JP2006170739A - Earthquake disaster prevention system using urgent earthquake prompt report - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To output a report on an earthquake and a control signal prior to the arrival of a main movement of the earthquake by using an urgent earthquake prompt report, to perform not only the prompt report but also high-accuracy evaluation in consideration of a predominant period of the ground and the vibration characteristics of a building by using a neural network, and to dramatically enhance the reliability of prediction by optimizing the neural network by learning, as to earthquake disaster prevention for facilities, buildings, etc. <P>SOLUTION: The quake or the level of damage at a target point is estimated prior to the arrival of the main movement of the earthquake by using the neural network that adopts, as input parameters, not only the magnitude at the seismic center, the position (longitude and latitude) of the seismic center, the depth of the seismic center, and the arrival time and size of earthquake motion, acquired from the urgent earthquake prompt report, but also dynamic characteristics such as the predominant period of the ground at an object point previously acquired by survey and, as to the interior of a building, the proper period, damping stories, etc. of the building, to perform reporting for earthquake disaster prevention or controlling on installations/equipment. Optimization is achieved by thereto adding a self-learning function. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention is an earthquake disaster prevention system installed in a facility or building, etc., using emergency earthquake bulletin, outputs effective information for taking measures for disaster prevention before the arrival of earthquake motion, The present invention relates to an earthquake disaster prevention system used for a control device for equipment.

  Conventional earthquake notification and control devices installed in facilities and buildings, in combination with seismometers, are often used to provide notification and control when seismometers of a certain level or higher are detected. is there. For example, in an elevator, when an acceleration of 80 Gal or higher is observed with a normal seismic device, the elevator is stopped at the nearest floor.

  Recently, an earthquake early warning (see Patent Document 1) has been transmitted from the Japan Meteorological Agency on a trial basis, and attempts to use it for notification and control have been made in various fields. The Earthquake Early Warning is earthquake information that is transmitted before the main motion (large shaking) arrives using information captured by the seismometer near the epicenter, and will be used in the near future. The earthquake disaster prevention notification and control equipment that uses emergency earthquake warnings to date estimates the arrival time and magnitude of the main motion using the established empirical relationship between the epicenter and observation points, and uses the results. However, due to the crustal structure in which seismic waves propagate from the epicenter to the target point, as well as the local surface strata and topography of the target point, errors cannot be estimated and the duration of the earthquake cannot be estimated. Furthermore, when the target location is inside a building, the shaking varies depending on the structure of the building and the number of floors. Since there was no examination of accuracy for using the earthquake early warning and evaluation and correction of actual earthquake motion, there was a limit to using it for more appropriate disaster prevention.

Moreover, there exists patent document 2 as a prior art document relevant to this invention. This invention is an early earthquake detection system with a self-learning function using a neural network, and the neural network is applied to the evaluation of magnitude, epicenter distance, and epicenter depth, which are the epicenter parameters performed in the observation point seismometer. To do. In addition, an earthquake is detected at one detection point, and the epicenter parameter is evaluated on the spot. Furthermore, the earthquake information is reported directly to the end user from the earthquake detection device.
JP 2002-277557 A JP 11-64533 A

  (1) The conventional earthquake alarm / control device triggers at the vibration level observed by a seismometer such as a seismometer installed inside the building. There is a possibility that damage cannot be prevented.

  (2) If earthquake early warning is used for earthquake notification and control equipment, earthquake information can be obtained before it starts to shake greatly, so it is possible to prevent damage more efficiently by taking measures in advance. However, if the information is used as it is, the magnitude of the shaking is processed uniformly based on empirical rules, so it becomes an estimated value on the average ground, and depending on the location, it is affected by the amplification characteristics of the local ground. There is a risk of errors. Furthermore, in a building, an accurate shake cannot be estimated under the influence of the response of the building, and the error increases.

  (3) If the evaluation error is large, a trigger for notification and control must be applied at a low vibration level on the safe side, and it will operate even in an earthquake that does not cause damage, so it is not efficient.

  The present invention has been made to solve such problems. The purpose of the present invention is to use an earthquake early warning in earthquake disaster prevention of facilities and buildings, etc., so that an earthquake notification can be made before the main motion of the earthquake arrives. And control signals can be output, and by using a neural network, it is possible to evaluate not only the earthquake early warning, but also the ground period and the vibration characteristics of the building with high accuracy. By optimizing, it is possible to dramatically improve the reliability of prediction, and to provide an earthquake disaster prevention system using emergency earthquake warning that can be effectively notified and controlled in the event of an earthquake.

  The invention according to claim 1 of the present invention is an earthquake disaster prevention system installed at a target point for earthquake disaster prevention, using a neural network having information from an emergency earthquake warning and information specific to the target point as input parameters. An earthquake disaster prevention system using earthquake early warning, which estimates the shaking or damage level of the target point before the main earthquake motion arrives, and controls earthquake disaster prevention or equipment. It is.

  In the present invention, in order to improve the evaluation accuracy of earthquake disaster prevention notification and control devices installed in facilities and buildings, not only emergency earthquake bulletins but also other information (information specific to the target point) is input. Is. That is, not only the magnitude of the epicenter obtained from the earthquake early warning, the location of the epicenter (latitude and longitude), the depth of the epicenter, the arrival time and the magnitude of the ground motion, but also the prevailing period of the ground at the target point obtained in advance. In addition, the building dynamic characteristics such as the natural period of the building, the natural period of the building, the attenuation rank, etc. are used as the input parameters of the neural network.

  The earthquake disaster prevention system of the present invention performs, as shown in FIG. 1, for example, an immediate earthquake information transmission device as an information processing device, an emergency earthquake warning input device such as a satellite antenna, and notification and control for earthquake disaster prevention. It can consist of output devices. The operation and storage device of the immediate earthquake information transmission device has an emergency earthquake warning data processing function, an evaluation input data setting function, a neural network evaluation function, and a neural network setting function. Using a neural network with the information and information specific to the target point as input parameters, before the main motion of the earthquake arrives, the shake or damage level of the target point is estimated, and the earthquake notification device or the control device for equipment To use.

  This earthquake notification device or control device is, for example, elevator or escalator control, industrial machine control, transportation stop, play facility stop, evacuation guidance for people in buildings (public facilities, schools, event venues, etc.) Used to open and close doors, gas and water valves.

  A neural network is an analysis tool that models a human nerve cell, and can be configured with a simple system. Therefore, the neural network is actively used for control in recent engineering fields. By using a neural network, there are a function for evaluating many parameters as input and a function for optimizing the network by a learning function.

  As an estimation system of epicenter information (seismic source position, epicenter depth, magnitude) using a neural network, “Earthquake early detection system with self-learning function by neural network” (Japanese Patent Laid-Open No. 11-64533 (Patent Document 2)) The present invention has been applied to the estimation of the vibration of the target point and the damage level.

  The invention according to claim 2 of the present invention is the earthquake disaster prevention system according to claim 1, which has a self-learning function of a neural network, and uses an earthquake early warning.

  For example, as shown in FIG. 2, this is a case where a neural network self-learning function and a learning data setting function are added to the arithmetic storage device of the immediate earthquake information transmission device. The neural network used for the evaluation is optimized by learning from the earthquake source information and past shaking data (for example, seismic intensity data and earthquake records taken nearby) by the neural network.

  The invention according to claim 3 of the present invention is the earthquake disaster prevention system according to claim 2, wherein seismometer data installed at a target point is used as learning data. System.

  For example, as shown in FIG. 2, a seismometer can be installed, and a seismic data storage device is added, and seismometer data is used as learning data. The seismometers are used together with the seismic source information published by the Japan Meteorological Agency to learn and optimize the neural network to improve accuracy.

  As a learning method, there are a method of automatically performing each occurrence of an earthquake and a method of performing manually after several earthquake data are accumulated.

  According to the present invention as described above, by using the earthquake early warning, it is possible to output an earthquake notification and a control signal before the main motion of the earthquake arrives. It is possible to estimate the ground motion level and damage level in consideration of not only the information from but also the information specific to the target location, and the evaluation accuracy is improved. Furthermore, if observation data can be accumulated by introducing a self-learning function, the accuracy can be further improved. As a result, notification and control signals for earthquake disaster prevention can be issued only in the case of a truly necessary earthquake, which is efficient.

  Since the present invention is configured as described above, the following effects can be obtained.

  (1) Because the earthquake early warning is used, it is possible to output earthquake notifications and control signals before the main motion of the earthquake arrives, and by using a neural network, not only the earthquake early warning but also the ground It is possible to evaluate with high accuracy in consideration of the dominant cycle and vibration characteristics of buildings. As a result, it is possible to output an accurate notification and control signal before the main motion of the earthquake reaches a facility or a building, and it is possible to reliably prevent damage caused by the earthquake.

  (2) By optimizing the neural network by learning, it can be applied to various places with this device alone, even without the attached ones. In other words, if data is accumulated, learning can be performed and the reliability of prediction can be dramatically improved, and efficient notification and control can be performed during an earthquake.

  (3) Furthermore, it is possible to set various notifications and triggers for control signals, and an extremely general earthquake disaster prevention system can be obtained.

  Hereinafter, the present invention will be described based on the illustrated embodiment. FIG. 1 is a block diagram showing a basic configuration of an immediate earthquake information transmission device of the earthquake disaster prevention system of the present invention. FIG. 2 is a block diagram showing a configuration in which a self-learning function and a seismometer are added to the basic configuration of FIG. FIG. 3 is a flow showing an evaluation method using the neural network of the present invention. FIG. 4 is a block diagram illustrating a specific example of the notification device and the control device of the present invention.

  As shown in FIG. 1, the immediate earthquake information transmission device 1 of the present invention estimates the magnitude and damage level of a target point using a neural network before the main motion arrives by using an emergency earthquake bulletin. It is a device that outputs information and control signals for disaster prevention, and comprises an arithmetic storage device 2, input / output interface units 3 and 4, a tuner 5, etc., which are in a box (set top box) of the device 1. Built in.

  A satellite antenna 10, the Internet, a normal telephone line, a mobile phone 11 and the like are connected to the input side of such an immediate earthquake information transmission device 1, and a display device 20, a notification device 21, and a control device are connected to the output side. 22, a communication cable 23 of a LAN (Ethernet (registered trademark), etc.) is connected, and thereby an earthquake disaster prevention system is configured. As for the output side, the device 1 and the display device 20 have the minimum device configuration, and the others are appropriately selected and connected.

  The earthquake early warning is received by the satellite antenna 10, a signal from the satellite antenna 10 is converted by the tuner 5, and is input to the arithmetic storage device 2 via the interface unit 3. In addition, signals from the Internet 11, a normal telephone line, a mobile phone 11 and the like are input via the interface unit 3. Here, when the earthquake early warning is obtained via the Internet or a normal telephone line, the information acquisition time depends on how the line is congested, and there is a possibility that the information cannot be obtained before the main motion arrives. If a notification or control signal is required before the arrival of the main motion, it is connected to a dedicated telephone line or satellite antenna 10 and various input / output correspondences are made so that emergency earthquake information can be obtained.

  The arithmetic storage device 2 has an emergency earthquake warning data processing function 30, an evaluation input data setting function 31, a neural network evaluation function 32, and a neural network setting function 33.

  As shown in Fig. 1 and Fig. 3, not only the magnitude obtained from the earthquake early warning, the location of the epicenter (latitude / longitude), the depth of the epicenter, the arrival time and magnitude of the earthquake motion, but also the object obtained by conducting a preliminary survey For the purpose of earthquake disaster prevention, the prevailing period of the ground at the point, and if it is inside the building, the dynamic characteristics such as the natural period of the building and the attenuation rank are used as input parameters and the neural network is used to estimate the shaking and damage level of the target point. Output for control of equipment and equipment. The display device 20 displays the earthquake early warning reception data and the estimation result by the neural network.

  The neural network is set from the outside through an interface. Learning for setting the neural network is performed manually outside so that the setting can be changed as needed. Neural network input data that takes into account ground and building information is also provided from the outside via an interface.

  In the embodiment of FIG. 2, a neural network self-learning function 34 and a learning data setting function 35 are added to the arithmetic storage device 2. Further, when the seismometer 12 is installed, a seismic data storage device 6 for storing records observed by the seismometer 12 is provided inside the device 1.

  The neural network used for the evaluation is optimized by learning using the Japan Meteorological Agency's published hypocenter information (seismic source location, depth, magnitude), earthquake motions of the target site or nearby ground, and building shaking data for past earthquakes. Shall. The observed data or the data obtained by simulation analysis is used for the ground motion and building shaking data.

  Neural network learning is based on a manual method when several earthquake data are accumulated, and it can be added automatically as an option to automatically update the neural network after every earthquake and update the neural network. Yes. However, in order to perform after each occurrence of an earthquake, it is necessary to input records observed with a seismometer as learning data. As a manual learning method, both on-site (local) PC connection and remote connection via the Internet are possible.

  When the seismometer 12 is installed, learning data for the neural network is created by analyzing the records stored in the seismic wave data storage device 6. Based on the data, self-learning is performed after an earthquake or periodically, and the neural network is updated to an optimal one. This is done automatically.

  As shown in FIG. 4, the notification device 21 is an alarm sound generation device, a display / indicator device, illumination (flash), a vibration device, or the like. That is, a warning is issued from the broadcasting device immediately before the shaking comes, and the danger is notified to the person by sound or voice. Similarly, use vibration, lighting (flash), and display devices to signal danger. Examples of installation locations include public facilities, schools, event venues, disaster prevention managers and executives.

  The control device 22 is a disaster prevention device, a disaster prevention system, or the like. For example, interlocking with an elevator control device, decelerating / stopping before swinging, and interlocking with a building management device (door unlocking / locking / opening / closing, shutter opening / closing, power on / off, general lighting / emergency lighting on / off, emergency Power up, gas / water valve open / close). IT equipment protection (storage protection circuit activation, network protection circuit activation, power on / off), traffic regulation, amusement facility deceleration / stop, emergency line reservation, sea sluice gate opening / closing, public institutions Triggers for switching to emergency (disaster prevention related organizations, fire departments), start-up and operation control of various sensors, start up emergency mode of disaster prevention radio equipment, control of city gas governor solenoid valve, emergency of dam facilities Establishment of a system.

It is a block diagram which shows the basic composition of the immediate earthquake information transmission apparatus of the earthquake disaster prevention system of this invention. It is a block diagram which shows the structure which added the self-learning function and the seismometer to the basic structure of FIG. It is a flow which shows the evaluation method using the neural network of this invention. It is the block diagram which illustrated the specific example of the alerting | reporting apparatus and control apparatus of this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 ... Immediate earthquake information transmission device 2 ... Operation storage device 3 ... Interface part 4 ... Interface part 5 ... Tuner 6 ... Earthquake wave data storage device 10 ... Satellite antenna 11 ... Internet, normal telephone line, mobile phone Etc. 12 ... Seismometer 20 ... Display device 21 ... Notification device 22 ... Control device 23 ... LAN
30 ... Earthquake early warning data processing function 31 ... Evaluation input data setting function 32 ... Neural network evaluation function 33 ... Neural network setting function 34 ... Neural network self-learning function 35 ... Learning data setting function

Claims (3)

An earthquake disaster prevention system installed at a target site for earthquake disaster prevention,
Using the neural network with the information from the emergency earthquake bulletin and the information specific to the target point as input parameters, before the main motion of the earthquake arrives, the level of shaking or damage of the target point is estimated and An earthquake disaster prevention system using earthquake early warning, which is characterized by reporting or controlling equipment.   The earthquake disaster prevention system according to claim 1, wherein the earthquake disaster prevention system uses an emergency earthquake bulletin having a self-learning function of a neural network. 3. The earthquake disaster prevention system according to claim 2, wherein data of a seismometer installed at a target point is used as learning data.

Images