JP7231452B2 - Sediment disaster prediction device - Google Patents

Description

The present invention relates to a sediment disaster prediction device.

Various systems have been proposed that detect a plurality of events that are precursors of sediment disasters, predict the occurrence of sediment disasters based on the detection results, and issue warnings.
In Patent Document 1, the detection results of rain gauges, sediment flow gauges, and landslide gauges installed at each observation point, or the rainfall data provided from the database of the disaster countermeasures headquarters of the municipal office, etc. are aggregated in a server via the Internet. Then, a system is disclosed that transmits the aggregated data to residents' terminals, or transmits an evacuation advisory to residents' terminals if a sediment disaster is predicted.

Patent No. 3762362

However, the above-described conventional technology requires installation of a rain gauge, a sediment flow gauge, and a landslide gauge, which is large-scale and time-consuming.
In addition, since the detection results of rain gauges, landslide gauges, and landslide gauges, as well as rainfall data provided from databases, etc., are aggregated on a server via the Internet, it is difficult to use on-site where the communication environment is not well-equipped. .
The present invention has been made in view of such circumstances, and its object is to be able to accurately predict a sediment disaster and issue a warning of a sediment disaster even at a site where the installation is simple and the communication environment is not in place. An object of the present invention is to provide a sediment disaster prediction device.

In order to achieve the above object, the present invention provides an image capturing unit that captures an image of a dangerous area where a sediment disaster is expected to occur and generates image information in chronological order, and a sign of a sediment disaster based on the image information. a risk score generation unit that evaluates a plurality of events and generates a risk score that increases as the risk of a sediment-related disaster increases for each event; an alarm mode setting unit for setting an alarm mode when the alarm mode is exceeded; and an event occurrence interval detection unit for detecting an occurrence interval of the plurality of events for each event based on the image information while the alarm mode is set; and at least and a warning issuing unit that issues a warning of a sediment-related disaster when the occurrence interval of one event falls below a warning threshold value set for each event.
Further, the present invention is a threshold change method for changing the caution threshold value to a lower value than before when the total score of the risk score exceeds the caution threshold value which is lower than the caution threshold value. It is characterized by further comprising a part.
Further, the present invention further comprises a rainfall score generator that detects rainfall based on the image information and generates a rainfall score that increases as the rainfall increases, and the warning mode setting unit detects the risk score. A value obtained by adding the rainfall score to the total score of is handled as the total score of the risk score.
In addition, the present invention provides a method when all of the events or occurrence intervals of the selected specific events during the setting of the alert mode exceed an alert cancellation threshold which is larger than the alarm threshold. It is characterized by further comprising a warning mode canceling unit that cancels the warning mode and returns the warning threshold value to an initial value.
Also, the present invention is characterized in that the event includes any one of falling rocks, outflow of earth and sand, collapse of the slope, cracks in the slope, and spring water on the slope.
Further, the present invention is characterized in that the risk score generation unit generates the risk score using a landslide disaster sign database that associates the image information, the plurality of events, and the risk score. do.

According to the present invention, a plurality of events that are signs of a sediment-related disaster are evaluated based on time-series image information obtained by imaging a dangerous area where a sediment-related disaster is expected to occur, and the risk of a sediment-related disaster is evaluated for each event. generates a risk score with a higher value as the value increases, sets an alert mode when the total score of the risk score exceeds the alert threshold, and during the alert mode setting, selects multiple images based on image information. The occurrence interval of events is detected for each event, and when the occurrence interval of at least one event falls below the threshold value set for each event, a sediment disaster warning is issued.
Therefore, if the imaging unit can image a dangerous area, the sediment disaster prediction device can be easily installed without performing installation work of various instruments as in the conventional art.
In addition, since the landslide disaster prediction device requires minimal equipment, it is advantageous in shortening the work time required for installation and removal from the site.
In addition, it is possible to accurately predict landslide disasters and issue landslide warnings even at sites where the communication environment such as the Internet is not ready as in the past.
Also, if the total risk score exceeds a caution threshold that is lower than the caution threshold, changing the caution threshold to a lower value will make it easier to set the caution mode. , it is advantageous for early prediction of sediment disasters and early issuing of sediment disaster warnings.
In addition, if a rainfall score is generated based on image information, and the value obtained by adding the rainfall score to the total risk score is treated as the total risk score, it is possible to obtain a risk score that takes rainfall into consideration. This is more advantageous in accurately predicting sediment-related disasters and issuing sediment-related disaster warnings.
In addition, when the occurrence interval of all events or selected specific events during setting of the alert mode exceeds the alert release threshold which is larger than the alarm threshold, the alert mode is canceled and the alert threshold is set. If the value is returned to the initial value, the warning mode will be canceled when the probability of occurrence of a sediment disaster is low, so it will be more advantageous for accurately predicting sediment disasters and issuing warnings for sediment disasters. .
In addition, by using small rockfalls, sediment runoff, slope failures, slope cracks, and slope springs as precursors of sediment-related disasters, it is possible to accurately predict sediment-related disasters and issue warnings. It will be more advantageous for reporting.
Also, if a risk score is generated using a sediment disaster prediction database that associates image information, a plurality of events, and a risk score, a sediment disaster can be predicted accurately, and a sediment disaster warning can be issued accurately. It will be more advantageous for reporting.

BRIEF DESCRIPTION OF THE DRAWINGS It is a block diagram which shows the structure of the sediment disaster prediction apparatus which concerns on embodiment. 3 is a block diagram of a computer that constitutes the control device; FIG. (A) is an explanatory diagram of risk scores, and (B) is an explanatory diagram of rainfall scores. FIG. 10 is an explanatory diagram of an event occurrence interval ΔT; It is an operation|movement flowchart of the sediment disaster prediction apparatus which concerns on embodiment.

Next, embodiments of the present invention will be described with reference to the drawings.
As shown in FIG. 1 , the landslide disaster prediction device 10 includes an imaging unit 12 , a control device 14 and an alarm device 16 .
The image capturing unit 12 captures images of dangerous areas where landslide disasters are expected to occur, such as mountains and cliffs, and generates image information in chronological order. The image information may be image information of moving images or image information of still images captured at regular time intervals.
A conventionally known imaging device such as a CCD camera can be used as the imaging unit 12 .
Also, the number of imaging units 12 may be one or plural.
The alarm device 16 issues a warning under the control of the control device 14 when the occurrence of a sediment disaster is predicted.
As the alarm device 16, for example, a siren that generates a warning sound, a warning light that lights up (blinks), an electric bulletin board that displays a warning message warning of the occurrence of a sediment disaster, or a combination thereof can be used. Various conventionally known devices can be used.

The control device 14 comprises a computer 18, which has a CPU 18A, a ROM 18B, a RAM 18C, a hard disk device 18D (or a RAM disk device), a mouse 18E, a keyboard 18F, a display device 18G, an input/output interface 18H, and the like. ing.
The ROM 18B is composed of, for example, a flash memory or the like, and stores control programs and the like, and the RAM 18C provides a working area.
The hard disk device 18D constitutes a storage section for storing various information.
The mouse 18E and keyboard 18F are for receiving operations by the operator.
The display device 18G displays various information.
The input/output interface 18H is connected to the alarm device 16, and supplies the alarm device 16 with a control signal instructing the alarm to be issued.
By executing the control program by the CPU 18A, the risk score generation unit 20, the rainfall score generation unit 22, the warning mode setting unit 24, the threshold value change unit 26, the event occurrence interval detection unit 28, and the warning issuing unit A unit 30 and an alert mode cancellation unit 32 are implemented.

The risk score generating unit 20 evaluates a plurality of events that are signs of a landslide disaster based on the image information of the dangerous area generated by the imaging unit 12, and for each event, the higher the risk of a landslide disaster, the higher the value. It produces an increasing risk score.
In the present embodiment, a case will be described in which events that are precursors of landslide disasters are falling rocks, outflow of earth and sand, slope collapses, cracks in slopes, and spring water on slopes. , and are not limited to these, and may be various conventionally known events.
As shown in FIG. 3A, a risk score is generated corresponding to the magnitude of risk for each event based on the image information.
Specifically, image information that is generated in chronological order, for example, when each frame of a still image that is the basis of a moving image is regarded as an image frame, the image frames that precede and follow in time are compared to create an image. , and based on the detection results, occurrence of events such as falling rocks, sediment runoff, slope failures, slope cracks, and slope springs, as well as the magnitude of the event (degree of risk) ), etc., and evaluate the events to generate a hazard score for each event.
In the present embodiment, if the event is a small rockfall, 0 points are given for no rockfall detection, 1 point for a small number of rockfalls, 2 points for medium-scale rockfalls, and 3 points for large-scale rockfalls. Generate a hazard score as
Similarly, a risk score between 0 and 3 is generated for each of sediment runoff, slope failure, slope cracks, and slope spring water.
Note that the risk score is not limited to four stages from 0 to 3, and may be, for example, two stages, three stages, or five or more stages.
Moreover, it is not necessary to set the risk score at the same level for all of a plurality of events, and it is optional to set different levels for each event.
The event evaluation and the risk score generation based on the image information by the risk score generation unit 20 as described above are based on the sediment disaster associated with the image information, the plurality of events, and the risk score stored in the hard disk device 18D. This is done using the prediction database (earth-slide prediction data set). In addition, the risk score generation unit 20 may use artificial intelligence and a landslide disaster sign database to evaluate events and generate risk scores based on image information.

The rainfall score generation unit 22 detects the amount of rainfall based on the image information of the dangerous area generated by the imaging unit 12, and generates a rainfall score having a higher value as the amount of rainfall increases.
Although rainfall is not a precursor of sediment-related disasters, it is an event that greatly affects the occurrence of sediment-related disasters.
In the present embodiment, as shown in FIG. 3B, a rainfall score is generated according to the amount of rainfall, with 0 points for no rainfall, 1 point for light rain, and 2 points for heavy rain.
Note that the rainfall score is not limited to three levels from 0 to 2, and may be, for example, two levels of rainfall and no rainfall, or four or more levels.
Also, the rainfall score does not necessarily have to be a positive value. For example, the rainfall score may be generated with -2 points for no rainfall, 2 points for light rain, and 4 points for heavy rain.
In addition, as a method of detecting the amount of rainfall and generating a rainfall score based only on image information, for example, reference image data of a plurality of stages such as no rain, light rain, and heavy rain are prepared in advance, and images of dangerous areas are captured. Various conventionally known methods such as a method of comparing information with a plurality of reference image data and calculating a rainfall score based on the comparison result can be employed.

The alert mode setting unit 24 sets the alert mode when the total risk score exceeds the alert threshold value. The value obtained by adding the rainfall score to the point is treated as the total hazard score.

The threshold change unit 26 changes the warning threshold to a lower value than before when the total score of the danger score exceeds the warning threshold which is lower than the warning threshold.
For example, if the initial warning threshold is set to 10 points and the caution threshold is set to 7 points, the total risk score (including the rainfall score) is 8 points and the caution threshold is 7 points. , the threshold changing unit 26 lowers the warning threshold from 10 points to, for example, 9 points.
In this way, when the total risk score exceeds the caution threshold value, the caution mode setting unit 24 can easily enter the caution mode.

The event occurrence interval detection section 28 detects the occurrence intervals of a plurality of events for each event based on the image information generated by the imaging section 12 while the alert mode is set.
Specifically, the event occurrence interval detection unit 28 records the occurrence time of each event in a storage unit such as the hard disk 18D, and detects the time between the occurrence times as the occurrence interval.
For example, as shown in FIG. 3, it is assumed that an event of falling rocks occurs at times T1, T2, and T3. In this case, the time between time points T1 and T2 and the time time between time points T2 and T3 are each set to an occurrence interval ΔT.
The longer the occurrence interval ΔT of events that are precursors of landslide disasters, the lower the probability that landslide disasters will actually occur. known to do.

The warning issuing unit 30 issues a warning of a landslide disaster when the occurrence interval of at least one event among a plurality of events falls below a threshold value set for each event. The threshold may be different or the same for each event.
Specifically, the warning issuing unit 30 gives a warning signal for issuing a warning to the reporting device 16, whereby the reporting device 16 operates to issue a warning.
For example, a siren is used to generate a warning sound, a warning light is turned on (blinks), and an electric bulletin board is used to display a warning message warning of the occurrence of a sediment disaster.
Note that the alarm threshold may be different for each event, or may be the same.

An alert mode canceling unit 32 cancels the alert mode when the occurrence interval ΔT of all events or a selected specific event during the alert mode setting exceeds an alert cancellation threshold that is larger than the alarm threshold. At the same time, the warning threshold value is returned to the initial value.
For example, if the probability of occurrence of a sediment-related disaster decreases as the weather improves and the amount of rainfall becomes zero or decreases with the passage of time, the occurrence interval ΔT of events that are signs of a sediment-related disaster becomes longer.
If the occurrence probability of sediment disasters is reduced in this way, the need to issue a warning is reduced, and therefore it is necessary to cancel the warning mode.
Therefore, the warning mode cancellation unit 32 cancels the warning mode when the occurrence interval ΔT exceeds the warning cancellation threshold value.
The warning threshold value, caution threshold value, warning cancellation threshold value, and warning cancellation threshold value described above may be set with reference to past landslide disaster data or simulation results of landslide disasters.

Next, the operation of the landslide disaster prediction device 10 of this embodiment will be described with reference to the flowchart of FIG.
It is assumed that the imaging unit 12 is installed in advance at a site where it is possible to capture an image of a dangerous area such as a mountain or a cliff where a sediment-related disaster is expected.
When the sediment disaster prediction device 10 is activated, image information is generated in chronological order from the imaging unit 12 (step S10).
Next, the risk score generator 20 generates a risk score based on the image information of the dangerous area (step S12).
A rainfall score is generated by the rainfall score generator 22 based on the image information of the dangerous area (step S14).

Next, the warning mode setting unit 24 determines whether or not the total score of the risk score (including the rainfall score) exceeds the warning threshold value (step S16).
If step S16 is affirmative, the caution mode is set by the caution mode setting unit 24 (step S18).
If step S16 is negative, the threshold changing unit 26 determines whether or not the total score of the risk score exceeds a caution threshold lower than the caution threshold (step S20).
If step S20 is negative, the process returns to step S10.
If step S20 is affirmative, the threshold changing unit 26 changes the warning threshold to a value lower than before (step S22), and the process returns to step S10.

Further, even after the alert mode is set in step S18, image information is continuously generated from the imaging unit 12 in chronological order (step S24).
The event occurrence interval detection unit 28 detects occurrence intervals of a plurality of events for each event based on the image information generated by the imaging unit 12 (step S26).
Next, the warning issuing unit 30 determines whether or not the occurrence interval of at least one of the plurality of events has fallen below the threshold value for issuing a warning set for each event (step S28).
If step S28 is affirmative, the warning issuing unit 30 operates the warning device 16 to issue a warning (step S30).
If step S28 is negative, the warning mode canceling unit 32 determines whether or not the occurrence interval ΔT of all the events or the selected specific event exceeds the warning canceling threshold which is larger than the alerting threshold. (Step S32).
If step S32 is affirmative, the warning mode is canceled by the warning mode cancellation unit 32 and the warning threshold value of the warning mode setting unit 24 is returned to the initial value (step S34), and the process returns to step S10.
If step S32 is negative, the process returns to step S24 and similar processing is performed.

As described above, according to the present embodiment, the imaging unit 12 images a dangerous area where a sediment-related disaster is expected to occur, generates image information in chronological order, and predicts a sediment-related disaster based on the image information. Multiple events are evaluated to generate a risk score for each event that increases with the risk of a sediment-related disaster. and detecting, for each event, a plurality of event intervals based on the image information during the alert mode setting, and at least one event interval falling below the alarm threshold set for each event A sediment-related disaster warning is issued in such cases.
Therefore, at a site where the imaging unit 12 can image a dangerous area, the sediment disaster prediction device 10 can be easily installed without performing installation work of various instruments as in the conventional art.
In addition, since the imaging unit 12, the control device 14, and the alarm device 16 are sufficient, it is advantageous in shortening the work time required for installation and withdrawal from the site.
In addition, it is possible to accurately predict landslide disasters and issue landslide warnings even at sites where the communication environment such as the Internet is not ready as in the past.

Further, in the present embodiment, when the total score of risk score exceeds the caution threshold which is lower than the caution threshold, the caution threshold is changed to a lower value than before. Once the risk score exceeds the caution threshold, the warning mode is likely to be set, which is advantageous in early prediction of landslide disasters and early issuance of landslide warnings.

In addition, in the present embodiment, a rainfall score is generated based on image information, and a value obtained by adding the rainfall score to the total risk score is treated as the total risk score. It is possible to obtain an accurate risk score, which is more advantageous in accurately predicting sediment disasters and issuing sediment disaster warnings accurately.

In addition, in the present embodiment, the warning mode is canceled when the occurrence interval of all events or selected specific events during setting of the warning mode exceeds the warning cancellation threshold which is larger than the alarm threshold. In addition, the warning threshold was returned to the initial value.
Therefore, since the alert mode is canceled when the sediment disaster occurrence probability is low, it is more advantageous in accurately predicting the sediment disaster and issuing the sediment disaster warning accurately.

In addition, in the present embodiment, falling rocks, sediment outflow, slope failure, slope cracks, and slope springs are used as events that are precursors of landslide disasters. This is more advantageous in accurately issuing a sediment-related disaster warning.

In addition, in the present embodiment, the risk score is generated using the landslide disaster prediction database that associates image information, a plurality of events, and the risk score. This is more advantageous in accurately issuing a sediment-related disaster warning.

In the present embodiment, the landslide disaster prediction device 10 is configured to include the alarm device 16, but sirens, warning lights, electronic bulletin boards, etc. are provided as existing equipment at the site. In that case, the alarm may be issued using those facilities, and in that case, the alarm device 16 can be omitted from the landslide disaster prediction device 10 .

10 Sediment disaster prediction device 12 Imaging unit 14 Control device 16 Alarm device 20 Risk score generation unit 22 Rainfall score generation unit 24 Alert mode setting unit 26 Threshold value change unit 28 Event occurrence interval detection unit 30 Warning issuing unit 32 Warning Mode release part

Claims (6)

an imaging unit that captures an image of a dangerous area where a sediment-related disaster is expected to occur and generates image information in chronological order;
a risk score generation unit that evaluates a plurality of events that are predictors of a sediment disaster based on the image information and generates a risk score that increases as the risk of a sediment disaster increases for each of the events;
a warning mode setting unit that sets a warning mode when the sum of the risk scores exceeds a warning threshold;
an event occurrence interval detection unit that detects occurrence intervals of the plurality of events for each event based on the image information during setting of the alert mode;
a warning issuing unit that issues a sediment-related disaster warning when the occurrence interval of at least one of the events falls below a warning threshold set for each of the events;
A sediment disaster prediction device comprising: Further comprising a threshold change unit that changes the caution threshold to a lower value when the sum of the risk scores exceeds a caution threshold that is lower than the caution threshold;
A sediment disaster prediction device according to claim 1, characterized in that: further comprising a rainfall score generation unit that detects rainfall based on the image information and generates a rainfall score that increases in value as the rainfall increases,
The alert mode setting unit treats a value obtained by adding the rainfall score to the total risk score as the total risk score.
A sediment disaster prediction device according to claim 1 or 2, characterized in that: When the occurrence interval of all of the events or selected specific events during setting of the alert mode exceeds an alert release threshold value larger than the alarm threshold value, the alert mode is released. and an alert mode canceling unit that returns the alert threshold to an initial value,
The sediment disaster prediction device according to any one of claims 1 to 3, characterized in that: The event includes any one of pebble rockfall, sediment outflow, slope collapse, crack in the slope, and spring water on the slope.
The sediment disaster prediction device according to any one of claims 1 to 4, characterized in that: The risk score generation unit generates the risk score using a landslide disaster sign database that associates the image information, the plurality of events, and the risk score.
The sediment disaster prediction device according to any one of claims 1 to 5, characterized in that:

Images