WO2020189679A1 - Debris flow notification system and debris flow sensor - Google Patents

Debris flow notification system and debris flow sensor Download PDF

Info

Publication number
WO2020189679A1
WO2020189679A1 PCT/JP2020/011724 JP2020011724W WO2020189679A1 WO 2020189679 A1 WO2020189679 A1 WO 2020189679A1 JP 2020011724 W JP2020011724 W JP 2020011724W WO 2020189679 A1 WO2020189679 A1 WO 2020189679A1
Authority
WO
WIPO (PCT)
Prior art keywords
debris flow
information
voice information
voice
compression
Prior art date
Application number
PCT/JP2020/011724
Other languages
French (fr)
Japanese (ja)
Inventor
小林 誠司
恭助 佐々木
Original Assignee
公立大学法人公立諏訪東京理科大学
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 公立大学法人公立諏訪東京理科大学 filed Critical 公立大学法人公立諏訪東京理科大学
Publication of WO2020189679A1 publication Critical patent/WO2020189679A1/en

Links

Images

Classifications

    • EFIXED CONSTRUCTIONS
    • E02HYDRAULIC ENGINEERING; FOUNDATIONS; SOIL SHIFTING
    • E02BHYDRAULIC ENGINEERING
    • E02B1/00Equipment or apparatus for, or methods of, general hydraulic engineering, e.g. protection of constructions against ice-strains
    • GPHYSICS
    • G01MEASURING; TESTING
    • G01DMEASURING NOT SPECIALLY ADAPTED FOR A SPECIFIC VARIABLE; ARRANGEMENTS FOR MEASURING TWO OR MORE VARIABLES NOT COVERED IN A SINGLE OTHER SUBCLASS; TARIFF METERING APPARATUS; MEASURING OR TESTING NOT OTHERWISE PROVIDED FOR
    • G01D21/00Measuring or testing not otherwise provided for
    • GPHYSICS
    • G08SIGNALLING
    • G08BSIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
    • G08B21/00Alarms responsive to a single specified undesired or abnormal condition and not otherwise provided for
    • G08B21/02Alarms for ensuring the safety of persons
    • G08B21/10Alarms for ensuring the safety of persons responsive to calamitous events, e.g. tornados or earthquakes
    • GPHYSICS
    • G10MUSICAL INSTRUMENTS; ACOUSTICS
    • G10LSPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
    • G10L19/00Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis

Definitions

  • the present invention relates to a debris flow notification system and a debris flow sensor.
  • contact type debris flow detectors using wires and non-contact type debris flow detectors using luminous flux are used. These debris flow detectors detect the occurrence of debris flow, etc. by detecting that the debris flow has cut the wire, or by detecting that the debris flow has blocked the luminous flux, and alert people living downstream. Is configured to emit.
  • these debris flow detectors may be erroneously detected by animals such as deer. Furthermore, with a debris flow detector that uses a wire, it is necessary to re-tension the wire installed in the mountains every time an animal cuts the wire, and maintenance costs are high. There is also the problem that there is not enough time to evacuate even if an alarm is issued after the debris flow occurs.
  • Non-Patent Paper 1 in Japanese sabo rivers, attempts to measure the amount of quicksand by analyzing the sound of quicksand colliding with a hydrophone (metal pipe) are being studied. Also, in Switzerland, plate-type geophones have been developed for a long time and are used in European countries.
  • the hydrophone extracts the sound of particles colliding with a metal tube installed on the riverbed in a specific frequency band, and measures the number of times the amplitude value exceeds the set amplitude threshold to estimate the amount of quicksand. is there. It is expected that the hydrophone can be used as a debris flow warning by capturing the phenomenon of gravel flowing before the debris flow occurs. In addition, there is no problem that the wire is cut by the animal, and there is an advantage that the data can be easily handled. However, if a large gravel collides with the pipe, the metal pipe will be deformed or damaged and will need to be replaced. It is also necessary to set the amplitude threshold appropriately depending on the river conditions, but there is also the problem that the correct set value cannot be known unless the actual debris flow occurs.
  • the geophone developed in Switzerland has a structure in which an iron plate is installed on the top plate of the river floor and the number of gravel is counted by measuring the vibration caused by the gravel passing over it. With this structure, the geophone will not be destroyed even if a debris flow occurs.
  • floor consolidation work is required in mountain rivers, which is difficult and extremely expensive.
  • the debris flow / mudflow generation detection device described in Patent Document 1 transmits ultrasonic waves and radio waves to the flowing body flowing down on the inclined surface, receives the reflected wave from the flowing body, and receives the spectrum of the received wave. Focusing on the behavior during distribution, the presence or absence of debris flow or mudflow is determined.
  • the debris flow detection system described in Patent Document 2 is a system in which a sound wave detector is installed on an object in contact with the water surface of a target river, the sound wave level is transmitted wirelessly, and the debris flow is determined from the received sound wave level.
  • This system has the advantage of being inexpensive to install.
  • the sound wave level increases due to a phenomenon other than debris flow (for example, when there is a lightning strike, when a helicopter flies near the detector, when a heavy machine such as a bulldozer runs, etc.)
  • an erroneous report is made. There is a problem that it ends up.
  • the present invention provides a debris flow reporting system and a debris flow sensor capable of appropriately determining and reporting the risk of debris flow by distinguishing between debris flow generated in a river in a mountainous area and noise other than debris flow.
  • the purpose is.
  • the earth and stone flow reporting system of the present invention includes a voice acquisition means for acquiring voice information, a compression means for compressing the voice information to reduce the amount of information, and a wireless transmission means for wirelessly transmitting the voice information compressed by the compression means.
  • a receiving means for receiving the voice information wirelessly transmitted by the wireless transmission means, a determination means for determining the state of the earth and stone flow using the voice information received by the receiving means, and an output of the determination means. It has a notification means for notifying the user.
  • the earth and stone flow sensor of the present invention includes a voice acquisition means for acquiring voice information, a compression means for compressing the voice information to reduce the amount of information, and a wireless transmission means for wirelessly transmitting the voice information compressed by the compression means.
  • the compression means compresses the amount of information by using the audio information according to a predetermined waveform model.
  • a microphone is installed at a position away from the target river to acquire the voice signal of the debris flowing through the river.
  • Voice information (amount of information) is compressed by detecting a signal peculiar to gravel collision from the acquired voice signal, and transmitted to a cloud computer via a receiver at the foot of the LPWA (Low Power Wide Area) radio.
  • LPWA wireless is a wireless communication technology that can be applied when the amount of information to be transmitted is small (approximately 100 bits), and is a technology that enables long-distance wireless transmission with low power consumption.
  • the cloud computer is equipped with a judgment engine, which makes judgments about debris flow using compressed audio information transmitted by LPWA and reports the degree of risk of debris flow.
  • the cloud computer accumulates compressed audio information transmitted by LPWA in a database.
  • the cloud computer updates (learns) the judgment engine by comparing it with the actual debris flow, so the performance of the judgment engine will gradually improve. As a result of repeating the learning of debris flow in this way, it is a system that correctly determines and notifies the degree of risk related to debris flow.
  • voice information in which gravel collides with gravel is detected at a place away from a river
  • voice information is compressed by using the voice waveform peculiar to gravel, and wireless to a receiver at the foot of the river.
  • the risk of debris flow can be appropriately determined and reported by distinguishing between debris flow and noise other than debris flow.
  • LPWA wireless technology
  • LPWA wireless technology
  • by updating (learning) the database of the judgment engine using the actual debris flow occurrence status as teacher data it is possible to provide a system that is far more reliable than the conventional debris flow detection system at a low cost. It becomes.
  • FIG. 1 is a diagram for explaining the overall configuration of the debris flow notification system 1.
  • the debris flow notification system 1 is a system that detects the state of debris flow that may occur in mountainous areas or the precursory phenomenon of debris flow and notifies the user terminal 8.
  • a plurality of sensor information compression transmitters 20 are installed at a location away from the river 14 in the upstream part of the river 14 in the mountainous area 13.
  • the sensor information compression transmitter 20 detects information such as voice information, odor information, and rainfall information around the installation location, compresses it into 128-bit payload data, and transmits it over a long distance to the receiving station 7 at the foot of the device by LPWA radio. ..
  • the LPWA receiver 7 installed in the urban area receives the payload data 30 (see FIG. 4 described later) transmitted from the sensor information compression transmitter 20, decodes it, and transmits it to the cloud computer 10 on the Internet.
  • the cloud computer 10 cuts out information of various sensors from the payload data 30 and stores it in the database 12.
  • the cloud computer 10 also supplies data obtained from various sensors to the judgment engine 11 (judgment means).
  • the judgment engine 11 is a recognition program executed by the cloud computer 10, and is realized by, for example, a neural network.
  • a mail is transmitted to the user terminal 8 to increase the risk of debris flow to local residents and local governments. Inform you that you are. This email contains GPS location information and information indicating the status of debris flow determined by the determination engine 11.
  • the cloud computer 10 updates (learns) the judgment engine 11. That is, the judgment engine 11 is configured to improve its judgment performance by feedback from the user. By sequentially improving (learning) the performance of the judgment engine 11 in this way, this system finally constructs a system that enables debris flow judgment with high accuracy. In the summer when heavy rains that cause debris flow occur frequently, it is desirable to update the judgment engine 11 frequently.
  • the sensor information compression transmitter 20 is installed at a location distant from the river 14 within the range where the sound can reach, and the sound waveform generated by the collision of the gravel and the gravel is extracted and compressed. By transmitting from the sensor information compression transmitter 20, the possibility that the sensor information compression transmitter 20 is broken is reduced even when a gravel flow occurs.
  • the voice waveform is compressed, low-bit and long-distance wireless technology such as LPWA can be applied, and the sensor information compression transmitter 20 can be installed in a remote mountain. It has become. Further, since the waveform in which the gravel collides with the gravel is extracted, there is a low possibility of malfunction due to noise unlike the conventional number of pulses, and a reliable judgment can be made.
  • FIG. 2 is a diagram showing a circuit configuration of the sensor information compression transmitter 20.
  • the sensor information compression transmitter 20 has a built-in battery 27 and supplies power to each part of the WakeUp circuit 26 and the sensor information compression transmitter 20.
  • the voice sensor 23 is composed of a small microphone and an AD converter, and detects voice information in which gravel collides with gravel in or near the river 11 and sends it to the CPU 24.
  • the odor sensor 22 is a so-called malodor sensing sensor that detects sulfur compound gas and the like, and AD-converts the detection output and sends it to the CPU 24.
  • the rainfall sensor 21 detects the amount of rainfall, converts it to AD, and sends it to the CPU 24.
  • the voice sensor 23 and the odor sensor 22 are useful for more accurately determining the danger of debris flow by capturing such a precursory phenomenon.
  • the GPS receiver 29 receives radio waves from a plurality of GPS satellites orbiting the earth, calculates latitude and longitude information of the place where the sensor information compression transmitter 20 is installed, and sends the radio waves to the CPU 24.
  • the wireless transmitter 25 transmits the 128-bit payload data generated by the CPU 24 as LPWA radio.
  • the CPU 24 compresses the information from the voice sensor 23 by an algorithm described later, adds GPS latitude / longitude information and an identification code (ID), and adds the information of the odor sensor 22 and the rainfall sensor 21 to obtain a 128-bit payload.
  • ID an identification code
  • the data 30 is created and sent to the wireless transmitter 25.
  • the WakeUp circuit 26 monitors the output level of the voice sensor 23, and when the level falls below a predetermined value, issues an instruction to the CPU 24 and the wireless transmitter 25 to put the sensor information compression transmitter 20 into sleep mode. Reduce power consumption.
  • FIG. 3 is a flowchart showing an operation algorithm of the CPU 24.
  • the CPU 24 executes a signal processing algorithm described later every minute.
  • the CPU 24 measures the voice level obtained from the voice sensor 23, and if the level is equal to or less than a predetermined value, the processing is interrupted to reduce the power consumption of the sensor information compression transmitter 20.
  • the signal level obtained from the voice sensor 23 exceeds a predetermined value, that is, if there is a possibility that a rustling sound due to rainfall or a sound of gravel rolling on a slope is detected, a step is taken.
  • a predetermined value that is, if there is a possibility that a rustling sound due to rainfall or a sound of gravel rolling on a slope is detected.
  • step SP4 the CPU 24 acquires the latitude and longitude information of the installation location from the GPS receiver 29. For example, information such as latitude 36.030160 degrees north and longitude 138.155298 degrees east can be obtained from the GPS receiver 29.
  • the CPU 24 compresses latitude and longitude 6-digit information (“030160” and “155298” in the above example) into 24-bit latitude information and longitude information by representing them in BCD (Binary Coded Decimal), respectively. ..
  • the digits above the decimal point of latitude and longitude (36 degrees north latitude, 138 degrees east longitude) can be restored by the receiver 7, so they may be deleted in this way.
  • the latitude and longitude information obtained in this way can be used to know the fact that the sensor information compression transmitter 20 has been "flowed" because the latitude and longitude change if it is washed away by a debris flow. , It is important information to show that the debris flow has already occurred.
  • step SP5 the data of the odor sensor 22 is processed. By converting the output value of the odor sensor 22 into 6 bits, it is added to the payload data 30 as odor level information SCENT.
  • the odor level may be a sign of debris flow.
  • wild animals such as bears also emit extremely strong odors, it is not appropriate to estimate the possibility of debris flow based only on the odor level. Therefore, in this system, the odor level is transmitted to the cloud computer 10 as one of the reference information, stored in the database 12, and combined with the information from other sensors and the past results such as whether debris flow occurred or not. Therefore, the determination engine 11 mounted on the cloud computer 10 is made to learn.
  • step SP6 the data of the rainfall sensor 21 is processed and added to the payload data 30 as 6-bit data (Prescription).
  • Precipitation is the root cause of debris flow, but whether or not debris flow occurs is also affected by soil composition and groundwater flow. Therefore, it is not appropriate to judge the occurrence of debris flow based only on the amount of rainfall. Therefore, in this system, the amount of rainfall is transmitted to the cloud computer 10 as one of the reference information, stored in the database 12, and combined with the information from other sensors and the actual information such as whether or not the debris flow occurred.
  • the determination engine 11 is configured to learn.
  • step SP7 the CPU 24 processes the information of the voice sensor 23.
  • the voice sensor 23 is composed of a small microphone and an AD converter, and detects surrounding voice information. If the conversion rate of the AD converter is 10 kHz and the number of quantization bits is 10 bits, the data will be as much as 6 megabits in 1 minute. Although this data contains various information and is useful, it cannot be transmitted by LPWA due to the large amount of data. Therefore, the CPU 24 extracts audio information useful for debris flow detection as will be described later in FIG. 5, and adds it to the payload data 30 as a 42-bit compressed audio information Sound.
  • step SP8 the 128-bit payload data 30 is configured as shown in FIG. 4, which will be described later.
  • the beginning of the payload data 30 is 16-bit ID information, and the unique number recorded in the internal non-volatile memory of the CPU 24 can be used.
  • status information (5 bits), GPS latitude information (24 bits) and longitude information (24 bits), ambient temperature information Temp (5 bits), rainfall information Precipitation (6 bits), and odor information.
  • the payload data is composed of SCENT (6 bits) and compressed audio information (42 bits).
  • FIG. 4 is a diagram showing a configuration of 128-bit payload data 30. As shown in the lower left of FIG. 4, the status information is composed of TEST (1 bit) and BAT information (4 bits) indicating the remaining battery level in 16 steps.
  • a test switch 28 is mounted on the sensor information compression transmitter 20. When confirming that the system operates correctly, the operation test of the system can be performed by pressing the test switch 20 and setting the TEST information of the payload data to "1".
  • step SP9 the payload data 30 is transmitted by LPWA radio.
  • This wireless signal is received by the LPWA receiver 7 installed in an urban area or the like and transmitted to the cloud computer 10.
  • the cloud computer 10 determines the risk of debris flow by the determination engine 11, and notifies the residents and local governments by sending an e-mail to the user terminal 8.
  • This email contains GPS location information and the state of debris flow determined by the determination engine 11.
  • the local government can make necessary decisions such as the start of evacuation preparations comprehensively by referring to not only the information sent by this e-mail but also the information such as the amount of precipitation in the entire area by the X-band radar.
  • FIG. 5 is a diagram showing a compression algorithm for voice information.
  • FIG. 6 is a diagram showing a gravel-to-grave collision waveform (A) and a modeled waveform (B).
  • step SP7 compression of information obtained from the voice sensor 23
  • the CPU 24 regards the entire waveform captured in one minute as noise and calculates the noise level LVN.
  • the noise level LVN is calculated by Equation 1.
  • LVN ⁇ (Au (n) ⁇ Au (n)) ⁇ ⁇ N ⁇ ⁇ Equation 1
  • is an operator that represents the sum of N points
  • is an operator that represents multiplication.
  • step SP21 the value of the counter CNT is reset to zero.
  • step SP22 the waveform data Au (n) (see FIG. 6A) is scanned to search for the peak of the waveform exceeding a predetermined level.
  • the peak of the waveform may be caused by the collision of gravel and gravel.
  • the processing after step SP24 is performed to extract a voice waveform due to a collision between gravel and gravel.
  • step SP24 1 is added to the counter CNT.
  • the value of the counter CNT represents a pulse in which the gravel seems to have collided with the gravel.
  • step SP25 waveform data Au (n) before and after the peak is cut out.
  • FIG. 6A shows an example of the collision waveform of the gravel and the gravel cut out in this way. It can be seen that the waveform peaks immediately after the collision and then decays rapidly, that is, it is a free vibration accompanied by damping. Such a waveform is completely different from the sound of an animal or a helicopter, and is modeled by the following equation 2.
  • R (n- ⁇ ) A ⁇ EXP (- ⁇ ⁇ n) ⁇ Sin (2 ⁇ ⁇ Fpeak ⁇ ⁇ n) ⁇ Equation 2
  • A is the amplitude of the peak
  • is the damping coefficient
  • Fpeak is the free vibration frequency
  • is the time delay.
  • step SP26 the CPU 24 calculates the spectrum by Fourier transforming the cut out voice information Au (n).
  • FIG. 7 is a diagram showing an example of the collision spectrum distribution between gravel and gravel.
  • FIG. 7 shows an example of the spectrum obtained by Fourier transforming the voice information Au (n).
  • the peak of the spectrum is observed at a frequency of 2200 Hz (Fpeak).
  • Fpeak the peak frequency
  • step SP27 the attenuation coefficient ⁇ and the time delay ⁇ are estimated. That is, the waveform R (n) is obtained by Equation 2 and the correlation coefficient with Au (n) is obtained by using ⁇ and ⁇ that are sequentially changed stepwise from a predetermined initial value. Find the combination of ⁇ and ⁇ that maximizes this correlation coefficient, and use it as the attenuation coefficient ⁇ and the time delay ⁇ . Further, as the peak amplitude A, the peak value of the cut-out voice waveform Au (n) can be used.
  • FIG. 6 (B) described above shows an example of R (n- ⁇ ) thus obtained. It can be seen that the waveform shown in FIG. 6 (B) is almost the same as the actually observed gravel-grave collision waveform (FIG. 6 (A)).
  • step SP28 the SNR is calculated according to Equations 3 and 4.
  • Err (n) Au (n) -R (n- ⁇ ) ... Equation 3
  • SNR ⁇ ⁇ R (n- ⁇ ) ⁇ R (n- ⁇ ) ⁇ ⁇ ⁇ ⁇ Err (n) ⁇ Err (n) ⁇ ⁇ ⁇ Equation 4
  • the SNR obtained in this way is a large value if it is a voice waveform due to a collision between gravel and gravel. On the contrary, if it is an animal bark, the SNR will be a small value.
  • step SP28 When step SP28 is completed, the process returns to SP22, the waveform data Au (n) is scanned again, and it is determined in step SP23 whether or not there is a waveform peak that has not yet been processed. If unprocessed peaks remain, SNR, ⁇ , and ⁇ are obtained by repeating the processes from steps SP24 to SP28.
  • step SP29 the payload data 30 is set. That is, as shown in the lower right of FIG. 4, the count CNT is stored as 10-bit data. Subsequently, SNR, peak frequency Fpeak, attenuation coefficient ⁇ , and noise level LVN are each stored as 8-bit data. When the counter CNT is 2 or more, Fpeak, ⁇ , and LVN when the SNR is maximized are stored. The accuracy of each value is adjusted so that it has a predetermined number of bits.
  • the voice information Au (n) which was 6 megabits, is compressed into 42 bits of information (Sound) and then transmitted.
  • Sound the waveform characteristics peculiar to the collision sound between gravel and gravel.
  • SNR is added as an evaluation index so that the determination engine 11 can make a more accurate determination.
  • FIG. 8 is a schematic diagram showing a configuration example of the determination engine 11.
  • FIG. 8 shows the configuration of the determination engine 11 that determines the degree of risk by processing the information of various sensors extracted from the payload data. By inputting the information of various sensors into a so-called feed-forward type neural network, the degree of risk is determined. The internal setting of the neural network is performed by learning (backpropagation) using the past sensor information stored in the database 12 and the teacher data. A detailed description of the feed-forward type neural network will be omitted.
  • a plurality of sensor information compression transmitters 20 are installed in a place away from the river 14 in the upstream part of the river 14 in the mountainous area 13, so that the hide has been conventionally used. It is possible to capture the risk of debris flow by compensating for the shortcomings of the phone.
  • the debris flow reporting system 1 makes it possible to utilize long-distance communication means such as LPWA by compressing and transmitting voice information generated when gravel collides with gravel using its characteristics, and is a mobile phone. The danger of debris flow can be detected even in mountainous areas where lines cannot be used. Further, in the debris flow reporting system 1, it is possible to construct a system that enables highly accurate debris flow judgment by accumulating the fed-back information as teacher data and updating (learning) the judgment engine 11. ..
  • is a parameter determined by the response speed of the voice sensor 23 or the like.
  • 1 earth and stone flow notification system 7 LPWA receiver, 8 user terminal, 10 cloud computer, 11 judgment engine, 12 database, 13 mountainous area, 14 river, 20 sensor information compression transmitter, 21 rain sensor, 22 odor sensor, 23 voice sensor , 24 CPU, 25 LPWA wireless transmitter, 26 WakeUp circuit, 27 battery, 28 test switch, 29 GPS receiver

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Audiology, Speech & Language Pathology (AREA)
  • Multimedia (AREA)
  • Civil Engineering (AREA)
  • Computational Linguistics (AREA)
  • Signal Processing (AREA)
  • Health & Medical Sciences (AREA)
  • Mechanical Engineering (AREA)
  • Human Computer Interaction (AREA)
  • Acoustics & Sound (AREA)
  • Structural Engineering (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Environmental & Geological Engineering (AREA)
  • General Life Sciences & Earth Sciences (AREA)
  • Geology (AREA)
  • Business, Economics & Management (AREA)
  • Emergency Management (AREA)
  • Testing Or Calibration Of Command Recording Devices (AREA)
  • Emergency Alarm Devices (AREA)

Abstract

This debris flow reporting system has: a sound acquisition means that acquires sound information; a compression means that reduces an information amount by compressing the sound information; a wireless transmission means that wirelessly transmits the sound information compressed by the compression means; a reception means that receives the sound information wirelessly transmitted by the wireless transmission means; an assessment means that uses the sound information received by the reception means to assess a debris flow situation; and a notification means that notifies a user of an output by the assessment means. The present invention provides a reporting system for debris flows that can appropriately assess and report on a debris flow hazard level by distinguishing between a debris flow occurring at a river in a mountainous area and noise not caused by a debris flow.

Description

土石流通知システム及び土石流センサDebris flow notification system and debris flow sensor
 本発明は、土石流通知システム及び土石流センサに関する。 The present invention relates to a debris flow notification system and a debris flow sensor.
 山地河川においては土石流により甚大な被害がもたらされることから、明治時代から砂防ダムの建設が続けられている。 しかし依然として砂防ダムが設けられていない河川も多く、国土交通省の発表によれば平成30年の土砂災害は1都2府41県で3、451件と、昭和57年に集計を開始してからの最多件数(平成16年に2、537件)を大幅に上回ってしまった。  Since debris flow causes enormous damage in mountain rivers, construction of sabo dams has been continued since the Meiji era. However, many rivers still do not have sabo dams, and according to the Ministry of Land, Infrastructure, Transport and Tourism, there were 3,451 landslide disasters in 1 metropolitan area, 2 prefectures and 41 prefectures, and the total was started in 1982. It has greatly exceeded the maximum number of cases from (2,537 cases in 2004).
 土石流等の発生を検知するために、ワイヤーを用いた接触型の土石流検知器や、光束を利用した非接触型の土石流検知器が用いられている。これらの土石流検知器は、土石流がワイヤーを切断したことを検出することによって、または、土石流が光束を遮断したことを検出することによって、土石流等の発生を検知して、下流に住む人々に警報を発するように構成されている。 In order to detect the occurrence of debris flow, contact type debris flow detectors using wires and non-contact type debris flow detectors using luminous flux are used. These debris flow detectors detect the occurrence of debris flow, etc. by detecting that the debris flow has cut the wire, or by detecting that the debris flow has blocked the luminous flux, and alert people living downstream. Is configured to emit.
 しかしながら、これら土石流検知器は鹿などの動物によって誤検出される可能性がある。さらにワイヤーを用いた土石流検知器では、動物がワイヤーを切断する度に山中に設置したワイヤーを張り直す作業が必要となり、維持費が高い。また土石流が発生してから警報が出ても、避難するための時間が充分にないという問題点もある。 However, these debris flow detectors may be erroneously detected by animals such as deer. Furthermore, with a debris flow detector that uses a wire, it is necessary to re-tension the wire installed in the mountains every time an animal cuts the wire, and maintenance costs are high. There is also the problem that there is not enough time to evacuate even if an alarm is issued after the debris flow occurs.
特開2001-281015号公報Japanese Unexamined Patent Publication No. 2001-281015 特開平11-230792号公報Japanese Unexamined Patent Publication No. 11-230792
 「非特許論文1」に記されているように、日本の砂防河川においてはハイドロフォン(金属管)に流砂が衝突する音を解析して,流砂量を計測する試みが研究されている。 またスイスにおいては古くからプレート式ジオフォンが開発され、ヨーロッパ諸国で用いられている。  As described in "Non-Patent Paper 1", in Japanese sabo rivers, attempts to measure the amount of quicksand by analyzing the sound of quicksand colliding with a hydrophone (metal pipe) are being studied. Also, in Switzerland, plate-type geophones have been developed for a long time and are used in European countries.
 ハイドロフォンは,川床に設置された金属管に粒子が衝突した音を特定の周波数帯で抽出して,その振幅値が設定した振幅閾値を超えた回数を計測して流砂量を推定するものである。 ハイドロフォンは、土石流が発生する前の段階で礫が流れてくる現象を捉えることにより、土石流の警報として使うことができると期待される。また、動物によりワイヤーが切断される問題が無く、データの取り扱いが容易であるという利点もある。しかし大きな礫がパイプに衝突した場合は金属管が変形・破損し、取り換えが必要になる。 また河川の状況によって振幅閾値を適切に設定する必要があるが、実際の土石流が発生しなければ正しい設定値が解らないという問題もある。 The hydrophone extracts the sound of particles colliding with a metal tube installed on the riverbed in a specific frequency band, and measures the number of times the amplitude value exceeds the set amplitude threshold to estimate the amount of quicksand. is there. It is expected that the hydrophone can be used as a debris flow warning by capturing the phenomenon of gravel flowing before the debris flow occurs. In addition, there is no problem that the wire is cut by the animal, and there is an advantage that the data can be easily handled. However, if a large gravel collides with the pipe, the metal pipe will be deformed or damaged and will need to be replaced. It is also necessary to set the amplitude threshold appropriately depending on the river conditions, but there is also the problem that the correct set value cannot be known unless the actual debris flow occurs.
 スイスで開発されたジオフォンは、河川の床固めの天板に鉄板を設置し、その上を通過する礫による振動を計測して礫の数を数える構造となっている。この構造であれば土石流が発生してもジオフォンが破壊されることは無い。しかし山岳地の河川において床固め工事が必要となり、これは困難であると同時に極めて費用がかかる。  The geophone developed in Switzerland has a structure in which an iron plate is installed on the top plate of the river floor and the number of gravel is counted by measuring the vibration caused by the gravel passing over it. With this structure, the geophone will not be destroyed even if a debris flow occurs. However, floor consolidation work is required in mountain rivers, which is difficult and extremely expensive.
 そこで特許文献1に記載の土石流・泥流発生検知装置では、傾斜面上を流下する流下体に対して超音波や電波を送信するとともに、流下体からの反射波を受信し、受信波のスペクトル分布中の挙動に着目して、土石流や泥流の発生の有無を判定している。 Therefore, the debris flow / mudflow generation detection device described in Patent Document 1 transmits ultrasonic waves and radio waves to the flowing body flowing down on the inclined surface, receives the reflected wave from the flowing body, and receives the spectrum of the received wave. Focusing on the behavior during distribution, the presence or absence of debris flow or mudflow is determined.
 特許文献1に記載の土石流・泥流発生検知装置によれば、比較的廉価な設置費用によって、土石流や泥流の発生を検知することができると考えられる。 しかし、スペクトル分布の広がりは、河川の構造により大きく異なる。 実際に土石流が発生するまで、スペクトル分布の判定基準を正しく設けることが難しい。 According to the debris flow / mudflow generation detection device described in Patent Document 1, it is considered that the debris flow / mudflow generation can be detected at a relatively low installation cost. However, the spread of the spectral distribution varies greatly depending on the structure of the river. It is difficult to correctly set the criteria for spectral distribution until a debris flow actually occurs.
 特許文献2に記載の土石流検知システムは、対象となる河川の水面に接する物体に音波検出器を設置し、音波レベルを無線で伝送し、受信した音波レベルから土石流を判定するシステムである。 このシステムは安価に設置できるメリットがある。 しかし、土石流以外の現象で音波レベルが増大した場合(例えば、落雷があった場合、検出器付近をヘリコプターが飛行した場合、ブルドーザなどの重機が走行した場合など)に、誤発報がなされてしまう問題点がある。 The debris flow detection system described in Patent Document 2 is a system in which a sound wave detector is installed on an object in contact with the water surface of a target river, the sound wave level is transmitted wirelessly, and the debris flow is determined from the received sound wave level. This system has the advantage of being inexpensive to install. However, when the sound wave level increases due to a phenomenon other than debris flow (for example, when there is a lightning strike, when a helicopter flies near the detector, when a heavy machine such as a bulldozer runs, etc.), an erroneous report is made. There is a problem that it ends up.
 そこで、本発明は、山間部の河川で発生する土石流と土石流以外の雑音とを区別することで土石流の危険度を適切に判断して通報することのできる、土石流通報システム及び土石流センサを提供することを目的とする。 Therefore, the present invention provides a debris flow reporting system and a debris flow sensor capable of appropriately determining and reporting the risk of debris flow by distinguishing between debris flow generated in a river in a mountainous area and noise other than debris flow. The purpose is.
 本発明の土石流通報システムは、音声情報を取得する音声取得手段と、前記音声情報を圧縮して情報量を減らす圧縮手段と、前記圧縮手段により圧縮された前記音声情報を無線伝送する無線伝送手段と、前記無線伝送手段により無線伝送された前記音声情報を受信する受信手段と、前記受信手段により受信された前記音声情報を使って土石流の状況を判断する判断手段と、前記判断手段の出力をユーザに通知する通知手段を持つものである。 The earth and stone flow reporting system of the present invention includes a voice acquisition means for acquiring voice information, a compression means for compressing the voice information to reduce the amount of information, and a wireless transmission means for wirelessly transmitting the voice information compressed by the compression means. A receiving means for receiving the voice information wirelessly transmitted by the wireless transmission means, a determination means for determining the state of the earth and stone flow using the voice information received by the receiving means, and an output of the determination means. It has a notification means for notifying the user.
 本発明の土石流センサは、音声情報を取得する音声取得手段と、前記音声情報を圧縮して情報量を減らす圧縮手段と、前記圧縮手段により圧縮された前記音声情報を無線伝送する無線伝送手段とを持ち、前記圧縮手段は、前記音声情報が所定の波形モデルに従うことを使って情報量を圧縮するものである。 The earth and stone flow sensor of the present invention includes a voice acquisition means for acquiring voice information, a compression means for compressing the voice information to reduce the amount of information, and a wireless transmission means for wirelessly transmitting the voice information compressed by the compression means. The compression means compresses the amount of information by using the audio information according to a predetermined waveform model.
 本発明の土石流通報システム及び土石流センサにおいては、対象となる河川から離れた位置にマイクロフォンを設置し、河川を流れる土石の音声信号を取得する。 取得された音声信号から礫の衝突に特有の信号を検出することにより音声情報(情報量)を圧縮し、LPWA(Low Power Wide Area)無線で麓にある受信機を介してクラウドコンピュータに伝送する。LPWA無線は伝送する情報量が少ない(概ね100ビット程度)場合に適用できる無線通信技術で、長距離の無線伝送を低消費電力で可能とする技術である。 In the debris flow reporting system and the debris flow sensor of the present invention, a microphone is installed at a position away from the target river to acquire the voice signal of the debris flowing through the river. Voice information (amount of information) is compressed by detecting a signal peculiar to gravel collision from the acquired voice signal, and transmitted to a cloud computer via a receiver at the foot of the LPWA (Low Power Wide Area) radio. .. LPWA wireless is a wireless communication technology that can be applied when the amount of information to be transmitted is small (approximately 100 bits), and is a technology that enables long-distance wireless transmission with low power consumption.
 クラウドコンピュータには判断エンジンが搭載され、LPWAにより伝送された圧縮音声情報を使って、土石流に関する判断を行い、土石流の危険度を通報する。 クラウドコンピュータは、LPWAにより伝送された圧縮音声情報などをデータベースに蓄積していく。 クラウドコンピュータは、実際の土石流発生と照らし合わせることにより判断エンジンの更新(学習)をおこなうので、判断エンジンの性能が順次向上していく。 このようにして土石流の学習を繰り返した結果として、土石流に関する危険度合を正しく判定し、通知するシステムである。 The cloud computer is equipped with a judgment engine, which makes judgments about debris flow using compressed audio information transmitted by LPWA and reports the degree of risk of debris flow. The cloud computer accumulates compressed audio information transmitted by LPWA in a database. The cloud computer updates (learns) the judgment engine by comparing it with the actual debris flow, so the performance of the judgment engine will gradually improve. As a result of repeating the learning of debris flow in this way, it is a system that correctly determines and notifies the degree of risk related to debris flow.
 本発明によれば、河川から離れた場所において礫と礫が衝突する音声情報を検出し、礫に特有の音声波形であることを使って音声情報をデータ圧縮し、麓にある受信機まで無線伝送することにより、土石流と土石流以外の雑音とを区別することで土石流の危険度を適切に判断して通報することができる。 また、無線伝送する情報量が少ないので長距離・低消費電力で動作する無線技術(LPWA)が適用可能となり、携帯電話が届かない山中の土石流による危険を判断することが可能となる。 また実際の土石流発生状況を教師データとして用いて、判断エンジンのデータベースを更新(学習)していくことにより、従来の土石流検知システムとは格段に信頼性の高いシステムを安価に提供することが可能となる。 According to the present invention, voice information in which gravel collides with gravel is detected at a place away from a river, voice information is compressed by using the voice waveform peculiar to gravel, and wireless to a receiver at the foot of the river. By transmitting, the risk of debris flow can be appropriately determined and reported by distinguishing between debris flow and noise other than debris flow. In addition, since the amount of information transmitted wirelessly is small, wireless technology (LPWA) that operates over long distances and with low power consumption can be applied, and it becomes possible to judge the danger due to debris flow in the mountains that mobile phones cannot reach. In addition, by updating (learning) the database of the judgment engine using the actual debris flow occurrence status as teacher data, it is possible to provide a system that is far more reliable than the conventional debris flow detection system at a low cost. It becomes.
土石流通報システム1の全体構成を説明する図である。It is a figure explaining the whole structure of the debris flow report system 1. センサ情報圧縮送信機20の回路構成を示す図である。It is a figure which shows the circuit structure of the sensor information compression transmitter 20. CPU24の動作アルゴリズムを示すフローチャートである。It is a flowchart which shows the operation algorithm of CPU 24. 128ビットのペイロードデータ30の構成を示す図である。It is a figure which shows the structure of the 128-bit payload data 30. 音声情報の圧縮アルゴリズムを示す図である。It is a figure which shows the compression algorithm of voice information. 礫と礫との衝突波形(A)及びモデル化された波形(B)を示す図である。It is a figure which shows the collision waveform (A) of gravel and gravel, and the modeled waveform (B). 礫と礫との衝突スペクトル分布の一例を示す図である。It is a figure which shows an example of the collision spectrum distribution of a gravel and a gravel. 判断エンジン11の構成例を示す模式図である。It is a schematic diagram which shows the structural example of the determination engine 11.
 以下、本開示を実施するための形態(以下実施の形態とする)について説明する。 Hereinafter, a mode for implementing the present disclosure (hereinafter referred to as an embodiment) will be described.
 <全体構成>
 図1は、土石流通報システム1の全体構成を説明する図である。
 土石流通知システム1は、図1に示すように、山岳地において発生することのある土石流の状態、あるいは土石流の予兆現象をとらえてユーザ端末8へ通知するシステムである。
<Overall configuration>
FIG. 1 is a diagram for explaining the overall configuration of the debris flow notification system 1.
As shown in FIG. 1, the debris flow notification system 1 is a system that detects the state of debris flow that may occur in mountainous areas or the precursory phenomenon of debris flow and notifies the user terminal 8.
 土石流通報システム1では、山岳地帯13にある河川14の上流部において、河川14から離れた場所にセンサ情報圧縮送信機20(圧縮手段)を複数設置する。 センサ情報圧縮送信機20は、設置場所周囲の音声情報、におい情報、降雨情報などの情報を検出し、128ビットのペイロードデータに圧縮し、LPWA無線によって麓にある受信局7に長距離伝送する。 In the debris flow notification system 1, a plurality of sensor information compression transmitters 20 (compression means) are installed at a location away from the river 14 in the upstream part of the river 14 in the mountainous area 13. The sensor information compression transmitter 20 detects information such as voice information, odor information, and rainfall information around the installation location, compresses it into 128-bit payload data, and transmits it over a long distance to the receiving station 7 at the foot of the device by LPWA radio. ..
 市街地に設置されたLPWA受信機7は、センサ情報圧縮送信機20から送信されたペイロードデータ30(後述する図4参照。)を受信して復号し、インターネット上のクラウドコンピュータ10に伝送する。クラウドコンピュータ10は、ペイロードデータ30から各種センサの情報を切り出してデータベース12に保管していく。 クラウドコンピュータ10はまた、各種センサから得られたデータを判断エンジン11(判断手段)に供給する。  The LPWA receiver 7 installed in the urban area receives the payload data 30 (see FIG. 4 described later) transmitted from the sensor information compression transmitter 20, decodes it, and transmits it to the cloud computer 10 on the Internet. The cloud computer 10 cuts out information of various sensors from the payload data 30 and stores it in the database 12. The cloud computer 10 also supplies data obtained from various sensors to the judgment engine 11 (judgment means).
 判断エンジン11はクラウドコンピュータ10で実行される認識プログラムであって、例えばニューラルネットワークにより実現される。 判断エンジン11により、土石流が発生している、あるいは発生する可能性が高いと判断された場合には、ユーザ端末8にメールを伝送することにより、地元住民や自治体に土石流発生の危険が高まっていることを知らせる。 このメールには、GPSによる位置情報と、判断エンジン11が判断した土石流の状況を表す情報が含まれる。 The judgment engine 11 is a recognition program executed by the cloud computer 10, and is realized by, for example, a neural network. When the determination engine 11 determines that a debris flow has occurred or is likely to occur, a mail is transmitted to the user terminal 8 to increase the risk of debris flow to local residents and local governments. Inform you that you are. This email contains GPS location information and information indicating the status of debris flow determined by the determination engine 11.
 ユーザ端末8に表示された情報を参考にして、自治体などは適宜避難の指示を出すことになる。 もちろんユーザ端末8に表示された通りに土石流が発生することが予想されるが、判断エンジン11の判断ミスにより、ユーザ端末8に表示されたのとは異なる結果となる可能性もある。 このとき、ユーザはユーザ端末8を操作して実際の状況をクラウドコンピュータ10にフィードバックする。 クラウドコンピュータ10は、フィードバックされた情報を教師データとしてデータベース12に蓄積していく。  With reference to the information displayed on the user terminal 8, local governments and others will issue evacuation instructions as appropriate. Of course, it is expected that debris flow will occur as it is displayed on the user terminal 8, but due to a judgment error of the judgment engine 11, the result may be different from that displayed on the user terminal 8. At this time, the user operates the user terminal 8 to feed back the actual situation to the cloud computer 10. The cloud computer 10 accumulates the fed-back information as teacher data in the database 12.
 ユーザから直接のフィードバックが得られない場合においては、降雨の前後においてドローンにより河川14の周囲を撮影し、河川敷にある礫の画像情報を比較することにより、下流に流れた礫の情報を知ることができる。 このような礫の移動情報も、クラウドコンピュータ10にフィードバックされ、教師データとして活用される。 When direct feedback cannot be obtained from the user, the surroundings of the river 14 are photographed by a drone before and after the rainfall, and the image information of the gravel on the riverbed is compared to know the information of the gravel flowing downstream. Can be done. Such gravel movement information is also fed back to the cloud computer 10 and used as teacher data.
 このようにしてデータベース12に各種センサの情報と教師データが蓄積されていく。 これらセンサ情報と教師データを用いて、クラウドコンピュータ10は判断エンジン11の更新(学習)をおこなう。 すなわち判断エンジン11は、ユーザからのフィードバックにより、その判断性能を向上させていくように構成されている。 本システムはこのようにして判断エンジン11の性能を順次改善(学習)させることにより、最終的に高い精度で土石流の判断を可能とするシステムを構築する。 土石流をもたらすような豪雨が頻繁に発生する夏場には、頻繁に判断エンジン11の更新を行うことが望まれる。 In this way, information on various sensors and teacher data are accumulated in the database 12. Using these sensor information and teacher data, the cloud computer 10 updates (learns) the judgment engine 11. That is, the judgment engine 11 is configured to improve its judgment performance by feedback from the user. By sequentially improving (learning) the performance of the judgment engine 11 in this way, this system finally constructs a system that enables debris flow judgment with high accuracy. In the summer when heavy rains that cause debris flow occur frequently, it is desirable to update the judgment engine 11 frequently.
 従来使われてきたハイドロフォンは、河川14の流域内に設置されてきたため、土石流が発生した場合にはハイドロフォンが壊れ、あるいは流されてしまう可能性が高かった。これに対して本実施例では、河川14から音声が届く範囲で離れた場所にセンサ情報圧縮送信機20を設置し、礫と礫が衝突することにより発生する音声波形を抽出し、圧縮してから伝送することにより、土石流が発生した場合においてもセンサ情報圧縮送信機20が壊れる可能性を低下させている。 Since the hydrophones that have been used in the past have been installed in the basin of the river 14, there was a high possibility that the hydrophones would break or be washed away if a debris flow occurred. On the other hand, in this embodiment, the sensor information compression transmitter 20 is installed at a location distant from the river 14 within the range where the sound can reach, and the sound waveform generated by the collision of the gravel and the gravel is extracted and compressed. By transmitting from the sensor information compression transmitter 20, the possibility that the sensor information compression transmitter 20 is broken is reduced even when a gravel flow occurs.
 また本実施例においては音声波形を圧縮しているのでLPWAのような低ビットかつ長距離の無線技術が適用可能となり、センサ情報圧縮送信機20を人里離れた山中に設置することが可能となっている。 また礫と礫が衝突する波形を抽出しているので、従来のパルス数のように雑音で誤動作する可能性が低く、確実な判断を行うことができる。 Further, in this embodiment , since the voice waveform is compressed, low-bit and long-distance wireless technology such as LPWA can be applied, and the sensor information compression transmitter 20 can be installed in a remote mountain. It has become. Further, since the waveform in which the gravel collides with the gravel is extracted, there is a low possibility of malfunction due to noise unlike the conventional number of pulses, and a reliable judgment can be made.
 <センサ情報圧縮送信機20の構成>
 図2は、センサ情報圧縮送信機20の回路構成を示す図である。
 センサ情報圧縮送信機20は、図2に示すように、バッテリー27を内蔵し、電源をWakeUp回路26及びセンサ情報圧縮送信機20の各部に供給する。 音声センサ23は、小型のマイクロフォンとAD変換器で構成され、河川11、あるいはその付近において、礫と礫が衝突する音声情報を検出してCPU24に送る。 臭いセンサ22は、硫黄化合物ガスなどを検出する所謂悪臭感知センサであり、その検出出力をAD変換してCPU24に送る。雨量センサ21は降雨量を検出し、AD変換してCPU24に送る。
<Structure of sensor information compression transmitter 20>
FIG. 2 is a diagram showing a circuit configuration of the sensor information compression transmitter 20.
As shown in FIG. 2, the sensor information compression transmitter 20 has a built-in battery 27 and supplies power to each part of the WakeUp circuit 26 and the sensor information compression transmitter 20. The voice sensor 23 is composed of a small microphone and an AD converter, and detects voice information in which gravel collides with gravel in or near the river 11 and sends it to the CPU 24. The odor sensor 22 is a so-called malodor sensing sensor that detects sulfur compound gas and the like, and AD-converts the detection output and sends it to the CPU 24. The rainfall sensor 21 detects the amount of rainfall, converts it to AD, and sends it to the CPU 24.
 過去の土石流において、土石流が発生する1時間程度前に、「臭い」や「石が転がる音」が観測されたことが報告されている。 音声センサ23と臭いセンサ22は、このような前兆現象を捉えることにより、土石流の危険をより正しく判断するのに有用である。 It has been reported that in the past debris flow, "smell" and "sound of rolling stones" were observed about 1 hour before the debris flow occurred. The voice sensor 23 and the odor sensor 22 are useful for more accurately determining the danger of debris flow by capturing such a precursory phenomenon.
 GPS受信機29は、地球を周回する複数のGPS衛星からの電波を受信し、センサ情報圧縮送信機20が設置された場所の緯度と経度情報を算出して、CPU24に送る。 無線送信機25は、CPU24により生成された128ビットのペイロードデータをLPWA無線として送信する。 The GPS receiver 29 receives radio waves from a plurality of GPS satellites orbiting the earth, calculates latitude and longitude information of the place where the sensor information compression transmitter 20 is installed, and sends the radio waves to the CPU 24. The wireless transmitter 25 transmits the 128-bit payload data generated by the CPU 24 as LPWA radio.
 CPU24は、後述するアルゴリズムにより、音声センサ23からの情報を圧縮し、GPS緯度経度情報と識別符号(ID)を付加し、臭いセンサ22と雨量センサ21の情報を加えることにより、128ビットのペイロードデータ30を作成し無線送信機25に送る。  The CPU 24 compresses the information from the voice sensor 23 by an algorithm described later, adds GPS latitude / longitude information and an identification code (ID), and adds the information of the odor sensor 22 and the rainfall sensor 21 to obtain a 128-bit payload. The data 30 is created and sent to the wireless transmitter 25.
 WakeUp回路26は、音声センサ23の出力レベルをモニタし、レベルが所定値を下回る場合には、CPU24及び無線送信機25に指示を出してスリープモードとすることにより、センサ情報圧縮送信機20の消費電力を低減させる。 The WakeUp circuit 26 monitors the output level of the voice sensor 23, and when the level falls below a predetermined value, issues an instruction to the CPU 24 and the wireless transmitter 25 to put the sensor information compression transmitter 20 into sleep mode. Reduce power consumption.
 図3は、CPU24の動作アルゴリズムを示すフローチャートである。
 ステップSP1において、CPU24は1分毎に後述する信号処理アルゴリズムを実行する。 ステップSP2において、CPU24は音声センサ23から得られる音声レベルを測定し、そのレベルが所定値以下である場合には処理を中断することにより、センサ情報圧縮送信機20の消費電力を低減させる。
FIG. 3 is a flowchart showing an operation algorithm of the CPU 24.
In step SP1, the CPU 24 executes a signal processing algorithm described later every minute. In step SP2, the CPU 24 measures the voice level obtained from the voice sensor 23, and if the level is equal to or less than a predetermined value, the processing is interrupted to reduce the power consumption of the sensor information compression transmitter 20.
 音声センサ23(音声取得手段)から得られる信号レベルが所定値を超えていた場合、すなわち降雨によるザーザーという音や、礫が斜面を転がる音などが検出された可能性がある場合には、ステップSP3において各種センサの電源がONとなり、以降の処理が実行されて土石流に関わる情報計測及びLPWA無線通信が行われる。 If the signal level obtained from the voice sensor 23 (voice acquisition means) exceeds a predetermined value, that is, if there is a possibility that a rustling sound due to rainfall or a sound of gravel rolling on a slope is detected, a step is taken. In SP3, the power of various sensors is turned on, the subsequent processing is executed, and information measurement related to the earth and stone flow and LPWA wireless communication are performed.
 ステップSP4において、CPU24はGPS受信機29から設置場所の緯度及び経度情報を取得する。 例えば、北緯36.030160度、東経138.155298度といった情報がGPS受信機29から得られる。 CPU24は、緯度及び経度の小数点以下6桁の情報(上の例では“030160”、“155298”)をBCD(Binary Coded Decimal)で表すことにより、それぞれ24ビットの緯度情報と経度情報に圧縮する。 In step SP4, the CPU 24 acquires the latitude and longitude information of the installation location from the GPS receiver 29. For example, information such as latitude 36.030160 degrees north and longitude 138.155298 degrees east can be obtained from the GPS receiver 29. The CPU 24 compresses latitude and longitude 6-digit information (“030160” and “155298” in the above example) into 24-bit latitude information and longitude information by representing them in BCD (Binary Coded Decimal), respectively. ..
 緯度と経度の小数点より上の桁(北緯36度、東経138度)は、受信機7において復元することができるので、このように削除して構わない。またこのようにして得られる緯度と経度の情報は、万が一センサ情報圧縮送信機20が土石流により流された場合に、緯度と経度が変化することから「流された」という事実を知ることができ、土石流が既に発生していることを示す重要な情報となる。 The digits above the decimal point of latitude and longitude (36 degrees north latitude, 138 degrees east longitude) can be restored by the receiver 7, so they may be deleted in this way. In addition, the latitude and longitude information obtained in this way can be used to know the fact that the sensor information compression transmitter 20 has been "flowed" because the latitude and longitude change if it is washed away by a debris flow. , It is important information to show that the debris flow has already occurred.
 ステップSP5において、臭いセンサ22のデータを処理する。臭いセンサ22の出力値を6ビットに変換することにより、臭いレベル情報SCENTとしてペイロードデータ30に付加する。  In step SP5, the data of the odor sensor 22 is processed. By converting the output value of the odor sensor 22 into 6 bits, it is added to the payload data 30 as odor level information SCENT.
 前述したように、過去の土石流発生においては1時間前から独特の臭いが報告されている。 そこで臭いレベルは土石流の予兆となる可能性がある。 ただし、例えば熊などの野生動物からも極めて強い臭いが発せられるので、臭いレベルだけ基づいて、土石流の発生可能性を推測するのは適当でない。 そこで本システムでは、臭いレベルを参考情報の一つとしてクラウドコンピュータ10に伝送し、データベース12に蓄積し、他のセンサからの情報、さらには土石流が起きたか起きなかったか、といった過去の実績と合わせて、クラウドコンピュータ10に実装された判断エンジン11に学習させる構成となっている。 As mentioned above, a peculiar odor has been reported from 1 hour before the past debris flow occurrence. Therefore, the odor level may be a sign of debris flow. However, since wild animals such as bears also emit extremely strong odors, it is not appropriate to estimate the possibility of debris flow based only on the odor level. Therefore, in this system, the odor level is transmitted to the cloud computer 10 as one of the reference information, stored in the database 12, and combined with the information from other sensors and the past results such as whether debris flow occurred or not. Therefore, the determination engine 11 mounted on the cloud computer 10 is made to learn.
 ステップSP6において、雨量センサ21のデータを処理して、6ビットのデータ(Precipitation)としてペイロードデータ30に付加する。 降雨は土石流発生の根本原因であるが、土石流が発生するか否かは土壌組成や地下水流などにも影響される。
 このため降雨量だけに依存して土石流発生を判断することは適当でない。 そこで本システムでは、降雨量を参考情報の一つとしてクラウドコンピュータ10に伝送し、データベース12に蓄積し、他のセンサからの情報、さらには土石流が起きたか起きなかったか、といった実績情報と合わせて判断エンジン11が学習するように構成されている。
In step SP6, the data of the rainfall sensor 21 is processed and added to the payload data 30 as 6-bit data (Prescription). Precipitation is the root cause of debris flow, but whether or not debris flow occurs is also affected by soil composition and groundwater flow.
Therefore, it is not appropriate to judge the occurrence of debris flow based only on the amount of rainfall. Therefore, in this system, the amount of rainfall is transmitted to the cloud computer 10 as one of the reference information, stored in the database 12, and combined with the information from other sensors and the actual information such as whether or not the debris flow occurred. The determination engine 11 is configured to learn.
 ステップSP7において、CPU24は音声センサ23の情報を処理する。先に述べたように音声センサ23は、小型のマイクロフォンとAD変換器で構成され、周囲の音声情報を検出する。 AD変換器の変換レートが10kHz、量子化ビット数が10ビットの場合、1分間で6メガビットものデータになってしまう。このデータは様々な情報を含んでいて有用であるが、データ量が多いためLPWAで伝送することはできない。 そこでCPU24は、図5において後述するように土石流検出に有用である音声情報を抽出し、42ビットの圧縮音声情報Soundとしてペイロードデータ30に付加する。 In step SP7, the CPU 24 processes the information of the voice sensor 23. As described above, the voice sensor 23 is composed of a small microphone and an AD converter, and detects surrounding voice information. If the conversion rate of the AD converter is 10 kHz and the number of quantization bits is 10 bits, the data will be as much as 6 megabits in 1 minute. Although this data contains various information and is useful, it cannot be transmitted by LPWA due to the large amount of data. Therefore, the CPU 24 extracts audio information useful for debris flow detection as will be described later in FIG. 5, and adds it to the payload data 30 as a 42-bit compressed audio information Sound.
 ステップSP8においては、128ビットのペイロードデータ30を後述する図4に示すように構成する。 ペイロードデータ30の先頭は16ビットのID情報であり、CPU24の内部不揮発メモリに記録されている固有番号を使うことができる。 次にステータス情報(5ビット)、GPSの緯度情報(24ビット)及び経度情報(24ビット)、周辺温度の情報Temp(5ビット)、降雨量をあらわす情報Precipitation(6ビット)、臭いをあらわす情報SCENT(6ビット)、そして圧縮音声情報(42ビット)で、ペイロードデータが構成される。 In step SP8, the 128-bit payload data 30 is configured as shown in FIG. 4, which will be described later. The beginning of the payload data 30 is 16-bit ID information, and the unique number recorded in the internal non-volatile memory of the CPU 24 can be used. Next, status information (5 bits), GPS latitude information (24 bits) and longitude information (24 bits), ambient temperature information Temp (5 bits), rainfall information Precipitation (6 bits), and odor information. The payload data is composed of SCENT (6 bits) and compressed audio information (42 bits).
 図4は、128ビットのペイロードデータ30の構成を示す図である。
 図4左下に示すように、ステータス情報はTEST(1ビット)、電池残量を16段階で表すBAT情報(4ビット)で構成される。 センサ情報圧縮送信機20にはテストスイッチ28が搭載されている。 システムが正しく動作することを確認する場合は、テストスイッチ20を押すことにより、ペイロードデータのTEST情報を“1”として、システムの動作試験を行うことができる。
FIG. 4 is a diagram showing a configuration of 128-bit payload data 30.
As shown in the lower left of FIG. 4, the status information is composed of TEST (1 bit) and BAT information (4 bits) indicating the remaining battery level in 16 steps. A test switch 28 is mounted on the sensor information compression transmitter 20. When confirming that the system operates correctly, the operation test of the system can be performed by pressing the test switch 20 and setting the TEST information of the payload data to "1".
 ステップSP9において、ペイロードデータ30をLPWA無線により送信する。 この無線信号は、市街地などに設置されたLPWA受信機7により受信され、クラウドコンピュータ10に伝送される。クラウドコンピュータ10は判断エンジン11によって土石流が発生する危険性を判断し、ユーザ端末8にメールを送ることにより、住民や自治体に通知する。 このメールには、GPSによる位置情報と、判断エンジン11が判断した土石流の状態が含まれる。自治体はこのメールで送られる情報だけでなく、Xバンドレーダによる地域全体の降水量などの情報を参考にして、総合的に避難準備開始など必要な判断を行うことができる。  In step SP9, the payload data 30 is transmitted by LPWA radio. This wireless signal is received by the LPWA receiver 7 installed in an urban area or the like and transmitted to the cloud computer 10. The cloud computer 10 determines the risk of debris flow by the determination engine 11, and notifies the residents and local governments by sending an e-mail to the user terminal 8. This email contains GPS location information and the state of debris flow determined by the determination engine 11. The local government can make necessary decisions such as the start of evacuation preparations comprehensively by referring to not only the information sent by this e-mail but also the information such as the amount of precipitation in the entire area by the X-band radar.
 図5は、音声情報の圧縮アルゴリズムを示す図である。
 図6は、礫と礫との衝突波形(A)及びモデル化された波形(B)を示す図である。
 図5において、ステップSP7の処理(音声センサ23から得られる情報の圧縮)を詳細に説明する。 ステップSP20において、CPU24は1分間で取り込んだ波形全体をノイズとみなして、ノイズレベルLVNを算出する。 時間間隔ΔでNポイントのAD変換結果をAu(n)(n=0、1、。。。N-1)とすると、ノイズレベルLVNは式1で算出される。

 LVN={ Σ(Au(n)・ Au(n))} ÷ N ・・ 式1

 ここで「Σ」はNポイントの総和を、「・」は乗算をあらわす演算子である。
FIG. 5 is a diagram showing a compression algorithm for voice information.
FIG. 6 is a diagram showing a gravel-to-grave collision waveform (A) and a modeled waveform (B).
In FIG. 5, the process of step SP7 (compression of information obtained from the voice sensor 23) will be described in detail. In step SP20, the CPU 24 regards the entire waveform captured in one minute as noise and calculates the noise level LVN. Assuming that the AD conversion result of N points at the time interval Δ is Au (n) (n = 0, 1, ... N-1), the noise level LVN is calculated by Equation 1.

LVN = {Σ (Au (n) ・ Au (n))} ÷ N ・ ・ Equation 1

Here, "Σ" is an operator that represents the sum of N points, and "・" is an operator that represents multiplication.
 ステップSP21においては、カウンタCNTの値をゼロにリセットする。カウンタCNTは、礫と礫が衝突したことによる音声パルスを数えるカウンタである。 CNT=0は、計測時間において衝突によるパルス音が一つも発生しなかったことを表す。 In step SP21, the value of the counter CNT is reset to zero. The counter CNT is a counter that counts voice pulses caused by collision of gravel with gravel. CNT = 0 indicates that no pulse sound due to collision was generated during the measurement time.
 ステップSP22において、波形データAu(n)(図6(A)参照。)をスキャンし、所定のレベルを超えている波形のピークを探索する。 波形のピークは、礫と礫の衝突により発生している可能性がある。 ステップSP24~SP28までの処理が施されていない波形ピークが見つかった場合には、ステップSP24以降の処理が実施されることにより、礫と礫の衝突による音声波形が抽出される。  In step SP22, the waveform data Au (n) (see FIG. 6A) is scanned to search for the peak of the waveform exceeding a predetermined level. The peak of the waveform may be caused by the collision of gravel and gravel. When a waveform peak that has not been processed in steps SP24 to SP28 is found, the processing after step SP24 is performed to extract a voice waveform due to a collision between gravel and gravel.
 ステップSP24において、カウンタCNTに1を加えられる。カウンタCNTの値は、礫と礫が衝突したと思われるパルスを表している。 In step SP24, 1 is added to the counter CNT. The value of the counter CNT represents a pulse in which the gravel seems to have collided with the gravel.
 ステップSP25において、ピーク前後の波形データAu(n)を切り出す。 図6(A)には、このようにして切り出された礫と礫の衝突波形の一例が示されている。 衝突直後に波形のピークを迎え、その後急速に減衰する波形、すなわち減衰を伴う自由振動であることがわかる。 このような波形は、動物の鳴き声や、ヘリコプターの音とは波形が全く異なり、次式2でモデル化される。

 R(n-τ)=
 A・EXP(―β・n)・ Sin(2π・Fpeak・Δ・n) ・・ 式2

 ここでAはピークの振幅、βは減衰係数、Fpeakは自由振動周波数、τは時間遅延をあらわす。
In step SP25, waveform data Au (n) before and after the peak is cut out. FIG. 6A shows an example of the collision waveform of the gravel and the gravel cut out in this way. It can be seen that the waveform peaks immediately after the collision and then decays rapidly, that is, it is a free vibration accompanied by damping. Such a waveform is completely different from the sound of an animal or a helicopter, and is modeled by the following equation 2.

R (n-τ) =
A ・ EXP (-β ・ n) ・ Sin (2π ・ Fpeak ・ Δ ・ n) ・ ・ Equation 2

Here, A is the amplitude of the peak, β is the damping coefficient, Fpeak is the free vibration frequency, and τ is the time delay.
 ステップSP26において、CPU24は切り出された音声情報Au(n)をフーリエ変換することによりスペクトルを算出する。 In step SP26, the CPU 24 calculates the spectrum by Fourier transforming the cut out voice information Au (n).
 図7は、礫と礫との衝突スペクトル分布の一例を示す図である。図7には音声情報Au(n)をフーリエ変換することにより得られたスペクトルの一例が示されている。図7において、周波数2200Hz(Fpeak)において、スペクトルのピークが観測されている。 一般に、大きな礫である程、ピーク周波数Fpeakが低下する傾向にある。    FIG. 7 is a diagram showing an example of the collision spectrum distribution between gravel and gravel. FIG. 7 shows an example of the spectrum obtained by Fourier transforming the voice information Au (n). In FIG. 7, the peak of the spectrum is observed at a frequency of 2200 Hz (Fpeak). In general, the larger the gravel, the lower the peak frequency Fpeak tends to be.
 ステップSP27において、減衰係数β及び時間遅延τの推定を行う。 すなわち、所定の初期値から順次ステップ的に変化させるβ及びτを用いて、式2により波形R(n)を求め、Au(n)との相関係数を求める。 この相関係数が最大となるようなβとτの組み合わせを求め、減衰係数β及び時間遅延τとして採用する。 またピーク振幅Aとして、切り出された音声波形Au(n)のピーク値を用いることができる。 In step SP27, the attenuation coefficient β and the time delay τ are estimated. That is, the waveform R (n) is obtained by Equation 2 and the correlation coefficient with Au (n) is obtained by using β and τ that are sequentially changed stepwise from a predetermined initial value. Find the combination of β and τ that maximizes this correlation coefficient, and use it as the attenuation coefficient β and the time delay τ. Further, as the peak amplitude A, the peak value of the cut-out voice waveform Au (n) can be used.
 上述した図6(B)には、このようにして求められたR(n-τ)の一例が示されている。 図6(B)に示された波形は、実際に観測された礫と礫の衝突波形(図6(A))と殆ど同じであることが判る。 FIG. 6 (B) described above shows an example of R (n-τ) thus obtained. It can be seen that the waveform shown in FIG. 6 (B) is almost the same as the actually observed gravel-grave collision waveform (FIG. 6 (A)).
 ステップSP28において、式3及び式4に従ってSNRが計算される。

 Err(n)=Au(n)- R(n-τ) ・・ 式3

 SNR=Σ{R(n―τ)・R(n-τ)}
     ÷ Σ{Err(n)・Err(n)} ・・ 式4
In step SP28, the SNR is calculated according to Equations 3 and 4.

Err (n) = Au (n) -R (n-τ) ... Equation 3

SNR = Σ {R (n-τ) · R (n-τ)}
÷ Σ {Err (n) ・ Err (n)} ・ ・ Equation 4
 このようにして求められるSNRは、礫と礫の衝突による音声波形であれば、大きな値となる。 これとは逆に、動物の鳴き声であった場合にはSNRが小さな値となる。 The SNR obtained in this way is a large value if it is a voice waveform due to a collision between gravel and gravel. On the contrary, if it is an animal bark, the SNR will be a small value.
 ステップSP28が終了すると処理はSP22に戻り、波形データAu(n)を再びスキャンし、未だ処理が行われていない波形ピークがあるかどうかを、ステップSP23において判断する。 未処理のピークが残っていた場合には、ステップSP24~SP28までの処理が繰り返し行われることにより、SNR、β、τが求められる。 When step SP28 is completed, the process returns to SP22, the waveform data Au (n) is scanned again, and it is determined in step SP23 whether or not there is a waveform peak that has not yet been processed. If unprocessed peaks remain, SNR, β, and τ are obtained by repeating the processes from steps SP24 to SP28.
 全ての波形データに対する処理が終了すると、音声波形Auの圧縮処理が終了したことになる。 ステップSP29においてペイロードデータ30へのセットが行われる。 すなわち図4右下に示すように、10ビットのデータとしてカウントCNTが格納される。 引き続いてSNR、ピーク周波数Fpeak、減衰係数βそしてノイズレベルLVNが、それぞれ8ビットデータとして格納される。 カウンタCNTが2以上の場合は、SNRが最大となった場合のFpeak,β、LVNが格納される。 またそれぞれの値は、所定のビット数となるように精度が調整される。 When the processing for all the waveform data is completed, the compression processing of the voice waveform Au is completed. In step SP29, the payload data 30 is set. That is, as shown in the lower right of FIG. 4, the count CNT is stored as 10-bit data. Subsequently, SNR, peak frequency Fpeak, attenuation coefficient β, and noise level LVN are each stored as 8-bit data. When the counter CNT is 2 or more, Fpeak, β, and LVN when the SNR is maximized are stored. The accuracy of each value is adjusted so that it has a predetermined number of bits.
 このようにして本実施例では、6メガビットあった音声情報Au(n)を、42ビットの情報(Sound)に圧縮してから伝送する。 データ圧縮する際に礫と礫との衝突音に特有の波形特性であることを使っている。さらに、圧縮処理が適切であったかどうかを知るために、評価指標としてSNRを付加することにより、判断エンジン11がより正しく判断できるように構成している。 In this way, in this embodiment, the voice information Au (n), which was 6 megabits, is compressed into 42 bits of information (Sound) and then transmitted. When compressing data, it uses the waveform characteristics peculiar to the collision sound between gravel and gravel. Further, in order to know whether or not the compression process is appropriate, SNR is added as an evaluation index so that the determination engine 11 can make a more accurate determination.
 <判断エンジン11の構成>
 図8は、判断エンジン11の構成例を示す模式図である。図8には、ペイロードデータから抽出された各種センサの情報を処理して、危険度を判断する判断エンジン11の構成が示されている。 各種センサの情報を、所謂フィードフォワード型のニューラルネットワークに入力することにより、危険度の判断を行う構成となっている。 ニューラルネットワークの内部設定は、データベース12に格納された過去のセンサ情報と、教師データを使った学習(バックプロパゲーション)により行われる。 フィードフォワード型のニューラルネットワークに関しては詳細説明を省略する。
<Structure of judgment engine 11>
FIG. 8 is a schematic diagram showing a configuration example of the determination engine 11. FIG. 8 shows the configuration of the determination engine 11 that determines the degree of risk by processing the information of various sensors extracted from the payload data. By inputting the information of various sensors into a so-called feed-forward type neural network, the degree of risk is determined. The internal setting of the neural network is performed by learning (backpropagation) using the past sensor information stored in the database 12 and the teacher data. A detailed description of the feed-forward type neural network will be omitted.
 上記の土石流通報システム1では、山岳地帯13にある河川14の上流部において、河川14から離れた場所にセンサ情報圧縮送信機20を複数設置する構成となっているので、従来使われてきたハイドフォンが持っていた欠点を補って、土石流の危険度を捉えることができる。 また土石流通報システム1では、礫と礫が衝突する際に発する音声情報を、その特徴を使って圧縮して伝送することにより、LPWAなどの長距離通信手段を活用することを可能とし、携帯電話回線が使えない山間部においても土石流の危険を察知するができる。 さらに土石流通報システム1では、フィードバックされた情報を教師データとして蓄積し、判断エンジン11の更新(学習)をおこなうことにより、高い精度で土石流の判断を可能とするシステムを構築することが可能である。 In the above-mentioned debris flow reporting system 1, a plurality of sensor information compression transmitters 20 are installed in a place away from the river 14 in the upstream part of the river 14 in the mountainous area 13, so that the hide has been conventionally used. It is possible to capture the risk of debris flow by compensating for the shortcomings of the phone. In addition, the debris flow reporting system 1 makes it possible to utilize long-distance communication means such as LPWA by compressing and transmitting voice information generated when gravel collides with gravel using its characteristics, and is a mobile phone. The danger of debris flow can be detected even in mountainous areas where lines cannot be used. Further, in the debris flow reporting system 1, it is possible to construct a system that enables highly accurate debris flow judgment by accumulating the fed-back information as teacher data and updating (learning) the judgment engine 11. ..
 以上の実施例では、衝突音が減衰を伴う自由振動であると仮定した。 実際には、音声センサ23には応答速度があり、あまり急速な信号変化には対応することができない。そこで音声センサ23の応答速度を考慮することにより、音声センサ23が検出する衝突音をより正しくモデル化することが可能になる。 この場合は、これまでの式2を、以下の式5のように変形することで対応できる。

 R(n-τ)=
 A{1-EXP(-α・n)}・EXP(―β・n)・Sin(2π・Fpeak・Δ・n)  ・・式5

 ここでαは、音声センサ23などの応答速度で定まるパラメータである。
In the above examples, it is assumed that the collision sound is a free vibration with damping. In reality, the voice sensor 23 has a response speed and cannot cope with a very rapid signal change. Therefore, by considering the response speed of the voice sensor 23, it becomes possible to more accurately model the collision sound detected by the voice sensor 23. In this case, it can be dealt with by modifying the conventional equation 2 as in the following equation 5.

R (n-τ) =
A {1-EXP (-α ・ n)} ・ EXP (-β ・ n) ・ Sin (2π ・ Fpeak ・ Δ ・ n) ・ ・ Equation 5

Here, α is a parameter determined by the response speed of the voice sensor 23 or the like.
 さらに式5を簡略化して、以下の式6で表すことも可能である。

 R(n-τ)=
 A(θ・n)・EXP(―β・n)・Sin(2π・Fpeak・Δ・n) ・・式6

 ここでθは、音声センサ23などの応答速度で定まるパラメータである。
Further, it is also possible to simplify the equation 5 and express it by the following equation 6.

R (n-τ) =
A (θ ・ n) ・ EXP (-β ・ n) ・ Sin (2π ・ Fpeak ・ Δ ・ n) ・ ・ Equation 6

Here, θ is a parameter determined by the response speed of the voice sensor 23 or the like.
 1 土石流通報システム、 7 LPWA受信機、8 ユーザ端末、 10 クラウドコンピュータ、 11 判断エンジン、 12 データベース、13 山岳地帯、 14 河川、 20 センサ情報圧縮送信機、21 雨量センサ、22 臭いセンサ、 23 音声センサ、 24 CPU、 25 LPWA無線送信機、26 WakeUp回路、 27 バッテリー、 28 テストスイッチ、 29 GPS受信機 1 earth and stone flow notification system, 7 LPWA receiver, 8 user terminal, 10 cloud computer, 11 judgment engine, 12 database, 13 mountainous area, 14 river, 20 sensor information compression transmitter, 21 rain sensor, 22 odor sensor, 23 voice sensor , 24 CPU, 25 LPWA wireless transmitter, 26 WakeUp circuit, 27 battery, 28 test switch, 29 GPS receiver

Claims (8)

  1.  音声情報を取得する音声取得手段と、
     前記音声情報を圧縮して情報量を減らす圧縮手段と、
     前記圧縮手段により圧縮された前記音声情報を無線伝送する無線伝送手段と、
     前記無線伝送手段により無線伝送された前記音声情報を受信する受信手段と、
     前記受信手段により受信された前記音声情報を使って土石流の状況を判断する判断手段と、
     前記判断手段の出力をユーザに通知する通知手段を持つことを特徴とする土石流通報システム。
    Voice acquisition means to acquire voice information and
    A compression means that compresses the voice information to reduce the amount of information,
    A wireless transmission means for wirelessly transmitting the voice information compressed by the compression means, and
    A receiving means for receiving the voice information wirelessly transmitted by the wireless transmission means, and
    A determination means for determining the state of debris flow using the voice information received by the reception means, and
    A debris flow notification system characterized by having a notification means for notifying the user of the output of the determination means.
  2.  前記圧縮手段は、前記音声情報が所定の波形モデルに従うことを使って情報量を圧縮することを特徴とする請求項1に記載の土石流通報システム。 The debris flow reporting system according to claim 1, wherein the compression means compresses an amount of information by using the voice information according to a predetermined waveform model.
  3.  前記波形モデルが減衰を伴う自由振動の方程式であることを特徴とする請求項2に記載の土石流通報システム。 The debris flow reporting system according to claim 2, wherein the waveform model is an equation of free vibration accompanied by damping.
  4.  前記圧縮手段は、前記音声情報を圧縮する際に、前記音声情報と所定の音声モデルとの類似度を評価する評価手段を含み、前記無線伝送手段は前記評価手段の評価結果を伝送することを特徴とする請求項1に記載の土石流通報システム。 The compression means includes an evaluation means for evaluating the similarity between the voice information and a predetermined voice model when compressing the voice information, and the wireless transmission means transmits the evaluation result of the evaluation means. The earth and stone flow reporting system according to claim 1, which is characterized.
  5.  前記判断手段は、過去の土石の移動状況に関する情報を蓄積し学習する学習手段を搭載し、前記学習手段により前記判断手段の内部状態が更新されることを特徴とする請求項1に記載の土石流通報システム。 The debris flow according to claim 1, wherein the determination means is equipped with a learning means for accumulating and learning information on a past movement state of earth and stone, and the internal state of the determination means is updated by the learning means. Reporting system.
  6.  音声情報を取得する音声取得手段と、
     前記音声情報を圧縮して情報量を減らす圧縮手段と、
     前記圧縮手段により圧縮された前記音声情報を無線伝送する無線伝送手段とを持ち、
     前記圧縮手段は、前記音声情報が所定の波形モデルに従うことを使って情報量を圧縮することを特徴とする土石流センサ。
    Voice acquisition means to acquire voice information and
    A compression means that compresses the voice information to reduce the amount of information,
    It has a wireless transmission means for wirelessly transmitting the voice information compressed by the compression means.
    The compression means is a debris flow sensor characterized in that the amount of information is compressed by using the voice information according to a predetermined waveform model.
  7.  前記所定のモデルが減衰を伴う自由振動方程式であることを特徴とする請求項6に記載の土石流センサ。 The debris flow sensor according to claim 6, wherein the predetermined model is a free vibration equation with damping.
  8.  前記圧縮手段は、前記音声情報と前記所定の波形モデルの合致度を表す評価指標を出力し、前記無線伝送手段は前記評価指標を伝送することを特徴とする請求項6に記載の土石流センサ。 The debris flow sensor according to claim 6, wherein the compression means outputs an evaluation index indicating a degree of matching between the voice information and the predetermined waveform model, and the wireless transmission means transmits the evaluation index.
PCT/JP2020/011724 2019-03-19 2020-03-17 Debris flow notification system and debris flow sensor WO2020189679A1 (en)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
JP2019-051975 2019-03-19
JP2019051975A JP6777949B2 (en) 2019-03-19 2019-03-19 Debris flow notification system and debris flow sensor

Publications (1)

Publication Number Publication Date
WO2020189679A1 true WO2020189679A1 (en) 2020-09-24

Family

ID=72520904

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/JP2020/011724 WO2020189679A1 (en) 2019-03-19 2020-03-17 Debris flow notification system and debris flow sensor

Country Status (2)

Country Link
JP (1) JP6777949B2 (en)
WO (1) WO2020189679A1 (en)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113380004A (en) * 2021-06-02 2021-09-10 成都山地环安科技有限公司 Mountain disaster monitoring terminal, mountain disaster self-adaptive monitoring and early warning method and debris flow/landslide self-adaptive monitoring and early warning method
CN115110491A (en) * 2022-08-25 2022-09-27 四川省公路规划勘察设计研究院有限公司 Alarming device for debris flow monitoring and early warning
CN115273407A (en) * 2022-07-22 2022-11-01 四川大学 Remote monitoring alarm rod for depth of accumulated water in torrential rain and torrential flood in small watershed of mountainous area
CN116665422A (en) * 2023-05-29 2023-08-29 广西交通设计集团有限公司 Highway side slope falling stone risk monitoring and early warning system

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113393645B (en) * 2021-05-08 2022-09-16 杭州戬威科技有限公司 Mountain landslide ultrasonic monitoring and early warning system

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH11230792A (en) * 1998-02-18 1999-08-27 Mitsubishi Heavy Ind Ltd Detection system for avalanche of sand and stone
JPH11304550A (en) * 1998-04-27 1999-11-05 Mitsubishi Heavy Ind Ltd System for detecting avalanche of sand and stone
JP2003161782A (en) * 2001-11-26 2003-06-06 Geoscience Research Laboratory Rock destruction observing method and rock bed slope degradation monitor
JP2004186987A (en) * 2002-12-03 2004-07-02 Mitsubishi Electric Corp Remote monitor device and remote monitor system using the same

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH11230792A (en) * 1998-02-18 1999-08-27 Mitsubishi Heavy Ind Ltd Detection system for avalanche of sand and stone
JPH11304550A (en) * 1998-04-27 1999-11-05 Mitsubishi Heavy Ind Ltd System for detecting avalanche of sand and stone
JP2003161782A (en) * 2001-11-26 2003-06-06 Geoscience Research Laboratory Rock destruction observing method and rock bed slope degradation monitor
JP2004186987A (en) * 2002-12-03 2004-07-02 Mitsubishi Electric Corp Remote monitor device and remote monitor system using the same

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113380004A (en) * 2021-06-02 2021-09-10 成都山地环安科技有限公司 Mountain disaster monitoring terminal, mountain disaster self-adaptive monitoring and early warning method and debris flow/landslide self-adaptive monitoring and early warning method
CN115273407A (en) * 2022-07-22 2022-11-01 四川大学 Remote monitoring alarm rod for depth of accumulated water in torrential rain and torrential flood in small watershed of mountainous area
CN115273407B (en) * 2022-07-22 2023-10-24 四川大学 Remote monitoring alarm rod for mountain area small-river basin storm mountain torrent ponding depth
CN115110491A (en) * 2022-08-25 2022-09-27 四川省公路规划勘察设计研究院有限公司 Alarming device for debris flow monitoring and early warning
CN116665422A (en) * 2023-05-29 2023-08-29 广西交通设计集团有限公司 Highway side slope falling stone risk monitoring and early warning system
CN116665422B (en) * 2023-05-29 2024-03-29 广西交通设计集团有限公司 Highway side slope falling stone risk monitoring and early warning system

Also Published As

Publication number Publication date
JP6777949B2 (en) 2020-10-28
JP2020153793A (en) 2020-09-24

Similar Documents

Publication Publication Date Title
WO2020189679A1 (en) Debris flow notification system and debris flow sensor
Gonzalez et al. Deep-ocean assessment and reporting of tsunamis (DART): Brief overview and status report
US9348010B1 (en) Acoustic survey methods in weapons location system
Munawar Flood disaster management: Risks, technologies, and future directions
US8077036B2 (en) Systems and methods for security breach detection
KR101983938B1 (en) System and method for warning lightning
CN104408915A (en) Traffic state parameter estimation method and system
US20150362465A1 (en) System for monitoring and/or surveying conduits
EP2955493A1 (en) System for monitoring and/or surveying conduits
US6232882B1 (en) Warning system and method for detection of tornadoes
CN111024043A (en) Satellite remote sensing method based on sensor triggering
Khan et al. Artificial Intelligence based Multi-modal sensing for flash flood investigation
US20080021657A1 (en) Utilizing rapid water displacement detection systems and satellite imagery data to predict tsunamis
CN116129612A (en) Distributed landslide monitoring and early warning system based on Internet of things
Khan et al. Prior investigation for flash floods and hurricanes, concise capsulization of hydrological technologies and instrumentation: A survey
CN106932028A (en) A kind of Intelligent Iron Tower safety detection device
US20230143027A1 (en) System for real-time recognition and identification of sound sources
Lincoln et al. A Preliminary Look at Using Rainfall Average Recurrence Interval to Characterize Flash Flood Events for Real-time Warning Forecasting.
JP4994769B2 (en) Radar equipment
JP2021092530A (en) Water level measurement device and water level prediction system
JP2008249486A (en) Real-time earthquake motion distribution estimating method using s-wave
Comey et al. Recent studies using infrasound sensors to remotely monitor avalanche activity
WO2023004182A9 (en) Location determination of deployed fiber cables using distributed fiber optic sensing
Pirenne et al. The web-enabled awareness research network (WARN) project early earthquake and tsunami warning at Ocean Networks Canada
Hendrikx et al. Evaluation of an infrasound detection system for avalanches in Rogers Pass, Canada

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20774237

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 20774237

Country of ref document: EP

Kind code of ref document: A1