JP2023013325A - Rock fall monitoring system and rock fall monitoring method - Google Patents

Abstract

To provide a rock fall monitoring system and a rock fall monitoring method which can contribute to improvement of management and countermeasures for rock fall.SOLUTION: A rock fall monitoring system 1 for monitoring occurrence of rock fall 8 comprises: an imaging unit 2 which images a monitoring area; a rock fall analysis unit which obtains quantitative first information about the rock fall 8 occurring in the monitoring area from a photographed image captured by the imaging unit 2 with image analysis using a frame difference method; and a rock fall prediction unit which predicts new rock fall occurring in the monitoring area from the first information and second information about an environment at a time point of capturing the photographed image. The rock fall prediction unit has a learned model that has performed machine learning so as to output third information about the new rock fall 8 by receiving input of the first information and the second information.SELECTED DRAWING: Figure 1

Description

The present invention relates to a rockfall monitoring system and a rockfall monitoring method.

There are techniques for detecting falling rocks and collapses that occur on cliffs and slopes (see Patent Documents 1 to 3, for example).
In Patent Document 1, a sensor optical fiber is laid for each rockfall monitoring target section, the sensor optical fiber for each section is connected in series or in parallel to configure an optical fiber system, and this optical fiber system is used for transmission. Techniques for detecting falling rocks from changes in light have been described (see, for example, paragraph 0013).
In Patent Document 2, falling rocks are guided to and collected in an accumulation field by a rockfall protection net stretched over a slope of a monitoring area, and changes in the volume of the accumulation field or changes in the weight of the collected rockfalls are replaced with changes in the strain of an optical fiber to emit light. A technique for detecting the presence or absence of falling rocks from changes in fiber strain is described (see, for example, paragraph 0006).
Patent Document 3 describes a rockfall detection device that is attached to a rockfall protection facility such as a rockfall protection net installed on a slope and immediately below the slope to detect a rockfall, a landslide, or the like (see claim 1, for example). . This rockfall detection device is equipped with a semiconductor sensor element that detects geomagnetism, pressure, inclination, acceleration, etc., and compares waveform information such as the amplitude and period of vibration caused by falling rocks and landslides with pre-stored information. to determine rockfalls and landslides.

Japanese Patent Application Laid-Open No. 2000-180219 Japanese Patent Application Laid-Open No. 2002-138418 JP 2011-047252 A

Although it is possible to give cautions and warnings against rockfalls that have occurred using the techniques of Patent Documents 1 to 3, it is not possible to give cautions and warnings in consideration of rockfalls that will occur in the near future.
Moreover, the techniques of Patent Documents 1 to 3 also have the problem that an optical fiber system, a falling rock protection net, and the like must be installed in advance in the range in which falling rocks are to be detected. For example, if the range where rockfall is expected is wide, it is necessary to install optical fibers, rockfall protection nets, etc. over the entire range, requiring a great deal of work in advance preparation.
From this point of view, the present invention provides a rockfall monitoring system and a rockfall monitoring method that can contribute to the improvement of rockfall management and countermeasures.

A rockfall monitoring system according to the present invention is a rockfall monitoring system that monitors the occurrence of rockfall. This rockfall monitoring system includes an imaging unit that captures an image of a monitoring area, a rockfall analysis unit, and a rockfall prediction unit.
A rockfall analysis unit obtains quantitative first information about a rockfall that has occurred in the monitoring area from the captured image captured by the imaging unit by image analysis using a frame difference method.
The falling rock prediction unit predicts a new falling rock that will occur in the monitoring area from the first information and the second information about the environment at the time the captured image is captured. This rockfall prediction unit has a learned model that has undergone machine learning so as to output third information about the new rockfall by inputting the first information and the second information.
The first information is, for example, any of the source of the rockfall, the scale of the rockfall, and the characteristics of the rockfall movement. A notification unit may be further provided for notifying the occurrence of the new falling rock.
By doing so, it is possible to issue cautions and warnings in consideration of falling rocks that will occur in the near future, thereby contributing to the improvement of management and countermeasures against falling rocks.
A falling rock detection unit may be provided for detecting falling rocks appearing in the captured image captured by the imaging unit by a frame difference method. In this case, the falling rock analysis section preferably obtains the first information from the frame difference image in which the falling rock is detected by the falling rock detection section.
In this way, rockfall can be detected and quantitative first information about rockfall can be obtained by analyzing the captured image, so the effort required for preparation in advance can be reduced, and rockfall can be monitored more easily than before.

A rockfall monitoring method according to the present invention is a rockfall monitoring method for monitoring the occurrence of rockfall. This rockfall monitoring method includes an imaging process of imaging a monitoring area, a rockfall analysis process, and a rockfall prediction process.
In the rockfall analysis step, quantitative first information about the rockfall occurring in the monitoring area is obtained from the captured image captured in the imaging step by image analysis using a frame difference method.
In the rockfall prediction step, a new rockfall that will occur in the monitoring area is predicted from the first information and the second information about the environment at the time the captured image was captured. In this rockfall prediction step, prediction is performed using a trained model that has undergone machine learning so as to output the third information about the new rockfall by inputting the first information and the second information.
By doing so, it is possible to issue cautions and warnings in consideration of falling rocks that will occur in the near future, thereby contributing to the improvement of management and countermeasures against falling rocks.

ADVANTAGE OF THE INVENTION According to this invention, it can contribute to the improvement of management and countermeasures with respect to falling rocks.

1 is an image diagram of a rockfall monitoring system according to an embodiment of the present invention; FIG. 1 is a schematic configuration diagram of a rockfall monitoring system according to an embodiment of the present invention; FIG. 1 is a block diagram of an image processing device that is part of a monitoring unit; FIG. 1 is a flow chart showing an operation image of a rockfall monitoring system according to an embodiment of the present invention; FIG. FIG. 3 is an image diagram of image analysis by the frame difference method; It is a figure for demonstrating the extraction process of the quantitative information (1st information) regarding falling rocks. It is a figure for demonstrating the estimation process of the generation source of falling rocks. It is a figure which shows an example of the observation data on which learning data are based. It is a figure which shows an example of a learning model. It is a figure which shows an example of the observation data on which learning data are based. It is a figure which shows an example of a learning model.

BEST MODE FOR CARRYING OUT THE INVENTION Hereinafter, embodiments for carrying out the present invention will be described in detail with appropriate reference to the drawings. Each drawing is merely a schematic representation to the extent that the present invention can be fully understood. Accordingly, the present invention is not limited to the illustrated examples only. In addition, in each figure, the same code|symbol is attached|subjected about a common component and a similar component, and those overlapping description is abbreviate|omitted.
<Configuration of rockfall monitoring system according to embodiment>
The configuration of a falling rock monitoring system 1 according to an embodiment will be described with reference to FIGS. 1 and 2. FIG. FIG. 1 is an image diagram of a rockfall monitoring system 1. As shown in FIG. FIG. 2 is a schematic configuration diagram of the rockfall monitoring system 1. As shown in FIG.

The rockfall monitoring system 1 has a function of detecting a rockfall from captured video (composed of continuous images) and notifying that the rockfall has been detected. The rockfall monitoring system 1 also has a function of predicting new rockfalls that will occur in the near future from the detected rockfalls.
The rockfall monitoring system 1 can be used in various situations that require rockfall monitoring, and the rockfall to be monitored is not particularly limited in terms of shape, size, color, and the like. Here, as shown in FIG. 1, it is assumed that the falling rock monitoring system 1 is used for monitoring cliffs and slopes, and at least a part (or the whole) of the cliffs and slopes is the monitoring area.
The rockfall monitoring system 1 uses AI (Artificial Intelligence) technology to learn the tendency of rockfalls from images of rockfalls, and predicts rockfalls that will occur in the near future from rockfalls that have already occurred. In learning and prediction, quantitative information about rockfalls is extracted from images of rockfalls by image analysis using the frame difference method (also called "interframe difference method"), and the quantitative information is used. In the frame difference method, a moving object is identified by generating a frame difference image between two consecutive frames and performing threshold processing. Details will be described later.

As shown in FIG. 1 , the rockfall monitoring system 1 includes an imaging unit 2 , a monitoring unit 3 and a notification unit 4 . As shown in FIG. 2 , the imaging unit 2 , the monitoring unit 3 and the reporting unit 4 are supplied with power from a power source 5 . The type and number of power supplies 5 are not particularly limited. For example, the imaging unit 2, the monitoring unit 3, and the reporting unit 4 can each be configured as an independent device.
As shown in FIG. 1, the imaging unit 2 is installed so as to be capable of imaging a place (here, a cliff face 9) where a falling rock 8 may occur. In FIG. 1 , the imaging unit 2 is installed on a bridge over a place away from the cliff face 9 . The notification unit 4 is placed in a place where a person to be alerted (for example, a person who receives a notification of the fact or prediction of the occurrence of a rockfall, such as a person passing near the cliff face 9 or a person working near the cliff face 9) can visually recognize it. Installed. In FIG. 1 , the reporting unit 4 is installed on the road leading to the cliff face 9 . The installation location of the monitoring unit 3 is not particularly limited.
The imaging unit 2 and the monitoring unit 3 are communicably connected by, for example, wireless communication, and an image of the cliff face 9 is transmitted from the imaging unit 2 to the monitoring unit 3 . The monitoring unit 3 and the reporting unit 4 are communicably connected by, for example, a communication cable. and whether or not a rockfall will occur after a predetermined time) is notified. Note that the configuration of the rockfall monitoring system 1 is not limited to that shown here, and for example, the monitoring unit 3 may have the functions of the imaging unit 2 and the reporting unit 4 (that is, the imaging unit 2 and the reporting unit 4 may be the monitoring unit 3).

The image capturing unit 2 is means for capturing an image of the monitoring area (here, the cliff face 9). The imaging unit 2 has a resolution that allows the falling rock 8 to be recognized, and is capable of continuous imaging to the extent that the movement of the falling rock 8 can be recognized. The image capturing unit 2 is, for example, a digital video camera, and captures several tens (for example, 40) images (hereinafter sometimes referred to as “frames”) per second. In this embodiment, since it is assumed that it is used for monitoring the cliff face 9 , the imaging unit 2 is installed so as to photograph the cliff face 9 . An image captured by the imaging unit 2 is sent to the monitoring unit 3 . The imaging unit 2 should preferably have a specification that can withstand long-term use outdoors.
The imaging unit 2 may have illumination means (not shown). The lighting means is a light source that emits light, such as LED (Light Emitting Diode) lighting and organic EL lighting (OLED). The lighting means is installed so as to illuminate a range (that is, a monitoring area) for monitoring the occurrence of falling rocks 8 . The number and shape of the illumination means are not particularly limited. The illumination means is a light source containing at least a wavelength component corresponding to the sensitivity (camera sensitivity) of the imaging section 2, and may be visible light, infrared light, or the like, for example. Note that the imaging unit 2 may have a zoom function.

The notification unit 4 is means for notifying danger when falling rocks 8 are detected. The notification unit 4 may notify the danger when predicting the occurrence of falling rocks 8 in the near future (for example, several minutes to several hours later). The notification unit 4 notifies, for example, that a falling rock 8 has been detected or that it will occur, to the person to be alerted via sight or hearing. The notification unit 4 may be, for example, a display board, a revolving light, or a buzzer. In addition, when work on the cliff face 9 is assumed, a display function and a buzzer for sounding a warning sound may be provided on a device (for example, a helmet) worn by the person to be warned.
The reporting unit 4 is installed at a place related to the object to be monitored (in FIG. 1, it is installed on the road passing in front of the cliff face 9). When notifying the prediction of the rockfall 8 , the notification unit 4 may issue an alarm according to the probability and scale of the rockfall 8 occurring and the time until the rockfall 8 occurs. When the notification unit 4 is installed outdoors, it is preferable that it has a specification that can withstand long-term use outdoors.

The monitoring unit 3 will be described with reference to FIG. 2 (see also FIG. 1 as needed). The monitoring unit 3 detects falling rocks 8 by performing image processing on the image captured by the imaging unit 2 (see FIG. 1). In addition, the monitoring unit 3 extracts quantitative information (first information) about falling rocks from the image captured by the imaging unit 2 by image analysis using the frame difference method, and inputs the quantitative information to the trained model. to predict the occurrence of rockfall 8 in the future.
As shown in FIG. 2, the monitoring unit 3 is composed of, for example, an image storage device 3A and an image processing device 3B. The image storage device 3A and the image processing device 3B may be PCs (Personal Computers) or servers. The monitoring unit 3 may be a single device combining the image storage device 3A and the image processing device 3B.
An image obtained by photographing the cliff face 9 is stored in the image storage device 3A. The images stored in the image storage device 3A are used for learning the learning model. The image storage device 3A may store images obtained by photographing a plurality of cliff faces 9 at different locations.
The image processing device 3B is a device that predicts whether or not new rockfalls will occur in the near future from quantitative information on rockfalls extracted by image analysis using the frame difference method. The image processing device 3B has a trained model, and predicts the occurrence of a rockfall by inputting quantitative information about falling rocks into the trained model. The image processing device 3B may have a function of learning a learning model by itself, or may acquire and use a trained model that has been trained by another device (for example, the image storage device 3A). .

The configuration of the image processing device 3B will be described with reference to FIG. FIG. 3 is a block diagram of the image processing device 3B. The image processing device 3B has, for example, a storage section 31 and a control section 32 .
The storage unit 31 stores information necessary for detecting and predicting falling rocks 8 (for example, shooting condition information 31a, detection target information 31b, falling rock movement information 31c, etc.). The shooting condition information 31a is, for example, information about the imaging unit 2 (angle of view, resolution, frame rate, etc.), information about the distance L from the cliff face 9 to the imaging unit 2, and the like. The detection target information 31b is, for example, information such as the size (minimum and maximum) of the falling rock 8 to be detected and the falling speed. The rockfall motion information 31c is information about the characteristics of the rockfall motion, and is information on the motion of the rockfall 8 for each falling pattern (falling type) such as jumping motion, sliding motion, and free-falling motion.
The control unit 32 is implemented by program execution processing by a CPU (Central Processing Unit), a dedicated circuit, or the like. When the controller 32 implements the program execution process, the storage unit 31 stores a program (falling rock monitoring program) for realizing the function of the controller 32 .

The control unit 32 mainly includes an image acquisition unit 33 , a rockfall detection unit 34 , a rockfall analysis unit 35 , a rockfall prediction unit 36 , and a rockfall notification unit 37 . It should be noted that each function of the control unit 32 shown in FIG. 3 is divided for convenience of explanation, and does not limit the present invention. Here, an outline of the processing will be described, and details of the processing will be described in the operation described later.
The image acquisition section 33 acquires an image captured by the imaging section 2 . The image acquisition unit 33 here receives an image from the imaging unit 2 via wireless communication, but communication means for acquiring the image is not particularly limited.
The falling rock detection unit 34 performs image processing on the image captured by the imaging unit 2, and detects the falling rock 8 from the image after the image processing. The method of image processing by the falling rock detection unit 34 is not particularly limited, and the falling rock 8 is detected using, for example, a frame difference method. In the frame difference method, a moving object is identified by generating a frame difference image between two consecutive frames and performing threshold processing. A falling rock detection method using the frame difference method is described, for example, in Japanese Patent Application Laid-Open No. 2019-203288.
The rockfall analysis unit 35 extracts quantitative information (first information) about the rockfall 8 by image analysis using the frame difference method. Quantitative information extracted by the rockfall analysis unit 35 is sent to the rockfall prediction unit 36 and used for learning the learning model and predicting the rockfall 8 that will occur in the near future.
The rockfall prediction unit 36 has a trained model, and inputs the quantitative information extracted by the rockfall analysis unit 35 to the trained model to predict the occurrence of future rockfalls 8 . The rockfall prediction unit 36 may use the quantitative information extracted by the rockfall analysis unit 35 to learn a learning model or re-learn a learned model.
When the rockfall notification unit 37 detects the rockfall 8 or predicts that the rockfall 8 will occur in the near future (for example, several minutes to several hours later), the rockfall notification unit 37 notifies the information to the notification unit 4 . When receiving information such as the occurrence of falling rocks 8 and its prediction, the notification unit 4 notifies it.

<Operation of Falling Rock Monitoring System According to Embodiment>
The operation of the falling rock monitoring system 1 according to the embodiment will be described with reference to FIG. 4 (see FIGS. 1 to 3 as appropriate). FIG. 4 is a flowchart showing an operation image of the rockfall monitoring system 1 according to the embodiment. As shown in FIG. 4, the operation of the rockfall monitoring system 1 mainly includes a "rockfall image recording process (S10)", an "environmental data recording process (S20)", a "rockfall analysis process S30", and a "rockfall learning process ( S40)” and a “falling rock prediction step (S50)”.
(Rockfall image recording step (S10))
An imaging unit 2 starts photographing a cliff face 9, and when a falling rock 8 occurs during photographing, stores continuous images at the occurrence of the falling rock 8 and the photographing time in an image storage device 3A. The image captured by the imaging unit 2 is sent to the image processing device 3B, and it is determined whether or not the falling rock 8 is captured in the image by the processing of the falling rock detection unit 34, and the image determined to include the falling rock 8 is an image. It may be sent to and stored in the storage device 3A.

(Environmental data recording step (S20))
An organization (for example, the Meteorological Agency) (not shown) stores environment data (second information) regarding the environment of the cliff face 9 (or its vicinity) to be monitored. Environmental data may be, for example, meteorological data relating to weather, seismic data relating to earthquakes, and the like. For example, the Japan Meteorological Agency constantly observes weather and earthquakes, and records meteorological data and earthquake data in association with the time of measurement. Meteorological data includes, for example, rainfall amount, temperature, average wind speed, maximum wind speed, etc. for each predetermined time. The earthquake data is, for example, the seismic intensity and the type of earthquake.
(Rockfall analysis step S30)
The rockfall analysis unit 35 of the image processing device 3B acquires the image stored in the image storage device 3A, and extracts quantitative information (first information) regarding the rockfall 8 by image analysis using the frame difference method. Image analysis by the frame difference method will be described with reference to FIG. FIG. 5 is an image diagram of image analysis by the frame difference method.
In the frame difference method, a frame difference image F of two successive captured images E in time series is generated for captured images E of the past N frames (N≧3, for example N=10 is preferable). , threshold processing is performed to extract the region of the moving object (here, the fallen rock 8) from each frame difference image F (see the upper and middle stages of FIG. 5). _{In FIG} . ₅ , a frame difference image F1 is generated from captured images E1 and E2 _{, a} frame difference image F2 is generated from captured images E2 and E3 _, and _a frame difference image F is generated from captured images E3 and _E4 . ₃ is generated.
Also, by synthesizing the frame difference images F, a frame difference synthesized image G is generated, and the union of the regions of the moving object is obtained (see the lower part of FIG. 5). FIG. 5 shows a case where a frame difference composite image G is generated from frame difference images F ₁ , F ₂ , F ₃ , .

The rockfall analysis unit 35 of the image processing device 3B obtains, as quantitative information (first information) about the rockfall 8, for example, (a) the source of the rockfall, (b) the scale of the rockfall, and (c) the characteristics of the rockfall movement. .
(A) Estimation of Rockfall Source Estimation of the rockfall source will be described with reference to FIGS. 6 and 7. FIG. FIG. 6 is a diagram for explaining the process of extracting quantitative information about falling rock 8. As shown in FIG. FIG. 7 is a diagram for explaining the process of estimating the source of falling rock 8. In FIG.
As shown in FIG. 6, when the frame difference images F are arranged in chronological order _, the frame difference images F are composed _only of black pixels during the period T1 before the occurrence of rockfall and the period T3 after the rockfall ends. On the other hand, during the period T ₂ during which the falling rock is occurring, the area of the falling rock 8 (marked by the symbol K) is marked with white pixels, and the frame difference images F ₁ to F _n are composed of black and white pixels. The rockfall analysis unit 35 identifies the source of the rockfall 8 using the frame difference image F (frame difference image F ₁ in FIG. 6) in which white pixels appear in time series. The rockfall analysis unit 35 may determine the position of a white pixel as the source of the rockfall 8, or divide the frame difference image F into a predetermined number of pixels as shown in FIG. A region H for distinguishing the source may be set, and it may be specified which region H the region of the fallen rock 8 (reference K) corresponds to. In FIG. 7, the identification information of the area H is written in the area H with a two-digit number.

(b) Estimating the Scale of Falling Rocks The falling rocks analysis unit 35 obtains the area of the falling rocks 8 on the frame difference image F (that is, the area of the white pixels indicated by the symbol K), thereby estimating the approximate scale of the falling rocks 8. presume. For estimating the scale of the fallen rock 8, the following formula (1) may be used, for example.
Further, the scale of the falling rock 8 may be expressed by further evaluating the value obtained by the formula (1) in stages. For example, if the scale S of the falling rock 8 is "1 square meter or more", the graded evaluation "5" corresponds, "0.8 square meters or more and less than 1 square meter" corresponds to the graded evaluation "4", and "0.6 square meters or more" , less than 0.8 square meters” is assigned a graded evaluation of “3”, “0.4 square meters or more and less than 0.6 square meters” is assigned a graded evaluation of “2”, and “0.2 square meters or more and 0.4 "Less than square meters" corresponds to a graded evaluation "1", and "0 square meters or more and less than 0.2 square meters" corresponds to a graded evaluation "0".

(c) Estimation of Rockfall Motion Features The rockfall analysis unit 35 estimates the rockfall motion features from the trajectory of the rockfall 8 represented in the frame difference composite image G shown in FIG. Features of falling rock motion are, for example, falling patterns (falling types) such as jumping motion, sliding motion, and free-falling motion. In the jumping motion, falling rock 8 collides with cliff face 9 and falls while bouncing. In the sliding motion, falling rock 8 falls while sliding on cliff face 9 . A free-fall motion is one in which the falling rock 8 falls without much contact with the cliff face 9 . The rockfall analysis unit 35 holds, for example, the characteristics of these movements in advance as information, and compares the held information with the trajectory of the rockfall 8 shown in the frame difference composite image G to estimate the characteristics of the rockfall movement.
Quantitative information about the rockfall 8 extracted by the rockfall analysis unit 35 is sent to the rockfall prediction unit 36 and used for learning the learning model and predicting the rockfall 8 that will occur in the near future.

(Falling rock learning step (S40))
The rockfall learning process consists of (A) quantitative information (first information) and environmental data (second information) about the rockfall 8, and (B) information about the occurrence of new rockfall 8 in the near future (third information). It is a step of learning the relationship between For example, the rockfall prediction unit 36 obtains quantitative information about the rockfall 8 based on at least one of the “source of the rockfall, the scale of the rockfall, and the characteristics of the rockfall movement” and the “precipitation at predetermined time intervals” acquired as environmental data. Machine learning is performed using at least one of "quantity, temperature, average wind speed, maximum wind speed, seismic intensity, earthquake type". The number and types of data used for learning are preferably selected in consideration of the geographical features of the place where the cliff face 9 is located.
The rockfall prediction unit 36 has a learning model, and includes (A) quantitative information (first information) and environmental data (second information) about the rockfall 8 and (B) new rockfall 8 in the near future. The learning model is machine-learned using learning data paired with information (third information) on the occurrence of . A learning model is a learning system in machine learning that compares the results of classification, prediction, judgment, etc. based on the given data with the actual results that are correct answers, and leads to good results by adjusting various parameters. becomes possible. The learning model is, for example, a neural network, and the present embodiment also assumes a neural network. A learning model is also called a “learning device” or “artificial intelligence (AI)”.

As is well known, a neural network is an information processing technique that was created for the purpose of making a computer imitate the movement of the human brain. Neural networks are good at complex nonlinear processing, and have the feature that they do not need to formulate the relationship between explanatory variables and target variables by using learning functions. A neural network generally consists of an input layer, an intermediate layer, and an output layer. An explanatory variable is input to the input layer, and an objective variable is output from the output layer. The middle layer plays a role of connecting the two. The number of intermediate layers and the number of neurons in each intermediate layer are not restricted and can be set arbitrarily.
The method of learning the learning model is not particularly limited, and the learning model is learned, for example, by error backpropagation. Specifically, (A) quantitative information (first information) and environmental data (second information) regarding the falling rock 8 that constitute the learning data are input to the input layer, and the results output from the output layer are combined with the learning data. The intermediate layer is adjusted based on the error with (B) information (third information) regarding the occurrence of new falling rocks 8 in the near future, which constitutes the data. In other words, the rockfall prediction unit 36 inputs (A) quantitative information (first information) and environmental data (second information) regarding the rockfall 8, and (B) predicts the occurrence of a new rockfall 8 in the near future. A learning model is machine-learned so as to output information (third information). The number of learning data to be learned is not particularly limited. For example, learning is terminated when the expected accuracy is reached.

(Rockfall prediction step (S50))
The rockfall prediction step is a step of (B) predicting the occurrence of a new rockfall 8 in the near future from (A) quantitative information (first information) on the rockfall 8 and environmental data (second information). The rockfall prediction unit 36 predicts the occurrence of a rockfall in the near future using the quantitative information and environmental data regarding the (A) rockfall 8 acquired in real time.
The rockfall prediction unit 36 has a learned learning model (learned model). The learned model is obtained by subjecting the learning model to machine learning through the falling rock learning step (S40). In other words, the learned model receives (A) quantitative information (first information) and environmental data (second information) regarding the falling rock 8, and (B) information regarding the occurrence of a new falling rock 8 in the near future. It is learned to output (third information). By inputting (A) quantitative information about the rockfall 8 and environmental data acquired in real time, the rockfall prediction unit 36 obtains (B) information about the occurrence of a new rockfall 8 that has not yet occurred (B) in the near future ( For example, "with or without falling rocks" or "time, source, scale, etc. of falling rocks") is output.

Next, construction of learning data and a learning model will be described. FIG. 8 shows an example of observation data on which learning data is based. The observation data shown in FIG. 8 were measured every hour. Each observation data consists of "date", "time", "number of rockfalls", "source of rockfall", "scale of rockfall", "precipitation [mm]", "temperature [℃]", "average wind speed [m /s]”, “maximum wind speed [m/s]”, and “earthquake (seismic intensity)”.
The number of rockfalls here is the number of rockfalls that occurred in one hour. The falling rock source is the identification information of region H in FIG. The scale of falling rocks is obtained by stepwise evaluation of the value obtained by the above formula (1). The amount of rainfall is the amount per hour (hourly rainfall), the average temperature and average wind speed are the average values for each hour, and the maximum wind speed is the instantaneous maximum wind speed in one hour. Earthquake is the seismic intensity of the earthquake that occurred within one hour.
FIG. 9 shows an example of a learning model. For example, information about falling rocks 8 at a certain first time (for example, information shifted by one hour such as 10 hours, 9 hours, 8 hours, 1 hour before the prediction time) and the first A learning model is learned using a set of information about falling rocks 8 after the passage of time (for example, information on the time of prediction). As a result, by inputting information about the fallen rock 8 and the environment into the learned model, it is possible to predict the "rockfall occurrence time", "rockfall source", and "rockfall scale" of the rockfall 8 that will occur in the near future. For learning data items, for example, "rockfall occurrence time", "rockfall source", "rockfall scale", and "precipitation" are given high priority. "Rockfall Occurrence Time", "Rockfall Occurrence Source", and "Rockfall Scale" can be estimated with high accuracy.

FIG. 10 shows an example of observation data on which learning data is based. The observation data shown in FIG. 10 were measured every hour. Each observation data consists of "date", "time", "presence or absence of rockfall", "number of rockfall", "scale of rockfall", "precipitation [mm]", "temperature [℃]", "average wind speed [m/ s]”, “maximum wind speed [m/s]”, and “earthquake (seismic intensity)”.
Here, the presence or absence of falling rocks is information as to whether or not the falling rocks 8 occurred within one hour, and the number of falling rocks is the number of falling rocks that occurred in one hour. The scale of falling rocks is obtained by stepwise evaluation of the value obtained by the above formula (1). The amount of rainfall is the amount per hour (hourly rainfall), the average temperature and average wind speed are the average values for each hour, and the maximum wind speed is the instantaneous maximum wind speed in one hour. Earthquake is the seismic intensity of the earthquake that occurred within one hour.
FIG. 11 shows an example of a learning model. For example, the input data are the number of rockfalls, scale of rockfall, rainfall, temperature, wind speed, and earthquake, and the output data is whether or not rockfall has occurred (with or without rockfall). For example, one hour is defined as "one unit", the past 10 hours are input data (see symbol "a" in FIG. 10), and the next one hour is output data (see symbol "b" in FIG. 10). For example, the scale of falling rocks 8 in the same place multiple times in one hour is a total value. This makes it possible to predict whether or not the rockfall 8 will occur in the next hour from the observation data for the past ten hours. In addition, learning may be performed so as to predict the probability of falling rocks 8 occurring one hour later. Further, observation data for several hours may be subjected to addition processing, statistical processing, or the like, and input to the learning model.

As described above, according to the rockfall monitoring system 1 according to the embodiment, the rockfall analysis unit 35 obtains quantitative first information about the rockfall 8 occurring in the monitoring area from the captured image by image analysis using the frame difference method. , the falling rock prediction unit 36 uses the first information to predict new falling rocks 8 occurring in the monitoring area. As a result, it is possible to issue cautions and warnings in consideration of falling rocks 8 that will occur in the near future, which contributes to the improvement of management and countermeasures against falling rocks 8 .
Although the embodiment of the present invention has been described above, the present invention is not limited to this, and can be implemented within the scope of the claims.

1 Falling rock monitoring system 2 Imaging unit 3 Monitoring unit 3A Image storage device 3B Image processing device 4 Reporting unit 5 Power source 8 Falling rock 9 Cliff surface (monitoring area)
33 Image acquisition unit 34 Falling rock detection unit 35 Falling rock analysis unit 36 Falling rock prediction unit 37 Falling rock notification unit

Claims (5)

A rockfall monitoring system for monitoring the occurrence of rockfall,
an imaging unit that captures an image of a monitored area;
a rockfall analysis unit that obtains quantitative first information about a rockfall that has occurred in the monitoring area from the captured image captured by the imaging unit by image analysis using a frame difference method;
a rockfall prediction unit that predicts a new rockfall that will occur in the monitoring area from the first information and second information about the environment at the time the captured image is captured;
The rockfall prediction unit has a learned model that has undergone machine learning so as to output third information about the new rockfall by inputting the first information and the second information.
A falling rock monitoring system characterized by: 2. The rockfall monitoring system according to claim 1, wherein the first information is any one of a rockfall source, a rockfall scale, and a rockfall movement feature. A falling rock detection unit that detects falling rocks appearing in the captured image captured by the imaging unit by a frame difference method,
the falling rock analysis unit obtains the first information from the frame difference image in which the falling rock is detected by the falling rock detection unit;
The falling rock monitoring system according to claim 1 or 2, characterized in that: 4. The rockfall monitoring system according to any one of claims 1 to 3, further comprising a notification unit that notifies the occurrence of the new rockfall. A rockfall monitoring method for monitoring the occurrence of rockfall,
an imaging step of imaging the monitored area;
A falling rock analysis step of obtaining quantitative first information about falling rocks occurring in the monitoring area from the captured image captured in the imaging step by image analysis using a frame difference method;
a rockfall prediction step of predicting a new rockfall that will occur in the monitoring area from the first information and second information about the environment at the time the captured image was captured;
In the rockfall prediction step, by inputting the first information and the second information, prediction is performed using a trained model that has undergone machine learning so as to output third information about the new rockfall.
A falling rock monitoring method characterized by:

Images