Determination device, determination system, determination method and program

Abstract

To easily determine whether a vehicle passing a bridge is an overweight vehicle.SOLUTION: A determination device includes: an acquisition unit that acquires, from a sensor device provided on a bridge on which a vehicle travels, time-series data representing a temporal change in a displacement amount at a position on the bridge where the sensor device is provided; a peak detection unit that detects a peak value representing a magnitude of a peak of a first-shape waveform in which the displacement amount increases and then decreases or decreases and then increases; a distribution calculation unit that calculates a frequency distribution representing the number of samples of the peak values detected during a learning period, included in each of a plurality of divided ranges obtained by dividing a possible value range of the displacement amount; a threshold calculation unit that calculates, as a determination threshold, a displacement amount at a boundary in the frequency distribution between a decreasing region, in which the number of samples decreases, and a non-change region, in which the amount of change in the number of samples is equal to or less than a predetermined value; and a determination unit that determines, as an overweight vehicle, a vehicle that causes a peak value exceeding the determination threshold in the time-series data during an observation period after the learning period.SELECTED DRAWING: Figure 3

Description

Technical Field

[0001] Embodiments of the present invention relate to a determination device, a determination system, a determination method, and a program.

Background Art

[0002] There is known a system for estimating the weight of a vehicle that has passed over a bridge from the amount of displacement of the bridge (for example, Patent Document 1 and Patent Document 2). The relationship between the amount of displacement of the bridge and the weight of the vehicle varies from bridge to bridge. For this reason, in such a system, a conversion model representing the correspondence between the amount of displacement of the bridge and the weight of the vehicle must be generated in advance by learning. For example, such a system acquires the amount of displacement of the bridge when a test vehicle of known weight passes by, and generates a conversion model.

[0003] In such a system, it is desirable to generate a conversion model by having the test vehicle run alone at a determined time and pass over the bridge a plurality of times. In addition, a method has also been proposed in which a camera for identifying the passing vehicle is provided on the bridge to detect the passing time of the test vehicle (for example, Patent Document 3).

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Patent Document 2

Patent Document 3

Patent Document 4

Summary of the Invention

Problems to be Solved by the Invention

[0005] However, in order to make the test vehicle pass over the bridge multiple times under single - vehicle running conditions at a specified time, it was necessary to implement traffic control or conduct the test during a time period with low traffic volume at night. Also, when installing cameras on public bridges to detect the passage of test vehicles, it was necessary to take into account personal information such as people shown in the cameras.

[0006] In addition, it is said that the deterioration of bridges is caused by overloaded large - sized vehicles, which account for about 0.3% of the total traffic volume. In particular, overweight vehicles that exceed the design load of bridges have a significant impact on bridge deterioration. Therefore, the detection of overweight vehicles passing over bridges is very important for understanding bridge deterioration. However, in such a system, when detecting overweight vehicles, it was also necessary to generate in advance a conversion model representing the correspondence between the displacement amount of the bridge and the weight of the overweight vehicle through learning.

[0007] For example, a system has been proposed (for example, Patent Document 4) in which the feature amounts calculated when each of a plurality of test vehicles passes over a bridge are used as a group of normal values, the boundary between the normal values and the outliers is detected, and the overweight vehicle is detected using the feature amount of the detected boundary as a threshold value. However, in this system as well, the ratio of outliers has to be set as an inherent parameter of the bridge. Also, when performing such processing in real - time by an edge computer provided near the bridge, since an edge computer having sufficient performance for executing the processing is required, the cost becomes high, and the burden of developing a program to be executed on the edge computer also becomes large.

[0008] The present invention has been made in view of the above, and provides a determination device, a determination system, a determination method, and a program that can determine whether a vehicle passing over a bridge is an overweight vehicle by a process with a small load that can be executed by a computer with a simple configuration without using parameters set for each bridge.

Means for Solving the Problems

[0009] The determination device according to the embodiment includes: an acquisition unit that acquires time-series data representing a temporal change in the amount of displacement at the position where the sensor device is provided on the bridge from the sensor device provided on the bridge on which the vehicle travels; a peak detection unit that detects a peak value representing the magnitude of the peak of the first-shaped waveform each time the first-shaped waveform in which the amount of displacement increases and then decreases or decreases and then increases as the vehicle passes over the bridge is included in the time-series data; a distribution calculation unit that calculates a frequency distribution representing the number of samples of the peak values detected during the learning period, each included in a plurality of divided ranges obtained by dividing the range of values that the amount of displacement can take; a threshold calculation unit that calculates, as a determination threshold, the amount of displacement at the boundary between a decreasing region in which the number of samples decreases in the frequency distribution and a non-changing region on the side where the amount of displacement is larger than the decreasing region and in which the change amount of the number of samples is equal to or less than a predetermined value; and a determination unit that determines, as an overweight vehicle, the vehicle that has generated the peak value larger than the determination threshold in the time-series data during the observation period after the learning period.

Brief Description of the Drawings

[0010]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12

Figure 13

Figure 14

Figure 15

Figure 16

Figure 17

Figure 18

Figure 19

Figure 20

Figure 21

DETAILED DESCRIPTION OF THE INVENTION

[0011] FIG. 1 is a diagram showing the configuration of the determination system 10 according to the embodiment. The determination system 10 includes a sensor device 20, an edge device 22, a warning output device 24, and a server device 26.

[0012] The sensor device 20 is provided on a structure through which the vehicle passes. In the present embodiment, the structure is a bridge. The structure is not limited to a bridge, and may be, for example, a plate-like object provided on the bridge as long as it expands, contracts, or displaces when the vehicle passes.

[0013] The sensor device 20 detects displacement data representing the amount of displacement at the position where the sensor device 20 is provided on the bridge at a predetermined sampling interval. In the present embodiment, the sensor device 20 detects displacement data representing the amount of displacement in the traveling direction of the vehicle. More specifically, the sensor device 20 detects the amount of expansion and contraction of the bridge in the traveling direction of the vehicle at the position where the sensor device 20 is provided on the bridge.

[0014] The sensor device 20 transmits the time-series displacement data detected at a predetermined sampling interval to the edge device 22. Note that the determination system 10 may include a plurality of sensor devices 20. In this case, each of the plurality of sensor devices 20 independently detects the time-series displacement data and transmits the time-series displacement data to the edge device 22.

[0015] The edge device 22 is an information processing device connected to the sensor device 20. For example, the edge device 22 is connected to the sensor device 20 by wire via a LAN (Local Area Network).

[0016] The edge device 22 collects the time-series displacement data detected by the sensor device 20 at a predetermined sampling interval. The edge device 22 generates a block that summarizes a plurality of displacement data collected during the transfer interval for each transfer interval that is longer than the sampling interval. Then, the edge device 22 transmits the generated block to the server device 26 via the network for each transfer interval.

[0017] The network can be wired, wireless, or a combination of wired and wireless. The network can be, for example, a LAN, a PAN (Personal Area Network), or a WAN (Wide Area Network), or a mixed network of PAN, LAN, and WAN. Also, the network may include a cellular communication line such as LTE (Long Term Evolution).

[0018] Also, the edge device 22 functions as a determination device that analyzes the collected time-series displacement data to determine whether an overweight vehicle with a predetermined weight or more has passed over the bridge where the sensor device 20 is provided. An overweight vehicle is, for example, a vehicle that exceeds the weight of a vehicle determined to be able to pass over the bridge. An overweight vehicle is, for example, an overloaded special vehicle that loads something heavier than the loadable weight of the vehicle. Then, the edge device 22 controls the warning output device 24 based on the determination result of whether an overweight vehicle has passed.

[0019] The warning output device 24 is a device that provides information to the vehicles passing over the bridge where the sensor device 20 is provided in response to the control by the edge device 22. The warning output device 24 is, for example, a display device provided on the road that the vehicle passes after passing over the bridge where the sensor device 20 is provided, and displays information to the driver of the vehicle passing on the road. The warning output device 24 may be, for example, an audio output device that outputs audio to the driver. Also, the warning output device 24 may be a communication device that wirelessly transmits information to the information processing device in the vehicle. Also, the warning output device 24 may be a vehicle control device that stops the running of the vehicle or guides the vehicle to a predetermined lane.

[0020] When the warning output device 24 obtains a determination result indicating that an overweight vehicle has passed through the edge device 22, it outputs warning information indicating that the vehicle is an overweight vehicle to the vehicle that has passed through the bridge. In the present embodiment, the warning output device 24 is a display device provided in front of the bridge in the traveling direction of the vehicle, and displays warning information to the driver riding in the vehicle that has passed through the bridge.

[0021] The server device 26 is an information processing device. The server device 26 may be a single computer, or may be configured by a plurality of computers such as a cloud system.

[0022] The server device 26 is connected to the edge device 22 via a network. The server device 26 receives a plurality of blocks each including a plurality of displacement data from the edge device 22. The server device 26 stores the received plurality of blocks. The server device 26 analyzes the plurality of displacement data included in the received plurality of blocks. In the present embodiment, the server device 26 calculates, for example, the degree of deterioration of the bridge by analyzing the plurality of displacement data included in the plurality of blocks.

[0023] FIG. 2 is a diagram showing the arrangement of the sensor device 20 and the like with respect to the bridge.

[0024] In the present embodiment, the sensor device 20 is provided, for example, on the end side rather than the center in the traveling direction of the bridge. Further, in the present embodiment, the sensor device 20 is provided at substantially the center of the lane in the width direction of the bridge. Also, in the present embodiment, the sensor device 20 is provided, for example, on the lower surface of the bridge in the vicinity of the abutment. Note that the sensor device 20 may be provided at other portions of the bridge. For example, the sensor device 20 may be provided on the main girder of the bridge, or may be provided at substantially the center in the traveling direction of the floor slab. Further, when a plurality of sensor devices 20 are provided on the bridge, one or a plurality of sensor devices 20 may be provided for each lane.

[0025] The sensor device 20 detects displacement data representing the amount of displacement in the traveling direction at the target portion where the sensor device 20 is provided on the bridge. For example, the sensor device 20 uses an optical scale to detect the amount of expansion and contraction in the traveling direction at the target portion as displacement data. The amount of expansion and contraction is, for example, a change in distance from several nanometers to about several hundred nanometers between two points at a distance of about several tens of centimeters when a passenger car travels on a concrete bridge.

[0026] The sensor device 20 may detect other physical quantities on the bridge instead of the amount of expansion and contraction in the traveling direction on the bridge. For example, the sensor device 20 may detect strain or acceleration at the target portion of the bridge.

[0027] Also, the edge device 22 is provided near the bridge. For example, the edge device 22 is provided at a position on the bridge that can be connected to the sensor device 20 by a wired LAN cable.

[0028] Also, the warning output device 24 is provided on the road through which the vehicle passes after passing the bridge. For example, the warning output device 24 is a display device, and is arranged above the road with the display surface facing the bridge side. Thereby, the warning output device 24 can notify the driver of the vehicle passing through the road of warning information indicating that it is an overweight vehicle.

[0029] FIG. 3 is a block diagram showing the functional configuration of the edge device 22.

[0030] The edge device 22 includes an acquisition unit 32, a peak detection unit 34, a distribution calculation unit 36, a threshold calculation unit 40, a threshold storage unit 42, a determination unit 44, and a notification control unit 46.

[0031] The acquisition unit 32 acquires time-series data representing the temporal change in the amount of displacement of the bridge at the position where the sensor device 20 is provided on the bridge from the sensor device 20. The acquisition unit 32 provides the acquired time-series data to the peak detection unit 34.

[0032] When a vehicle passes over the bridge and the displacement amount increases and then decreases or decreases and then increases, the peak detection unit 34 detects a peak value representing the magnitude of the peak of the first-shaped waveform every time the first-shaped waveform is included in the time-series data. The magnitude of the peak value is, for example, the absolute value of the difference in the displacement amount from a predetermined reference value to the peak point. The peak detection unit 34 detects a peak value for one sample every time one first-shaped waveform is included in the time-series data. Therefore, the number of samples of the peak values detected by the peak detection unit 34 during a predetermined period represents the number of vehicles that have passed over the bridge during the predetermined period.

[0033] During the observation period for detecting overweight vehicles, every time the peak detection unit 34 detects a peak value, it provides the detected peak value to the determination unit 44. Also, during the preprocessing period for performing preprocessing for detecting overweight vehicles before the observation period, every time the peak detection unit 34 detects a peak value, it provides the detected peak value to the distribution calculation unit 36.

[0034] Note that the preprocessing period is divided into a learning period and a range determination period. The range determination period is a period that is temporally prior to the learning period. The learning period is a period that is longer than or the same as the range determination period.

[0035] The distribution calculation unit 36 identifies the maximum peak value, which is the peak value that is the maximum among the peak values detected during the range determination period. Then, based on the maximum peak value, the distribution calculation unit 36 determines a plurality of divided ranges obtained by dividing the range within which the displacement amount can take values. The range within which the displacement amount can take values is the range from the minimum value of the numerical values that the displacement amount can take to the maximum value of the numerical values that the displacement amount can take.

[0036] For example, the distribution calculation unit 36 determines N divided ranges obtained by dividing the range from 0 to a value corresponding to the maximum peak value into N parts. In this case, the distribution calculation unit 36 changes the upper limit value of the divided range with the largest displacement amount among the N divided ranges to the maximum value of the numerical values that the displacement amount can take. Note that the value based on the maximum peak value is, for example, a value obtained by rounding up or down the maximum peak value to a predetermined number of digits.

[0037] Then, the distribution calculation unit 36 calculates a frequency distribution representing the number of samples of peak values detected during the learning period, each included in a plurality of divided ranges. For example, every time a peak value is detected during the learning period, the distribution calculation unit 36 identifies which divided range among the plurality of divided ranges the detected peak value is included in. Then, during the learning period, the distribution calculation unit 36 calculates the frequency distribution by counting the number of peak values included in each of the plurality of divided ranges.

[0038] After the learning period ends, the distribution calculation unit 36 provides the calculated frequency distribution to the threshold calculation unit 40.

[0039] Based on the frequency distribution calculated by the distribution calculation unit 36, the threshold calculation unit 40 calculates a determination threshold for determining whether a vehicle passing through the bridge is an overweight vehicle. The threshold calculation unit 40 calculates, as the determination threshold, the displacement amount at the boundary between a decreasing region in the frequency distribution where the number of samples decreases and a non-changing region on the side where the displacement amount is larger than that of the decreasing region and where the change amount of the number of samples is equal to or less than a predetermined value. More specifically, the decreasing region is a region of the displacement amount in the frequency distribution where the number of samples decreases as the displacement amount increases. The non-changing region is a region of the displacement amount in the frequency distribution where the change amount of the number of samples is equal to or less than a predetermined value even when the displacement amount increases or decreases. The threshold calculation unit 40 writes the calculated determination threshold into the threshold storage unit 42.

[0040] The threshold storage unit 42 stores the determination threshold.

[0041] During the observation period, the determination unit 44 acquires peak values from the peak detection unit 34. Every time the determination unit 44 acquires a peak value, it determines whether the acquired peak value is larger than the determination threshold stored in the threshold storage unit 42. Then, during the observation period, the determination unit 44 determines a vehicle that has generated a peak value larger than the determination threshold in the time-series data as an overweight vehicle.

[0042] For example, the determination unit 44 determines whether the peak value acquired while the vehicle is passing through or immediately after passing through the bridge is greater than the determination threshold value. Thereby, when a peak value greater than the determination threshold value occurs, the determination unit 44 can determine that the vehicle immediately after passing through the bridge is an overweight vehicle. The determination unit 44 provides a determination result indicating whether the vehicle that has passed through the bridge is an overweight vehicle to the notification control unit 46.

[0043] When the notification control unit 46 determines that the vehicle that has passed through the bridge is an overweight vehicle by the determination unit 44, the notification control unit 46 notifies the overweight vehicle that has passed through the bridge of warning information indicating that it is an overweight vehicle. For example, when the determination unit 44 determines that the vehicle that has passed through the bridge is an overweight vehicle, the notification control unit 46 causes the warning output device 24 to display the warning information. In this case, the warning output device 24 is a display device that is provided on the road through which the vehicle passes after passing through the bridge and displays information to the driver of the vehicle passing through the road.

[0044] Note that the sensor device 20 may be provided for each lane of the bridge. In this case, the warning output device 24 notifies warning information to the passing vehicles for each lane. Also, in this case, the acquisition unit 32 acquires time-series data for each lane. The peak detection unit 34 detects a peak value for each lane. The distribution calculation unit 36 calculates a frequency distribution and a determination threshold value for each lane. The determination unit 44 determines, for each lane, a vehicle that has generated a peak value greater than the determination threshold value in the time-series data as an overweight vehicle. Then, the notification control unit 46 notifies, for each lane, the overweight vehicle that has passed through the bridge of warning information indicating that it is an overweight vehicle.

[0045] FIG. 4 is a diagram showing an example of time-series data of the observed values of the sensor device 20 and time-series data of the moving average values when the vehicle passes through the bridge.

[0046] The time-series data of the observed values of the sensor device 20 generates a first-shaped waveform that increases and then decreases or decreases and then increases over time when the vehicle passes over the bridge. Also, the time-series data of the observed values of the sensor device 20 has an offset component added according to the external environment such as the ambient temperature or ambient humidity. Therefore, the peak value of the first-shaped waveform generated when the vehicle passes over the bridge varies depending on the external environment such as the ambient temperature or ambient humidity even when vehicles of the same weight pass by.

[0047] This offset component changes with a period that is gentler than the first-shaped waveform generated when the vehicle passes over the bridge. Thus, in the present embodiment, the acquisition unit 32 calculates time-series data of difference values representing the difference between the observed values of the sensor device 20 and the moving average values obtained by moving-averaging the observed values using a preset time window. Then, the acquisition unit 32 outputs the calculated time-series data of the difference values as time-series data representing the time change of the displacement amount of the bridge. Thereby, the acquisition unit 32 can output highly accurate time-series data of the displacement amount with the offset component due to changes in the external environment such as the ambient temperature or ambient humidity removed.

[0048] FIG. 5 is a flowchart showing the processing flow of the edge device 22. The edge device 22 executes processing in the flow shown in FIG. 5.

[0049] First, in S11, the edge device 22 executes preprocessing for calculating a determination threshold value for determining whether to collect the peak value of the first-shaped waveform generated when the vehicle passes over the bridge and determine it as an overweight vehicle. The preprocessing period for executing the preprocessing is, for example, about two months in length.

[0050] After performing the preprocessing, in S12, the edge device 22 executes the observation process. In the observation process, each time a vehicle passes over the bridge, based on the peak value of the first shape waveform generated by the vehicle passing over the bridge, the edge device 22 determines whether the passed vehicle is an overweight vehicle. And if the edge device 22 determines that the passed vehicle is an overweight vehicle, it notifies the passed vehicle of warning information. The observation period for executing the observation process is longer than the preprocessing period, for example, about 1 year in length.

[0051] Subsequently, in S13, the edge device 22 determines whether the observation period has ended. If the observation period has not ended (No in S13), the edge device 22 continues to execute the observation process. If the observation period has ended (Yes in S13), the process returns to S11 and repeats the process from the preprocessing.

[0052] Note that the preprocessing periods after the second time may overlap with the immediately preceding observation period. For example, if the preprocessing period is 2 months and the observation period is 1 year, the preprocessing periods after the second time may overlap with the last 2 months in the immediately preceding observation period. Thereby, the edge device 22 can update the determination threshold value for each observation period and can continuously execute the observation process without a blank period.

[0053] FIG. 6 is a flowchart showing the flow of the preprocessing by the edge device 22. In the preprocessing of S11 shown in FIG. 5, the edge device 22 executes the process in the flow shown in FIG. 6.

[0054] First, in S21, the edge device 22 executes a range determination process. The range determination period for executing the range determination process is, for example, about two weeks long. During the range determination period, the edge device 22 detects peak values every time a vehicle passes over the bridge. The edge device 22 identifies the maximum peak value, which is the largest among the peak values detected during the range determination period. Then, based on the maximum peak value during the range determination period, the edge device 22 determines a plurality of divided ranges obtained by dividing the range in which the displacement amount can be taken.

[0055] Subsequently, in S22, the edge device 22 executes a frequency distribution generation process. The learning period for executing the frequency distribution generation process is, for example, about one to two months long. During the learning period, the edge device 22 detects peak values every time a vehicle passes over the bridge. During the learning period, the edge device 22 calculates a frequency distribution representing the number of samples of peak values included in each of the plurality of divided ranges. For example, every time a peak value is detected during the learning period, the edge device 22 identifies which divided range among the plurality of divided ranges the detected peak value is included in. Then, during the learning period, the edge device 22 calculates the number of samples for each of the plurality of divided ranges by counting the number of peak values included in each of the plurality of divided ranges.

[0056] Subsequently, in S23, the edge device 22 calculates a determination threshold value for determining whether to determine an overloaded vehicle based on the calculated frequency distribution. The threshold value calculation unit 40 calculates, as the determination threshold value, the displacement amount at the boundary between a decreasing region where the number of samples decreases in the frequency distribution and a non-changing region where the displacement amount is larger than that of the decreasing region and the change in the number of samples is equal to or less than a predetermined value.

[0057] Subsequently, in S24, the edge device 22 writes the calculated determination threshold value into the threshold value storage unit 42.

[0058] FIG. 7 is a flowchart showing the flow of the range determination process. The distribution calculation unit 36 of the edge device 22 executes the process in the flow shown in FIG. 7 in the range determination process of S21 shown in FIG. 6.

[0059] First, in S31, the distribution calculation unit 36 initializes the maximum peak value by substituting 0 for the maximum peak value.

[0060] Subsequently, in S32, the distribution calculation unit 36 determines whether the range determination period has ended. If the range determination period has not ended (No in S32), the distribution calculation unit 36 advances the process to S33.

[0061] In S33, the distribution calculation unit 36 determines whether a peak value has been detected from the time-series data representing the time change of the displacement amount of the bridge. If no peak value is detected (No in S33), the distribution calculation unit 36 returns the process to S32 and repeats the processes of S32 and S33 until the range determination period ends or a peak value is detected. If a peak value is detected (Yes in S33), the distribution calculation unit 36 advances the process to S34.

[0062] In S34, the distribution calculation unit 36 determines whether the detected peak value is greater than the current maximum peak value. If the detected peak value is not greater than the current maximum peak value (No in S34), the distribution calculation unit 36 returns the process to S32 and repeats the processes of S32 and S33 until the range determination period ends or a peak value is detected. If the detected peak value is greater than the current maximum peak value (Yes in S34), the distribution calculation unit 36 advances the process to S35.

[0063] In S35, the distribution calculation unit 36 updates the maximum peak value by substituting the detected peak value for the maximum peak value. When the process of S35 ends, the distribution calculation unit 36 returns the process to S32 and repeats the processes of S32 and S33 until the range determination period ends or a peak value is detected.

[0064] When the range determination period ends (Yes in S32), the distribution calculation unit 36 advances the process to S36.

[0065] In S36, the distribution calculation unit 36 determines a plurality of divided ranges obtained by dividing the range in which the displacement amount can take based on the maximum peak value. When the process of S36 ends, the distribution calculation unit 36 ends the range determination process and advances the process to S22 shown in FIG. 6.

[0066] FIG. 8 is a diagram showing an example of the maximum peak value for each day during the range determination period. As a result of executing the range determination process, at the time when the range determination period ends, the maximum peak value is the maximum value among the peak values detected during the range determination period. For example, in the example of FIG. 8, the maximum peak value is 6936.

[0067] When the distribution calculation unit 36 determines a plurality of divided ranges based on the maximum peak value, the actually obtained maximum peak value may be rounded up or down to a predetermined number of digits. In the example of FIG. 8, the distribution calculation unit 36 rounds up the lower two digits of the detected maximum peak value and sets the maximum peak value to 7000.

[0068] The distribution calculation unit 36 divides the maximum peak value rounded up or down to a predetermined number of digits by N, which is a preset number of divisions, to determine each of the N divided ranges. In the example of FIG. 8, N is set to 100. Therefore, the distribution calculation unit 36 divides the maximum peak value rounded up or down to a predetermined number of digits by 100 to determine each of the N divided ranges.

[0069] In the example of FIG. 8, the first divided range (♯1) where the displacement amount is the smallest is 0 or more and less than 70. The second divided range (♯2) is 70 or more and less than 140. The third divided range (♯3) is 140 or more and less than 210. Also, the 99th divided range (♯99) is 6860 or more and less than 6930.

[0070] Note that the upper limit value of the largest range among the N divided ranges is the maximum value of the numerical values that the displacement amount can take. In the example of FIG. 8, the maximum value of the numerical values that the displacement amount can take is 9999. Therefore, the 100th divided range (♯100) with the largest displacement amount is 6930 or more and 9999 or less.

[0071] FIG. 9 is a flowchart showing the flow of the frequency distribution generation process by the distribution calculation unit 36. The distribution calculation unit 36 of the edge device 22 executes the process in the flow shown in FIG. 9 in the frequency distribution generation process of S22 shown in FIG. 6.

[0072] First, in S41, the distribution calculation unit 36 initializes the distribution frequency by substituting 0 for the number of samples in each of the plurality of divided ranges.

[0073] Subsequently, in S42, the distribution calculation unit 36 determines whether the learning period has ended. If the learning period has not ended (No in S42), the distribution calculation unit 36 advances the process to S43.

[0074] In S43, the distribution calculation unit 36 determines whether a peak value has been detected. If a peak value is not detected (No in S43), the distribution calculation unit 36 returns the process to S42 and repeats the processes of S42 and S43 until the learning period ends or a peak value is detected. If a peak value is detected (Yes in S43), the distribution calculation unit 36 advances the process to S44.

[0075] In S44, the distribution calculation unit 36 identifies which of the plurality of divided ranges the detected peak value is included in.

[0076] Subsequently, in S45, the distribution calculation unit 36 increases the number of samples in the identified divided range in the frequency distribution by 1. Thereby, during the learning period, the distribution calculation unit 36 can calculate the number of samples for each of the plurality of divided ranges by counting the number of peak values included in each of the plurality of divided ranges.

[0077] When the process of S45 ends, the distribution calculation unit 36 returns the process to S42 and repeats the processes of S42 and S43 until the learning period ends or a peak value is detected.

[0078] When the learning period ends (Yes in S42), the distribution calculation unit 36 ends the frequency distribution generation process and advances the process to the determination threshold calculation process of S23 shown in FIG. 6.

[0079] FIG. 10 is a diagram showing an example of a frequency distribution. The frequency distribution generated by the distribution calculation unit 36 is represented by a graph with the displacement amount on the horizontal axis and the number of samples on the vertical axis, as shown in FIG. 10 for example. The majority of vehicles passing through the bridge have peak values within a certain range.

[0080] FIG. 11 is a diagram showing an example of time-series data of displacement amounts when a small vehicle, a medium-sized vehicle, and a large vehicle pass through.

[0081] As shown in FIG. 11, for the displacement amount of the bridge, the peak value of the first shape waveform becomes larger as the vehicle is heavier. For example, assume that vehicles are classified into small vehicles, medium-sized vehicles, or large vehicles according to their weight. In this case, the peak value becomes larger in the order of small vehicles, medium-sized vehicles, and large vehicles. And many overweight vehicles are large vehicles loaded with significantly more luggage than the legal loading capacity and are significantly heavier than large vehicles. Therefore, the edge device 22 can detect overweight vehicles by calculating a determination threshold larger than the peak value of the first shape waveform generated when a large vehicle passes through the bridge.

[0082] FIG. 12 is a diagram showing an example of the distribution range of normal values and the range of outliers in the frequency distribution.

[0083] Overweight vehicles passing through a bridge are generally said to account for about 0.3% of all vehicles passing through the bridge. In addition, many overweight vehicles are significantly heavier than large vehicles. Therefore, the peak value detected when an overweight vehicle passes through a bridge appears as an outlier with respect to the distribution range of the peak values caused by vehicles of normal weight. From this, the edge device 22 detects the boundary between the distribution range of the normal values of the peak values caused by vehicles of normal weight and the range in which outliers occur with respect to the distribution range of the normal values, and sets the displacement amount at the boundary as the determination threshold value, thereby detecting the passage of an overweight vehicle.

[0084] FIG. 13 is a diagram showing an example of a determination threshold value. The boundary between the distribution range of the normal values and the range of the outliers becomes the boundary between the decreasing region and the non-changing region in the frequency distribution. The decreasing region is a region of the displacement amount in the frequency distribution where the number of samples decreases when the displacement amount is increased. The non-changing region is the region on the side where the displacement amount is larger than the decreasing region, and is a region where the change amount of the number of samples is below a predetermined value even when the displacement amount is increased or decreased.

[0085] Therefore, the edge device 22 can calculate a determination threshold value for detecting the passage of an overweight vehicle by detecting the displacement amount of the boundary between the decreasing region and the non-changing region in the frequency distribution.

[0086] FIG. 14 is a diagram showing the relationship between the difference in the number of samples in adjacent divided ranges and the determination threshold value. The edge device 22 detects the boundary between the decreasing region and the non-changing region in the frequency distribution by calculating the change amount of the number of samples with respect to the displacement amount.

[0087] In the present embodiment, the edge device 22 calculates the difference in the number of samples in adjacent divided ranges as the change amount of the number of samples with respect to the displacement amount as shown in FIG. 14. Note that the edge device 22 may generate an approximate curve representing the number of samples with respect to the displacement amount or a function representing the number of samples with respect to the displacement amount, and use the slope of the generated approximate curve or the differential value of the generated function as the change amount of the number of samples with respect to the displacement amount.

[0088] Based on the change amount of the number of samples with respect to such a displacement amount, the edge device 22 detects the boundary between the decreasing region and the non-changing region in the frequency distribution, and calculates a determination threshold value based on the detected boundary.

[0089] FIG. 15 is a flowchart showing the process of calculating the determination threshold value by the threshold value calculation unit 40. The threshold value calculation unit 40 calculates the determination threshold value, for example, according to the flow shown in FIG. 15.

[0090] First, in S61, the threshold value calculation unit 40 specifies a boundary division range where the boundary between the decreasing region and the non-changing region in the frequency distribution is located among a plurality of division ranges.

[0091] Subsequently, in S62, the threshold value calculation unit 40 calculates a determination threshold value based on the range of the displacement amount in the boundary division range. For example, the threshold value calculation unit 40 calculates the central displacement amount in the boundary division range, the minimum displacement amount in the boundary division range, and the maximum displacement amount in the boundary division range as the determination threshold value. Note that the threshold value calculation unit 40 may calculate any displacement amount within the displacement amount included in the boundary division range as the determination threshold value.

[0092] Here, the threshold value calculation unit 40 may specify different division ranges as the boundary division range by a detection algorithm for detecting the division range of the position of the boundary between the decreasing region and the non-changing region. For this reason, the threshold value calculation unit 40 may specify a plurality of candidate division ranges that are candidates for the position of the boundary between the decreasing region and the non-changing region by a plurality of detection algorithms, and specify the boundary division range based on the plurality of specified candidate division ranges. For example, the threshold value calculation unit 40 may use the candidate division range with the minimum displacement amount among the plurality of candidate division ranges as the boundary division range, or may use the candidate division range with the maximum displacement amount among the plurality of candidate division ranges as the boundary division range, or may use the candidate division range with the average displacement amount among the plurality of candidate division ranges as the boundary division range.

[0093] In the present embodiment, the threshold value calculation unit 40 specifies the boundary division range by using at least one of a first detection algorithm, a second detection algorithm, and a third detection algorithm.

[0094] In the first detection algorithm, the threshold calculation unit 40 sequentially searches the plurality of divided ranges one by one from the divided range with the larger displacement amount, and specifies the first divided range detected first as the boundary divided range.

[0095] The first divided range is a divided range in which the amount of change in the number of samples is equal to or less than a predetermined value, and all of a predetermined number of consecutive divided ranges adjacent to the side with the larger displacement amount are equal to or less than a predetermined ratio of the number of samples of itself. For example, the first divided range is a divided range in which the difference in the number of samples between the adjacent divided range on the side with the larger displacement amount is 1 / 2 or less, and the number of samples of all of a predetermined number of consecutive divided ranges on the side with the larger displacement amount is 1 / 2 or less of the number of samples of itself.

[0096] By using such a first detection algorithm, the threshold calculation unit 40 can detect the divided range at the position of the boundary between the decreasing region and the non-changing region.

[0097] Also, in the second detection algorithm, the threshold calculation unit 40 sequentially searches the plurality of divided ranges one by one from the divided range with the smaller displacement amount, and specifies the first detected second divided range as the boundary divided range.

[0098] The second divided range is the divided range with the largest displacement amount in a region including a predetermined number of divided ranges with the same amount of change in the number of samples. For example, the second divided range is the divided range with the largest displacement amount in a region including a predetermined number of divided ranges with the same difference in the number of samples between the adjacent divided range on the side with the smaller displacement amount.

[0099] By using such a second detection algorithm, the threshold calculation unit 40 can detect the divided range at the position of the boundary between the decreasing region and the non-changing region.

[0100] Also, in the third detection algorithm, the threshold calculation unit 40 sequentially searches for a plurality of divided ranges one by one from the divided range with a smaller displacement amount, and specifies the first detected third divided range as the boundary divided range.

[0101] The third divided range is the divided range with the largest displacement amount in a region including a predetermined number of divided ranges where the change amount of the number of samples is 0. For example, the third divided range is the divided range with the largest displacement amount in a region including a predetermined number of divided ranges where the difference in the number of samples between the divided range adjacent to the side with a smaller displacement amount is 0.

[0102] By using such a third detection algorithm, the threshold calculation unit 40 can detect the divided range at the position of the boundary between the decreasing region and the non-changing region.

[0103] FIG. 16 is a flowchart showing the flow of the calculation process of the first determination threshold by the first detection algorithm. The threshold calculation unit 40 may execute the flow shown in FIG. 16 as the first detection algorithm to calculate the first determination threshold.

[0104] First, in S101, the threshold calculation unit 40 substitutes max for A, which is the first determination threshold. max is the maximum peak value.

[0105] Subsequently, in S102, the threshold calculation unit 40 substitutes N, which represents the number of a plurality of divided ranges, for n that specifies the divided range to be searched. Each of the plurality of divided ranges in the frequency distribution is assigned an identification number represented by an integer from 1 to N in order from the divided range with a smaller displacement amount. Thereby, the threshold calculation unit 40 can sequentially search for a plurality of divided ranges one by one from the divided range with a larger displacement amount.

[0106] Subsequently, in S103, the threshold calculation unit 40 determines whether n ≤ 1. If n ≤ 1 (Yes in S103), assuming that the first divided range cannot be detected even if all of the plurality of divided ranges are searched, this flow ends. In this case, the first determination threshold value becomes the maximum peak value. If n > 1 (No in S103), the threshold calculation unit 40 proceeds to S104 for processing.

[0107] Subsequently, in S104, the threshold calculation unit 40 subtracts 1 from n. Thereby, the threshold calculation unit 40 can change the divided range to be searched to the divided range with a smaller displacement amount by one.

[0108] Subsequently, in S105, the threshold calculation unit 40 obtains S(n), which is the number of samples in the divided range identified by n, and S(n + 1), which is the number of samples in the divided range identified by (n + 1), from the frequency distribution. S(n) represents the number of samples in the divided range to be searched. S(n + 1) represents the number of samples in the divided range adjacent to the divided range to be searched on the side with a larger displacement amount.

[0109] Subsequently, in S106, the threshold calculation unit 40 determines whether S(n + 1) ≤ {S(n) / 2}. Thereby, the threshold calculation unit 40 can determine whether the difference in the number of samples between the divided range to be searched and the divided range adjacent to the divided range to be searched on the side with a larger displacement amount is 1 / 2 or less. That is, the threshold calculation unit 40 can determine whether the change amount of the number of samples in the divided range to be searched is a predetermined value or less.

[0110] If S(n + 1) > {S(n) / 2} (No in S106), the threshold calculation unit 40 returns to S103 and repeats the process from S103. If S(n + 1) ≤ {S(n) / 2} (Yes in S106), the threshold calculation unit 40 proceeds to S107 for processing.

[0111] In S107, the threshold calculation unit 40 substitutes {S(n) / 2} for x representing the reference value. The reference value x represents 1 / 2 of the number of samples in the divided range to be searched. That is, the reference value x represents a predetermined ratio with respect to the number of samples in the divided range to be searched.

[0112] Subsequently, in S108, the threshold calculation unit 40 substitutes n + 1 for m that identifies the comparison divided range.

[0113] Subsequently, in S109, the threshold calculation unit 40 substitutes 0 for ct which is the condition satisfaction counter value.

[0114] Subsequently, in S110, the threshold calculation unit 40 determines whether m ≥ N. When m ≥ N (Yes in S110), the threshold calculation unit 40 returns the process to S103 and repeats the process from S103. When m < N (No in S110), the threshold calculation unit 40 proceeds to S111.

[0115] Subsequently, in S111, the threshold calculation unit 40 adds 1 to m.

[0116] Subsequently, in S112, the threshold calculation unit 40 obtains S(m), which is the number of samples in the divided range identified by m, from the frequency distribution. S(m) represents the number of samples in the comparison divided range.

[0117] Subsequently, in S113, the threshold calculation unit 40 determines whether S(m) ≤ x. Thereby, the threshold calculation unit 40 can determine whether the number of samples in the comparison divided range is 1 / 2 or less of the number of samples in the divided range to be searched. That is, the threshold calculation unit 40 can determine whether the number of samples in the comparison divided range is a predetermined ratio or less of the number of samples in the divided range to be searched.

[0118] When S(m) > x (No in S113), the threshold calculation unit 40 returns the process to S103 and repeats the process from S103. When S(m) ≤ x (Yes in S113), the threshold calculation unit 40 proceeds to S114.

[0119] In S114, the threshold calculation unit 40 adds 1 to ct.

[0120] Subsequently, in S115, the threshold calculation unit 40 determines whether ct ≥ a. Here, a represents a preset number. Thereby, the threshold calculation unit 40 can determine whether the total number of samples in all the divided ranges of the predetermined number that are continuous on the side with the larger displacement amount is less than or equal to 1 / 2 of the number of samples in the divided range to be searched. That is, the threshold calculation unit 40 can determine whether the total number of samples in all the divided ranges of the predetermined number that are continuous on the side with the larger displacement amount is less than or equal to a predetermined ratio of the number of samples in the divided range to be searched.

[0121] When ct ≥ a is not satisfied (No in S115), the threshold calculation unit 40 returns the process to S110 and repeats the process from S110. When ct ≥ a is satisfied (Yes in S115), the threshold calculation unit 40 proceeds to S116.

[0122] By determining Yes in S115, the threshold calculation unit 40 can specify that the divided range to be searched identified by n is the first divided range that is first detected by sequentially searching one by one from the divided range with the larger displacement amount among the plurality of divided ranges. That is, by determining Yes in S115, the threshold calculation unit 40 can specify that the divided range to be searched identified by n is the boundary divided range.

[0123] Then, in S116, the threshold calculation unit 40 substitutes the displacement amount based on the divided range identified by n into A, which is the first determination threshold. The displacement amount based on the divided range identified by n may be, for example, the minimum displacement amount, the maximum displacement amount, or the central displacement amount in the divided range identified by n. When the threshold calculation unit 40 finishes the process of S116, it ends this flow.

[0124] By executing such processing, the threshold calculation unit 40 can calculate the first determination threshold using the first detection algorithm.

[0125] FIG. 17 is a flowchart showing the flow of the calculation process of the second determination threshold by the second detection algorithm. The threshold calculation unit 40 may execute the flow shown in FIG. 17 as the second detection algorithm to calculate the second determination threshold.

[0126] First, in S121, the threshold calculation unit 40 substitutes max into B, which is the second determination threshold. max is the maximum peak value.

[0127] Subsequently, in S122, the threshold calculation unit 40 substitutes 1, which identifies the division range with the smallest displacement amount among the plurality of division ranges, into n, which specifies the division range to be searched. Thereby, the threshold calculation unit 40 can sequentially search the plurality of division ranges one by one from the division range with the smaller displacement amount.

[0128] Subsequently, in S123, the threshold calculation unit 40 substitutes 0 into ct, which is the condition satisfaction counter value.

[0129] Subsequently, in S124, the threshold calculation unit 40 substitutes 0 into last_def.

[0130] Subsequently, in S125, the threshold calculation unit 40 determines whether n≥N. If n≥N (Yes in S125), the threshold calculation unit 40 ends this flow on the assumption that the second division range could not be detected even after searching all of the plurality of division ranges. In this case, the second determination threshold is the maximum peak value. If n≥N is not satisfied (No in S125), the threshold calculation unit 40 proceeds to S126.

[0131] Subsequently, in S126, the threshold calculation unit 40 adds 1 to n. Thereby, the threshold calculation unit 40 can change the division range to be searched to one with a larger displacement amount.

[0132] Subsequently, in S127, the threshold calculation unit 40 obtains S(n - 1), which is the number of samples in the division range identified by (n - 1), and S(n), which is the number of samples in the division range identified by n, from the frequency distribution. S(n - 1) represents the number of samples in the division range adjacent to the division range of the search target on the side with a smaller displacement amount. S(n) represents the number of samples in the division range of the search target.

[0133] Subsequently, in S128, the threshold calculation unit 40 calculates def = {S(n - 1) - S(n)}.

[0134] Subsequently, in S129, the threshold calculation unit 40 determines whether def = last_def.

[0135] Here, def represents the difference in the number of samples between the division range adjacent to the division range of the search target on the side with a smaller displacement amount and the division range of the search target. That is, def represents the change amount of the number of samples in the division range of the search target. Also, last_def is the def calculated in the process of processing the division range of the search target one loop before. Therefore, in S129, the threshold calculation unit 40 can determine whether the difference in the number of samples is the same between the division range of the search target and the division range adjacent to the division range of the search target on the side with a smaller displacement amount. That is, in S129, the threshold calculation unit 40 can determine whether the change amount of the number of samples is the same between the division range of the search target and the division range adjacent to the division range of the search target on the side with a smaller displacement amount.

[0136] If def ≠ last_def (No in S129), the threshold calculation unit 40 advances the process to S132.

[0137] If def = last_def (Yes in S129), the threshold calculation unit 40 advances the process to S130.

[0138] In S130, the threshold calculation unit 40 adds 1 to ct.

[0139] Subsequently, in S131, the threshold calculation unit 40 determines whether ct ≥ b holds. Here, b represents a preset number. As a result, the threshold calculation unit 40 can determine whether the divided region to be searched is the divided range with the largest displacement amount in the region including the divided ranges with the same difference in the number of samples between the divided region to be searched and the divided range adjacent to the side with a smaller displacement amount. That is, the threshold calculation unit 40 can determine whether the divided region to be searched is the divided range with the largest displacement amount in the region including the preset number of divided ranges with the same change amount in the number of samples.

[0140] If the threshold calculation unit 40 determines that ct ≥ b does not hold (No in S131), the process proceeds to S132.

[0141] In S132, the threshold calculation unit 40 substitutes def for last_def. After finishing the process of S132, the threshold calculation unit 40 returns the process to S125 and repeats the process from S125.

[0142] If the threshold calculation unit 40 determines that ct ≥ b holds (Yes in S131), the process proceeds to S133.

[0143] By determining Yes in S131, the threshold calculation unit 40 can specify that the divided range to be searched identified by n is the second divided range first detected by sequentially searching the plurality of divided ranges one by one from the divided range with a smaller displacement amount. That is, by determining Yes in S131, the threshold calculation unit 40 can specify that the divided range to be searched identified by n is the boundary divided range.

[0144] Then, in S133, the threshold calculation unit 40 substitutes the displacement amount based on the divided range identified by n into B, which is the second determination threshold. After finishing the process of S133, the threshold calculation unit 40 ends this flow.

[0145] By executing such a process, the threshold calculation unit 40 can calculate the second determination threshold using the second detection algorithm.

[0146] Figure 18 is a flowchart showing the flow of the calculation process of the third determination threshold by the third detection algorithm. The threshold calculation unit 40 may execute the flow shown in Figure 18 as the third detection algorithm to calculate the third determination threshold.

[0147] First, in S141, the threshold calculation unit 40 substitutes max for C, which is the third determination threshold. max is the maximum peak value.

[0148] Subsequently, in S142, the threshold calculation unit 40 substitutes 1, which identifies the division range with the smallest displacement amount among the plurality of division ranges, for n, which specifies the division range to be searched. Thereby, the threshold calculation unit 40 can sequentially search the plurality of division ranges one by one from the division range with the smaller displacement amount.

[0149] Subsequently, in S143, the threshold calculation unit 40 substitutes 0 for ct, which is the condition satisfaction counter value.

[0150] Subsequently, in S144, the threshold calculation unit 40 determines whether n≥N. If n≥N (Yes in S144), the threshold calculation unit 40 ends this flow on the assumption that the third division range could not be detected even after searching all of the plurality of division ranges. In this case, the third determination threshold is the maximum peak value. If n≥N is not satisfied (No in S144), the threshold calculation unit 40 proceeds to S145.

[0151] Subsequently, in S145, the threshold calculation unit 40 adds 1 to n. Thereby, the threshold calculation unit 40 can change the division range to be searched to one with a larger displacement amount.

[0152] Subsequently, in S146, the threshold calculation unit 40 obtains S(n - 1), which is the number of samples in the division range identified by (n - 1), and S(n), which is the number of samples in the division range identified by n, from the frequency distribution. S(n - 1) represents the number of samples in the division range adjacent to the division range being searched for on the side with a smaller displacement amount. S(n) represents the number of samples in the division range being searched for.

[0153] Subsequently, in S147, the threshold calculation unit 40 calculates def = {S(n - 1) - S(n)}.

[0154] Subsequently, in S148, the threshold calculation unit 40 determines whether def = 0.

[0155] Here, def represents the difference in the number of samples between the division range adjacent to the division range being searched for on the side with a smaller displacement amount and the division range being searched for. That is, def represents the change amount of the number of samples in the division range being searched for. Therefore, in S148, the threshold calculation unit 40 can determine whether the difference in the number of samples between the division range being searched for and the division range adjacent to the division range being searched for on the side with a smaller displacement amount is 0. That is, in S148, the threshold calculation unit 40 can determine whether the change amount of the number of samples in the division range being searched for is 0.

[0156] When def ≠ 0 (No in S148), the threshold calculation unit 40 returns the process to S144 and repeats the process from S144.

[0157] When def = 0 (Yes in S148), the threshold calculation unit 40 advances the process to S149.

[0158] In S149, the threshold calculation unit 40 adds 1 to ct.

[0159] Subsequently, in S150, the threshold calculation unit 40 determines whether ct ≥ c. Here, c represents a preset number. As a result, the threshold calculation unit 40 can determine whether the divided region to be searched is the divided range with the largest displacement amount in the region including the divided ranges with a preset number of samples where the difference in the number of samples between the divided region to be searched and the divided range adjacent to the side with a smaller displacement amount is 0. That is, the threshold calculation unit 40 can determine whether the divided region to be searched is the divided range with the largest displacement amount in the region including the divided ranges with a preset number of samples where the change amount of the number of samples is 0.

[0160] If the threshold calculation unit 40 determines that ct < c (No in S150), the process returns to S144, and the process is repeated from S144.

[0161] If the threshold calculation unit 40 determines that ct ≥ c (Yes in S150), the process proceeds to S151.

[0162] By determining Yes in S150, the threshold calculation unit 40 can specify that the divided range to be searched identified by n is the third divided range first detected by sequentially searching the plurality of divided ranges one by one from the divided range with a smaller displacement amount. That is, by determining Yes in S150, the threshold calculation unit 40 can specify that the divided range to be searched identified by n is the boundary divided range.

[0163] Then, in S151, the threshold calculation unit 40 substitutes the displacement amount based on the divided range identified by n into C, which is the third determination threshold. When the threshold calculation unit 40 finishes the process of S151, it ends this flow.

[0164] By executing such a process, the threshold calculation unit 40 can calculate the third determination threshold using the third detection algorithm.

[0165] FIG. 19 is a diagram showing an example of a detection algorithm for generating a determination threshold based on the first determination threshold, the second determination threshold, and the third determination threshold.

[0166] The threshold calculation unit 40 may calculate a first determination threshold, a second determination threshold, and a third determination threshold according to the first detection algorithm shown in FIG. 16, the second detection algorithm shown in FIG. 17, and the third detection algorithm shown in FIG. 18. In this case, the threshold calculation unit 40 may calculate the minimum value, the maximum value, the intermediate value, or the average value of the first determination threshold, the second determination threshold, and the third determination threshold as the final determination threshold.

[0167] For example, when the threshold calculation unit 40 sets the minimum value among the first determination threshold, the second determination threshold, and the third determination threshold as the determination threshold, it may execute the process according to the flow shown in FIG. 19.

[0168] First, in S161, the threshold calculation unit 40 determines whether A, which is the first determination threshold, is less than or equal to B, which is the second determination threshold. If A≤B (Yes in S161), the threshold calculation unit 40 proceeds to S162. If A>B (No in S161), the threshold calculation unit 40 proceeds to S165.

[0169] In S162, the threshold calculation unit 40 determines whether A, which is the first determination threshold, is less than or equal to C, which is the third determination threshold. If A≤C (Yes in S162), the threshold calculation unit 40 proceeds to S163. In S163, the threshold calculation unit 40 substitutes the first determination threshold as the determination threshold on the assumption that the first determination threshold is the minimum value, and ends the process. If A>C (No in S162), the threshold calculation unit 40 proceeds to S164. In S164, the threshold calculation unit 40 substitutes the third determination threshold as the determination threshold on the assumption that the third determination threshold is the minimum value, and ends the process.

[0170] In S165, the threshold calculation unit 40 determines whether B, which is the second determination threshold, is less than or equal to C, which is the third determination threshold. When B ≤ C (Yes in S165), the threshold calculation unit 40 advances the process to S166. In S166, the threshold calculation unit 40 substitutes the second determination threshold as the determination threshold on the assumption that the second determination threshold is the minimum value, and ends the process. When B > C (No in S165), the threshold calculation unit 40 advances the process to S167. In S167, the threshold calculation unit 40 substitutes the third determination threshold as the determination threshold on the assumption that the third determination threshold is the minimum value, and ends the process.

[0171] By executing such processing, the threshold calculation unit 40 can set the minimum value among the first determination threshold, the second determination threshold, and the third determination threshold as the determination threshold.

[0172] As described above, the edge device 22 according to the present embodiment can determine whether a vehicle passing over a bridge is an overweight vehicle without calibration. Further, the edge device 22 according to the present embodiment can determine whether a vehicle passing over a bridge is an overweight vehicle by a process with a small load that can be executed by a computer with a simple configuration without using parameters set for each bridge. Therefore, the edge device 22 according to the present embodiment can be realized without restrictions such as a development environment and a development language, and thus the development cost can be reduced.

[0173] FIG. 20 is a diagram showing the frequency distributions of two different years.

[0174] The server device 26 may acquire the peak value detected by the edge device 22 via the network. Then, the server device 26 may generate a frequency distribution based on a plurality of peak values collected during a predetermined period. Then, the server device 26 may calculate the degree of deterioration of the bridge based on the frequency distribution based on the plurality of peak values collected during the first period and the frequency distribution based on the plurality of peak values collected during the second period after the first period.

[0175] Due to aging deterioration, when a vehicle of the same weight passes, the displacement amount becomes larger over time. Therefore, the server device 26 may calculate a numerical value indicating the degree of deterioration of the bridge by comparing a parameter such as the median or the mode representing the frequency distribution based on the peak values of the number collected in the first period with the same parameter representing the frequency distribution based on the peak values of the number collected in the second period.

[0176] FIG. 21 is a diagram showing an example of the hardware configuration of the edge device 22. The edge device 22 is realized by a device having a hardware configuration similar to that of a general information processing device as shown in FIG. 21 as an example. The information processing device includes a CPU (Central Processing Unit) 301, an operation device 302, a display device 303, a main storage device 305, an auxiliary storage device 306, a communication device 307, and a bus 309. Each part is connected by the bus 309.

[0177] The CPU 301 executes various processes in cooperation with various programs stored in advance in the auxiliary storage device 306 and the like with a predetermined area of the main storage device 305 as a work area, and comprehensively controls the operations of each part constituting the edge device 22. Also, the CPU 301 operates the operation device 302, the display device 303, the communication device 307, etc. in cooperation with the program.

[0178] The operation device 302 is an input device such as a touch panel, a mouse, or a keyboard. It receives the information input by the user's operation as an instruction signal and outputs the instruction signal to the CPU 301. The display device 303 displays various information based on the display signal from the CPU 301.

[0179] The main storage device 305 is a volatile storage medium such as an SDRAM (Synchronous Dynamic Random Access Memory). The main storage device 305 functions as a work area for the CPU 301.

[0180] The auxiliary storage device 306 is a rewritable recording device such as a semiconductor memory medium like a flash memory, or a magnetically or optically recordable memory medium. The auxiliary storage device 306 stores programs used for control. The communication device 307 transmits and receives data with other devices.

[0181] The program executed on the edge device 22 is stored, for example, on a computer connected to a network such as the Internet and provided by being downloaded via the network. Also, the program executed on the edge device 22 may be provided by being pre - incorporated into a portable storage medium or the like.

[0182] The program executed on the edge device 22 has a module configuration including an acquisition module, a peak detection module, a distribution calculation module, a threshold calculation module, a determination module, and a notification control module. The CPU 301 reads such a program from a storage medium or the like and loads each of the above - mentioned modules into the main storage device 305. Then, by executing such a program, the CPU 301 functions as an acquisition unit 32, a peak detection unit 34, a distribution calculation unit 36, a threshold calculation unit 40, a determination unit 44, and a notification control unit 46. Also, by executing such a program, the CPU 301 causes the main storage device 305 or the auxiliary storage device 306 to function as a threshold storage unit 42. Note that part or all of the acquisition unit 32, the peak detection unit 34, the distribution calculation unit 36, the threshold calculation unit 40, the determination unit 44, and the notification control unit 46 may be configured by hardware.

[0183] As described above, the embodiments of the present invention have been explained. However, these embodiments are presented as examples and are not intended to limit the scope of the invention. The embodiments can be modified in various ways.

Explanation of Reference Numerals

[0184] 10 Determination system, 20 Sensor device, 22 Edge device, 24 Warning output device, 26 Server device, 32 Acquisition unit, 34 Peak detection unit, 36 Distribution calculation unit, 40 Threshold calculation unit, 42 Threshold storage unit, 44 Determination unit, 46 Notification control unit

Claims

1. An acquisition unit that acquires time-series data representing the temporal change in the displacement amount at the position where the sensor device is provided on the bridge from the sensor device provided on the bridge on which the vehicle travels; A peak detection unit that detects a peak value representing the magnitude of the peak of the first-shaped waveform each time the first-shaped waveform in which the displacement amount increases and then decreases or decreases and then increases as the vehicle passes over the bridge is included in the time-series data; A distribution calculation unit that calculates a frequency distribution representing the number of samples of the peak values detected during the learning period, each included in a plurality of divided ranges obtained by dividing the range of values that the displacement amount can take; A threshold calculation unit that calculates, as a determination threshold, the displacement amount at the boundary between a decreasing region in which the number of samples decreases in the frequency distribution and a non-changing region on the side where the displacement amount is larger than the decreasing region and the change amount of the number of samples is equal to or less than a predetermined value; A determination unit that determines, as an overweight vehicle, the vehicle that generates the peak value larger than the determination threshold in the time-series data during an observation period after the learning period; A determination device comprising the above.

2. The distribution calculation unit: Determines the plurality of divided ranges based on the maximum peak value, which is the maximum among the peak values detected during a range determination period before the learning period; Each time a peak value is detected during the learning period, identifies which of the plurality of divided ranges the detected peak value is included in; Calculates the frequency distribution by counting the number of peak values included in each of the plurality of divided ranges during the learning period. The determination device according to Claim 1.

3. The learning period is longer than or the same as the range determination period. The determination device according to Claim 2.

4. The threshold calculation unit: Identifies one boundary divided range among the plurality of divided ranges; Calculates the determination threshold based on the displacement amount in the boundary divided range. The determination device according to Claim 1.

5. The threshold calculation unit sequentially searches the plurality of divided ranges one by one from the divided range with the larger displacement amount, and identifies the first divided range detected first as the boundary divided range. The first divided range is a divided range in which the change amount of the number of samples is equal to or less than a predetermined value, and all of a predetermined number of consecutive divided ranges adjacent to the side where the displacement amount is large are equal to or less than a predetermined ratio of their own number of samples. The determination device according to claim 4.

6. The first divided range is a divided range in which the difference in the number of samples between the divided range adjacent to the side where the displacement amount is large is equal to or less than 1 / 2, and the number of samples of all of a predetermined number of consecutive divided ranges on the side where the displacement amount is large is equal to or less than 1 / 2 of its own number of samples. The determination device according to claim 5.

7. The threshold calculation unit sequentially searches the plurality of divided ranges one by one from the divided range with the smaller displacement amount, and specifies the first detected second divided range as the boundary divided range. The second divided range is the divided range with the largest displacement amount in a region including a predetermined number of divided ranges with the same change amount of the number of samples. The determination device according to claim 4.

8. The second divided range is the divided range with the largest displacement amount in a region including a predetermined number of divided ranges with the same difference in the number of samples between the divided range adjacent to the side where the displacement amount is small. The determination device according to claim 7.

9. The threshold calculation unit sequentially searches the plurality of divided ranges one by one from the divided range with the smaller displacement amount, and specifies the first detected third divided range as the boundary divided range. The third divided range is the divided range with the largest displacement amount in a region including a predetermined number of divided ranges with a change amount of the number of samples of 0. The determination device according to claim 4.

10. The third divided range is the divided range with the largest displacement amount in a region including a predetermined number of divided ranges with a difference in the number of samples of 0 between the divided range adjacent to the side where the displacement amount is small. The determination device according to claim 9.

11. When it is determined that the vehicle is an overweight vehicle, a notification control unit that notifies the overweight vehicle that has passed the bridge of warning information indicating that it is an overweight vehicle The determination device according to claim 1, further comprising.

12. When it is determined that the vehicle is an overweight vehicle, the notification control unit causes a display device provided on a road through which the vehicle passes after passing the bridge and displaying information to a driver of the vehicle passing through the road to display the warning information. The determination device according to claim 11.

13. The sensor device detects the amount of expansion and contraction of the bridge in the traveling direction of the vehicle at the position where the sensor device is provided on the bridge. The determination device according to claim 1.

14. The acquisition unit acquires time series data representing a difference between an observed value detected by the sensor device and a moving average value obtained by moving-averaging the observed value using a preset time window, as the time series data representing the temporal change of the displacement amount of the bridge. The determination device according to claim 13.

15. The sensor device is provided for each lane of the bridge, The acquisition unit acquires the time series data for each lane, The determination unit determines, for each lane, a vehicle that generates a peak value greater than the determination threshold value in the time series data as the overweight vehicle. The determination device according to claim 1.

16. A sensor device provided on a bridge on which a vehicle travels, A determination device, Comprising: The determination device includes: An acquisition unit that acquires time series data representing a temporal change in the displacement amount at the position where the sensor device is provided on the bridge from the sensor device; A peak detection unit that detects a peak value representing the magnitude of a peak of a first shape waveform in which the displacement amount increases and then decreases or decreases and then increases when the vehicle passes over the bridge, every time the first shape waveform is included in the time series data; A distribution calculation unit that calculates a frequency distribution representing the number of samples of the peak values detected during a learning period, each of which is included in a plurality of divided ranges obtained by dividing the range of values that the displacement amount can take; A threshold calculation unit that calculates, as a determination threshold value, the displacement amount at the boundary between a decreasing region in which the number of samples decreases in the frequency distribution and a non-changing region on the side where the displacement amount is larger than the decreasing region and the change amount of the number of samples is equal to or less than a predetermined value; A determination unit that determines, during an observation period after the learning period, a vehicle that generates a peak value greater than the determination threshold value in the time series data as an overweight vehicle; Having A determination system.

17. An information processing device acquires time series data representing a temporal change in the displacement amount at the position where the sensor device is provided on the bridge from a sensor device provided on a bridge on which a vehicle travels. Each time the information processing apparatus detects that the time-series data includes a first-shaped waveform in which the displacement amount increases and then decreases or decreases and then increases as the vehicle passes over the bridge, the information processing apparatus detects a peak value representing the magnitude of the peak of the first-shaped waveform. The information processing apparatus calculates a frequency distribution representing the number of samples of the peak value detected during a learning period, each of which is included in a plurality of divided ranges obtained by dividing the range of values that the displacement amount can take. The information processing apparatus calculates, as a determination threshold value, the displacement amount at the boundary between a decreasing region in which the number of samples decreases in the frequency distribution and a non-changing region on the side where the displacement amount is larger than the decreasing region and the change amount of the number of samples is equal to or less than a predetermined value. A determination method in which the information processing apparatus determines, as an overweight vehicle, a vehicle that generates a peak value larger than the determination threshold value in the time-series data during an observation period after the learning period.

18. A program for causing an information processing apparatus to function as a determination apparatus, the information processing apparatus being caused to function as an acquisition unit that acquires time-series data representing a temporal change in a displacement amount at a position where a sensor device is provided on a bridge on which a vehicle travels, from the sensor device provided on the bridge; function as a peak detection unit that detects, each time the time-series data includes a first-shaped waveform in which the displacement amount increases and then decreases or decreases and then increases as the vehicle passes over the bridge, a peak value representing the magnitude of the peak of the first-shaped waveform; function as a distribution calculation unit that calculates a frequency distribution representing the number of samples of the peak value detected during a learning period, each of which is included in a plurality of divided ranges obtained by dividing the range of values that the displacement amount can take; function as a threshold calculation unit that calculates, as a determination threshold value, the displacement amount at the boundary between a decreasing region in which the number of samples decreases in the frequency distribution and a non-changing region on the side where the displacement amount is larger than the decreasing region and the change amount of the number of samples is equal to or less than a predetermined value; and function as a determination unit that determines, as an overweight vehicle, a vehicle that generates a peak value larger than the determination threshold value in the time-series data during an observation period after the learning period. A program for causing the above functions.

Images