information generation unit 31 performs preprocessing of the biological information 41 by excluding or interpolating missing intervals in the heart rate data (RRI data). The time-series biological-environmental-business information generation unit 31 can exclude heart rate data in an interval if the length of the missing interval exceeds a predetermined threshold Thf, and can perform interpolation if the length of the missing interval is less than or equal to the predetermined threshold Thf. Next, the time-series biological-environmental-business information generation unit 31 calculates autonomic nervous system function index (ANF information) from the preprocessed heart rate data as described later.

[0048] The time-series biological-environmental-business information generation unit 31 generates pre-processed information for environmental information 42, business information 43, and operation information 44 by excluding or interpolating missing sections, similar to the biological information 41.

[0049] Then, as will be described later, the time-series biological-environment-business information generation unit 31 combines the pre-processed environmental information 42, business information 43, and operation information 44 corresponding to the time series of ANF information to generate time-series biological-environment-business information 47.

[0050] Next, the accident risk prediction unit 33 inputs the generated time-series biological-environmental-business information 47 and past accident judgment information 52 into the trained accident risk prediction model 48 to calculate accident risk information 45 (S4).

[0051] Next, the accident risk factor calculation unit 34 inputs the accident risk information 45 calculated by the accident risk prediction unit 33 and the time-series biological-environmental-operational information 47 into a pre-configured accident risk factor calculation model 49 to calculate accident risk factor information 46 (S5). The accident risk factor information 46 is the estimated result of the factors of the accident risk information 45 predicted by the accident risk prediction model 48.

[0052] Next, if the prediction result (probability) calculated by the accident risk prediction unit 33 exceeds a predetermined threshold Th1, the prediction result presentation unit 35 retrieves a message set in the presentation content dictionary 53 based on the accident risk factor information 46 and sends a warning message containing the accident risk information 45 and the accident risk factor information 46 to the prediction result display terminal 90 of the relevant driver (S6). In addition, if the prediction result calculated by the accident risk prediction unit 33 (probability of accident risk information 45) exceeds a predetermined threshold Th1, the prediction result presentation unit 35 can also notify the administrator and users of the operation support server 1 of the message.

[0053] When the accident risk information 45 exceeds the threshold Th1, the notification should include not only the content of the accident risk information 45 but also the reasoning behind issuing the warning, thereby providing a more convincing warning for the driver.

[0054] The prediction result display unit 35 sends a message to the prediction result display terminal 90 prompting the driver to input information about the accident risk factors 46 transmitted to the prediction result display terminal 90, and can receive driver input (factor labels) from the factor label input unit 91 of the prediction result display terminal 90 (S7).

[0055] By having the driver input the factors they recognize in response to the warnings notified by the operation support server 1 and storing them as factor labels in the accident risk factor information 46, the warnings output by the operation support server 1 can be made to be information that is not unusual for the driver. Note that input of factor labels from the factor label input unit 91 may be done after the end of driving or after the end of work. In addition, input of factor labels from the factor label input unit 91 may be performed by an operations manager or other person on behalf of the driver.

[0056] When the prediction result presentation unit 35 receives input of a factor label from the prediction result display terminal 90, it sets the received factor label in the accident risk factor information 46 and updates it. The accident risk prediction model training unit 32 can feed back the contents of the accident risk factor information 46 with the updated factor label to the accident judgment information 52 and reflect it in the accident risk prediction model 48 and the accident risk factor calculation model 49.

[0057] When training the accident risk prediction model 48 or the accident risk factor calculation model 49, the accident risk prediction model training unit 32 uses the accident judgment information 52 with updated factor labels, thereby feeding back the accident risks actually encountered by drivers to the accident risk prediction model 48. The accident risk prediction model training unit 32 conducts training of the accident risk prediction model 48 and the accident risk factor calculation model 49 at predetermined intervals (for example, every month).

[0058] Figure 21 shows an example of outputting a warning from the operation support server 1 via voice from the prediction result display terminal 90. In the example shown, the prediction result display terminal 90 outputs a warning message 901 stating that there is an 80 percent probability that an incident (near miss in the figure) will occur within 30 minutes, and a cause message 902 stating that the factors that caused the warning are 6 consecutive days of work and that the current weather is rainy.

[0059] The driver can understand from warning message 901 that the cause of the warning is explained in cause message 902, and can take the warning received while driving into consideration without feeling any discomfort. Furthermore, suggestions that contribute to risk reduction may be added along with the warning.

[0060] Figure 21 shows an example of a screen for inputting factor labels displayed on the prediction result display terminal 90. Screen 910 is the input screen for factor labels output to the display device of the prediction result display terminal 90. This screen 910 is output by the factor label input unit 91 of the prediction result display terminal 90.

[0061] Screen 910 includes a graph 911 of accident risk information 45 consisting of time and accident probability, a biometric information 41 (changes in physical condition and fatigue) 916, an area for displaying changes in work information 43 and environmental information 42 (number of consecutive work days 912, weather 913, delay status 914), and an input section 915 for factor labels.

[0062] In the illustrated example, the user is prompted to input the name of the accident risk factor at time Tx, when the probability of an accident exceeds a predetermined threshold Th1. The driver inputs the factor label using the input device (not shown) of the prediction result display terminal 90. The factor label input unit 91 transmits the input factor label to the operation support server 1.

[0063] In addition to entering text into the input unit 915 for factor labels, as shown in Figure 22, multiple factor labels may be displayed as buttons 940, and a factor label may be selected from these buttons 940.

[0064] Furthermore, in addition to the elements in Figure 21, the screen 910 in Figure 22 includes a video window 930 that shows the driving state before the accident risk information 45 exceeded the threshold Th1, in addition to the button 940 mentioned above. The video window 930 can play back videos taken by the camera 15 of the vehicle 8, showing the video from the time when the accident risk information 45 exceeded the threshold Th1 (the time when the accident risk information 45 occurred) up to a predetermined time. The videos taken by the camera 15 are stored in the auxiliary storage device 4 along with the driving information 44.

[0065] Furthermore, video footage of the driver taken from the time of the accident risk information 45 up to a specified time may be provided as accident risk factor information 46.

[0066] In this embodiment, the operation support server 1 integrates the measurement interval granularity of the biometric information 41, environmental information 42, business information 43, and driving information 44 to match the calculation interval of the biometric information 41 (ANF information) to create time-series biometric-environmental-business information 47, thereby enabling it to present the background of accident risk information 45 from the biometric information 41 (ANF information) and driving information 44.

[0067] In other words, when the probability of an accident occurring, as indicated by the accident risk information 45, increases, the operation support server 1 has the accident risk factor calculation model 49 predict the business information 43 and environmental information 42 that are contributing factors, and when issuing a warning, it notifies the driver of the factors causing the warning, thereby enabling the server to issue a warning that the driver can understand and accept.

[0068] If a driver is suddenly notified only that the probability of an accident occurring has increased while driving vehicle 8, it is difficult for them to immediately understand why the warning was issued. Therefore, in this embodiment, when issuing a warning, the operation support server 1 presents business information 43 and environmental information 42 that are the background factors that led to the warning, making it possible to notify the driver of information that is easier for them to understand.

[0069] <Data Details> Next, we will explain the details of the data used by the operation support server 1.

[0070] Figure 3 shows an example of biometric information 41 measured by the biometric information collection device 60. The biometric information 41 includes user ID 411, date and time 412, heart rate interval data 413, blood pressure 414, body temperature 415, and interview results 416 in a single record.

[0071] User ID 411 stores the driver's identifier. In this embodiment, the identifier pre-set in the biometric information collection device 60 is used. Date and time 412 stores the date and time when the data was measured by the biometric information collection device 60.

[0072] Heart rate interval data 413 stores the heart rate interval (RRI data) measured by the heart rate monitor 61. Blood pressure 414 stores the blood pressure measured by the blood pressure monitor 63. Body temperature 415 stores the body temperature measured by the thermometer 62. Medical interview results 416 stores the results of medical interviews, such as those conducted at the start of work. Note that "Not measured" is stored for items for which data was not measured.

[0073] Figure 4 shows an example of environmental information 42 measured by the environmental information collection device 70. The environmental information 42 includes area ID 421, date 422, day of the week 423, time zone 424, weather 425, temperature 426, and atmospheric pressure 427 in a single record.

[0074] Area ID 421 stores the identifier of the region (prefecture, etc.) from which the data was obtained. Date 422 stores the year, month, and day the data was obtained. Day of the week 423 stores the day of the week from which the data was obtained. Time zone 424 stores the start and end times of the time from which the data was obtained.

[0075] Weather 425 stores the weather information obtained for each area. Temperature 426 stores the temperature measured by thermometer 71. Atmospheric pressure 427 stores the atmospheric pressure measured by barometer 72.

[0076] Figure 5 shows an example of attendance data 431 from the business information 43 received by the business information collection device 80. Attendance data 431 includes user ID 4311, start date and time 4312, previous departure date and time 4313, consecutive attendance days 4314, and break time 4315 in a single record.

[0077] User ID 4311 stores the driver's identifier. In this embodiment, the identifier of biometric information 41 is used. The clock-in date and time 4312 and the previous clock-out date and time 4313 store the clock-in and clock-out dates and times. The number of consecutive working days 4314 stores the number of consecutive days worked. The break time 4315 stores the break time taken by the driver.

[0078] Figure 6 shows an example of delivery data from the business information 43 received by the business information collection device 80. The delivery data 432 includes user ID 4321, date and time 4322, area ID 4323, transported items 4324, delivery route 4325, and whether or not there is a delay 4326 in a single record.

[0079] User ID 4321 stores the driver's identifier. In this embodiment, the identifier of biometric information 41 is used. Date and time 4322 stores the date and time when delivery started. Items to be delivered 4324 stores the type of goods to be delivered. Delivery route 4325 stores the delivery route. Delay status 4326 stores whether or not a delay was reported during delivery. Furthermore, the driving information 44, which includes the location information or driving route of vehicle 8, may be included in the delivery data 432 and treated as business information 43.

[0080] Figure 12 shows an example of time-series biological-environment-business information 47 generated by the time-series biological-environment-business information generation unit 31. Past time-series biological-environment-business information 51 has a similar configuration.

[0081] The time-series biological-environmental-work information 47 includes user ID 471, date and time 472, autonomic nerve LF / HF 473, body temperature 474, ambient temperature 475, consecutive days of work 476, delay 477, and driving status 478 in a single record.

[0082] User ID 471 stores the driver's identifier. In this embodiment, the identifier of biometric information 41 is used. Date and time 472 stores the date and time when the heart rate data, which is the starting point of the analysis window of the autonomic nerve LF / HF 473 constituting the biometric information 41, was measured by the biometric information collection device 60.

[0083] As described later, the autonomic nervous system LF / HF473 is stored as a value indicating the balance of the autonomic nervous system (sympathetic and parasympathetic nerves), representing the ratio of low-frequency (LF) to high-frequency (HF) components of the power spectral density of the R-wave interval (RRI) of heart rate data. The low-frequency component indicates sympathetic nervous system activity, while the high-frequency component indicates parasympathetic nervous system activity.

[0084] Body temperature 474 stores the body temperature measured by the biometric information collection device 60. Air temperature 475 stores the air temperature from the environmental information 42. Consecutive attendance days 476 stores the value of consecutive attendance days 4314 from the attendance data 431 of the work information 43.

[0085] Delay 477 stores the value 4326 indicating whether there is a delay in the delivery data 432 of the business information 43. Driving status 478 stores the driving status based on the speed and location of the driving information 44. In this embodiment, if the vehicle is driving, the type of road is stored, and if it is stopped, "Stopped" is stored.

[0086] Furthermore, if the operation support server 1 predicts accident risk based on the driver's biometric information 41, environmental information 42, and work information 43, regardless of the vehicle's driving status, it is not necessary to combine the values ​​of the driving information 44 with the time-series biometric-environmental-work information 47.

[0087] Figure 14 shows an example of accident determination information 52. Accident determination information 52 stores information on accidents or incidents that have occurred in the past. Accident determination information 52 includes user ID 521, detection date and time 522, vehicle speed 523, acceleration 524, distance between vehicles 525, camera information 526, presence or absence of incident 527, and incident cause 528 in a single record.

[0088] User ID 521 stores the driver's identifier. In this embodiment, the identifier of biometric information 41 is used. Detection date and time 522 stores the date and time when the accident or incident occurred. Vehicle speed 523, acceleration 524, and following distance 525 store the detected values ​​of vehicle speed, acceleration, and following distance, respectively, at the time the accident or incident occurred. Camera information 526 indicates the image information at the time the accident or incident occurred.

[0089] The Incident Status 527 stores whether or not an incident (or accident) occurred. It stores "1" if an incident (or accident) occurred, and "0" if there is no incident (or accident). The Incident Factor 528 stores the label of the factor that caused the incident (or accident).

[0090] Incident (or accident) detection may be performed automatically using a program or machine learning model that does not illustrate information about times when an incident (or accident) is likely to occur, such as sudden braking, based on driving information 44 from a distance sensor 12, speed sensor 13, acceleration sensor 14, etc., installed on the vehicle 8.

[0091] For detected incidents (or accidents), for example, the administrator of the operation support server 1 refers to camera information before and after the detection date and time 522 to set whether an incident exists 527, determine the incident cause 528, and input a label as text or other format. Note that the setting of whether an incident exists 527 and the determination and setting of the incident cause 528 may be performed by a pre-configured machine learning model.

[0092] As shown in Figure 23, an incident (or accident) is detected when the distance between vehicles D1 corresponding to the vehicle speed S1 is below a predetermined threshold or when the acceleration A1 is below a threshold during deceleration. The operation support server 1 acquires video footage taken by the camera 15 from the driving information 44 around the time the incident was detected and displays it on the output device 7. The administrator of the operation support system determines the cause of the incident (or accident) from the video footage before and after the incident on the output device 7 and inputs a cause label from the input device 6.

[0093] Figure 16 shows an example of accident risk information 45. Accident risk information 45 stores the prediction results calculated by the accident risk prediction unit 33. Accident risk information 45 includes user ID 451, measurement time 452, time-series biological-environmental-business information 453, prediction target time period 454, and accident occurrence probability 455 in a single record.

[0094] User ID 451 stores the driver's identifier. In this embodiment, the identifier of biometric information 41 is used. Measurement time 452 stores the date and time the prediction was made. Time-series biometric-environmental-business information 453 stores a pointer that identifies the time-series biometric-environmental-business information 47 used in the prediction.

[0095] The prediction time period 454 stores the start and end points of the time for which accident risk is predicted. The end point is a predetermined time period predicted by the accident risk prediction model 48. The accident probability 455 stores a value representing the probability of an accident or incident occurring as a percentage.

[0096] Figure 18 shows an example of accident risk factor information 46. Accident risk factor information 46 stores the factors calculated by the accident risk factor calculation unit 34. Accident risk factor information 46 includes user ID 461, accident risk 462, first accident risk factor 463, second accident risk factor 464, and factor label 465 in a single record.

[0097] User ID 461 stores the driver's identifier. In this embodiment, the identifier of biometric information 41 is used. Accident risk stores a pointer to the corresponding accident risk information 45. First accident risk factor 463 stores the element that was the biggest contributing factor to the accident or incident, as output by the accident risk factor calculation model 49. Second accident risk factor 464 stores the element that was the second biggest contributing factor, as output by the accident risk factor calculation model 49. Factor label 465 stores the factor label received from the prediction result notification device 9.

[0098] The accident risk factor calculation model 49 extracts items and values ​​that are factors in the probability of an accident occurring 455 from the driver's environmental information 42 and work information 43 for which accident risk information 45 has been calculated, and outputs the first accident risk factor 463 and the second accident risk factor 464.

[0099] Furthermore, the distinction between the first accident risk factor 463 and the second accident risk factor 464 can be made, for example, by designating the item with the highest probability of contributing to the accident probability 455 as the first accident risk factor 463, and the item with the next highest probability as the second accident risk factor 464.

[0100] Furthermore, the items in the accident risk factor calculation unit 34 are not limited to those shown in the diagram, but can be any items included in the biological information 41, environmental information 42, business information 43, and driving information 44, and the items to be integrated as time-series biological-environmental-business information 47 should be set in advance.

[0101] Although not illustrated, the content dictionary 53 consists of templates for messages to be sent to the driver, and pre-set example sentences corresponding to the magnitude of the accident probability 455 and the content of the first accident risk factor 463 and the second accident risk factor 464.

[0102] Furthermore, while the accident risk factor labels 54 are not illustrated, any information that associates the factor label with either the first accident risk factor 463 or the second accident risk factor 464 is acceptable.

[0103] <Processing details> The details of the process shown in Figure 2 will be explained below. Figure 7 is a flowchart showing an example of the process performed in the time-series biological-environmental-business information generation unit 31. This process is the same as the process performed in step S3 of Figure 2.

[0104] The time-series biological-environmental-business information generation unit 31 first extracts heart rate interval data 413 from the biological information 41 acquired in step S2 of Figure 2 (S11). The time-series biological-environmental-business information generation unit 31 may also extract blood pressure 414 and body temperature 415 data corresponding to the heart rate interval data 413 of the biological information 41.

[0105] Next, the time-series biological-environmental-business information generation unit 31 performs preprocessing such as excluding or interpolating missing intervals on the extracted heart rate interval data 413, the environmental information 42, the business information 43, and the driving information 44 acquired in step S2, to generate preprocessed data 450 (S12). The preprocessed data 450 consists of preprocessed heart rate interval data 413A, preprocessed environmental information 42A, preprocessed business information 43A, and preprocessed driving information 44A.

[0106] Furthermore, since the measurement (or acquisition) intervals for the data differ for each of the biological information 41 to the driving information 44, the threshold Thf used to determine missing intervals can be set to different values ​​for each of the following: heart rate interval data 413, environmental information 42, work information 43, and driving information 44.

[0107] Next, the time-series biological-environmental-business information generation unit 31 calculates the autonomic nervous system function LF / HF as an autonomic nervous system function index from the pre-processed heart rate interval data 413A within a predetermined analysis window (time width), as described later, and stores it in the pre-processed data 450 as ANF information 473 (S13).

[0108] The calculation of ANF information 473 is performed as follows. The time-series biological-environmental-business information generation unit 31 calculates heart rate interval data (RRI data) for an analysis window (predetermined period) ΔTw from the pre-processed heart rate interval data 413A shown in Figure 8 as heart rate variability time-series data, and further calculates fluctuations from the heart rate variability time-series data. Figure 9 is a graph showing an example of fluctuations (heart rate variability) of heart rate interval data calculated by the time-series biological-environmental-business information generation unit 31. The RRI of the heart rate interval data is not constant and fluctuates due to autonomic nervous system activity, etc.

[0109] The time-series biological-environmental-business information generation unit 31 performs frequency spectral analysis on the heart rate variability time-series data and calculates the power spectral density (PSD). The power spectral density can be calculated using well-known methods.

[0110] Next, the time-series biological-environmental-business information generation unit 31 calculates the intensity LF of the low-frequency component and the intensity HF of the high-frequency component of the power spectral density. Figure 10 is a graph showing an example of the frequency domain of the power spectral density of heart rate variability.

[0111] The time-series biological-environmental-business information generation unit 31 calculates the total autonomic nerve power by summing the intensity (integral value) LF of the low-frequency component region (0.05Hz to 0.15Hz) and the intensity (integral value) HF of the high-frequency component region (up to 0.15Hz to 0.40Hz) of the power spectrum (LF + HF), as shown in Figure 10.

[0112] Furthermore, the time-series biological-environmental-business information generation unit 31 calculates the ratio of the intensity LF of the low-frequency component of the power spectrum to the intensity HF of the high-frequency component (autonomic nerve LF / HF) as ANF information 473.

[0113] Through the above process, the operation support server 1 calculates heart rate variability time series data for each analysis window ΔTw from the heart rate interval data of the biometric information 41, and further calculates the ratio of the intensity of the low-frequency component to the high-frequency component as ANF information 473.

[0114] The high-frequency component of ANF information 473 appears in heart rate variability when the parasympathetic nervous system is activated (tension), while the low-frequency component appears in heart rate variability when both the sympathetic and parasympathetic nervous systems are activated (tension).

[0115] Since it is known that an activated sympathetic nervous system indicates a state of stress, and an activated parasympathetic nervous system indicates a state of relaxation, it is possible to determine whether a driver is in a stressed or relaxed state by analyzing the intensity of the low-frequency component (LF) and the high-frequency component (HF).

[0116] Next, the time-series biological-environment-operation information generation unit 31 aligns the time widths of the ANF information 473, pre-processed environment information 42A, pre-processed operation information 43A, and pre-processed operation information 44A, and combines them to generate time-series biological-environment-operation information 47 (S14).

[0117] ANF ​​information 473 is generated for each predetermined analysis window ΔTw from which preprocessed heart rate interval data 413A is acquired, with the time interval of the analysis window ΔTw being, for example, 1 minute. On the other hand, preprocessed environmental information 42A is acquired at time intervals such as every hour, preprocessed driving information 44A is collected at the measurement interval of the vehicle 8's sensors (for example, every second), and preprocessed work information 43A is recorded irregularly according to the divisions of the driver's work.

[0118] As shown in Figure 11, the measurement (or acquisition) timing of the pre-treated environmental information 42A to the pre-treated operation information 44A differs from that of the calculation interval (analysis window ΔTw) of the ANF information 473, and the granularity of the measurement interval (acquisition interval) also differs for each.

[0119] Therefore, the time-series biological-environment-business information generation unit 31 formats the data from pre-processed environmental information 42A to pre-processed operational information 44A in accordance with the calculation interval of the ANF information 473, which is the biological information 41 of the monitored target for issuing warnings, and then combines them to generate a single record of time-series biological-environment-business information 47.

[0120] In the case of pre-processed operational information 44A with a time interval shorter than the calculation interval (analysis window ΔTw) of ANF information 473, the time-series biological-environmental-operational information generation unit 31 calculates a representative value such as the average value within the time interval of analysis window ΔTw as the value corresponding to ANF information 473.

[0121] On the other hand, in the case of pre-processed environmental information 42A or pre-processed business information 43A, where the time interval is longer than the calculation interval (analysis window ΔTw) of ANF information 473, the most recent (or immediately preceding) data in the analysis window ΔTw is acquired as the data corresponding to ANF information 473.

[0122] As described above, the time-series biological-environment-business information generation unit 31 combines the pre-processed environmental information 42A, pre-processed business information 43A, and pre-processed operation information 44A data into a single record after adjusting them to the interval of the analysis window ΔTw of the ANF information 473 to generate time-series biological-environment-business information 47.

[0123] As a result, the time-series biological-environmental-operational information 47 can be generated without any time-series discrepancies, as shown in Figure 12, by combining the pre-processed environmental information 42A, pre-processed operational information 43A, and pre-processed operational information 44A corresponding to the time series of the ANF information 473, based on the time interval of the ANF information 473 calculated for each analysis window ΔTw in which heart rate interval data 413 is acquired.

[0124] Furthermore, the data accumulated in the time-series biological-environmental-business information 47 is reflected in the past accident risk prediction model training data 50 at predetermined timings.

[0125] Figure 13 is a flowchart showing an example of the process performed in the accident risk prediction model training unit 32. This process is the same as the process performed in step S1 of Figure 2.

[0126] The accident risk prediction model training unit 32 receives a specified period for the data to be used for training, and extracts data from the past time-series biological-environmental-business information 51 of the accident risk prediction model training data 50 within the specified period to generate model input data 500 (S21).

[0127] The specified period can be received from input device 6 or an external computer, and is entered by the user or administrator of the operation support system. Furthermore, the specified period should preferably be between 2 minutes and several hours.

[0128] Next, the accident risk prediction model training unit 32 acquires pre-configured accident judgment information 52, which is past information in which an administrator or computer has judged an actual accident or an event (incident) that could lead to an accident. It then extracts information on whether or not an event occurred from past time-series biological-environmental-business information 51 within a specified period and uses it as a training label 55 (S22).

[0129] This process predicts future events from past time-series biological-environmental-business information 51, meaning that the future period is within or approximately the same as the specified period. For example, if the specified period is 30 minutes, and the time of the event occurrence in the accident judgment information 52 is within 30 minutes from the start time of the time-series biological-environmental-business information 51, then the event occurrence label is assigned to that time and used as the training label 55. In other words, the incident factor 528 corresponding to the presence or absence of an incident 527 registered in the accident judgment information 52 is linked to past time-series biological-environmental-business information 51 for an arbitrary time range (e.g., 30 minutes) preceding the corresponding time, as the training label for the accident risk prediction model 48.

[0130] Next, using the model input data 500 extracted in step S21 and the training labels 55 generated in step S22, an accident risk prediction model 48 is trained to output the accident risk in the future (after a predetermined time: for example, 30 minutes later) (S28).

[0131] Figure 24 shows an example of training the accident risk prediction model 48. While the presence or absence of an incident corresponding to the training label 55 is a binary value of "0" or "1", the value output by the accident risk prediction model 48 is a continuous value that falls within a range such as 0 to 1. When applying the accident risk prediction model 48, the predicted continuous value is used as is, or processing is performed by setting a threshold and converting it to calculate the probability of an accident occurring from 0 to 100%.

[0132] The above training can be improved by using accident judgment information 52, which is created by adding new time-series biological-environmental-operational information 47 to past time-series biological-environmental-operational information 51, and also by adding accident risk factor labels 54. This improves the prediction accuracy of the accident risk prediction model 48.

[0133] Figure 15 is a flowchart showing an example of the processing performed by the accident risk prediction unit 33. This processing is the same as the processing performed in step S4 of Figure 2. The accident risk prediction unit 33 acquires time-series biological-environmental-work information 47 and inputs the time-series biological-environmental-work information 47 into the trained accident risk prediction model 48 to predict the driver's accident risk after a predetermined time. The accident risk prediction unit 33 stores the accident occurrence probability output by the accident risk prediction model 48 into the accident risk information 45 (S31).

[0134] The accident risk prediction unit 33 stores the user ID 471 of the time-series biological-environment-business information 47 in the user ID 451 of the accident risk information 45, similarly stores the date and time 472 of the time-series biological-environment-business information 47 in the measurement time 451, stores a pointer to identify the record of the time-series biological-environment-business information 47 in the time-series biological-environment-business information 453, stores the predicted time range output by the accident risk prediction model 48 in the prediction target time period 454, and stores the accident occurrence probability output by the accident risk prediction model 48 in the accident occurrence probability 455.

[0135] The data that the accident risk prediction unit 33 inputs to the accident risk prediction model 48 is the unprocessed data from the time-series biological-environmental-business information 47.

[0136] Through the above process, time-series biological-environmental-work information 47 is input into a trained accident risk prediction model 48, and accident risk information 45 for a predetermined time period is output for each driver.

[0137] Furthermore, the accident risk prediction unit 33 may omit generating accident risk information 45 if the accident probability output by the accident risk prediction model 48 is less than or equal to a predetermined threshold Th2 (for example, 5%).

[0138] Figure 17 is a flowchart showing an example of the process performed in the accident risk factor calculation unit 34. This process is the same as the process performed in step S5 of Figure 2.

[0139] The accident risk factor calculation unit 34 acquires the accident risk information 45 output by the accident risk prediction unit 33, the accident judgment information 52 collected from past cases, and the time-series biological-environmental-business information 47 input to the accident risk prediction unit 33, and inputs these into a pre-configured accident risk factor calculation model 49 to generate accident risk factor information 46 (S41).

[0140] The accident risk factor calculation unit 34 stores the user ID 451 of the accident risk information 45 in the user ID 461 of the accident risk factor information 46, stores a pointer to identify the record in the accident risk information 45 in accident risk 462, stores the first and second accident risk factors output by the accident risk prediction model 48 in first accident risk factor 463 and second accident risk factor 464, respectively, and stores the incident factor 528 of the accident judgment information 52 in factor label 465.

[0141] As a result of the above processing, for drivers for whom accident risk information 45 has been generated, the accident risk factor calculation model 49 predicts the factors from the current time-series biological-environmental-work information 47 and the incident factors 528 of past accident judgment information 52, and these factors are generated as the first accident risk factor 463, the second accident risk factor 464, and factor labels 465.

[0142] Furthermore, the accident risk factor calculation unit 34 sets "None" to the factor label 465 in the accident risk factor information 46 for data where the incident factor 528 in the accident judgment information 52 is blank. In addition, the first accident risk factor 463 indicates the main factors that caused the accident risk, and the second accident risk factor 464 presents the factors that contributed to the increase in accident risk.

[0143] Figure 18 shows an example where the first accident risk factor 463 is estimated to be primarily due to "more than 6 consecutive working days" based on work information 43, and the second accident risk factor 464 is estimated to be due to rainy weather based on environmental information 42.

[0144] Figure 19 is a flowchart showing an example of the processing performed in the prediction result presentation unit 35. This processing is performed in steps S6 and S7 of Figure 2.

[0145] The prediction result presentation unit 35 obtains the prediction target time period 454 and the accident occurrence probability 455 from the accident risk information 45, and if the accident occurrence probability 455 exceeds a predetermined threshold Th1, it performs the following processing.

[0146] The prediction result presentation unit 35 obtains the first accident risk factor 463, the second accident risk factor 464, and the factor label 465 from the accident risk factor information 46 corresponding to the accident risk information 45. The prediction result presentation unit 35 searches the presentation content dictionary 53 using the accident probability 455 and the first accident risk factor 463 or the second accident risk factor 464 to obtain a text template. The prediction result presentation unit 35 inserts the accident probability 455, the prediction target time period 454, and the first accident risk factor 463 or the second accident risk factor 464 into the obtained template to generate a warning or caution message and transmits it to the prediction result display terminal 90 used by the driver or administrator (S51).

[0147] Furthermore, the prediction result display unit 35 transmits information to accept factor labels in addition to the message, and when it receives a factor label entered by the driver or manager, etc., on the prediction result display terminal 90, it updates (or adds) the factor label 465 of the accident risk factor information 46 (S52).

[0148] The prediction result presentation unit 35 can feed back the accident judgment information 52 when the accident risk factor information 46 is updated, and add the content of the factor label 465 to the incident factor 528. This allows the incident factor 528 set by the driver or others to be reflected when training the accident risk prediction model 48. The operation support server 1 can then generate and send a message that is not unusual for the driver.

[0149] As described above, the operation support server 1 of this embodiment combines information with different measurement and acquisition intervals, such as environmental information 42, business information 43, and driving information 44, to match the time interval for calculating biological information 41, and then combines this information to generate time-series biological-environmental-business information 47, which is stored in chronological order. The operation support server 1 then uses the time-series biological-environmental-business information 47, which is matched to the calculation interval of biological information 41, to predict the probability of an accident occurring 455 (accident risk information 45) after a predetermined time (in the future), and sends a warning or caution message if the probability of an accident occurring 455 exceeds the threshold Th1.

[0150] This allows for the integration of environmental information 42, operational information 43, and driving information 44, which influence the accident probability 455, into information that matches the calculation interval of biometric information 41, thereby improving the accuracy of accident or incident prediction.

[0151] Furthermore, by inputting the predicted accident risk information 45, past accident judgment information 52, and new time-series bio-environment-work information 51 into the accident risk factor calculation model 49, accident risk factor information 46 can be calculated, and by including the accident risk factors in the message, the driver can be presented with the reason for the warning or alert. By adding accident risk factors in addition to the warning or alert, a message that does not feel unnatural can be sent.

[0152] Furthermore, by integrating the time-series data of environmental information 42, operational information 43, and driving information 44 to match the calculation interval of biological information 41, and inputting the resulting time-series biological-environmental-operational information 47 into the accident risk factor calculation model 49, it is possible to calculate accident risk factor information 46 without time-series discrepancies.

[0153] Furthermore, the operation support server 1 can receive labels for accident risk factors from the prediction result display terminal 90 in response to the accident risk information 45 transmitted to the prediction result display terminal 90, and feed this back to the accident risk prediction model 48 and the accident risk factor calculation model 49. This enables the provision of warnings that are not perceived as unusual by the driver.

[0154] Furthermore, the operation support server 1 can provide drivers with notifications that do not cause them any discomfort by offering video footage of the driving conditions prior to the time of the occurrence of the accident risk information 45 as accident risk factor information 46.

[0155] <Conclusion> As described above, the above embodiment can be configured as follows.

[0156] (1) A computer (operation support server 1) having a processor (2) and memory (3) supports the operation of a vehicle (8), comprising: a first step (biometric information collection device 60) in which the computer (1) acquires biometric information (41) of the driver operating the vehicle (8); a second step (environmental information collection device 70) in which the computer acquires environmental information (42) of the driver; a third step (business information collection device 80) in which the computer acquires business information (43) of the driver; a fourth step (S31) in which the computer generates biometric indicator data from the biometric information (41); and the computer processes the biometric indicator data (ANF information 473), environmental information (42), and business information in a time series. A method for supporting operations, comprising: a fifth step (time-series biological-environmental-operational information generation unit 31) for aligning and combining data to generate integrated information (time-series biological-environmental-operational information 47); a sixth step (S4) in which the computer inputs the integrated information (31) into a pre-set accident risk prediction model (48) to calculate accident risk information (45) after a predetermined time; and a seventh step (accident risk factor calculation unit 34) in which the computer inputs the accident risk information (45), the integrated information (31), and pre-set accident judgment information (52) into a pre-set factor calculation model (accident risk factor calculation model 49) to calculate factor information (accident risk factor information 46) for the accident risk information (45).

[0157] With the above configuration, the operation support server 1 can not only predict accident risk information 45 after a predetermined time, but also output accident risk factor information 46, thereby presenting information that is not unnatural to the driver.

[0158] (2) The operation support method described in (1) above, further comprising an eighth step (S6) in which the computer outputs the accident risk information (45) and the factor information (46) if the accident risk information (45) satisfies a predetermined condition (exceeds threshold Th1).

[0159] With the above configuration, if the accident risk information 45 after a predetermined time meets the predetermined conditions, the operation support server 1 outputs the accident risk information 45 and accident risk factor information 46 to the prediction result display terminal 90, thereby notifying the driver that the accident risk has increased, as well as the factors of the accident risk.

[0160] (3) The operation support method described in (2) above, further comprising a ninth step (S7) in which the computer updates the factor information (46) with the factor label when it receives a factor label for the output factor information (46).

[0161] With the above configuration, the accident judgment information 52 can be updated with factor labels set by the driver or manager, thereby reflecting the accident risks actually encountered by the driver.

[0162] (4) An operation support method as described in (3) above, further comprising a tenth step (S48) in which the computer updates the factor labels of the factor information (46) and reflects them in the accident judgment information (52), and then performs training of the factor calculation model (49) with the accident judgment information (52) that reflects the factor labels.

[0163] With the above configuration, when retraining the accident risk prediction model 48, the accident judgment information 52 with updated factor labels can be used to feed back the accident risk actually encountered by the driver to the accident risk prediction model 48.

[0164] (5) The operation support method described in (1) above, wherein the fifth step is characterized in that when integrating the biometric information (41), the environmental information (42), and the business information, the values ​​of the environmental information (42) and the business information are obtained based on the calculation interval of the biometric information (41).

[0165] With the above configuration, by acquiring environmental information 42 and business information 43 in accordance with the calculation interval of information calculated from biological information 41 (for example, ANF information 473), the time-series biological-environmental-business information 47 becomes information with a consistent time-series granularity. This improves the calculation accuracy of the accident risk prediction model 48 and the accident risk factor calculation model 49 that use the time-series biological-environmental-business information 47.

[0166] It should be noted that the present invention is not limited to the embodiments described above, and various modifications are included. For example, the embodiments described above are described in detail to make the present invention easier to understand, and are not necessarily limited to including all the configurations described. Furthermore, it is possible to replace parts of the configuration of one embodiment with the configuration of another embodiment, and it is also possible to add configurations from other embodiments to the configuration of one embodiment. In addition, for parts of the configuration of each embodiment, the addition, deletion, or substitution of other configurations can be applied individually or in combination.

[0167] Furthermore, each of the above-mentioned configurations, functions, processing units, and processing means may be implemented in hardware, in whole or in part, for example, by designing them as integrated circuits. Alternatively, each of the above-mentioned configurations and functions may be implemented in software by having the processor interpret and execute programs that realize each function. Information such as programs, tables, and files that realize each function can be stored in memory, a recording device such as a hard disk or SSD (Solid State Drive), or a recording medium such as an IC card, SD card, or DVD.

[0168] Furthermore, the control lines and information lines shown are those deemed necessary for explanatory purposes, and not all control lines and information lines are necessarily shown in the actual product. In reality, it is safe to assume that almost all components are interconnected. [Explanation of Symbols]

[0169] 1. Operation support server 2 processors 3 memory 4 Auxiliary storage 10. Operation Information Collection Device 31. Time-Series Biological-Environmental-Business Information Generation Department 32 Accident Risk Prediction Model Training Department 33 Accident Risk Prediction Unit 33 34. Accident Risk Factor Calculation Unit 35 Prediction Result Presentation Section 41. Biometric Information 42 Environmental information 43 Business Information 44. Driving Information 45. Accident Risk Information 46. ​​Information on accident risk factors 47, 51 Time-series biological-environmental-business information 48. Accident Risk Prediction Model 49 Accident Risk Factor Calculation Model 50 Accident risk prediction model training data 52 Accident Judgment Information 53. Dictionary of presented content 54 Accident Risk Factor Labels 60. Biological Information Acquisition Device 61 Heart rate monitor 70 Environmental Information Collection Device 80 Business Information Collection Device 90 Prediction result display terminal

Claims

1. A computer having a processor and memory provides a method for assisting the operation of a vehicle, The first step is for the computer to acquire biometric information of the driver while the vehicle is in operation, The computer performs a second step of acquiring the driver's environmental information, The computer then performs a third step of acquiring the driver's work information, The computer performs a fourth step of generating biometric data from the biometric information, The fifth step involves the computer combining the biometric data, environmental information, and business information, aligning their time-series granularity, to generate integrated information. The sixth step involves the computer inputting the integrated information into a pre-configured accident risk prediction model to calculate accident risk information after a predetermined time. The seventh step involves the computer inputting the accident risk information, the integrated information, and the pre-configured accident judgment information into a pre-configured factor calculation model to calculate the factor information for the accident risk information. A method for supporting vehicle operation, characterized by including the following:

2. A method for supporting the operation of an operator according to claim 1, An operation support method further comprising an eighth step in which the computer outputs the accident risk information and the factor information if the accident risk information satisfies predetermined conditions.

3. A method for supporting the operation according to claim 2, A method for supporting operations, further comprising a ninth step in which, if the computer receives a factor label for the factor information output, the computer updates the factor information with the factor label.

4. A method for supporting the operation of an operator according to claim 3, An operation support method characterized in that the computer further includes a tenth step of reflecting the updated factor labels in the accident judgment information and training the accident risk prediction model with the accident judgment information reflecting the factor labels.

5. A method for supporting the operation of an operator according to claim 1, The fifth step described above is: A method for supporting operations, characterized in that when integrating the aforementioned biological information, environmental information, and operational information, the values ​​of the environmental information and operational information are obtained based on the calculation interval of the biological information.

6. A server having a processor and memory, A biometric information collection device connected to the aforementioned server to acquire the driver's biometric information, An environmental information collection device connected to the aforementioned server to acquire the driver's environmental information, A business information collection device connected to the aforementioned server to acquire business information of the aforementioned driver, An operation support system that supports the operation of a vehicle, including a terminal connected to the server and outputting messages, The aforementioned server, An integrated information generation unit acquires biometric information of the driver operating the vehicle from the biometric information collection device, acquires environmental information of the driver from the environmental information collection device, acquires business information of the driver from the business information collection device, generates biometric indicator data from the biometric information, and combines the biometric indicator data, the environmental information, and the business information after aligning their time-series granularity to generate integrated information. An accident risk prediction unit inputs the aforementioned integrated information into a pre-configured accident risk prediction model to calculate accident risk information after a predetermined time, A factor calculation unit inputs the aforementioned accident risk information, the aforementioned integrated information, and the pre-set accident judgment information into a pre-set factor calculation model to calculate factor information for the aforementioned accident risk information. An operation support system characterized by having the following features.

7. The operation support system according to claim 6, The aforementioned server, An operation support system further comprising a display unit that outputs the accident risk information and the factor information to the terminal when the accident risk information meets predetermined conditions.

8. The operation support system according to claim 7, The aforementioned display unit is, An operation support system characterized in that, upon receiving a factor label for the output factor information, the factor information is updated with the factor label.

9. The operation support system according to claim 8, The aforementioned server, An operation support system further comprising a training unit that reflects the updated factor labels in the accident judgment information and uses the accident judgment information reflecting the factor labels to train the accident risk prediction model.

10. The operation support system according to claim 6, The integrated information generation unit, An operation support system characterized in that, when integrating the aforementioned biological information, environmental information, and business information, the values ​​of the environmental information and business information are obtained based on the calculation interval of the biological information.

11. A server having a processor and memory to support the operation of a vehicle, The aforementioned server, An integrated information generation unit that acquires biometric information of the driver while the vehicle is in operation, acquires environmental information of the driver, acquires work information of the driver, generates biometric indicator data from the biometric information, and combines the biometric indicator data, environmental information, and work information after aligning their time-series granularity to generate integrated information. An accident risk prediction unit inputs the aforementioned integrated information into a pre-configured accident risk prediction model to calculate accident risk information after a predetermined time, A factor calculation unit inputs the aforementioned accident risk information, the aforementioned integrated information, and the pre-set accident judgment information into a pre-set factor calculation model to calculate factor information for the aforementioned accident risk information. A server characterized by having the following features.

12. A server according to claim 11, A server further comprising a display unit that outputs the accident risk information and the factor information when the accident risk information meets predetermined conditions.

13. The server according to claim 12, The aforementioned display unit is, A server characterized by updating the factor information with the factor label when it receives a factor label for the factor information output.

14. The server according to claim 13, A server further comprising a training unit that reflects the updated factor labels in the accident judgment information and trains the accident risk prediction model with the accident judgment information that reflects the factor labels.

15. The server according to claim 11, The integrated information generation unit, A server characterized in that, when integrating the aforementioned biological information, environmental information, and business information, it acquires the values ​​of the environmental information and business information based on the calculation interval of the biological information.

Images