Seismic intensity estimation system

Abstract

This invention provides a seismic intensity estimation system that can improve the accuracy of estimation when estimating the seismic intensity class based on the acceleration waveform obtained by the acceleration sensor of a smart device, while the smart device is placed on a stand on the floor. [Solution] The seismic intensity estimation system 1 includes: an acceleration waveform collection unit 23 that collects acceleration waveforms from each of a plurality of smart devices 10 during an earthquake; an acceleration waveform adjustment unit 24 that rotates the acceleration waveforms of the plurality of smart devices 10 so that their orientations in the horizontal plane coincide with each other to generate orientation-adjusted acceleration waveforms; a frame vibration characteristic derivation unit 25 that derives the vibration characteristics of at least one frame based on the orientation-adjusted acceleration waveforms of at least one pair of smart devices 10; and a floor seismic intensity class estimation unit 26 that derives the Fourier spectrum of the floor based on the acceleration waveform on the frame and the vibration characteristics of the frame to estimate the seismic intensity class of the floor.

Description

[Technical Field]

[0001] This invention relates to a seismic intensity estimation system using a smart device with a built-in acceleration sensor. [Background technology]

[0002] When an earthquake occurs, the seismic intensity is estimated at each of multiple locations located over a wide area. For example, Patent Document 1 discloses a seismic intensity information providing device comprising: a storage unit for storing multiple tile image files; an earthquake information table for storing earthquake ID type codes, basic earthquake data, and storage locations of tile image files; an estimated seismic intensity distribution table for storing earthquake ID type codes, measured seismic intensity, and data representing the area where the measured seismic intensity is estimated; an observation point seismic intensity table for storing earthquake ID type codes, measured seismic intensity, and data representing the observation point of the measured seismic intensity; a tile image file generation unit for generating multiple tile image files and expanding them into the storage unit when the estimated seismic intensity distribution table or the observation point seismic intensity table is updated; an earthquake search unit for outputting earthquake ID type codes and data representing storage locations; and a seismic intensity search unit for outputting earthquake ID type codes and data representing measured seismic intensity. In Patent Document 1, the tile images are colored with a color representing the estimated seismic intensity value for the range of each tile image. If the range of the tile image includes an observation point where seismic intensity was observed, a figure containing a number representing the seismic intensity value is drawn at the position corresponding to that observation point. The tile images representing seismic intensity are displayed on the user terminal overlaid on the map image.

[0003] When an earthquake occurs, for example, to estimate the extent of damage to a building, it is desirable to obtain the seismic intensity at the location where the building is located with greater accuracy. In this regard, in the configuration of Patent Document 1, even if the location where a building is located is input, the seismic intensity of the entire large mesh including that location is obtained, rather than the seismic intensity of that specific location. Therefore, in the configuration of Patent Document 1, a value with sufficiently high accuracy cannot be obtained as the seismic intensity used to estimate the extent of damage to a building. In response to this, one might consider installing many seismometers at a high density to accurately estimate seismic intensity at any given location. However, installing many seismometers is costly. Therefore, it is being considered to increase the density of earthquake observation by using acceleration sensors, which are equipped in smart devices such as smartphones and tablet terminals that are owned by many people, as a substitute for seismometers, and by using smart devices to detect earthquakes.

[0004] For example, Patent Document 2 discloses an evaluation system for evaluating earthquake shaking at an evaluation target. This evaluation system includes an acquisition means for acquiring first information indicating earthquake shaking at the evaluation target, which is identified by a terminal device installed at the evaluation target, and second information indicating earthquake shaking around the evaluation target, which is identified by terminal devices installed around the evaluation target; and an evaluation means for performing an evaluation of the earthquake shaking at the evaluation target based on the first and second information acquired by the acquisition means. The terminal device is, for example, a smartphone or a tablet device. Furthermore, Patent Document 3 discloses a configuration comprising: a number of altitude alarms arranged in the vertical direction within a multi-story structure to notify nearby mobile terminals of their respective altitude positions; a number of smartphones equipped with sensors to detect their own vibrations and a location information acquisition unit to acquire their current location and altitude; an observation control unit that detects the orientation of the smartphones and initiates vibration detection by the sensors based on the duration of that orientation; an information collection unit that acquires the number of smartphones that have started detecting vibrations and their respective current locations, and collects the detection results from the sensors of each mobile terminal; and a determination unit that determines, based on the detection results collected by the information collection unit, that the vibrations are caused by an earthquake when the number of smartphones that have detected vibrations exceeds a predetermined number.

[0005] Incidentally, in order to accurately grasp the extent of damage on each floor of a building, it is essential to estimate the seismic intensity level at each floor. On the other hand, smart devices are not usually placed directly on the floor, but rather on stands such as desks or tables that are set up on the floor. The acceleration waveform measured by smart devices in this state may include the effect of vibrations of the stand. Therefore, the seismic intensity level estimated based on such acceleration waveforms may contain errors and may differ from the seismic intensity at the floor. When a smart device is placed on a stand on the floor, there is a need to improve the accuracy of the estimation when estimating the seismic intensity scale based on the acceleration waveform obtained by the smart device's acceleration sensor. [Prior art documents] [Patent Documents]

[0006] [Patent Document 1] Japanese Patent Publication No. 2017-133996 [Patent Document 2] Japanese Patent Publication No. 2020-193935 [Patent Document 3] International Publication No. 2018 / 174296 [Overview of the Initiative] [Problems that the invention aims to solve]

[0007] The problem that this invention aims to solve is to provide a seismic intensity estimation system that can improve the estimation accuracy when estimating the seismic intensity class based on the acceleration waveform obtained by the acceleration sensor of a smart device when the smart device is placed on a stand on the floor. [Means for solving the problem]

[0008] The inventors developed a seismic intensity estimation system using a smart device (with a built-in acceleration sensor). The smart device is placed on a platform inside a building to acquire acceleration waveforms. The collected acceleration waveforms are then oriented to match their orientation in the horizontal plane. Based on the oriented acceleration waveforms and the vibration characteristics of the platform, the Fourier spectrum of the floor is derived, and the seismic intensity class of the floor is estimated. By using the acceleration sensor in the smart device as a substitute for a seismometer, it becomes possible to collect and analyze earthquake data with high accuracy, and the seismic intensity class of the floor can be accurately estimated at multiple locations. This led to the present invention. To solve the above problems, the present invention employs the following means. That is, the present invention is a seismic intensity estimation system using smart devices with built-in earthquake information acquisition sensors, wherein a plurality of frames are provided on the same floor, each of the plurality of smart devices is placed on one of the frames, the earthquake information acquisition sensor is equipped with an acceleration sensor, and the present invention includes an acceleration waveform collection unit that collects acceleration waveforms acquired by the acceleration sensor during an earthquake from each of the plurality of smart devices, and an acceleration waveform adjustment unit that rotates the acceleration waveforms of the plurality of smart devices so that their orientations in the horizontal plane coincide with each other to generate an orientation-adjusted acceleration waveform. The present invention provides a seismic intensity estimation system comprising: a frame vibration characteristic derivation unit that derives the vibration characteristics of at least one frame based on the orientation-adjusted acceleration waveforms (for example, orientation-adjusted acceleration waveform 1 and orientation-adjusted acceleration waveform 2) of each of the pair of smart devices among a plurality of smart devices; and a floor seismic intensity class estimation unit that derives the Fourier spectrum of the floor based on the acceleration waveforms of the smart devices placed on the frame from which the vibration characteristics have been derived, and the vibration characteristics of the frame, and estimates the seismic intensity class of the floor. With the above configuration, multiple frames are provided on the same floor, and each of the multiple smart devices is placed on one of the frames. Acceleration waveforms acquired by acceleration sensors during an earthquake are collected from each of the multiple smart devices. Next, the acceleration waveforms of the multiple smart devices are rotated so that their orientations in the horizontal plane coincide with each other to generate orientation-adjusted acceleration waveforms. Then, for at least one pair of the multiple smart devices, the vibration characteristics of at least one frame are derived based on the orientation-adjusted acceleration waveforms of each of the smart devices in that pair (for example, orientation-adjusted acceleration waveform 1 and orientation-adjusted acceleration waveform 2). For example, if a pair of smart devices includes one smart device and another smart device, the Fourier spectrum of the orientation-adjusted acceleration waveform of one smart device (e.g., orientation-adjusted acceleration waveform 1) can be expressed as the product of the Fourier spectrum of the floor acceleration waveform and the transfer function of the platform on which one smart device is mounted. Similarly, the Fourier spectrum of the orientation-adjusted acceleration waveform of the other smart device (e.g., orientation-adjusted acceleration waveform 2) can be expressed as the product of the Fourier spectrum of the floor acceleration waveform and the transfer function of the platform on which the other smart device is mounted. Therefore, by dividing the Fourier spectrum of the orientation-adjusted acceleration waveform of one smart device (e.g., orientation-adjusted acceleration waveform 1) by the Fourier spectrum of the orientation-adjusted acceleration waveform of the other smart device (e.g., orientation-adjusted acceleration waveform 2), the ratio of the transfer functions between the platform on which one smart device is mounted and the platform on which the other smart device is mounted can be calculated. By extracting the peak frequency within this ratio of transfer functions, it is possible to derive the natural frequency of a mounting frame on which a single smart device is mounted, as the vibration characteristic of the mounting frame on which a single smart device is mounted. As described above, if the vibration characteristics, such as the natural frequency, of at least one mounting structure (for example, a mounting structure on which one smart device is placed) are known, then, for example, by considering the response of the mounting structure as the response of a single-mass system, calculating the transfer function of the single-mass system based on the vibration characteristics of the mounting structure, and then dividing the Fourier spectrum of the acceleration waveform of the smart device by the transfer function of the single-mass system, the Fourier spectrum of the floor can be derived. The Fourier spectrum of the floor derived in this way excludes the influence of the mounting structure's vibration. Therefore, using such a Fourier spectrum of the floor, the seismic intensity class of the floor can be estimated with high accuracy. In this way, when the smart device is placed on a stand on the floor, it becomes possible to improve the accuracy of the estimation when estimating the seismic intensity class based on the acceleration waveform obtained by the smart device's acceleration sensor.

[0009] In one embodiment of the present invention, the vibration characteristic is a natural frequency, and the mount vibration characteristic derivation unit calculates the ratio of the transfer functions between the mount on which the one smart device is mounted and the mount on which the other smart device is mounted by dividing the Fourier spectrum of the azimuth-adjusted acceleration waveform (e.g., azimuth-adjusted acceleration waveform 1) of the one smart device by the Fourier spectrum of the azimuth-adjusted acceleration waveform (e.g., azimuth-adjusted acceleration waveform 2) of the other smart device, and sets the frequency that is the peak value in the ratio of the transfer functions as the natural frequency of the mount on which the one smart device is mounted. With the configuration described above, the ratio of the transfer functions between the rigging platform on which one smart device is mounted and the rigging platform on which the other smart devices are mounted is calculated by dividing the Fourier spectrum of the azimuth-adjusted acceleration waveform of one smart device (e.g., azimuth-adjusted acceleration waveform 1) by the Fourier spectrum of the azimuth-adjusted acceleration waveform of another smart device (e.g., azimuth-adjusted acceleration waveform 2). By extracting the peak frequency from this ratio of transfer functions, the natural frequency of the rigging platform on which one smart device is mounted can be derived. Using the natural frequency of the rigging platform derived in this way, the seismic intensity class of the floor can be estimated with high accuracy, as already explained.

[0010] In another embodiment of the present invention, the vibration characteristics are natural frequencies, and the floor seismic intensity class estimation unit considers the response by the frame to be the response by a single-mass system, calculates the transfer function of the single-mass system based on the natural frequency of the frame, and derives the Fourier spectrum of the floor by dividing the Fourier spectrum of the acceleration waveform of the smart device placed on the frame from which the vibration characteristics were derived by the transfer function of the single-mass system. With the above configuration, if the vibration characteristics such as the natural frequency of the support structure are known, the response of the support structure can be considered as the response of a single-mass system. Based on the vibration characteristics of the support structure, the transfer function of the single-mass system can be calculated, and the Fourier spectrum of the floor can be derived by processing such as dividing the Fourier spectrum of the acceleration waveform of the smart device placed on the support structure (from which the vibration characteristics have been derived) by the transfer function of the single-mass system. The Fourier spectrum of the floor derived in this way excludes the influence of the support structure's vibration. Therefore, using such a Fourier spectrum of the floor, the seismic intensity class of the floor can be estimated with high accuracy.

[0011] In another aspect of the present invention, the present invention includes: a sliding section identification unit that identifies a sliding section, which is a time interval when the smart device slides on the platform, based on the measurement results from the earthquake information acquisition sensor; a sliding acceleration threshold calculation unit that generates an envelope for the sliding section waveform, which is the waveform of the acceleration waveform of the smart device corresponding to the sliding section, and calculates the mode of the values ​​on the envelope as the sliding acceleration threshold, which is the threshold of the acceleration at which the smart device slides on the platform; and an acceleration waveform correction unit that corrects the portion of the acceleration waveform whose absolute value is greater than the sliding acceleration threshold to generate a corrected acceleration waveform, wherein the acceleration waveform adjustment unit uses the corrected acceleration waveform generated for the smart device as the acceleration waveform of the smart device to generate the orientation-adjusted acceleration waveform. A smart device that is stationary and placed on a stand may slide against the mounting surface during an earthquake if the earthquake is strong enough to exert a force on the smart device that exceeds the static frictional force between the smart device and the mounting surface on which it is placed. In this state, a portion of the acceleration waveform will appear where the amplitude becomes constant and plateaus, resulting in a loss of acceleration data in this portion. In contrast, with the configuration described above, the sliding section identification unit first identifies the sliding section, which is the time interval when the smart device slides on the platform, based on the measurement results from the earthquake information acquisition sensor. Next, the sliding acceleration threshold calculation unit generates an envelope for the sliding section waveform, which is the waveform of the smart device's acceleration waveform that corresponds to the sliding section identified as described above. Such an envelope generally takes the shape of connecting the parts of the sliding section waveform that have maximum values. Here, when the smart device slides on the platform and a part occurs in the acceleration waveform where the amplitude becomes constant and plateaus, it is considered highly likely that the acceleration value of this plateauing part will be the mode of the frequency distribution when values ​​are acquired on this envelope at regular time intervals and a frequency distribution is created from the acquired values. Therefore, if the mode of the values ​​on the envelope generated as described above is calculated, that value can be considered to be the acceleration value of the part where sliding occurs and the amplitude of the acceleration waveform becomes constant and plateaus. Based on this idea, the sliding acceleration threshold calculation unit calculates the mode of the values ​​on the envelope as the sliding acceleration threshold, which is the threshold for the acceleration at which the smart device slides on the platform. Then, the acceleration waveform correction unit corrects the portion of the acceleration waveform whose absolute value is greater than the sliding acceleration threshold, and generates a corrected acceleration waveform. With the above configuration, the sliding acceleration threshold, which is the threshold for the acceleration at which the smart device slides on the platform, is set with high accuracy, so that even when the smart device slides, the acceleration waveform correction unit can accurately estimate and generate the acceleration waveform of a non-sliding state. In this way, the accurately estimated and generated corrected acceleration waveform is used as the acceleration waveform of the smart device, and subsequent processing, including the generation of the azimuth-adjusted acceleration waveform, is performed. Therefore, even when the smart device slides, it becomes possible to further improve the accuracy of seismic intensity scale estimation. [Effects of the Invention]

[0012] According to the present invention, when a smart device is placed on a stand on the floor, it is possible to provide a seismic intensity estimation system that can further improve the estimation accuracy when estimating the seismic intensity class based on the acceleration waveform obtained by the acceleration sensor of a smart device. [Brief explanation of the drawing]

[0013] [Figure 1] This is an explanatory diagram showing a state in which multiple stands are installed on the same floor, and each of the multiple smart devices is placed on one of the stands. [Figure 2] This table shows the relationship between seismic waves, the seismic intensity scale estimated based on acceleration waveforms acquired on the floor for those seismic waves, and the seismic intensity scale estimated based on acceleration waveforms acquired on a platform installed on the floor. [Figure 3] This graph shows the characteristics of the seismic intensity filter used when calculating the seismic intensity scale. [Figure 4] This graph shows the Fourier spectra of the acceleration waveform acquired on the floor and the acceleration waveform acquired on the support structure, for a Kumamoto wave normalized to 400 gal, in each of two orthogonal directions in the horizontal plane. [Figure 5] This graph shows the Fourier spectra of the acceleration waveform acquired on the floor and the acceleration waveform acquired on the support structure, for a Tsukidate wave normalized to 200 gal, in each of two orthogonal directions in the horizontal plane. [Figure 6] This graph shows the Fourier spectra of the acceleration waveform acquired on the floor and the acceleration waveform acquired on the support structure, for the Tsukidate wave normalized at 400 gal, in each of two orthogonal directions in the horizontal plane. [Figure 7] This is a block diagram of a seismic intensity estimation system according to an embodiment of the present invention. [Figure 8] This is an explanatory diagram of a smart device. [Figure 9] This is an explanatory diagram showing multiple smart devices placed facing different directions. [Figure 10] Figure 9 is an explanatory diagram showing the state in which the orientations of the acceleration waveforms acquired by each smart device are adjusted to match each other. [Figure 11] This table shows the natural frequencies of the mounting frame used in the verification of the above embodiment, and the orientation of the smart device installed on the mounting frame. [Figure 12] This graph shows the result of calculating the ratio of transfer functions by dividing the Fourier spectrum of the azimuth-adjusted acceleration waveform of smart device 10-1 in Figure 11 by the Fourier spectrum of the azimuth-adjusted acceleration waveform of other smart devices, and then calculating the average of these ratios among the other smart devices. [Figure 13] This graph shows the results of applying the above embodiment to the Fourier spectrum of the acceleration waveform acquired on the mounting platform for the Tsukidate wave normalized at 200 gal. [Figure 14] This table shows the seismic intensity class estimated by applying each of the above-described processes to the acceleration waveforms acquired on the platform for each case shown in Figure 2, with the values ​​added to Figure 2. [Figure 15] This is a flowchart of the seismic intensity estimation method using the above seismic intensity estimation system. [Figure 16] (a) is the horizontal component of the acceleration waveform acquired by a smart device fixed on a pedestal, (b) is an illustrative diagram of the horizontal component of the acceleration waveform that may be acquired when a smart device not fixed on a pedestal slides relative to the pedestal, and (c) is an example of the horizontal component of the acceleration waveform actually acquired when a smart device not fixed on a pedestal slides relative to the pedestal. [Figure 17] This is a block diagram of a seismic intensity estimation system according to a modified embodiment described above. [Figure 18] In the event of a certain earthquake, (a) is the horizontal component of the acceleration waveform, and (b) is the horizontal component of the time history waveform of the magnetic flux density acquired by a magnetic sensor. [Figure 19]In the case of the same earthquake as in Figure 18, (a) is the vertical component of the acceleration waveform, and (b) is the vertical component of the time history waveform of the magnetic flux density acquired by the magnetic sensor. [Figure 20] This graph shows an example of a waveform u(t). [Figure 21] This is a graph showing the absolute value |u(t)| of the waveform u(t) in Figure 20. [Figure 22] This is an explanatory diagram regarding the Hilbert transformation. [Figure 23] This graph shows the Hilbert transform pairs uH(t) of the waveform u(t) superimposed on Figure 20. [Figure 24] Figure 21 is a graph that shows the absolute value |uH(t)| of the Hilbert transform pair uH(t) of the waveform u(t) and the absolute value |a(t)| of the analyzed signal a(t) superimposed on each other. [Figure 25] This is an explanatory diagram showing an example of acceleration waveforms when an earthquake occurs and a smart device is in a sliding state, and an example of a corrected acceleration waveform after applying corrections to it. [Figure 26] This is a flowchart of the seismic intensity estimation method using the seismic intensity estimation system in the modified example described above. [Figure 27] This figure shows the first example of calculating the sliding acceleration threshold. [Figure 28] This figure shows a second example of calculating the sliding acceleration threshold. [Figure 29] This figure shows a third example of calculating the sliding acceleration threshold. [Modes for carrying out the invention]

[0014] The present invention relates to a seismic intensity estimation system using smart devices with built-in acceleration sensors, comprising: a platform vibration characteristic derivation unit that places multiple smart devices on a platform, collects acceleration waveforms during an earthquake, performs orientation adjustment on the observed acceleration waveforms, and then derives the vibration characteristics of the platform; and a floor seismic intensity class estimation unit that derives the Fourier spectrum of the floor based on this and estimates the seismic intensity class. Embodiments of the present invention will be described in detail below with reference to the drawings. Figure 1 is an explanatory diagram showing a state in which multiple stands are installed on the same floor, and each of the multiple smart devices is placed on one of the stands. The seismic intensity estimation system of this embodiment uses a smart device 10, which has a built-in earthquake information acquisition sensor including an acceleration sensor, to acquire an acceleration waveform, and estimates the seismic intensity class at the location where the smart device 10 is located based on the acceleration waveform. Smart devices 10 are widely distributed, owned by many people, and are usually distributed at a high density over a wide area. Therefore, if sufficiently accurate seismic motion is observed at each of them, it becomes possible to accurately observe seismic motion at any location.

[0015] To accurately understand the extent of damage on each floor of building 110, it is essential to estimate the seismic intensity level at the floor 111 of each floor of building 110. On the other hand, smart devices 10 are not usually placed directly on the floor 111, but rather on a stand T, such as a desk or table, that is placed on the floor 111. Here, on the stand T, the acceleration waveform observed is the result of the acceleration waveform observed on the floor 111 being transmitted through the stand T. Therefore, the seismic intensity level calculated based on the acceleration waveform observed on the stand T may have a different value from the seismic intensity level calculated based on the acceleration waveform observed on the floor 111, which is necessary to accurately understand the extent of damage to building 110.

[0016] Figure 2 is a table showing the relationship between seismic waves, the seismic intensity scale estimated based on acceleration waveforms acquired on the floor for those seismic waves (labeled "floor seismic intensity"), and the seismic intensity scale estimated based on acceleration waveforms acquired on a platform installed on the floor (labeled "platform seismic intensity"). The floor seismic intensity and platform seismic intensity are calculated for cases where the Kumamoto wave normalized at 400 gal, the Tsukidate wave normalized at 200 gal, and the Tsukidate wave normalized at 400 gal are input to building 110. As shown in Figure 2, the seismic intensity of the support structure tends to be higher than that of the floor. Therefore, if we attempt to evaluate the damage to Building 110 based on the seismic intensity of the support structure, there is a possibility that the damage will be overestimated compared to when evaluating the damage to Building 110 based on the seismic intensity of the floor.

[0017] Figure 3 is a graph showing the characteristics of the seismic intensity filter used when calculating the seismic intensity scale. As shown in Figure 3, the seismic intensity filter is set so that its value (weight) is large in the frequency band R1 of approximately 0.5 to 5 Hz. In contrast, the natural frequency of the mounting structure T is often in the frequency band R2, which includes, for example, 10 to 20 Hz. Thus, the frequency band R1, in which the seismic intensity filter value is set to be large, and the frequency band R2, which includes the natural frequency of the mounting structure T, are likely to be in different ranges.

[0018] Figure 4 is a graph showing the Fourier spectra of the acceleration waveform acquired on the floor and the acceleration waveform acquired on the platform for the Kumamoto wave normalized at 400 gal, in each of two orthogonal directions in the horizontal plane. Figure 5 is a graph showing the Fourier spectra of the acceleration waveform acquired on the floor and the acceleration waveform acquired on the platform for the Tsukidate wave normalized at 200 gal, in each of two orthogonal directions in the horizontal plane. Figure 6 is a graph showing the Fourier spectra of the acceleration waveform acquired on the floor and the acceleration waveform acquired on the platform for the Tsukidate wave normalized at 400 gal, in each of two orthogonal directions in the horizontal plane. In each of the graphs from Figures 4 to 6, the Fourier spectra of the acceleration waveform acquired on floor 111 are denoted with the sign SPF, and the Fourier spectra of the acceleration waveform acquired on platform T are denoted with the sign SPT. For example, as in the Kumamoto wave shown in Figure 4, if the frequency band R1, where the seismic intensity filter value is set to a large value, contains a sufficient number of frequencies where the peak value of the seismic wave occurs, then even if the Fourier spectrum SPT of the acceleration waveform acquired on the support structure T deviates significantly from the Fourier spectrum SPF of the acceleration waveform acquired on the floor 111 in the frequency band R2, which contains the natural frequency of the support structure T, the influence of the vibration of the support structure T on calculating the seismic intensity scale based on the acceleration waveform acquired on the support structure T will be limited. Therefore, as shown in Figure 2, the seismic intensity scale calculated based on the acceleration waveform acquired on the support structure T will be close to the seismic intensity scale calculated based on the acceleration waveform acquired on the floor 111.

[0019] However, as shown in Figures 5 and 6 for the Tsukidate wave, if the peak frequency of the seismic wave is not sufficiently contained within the frequency band R1 where the seismic intensity filter value is set to a large value, the deviation of the Fourier spectrum SPT of the acceleration waveform acquired on the support structure T from the Fourier spectrum SPF of the acceleration waveform acquired on the floor 111 will greatly affect the calculation of the seismic intensity scale based on the acceleration waveform acquired on the support structure T. In this way, when calculating the seismic intensity scale based on the acceleration waveform acquired on the support structure T, there is a possibility that the seismic intensity scale may be overestimated as a result of being affected by the vibration of the support structure T. The seismic intensity estimation system of this embodiment calculates the seismic intensity class based on the acceleration waveform acquired on the support structure T, and in order to reduce the influence of vibration of the support structure T, it estimates the seismic intensity class to be close to the seismic intensity class calculated based on the acceleration waveform acquired on the floor 111.

[0020] Figure 7 is a block diagram of the seismic intensity estimation system. The seismic intensity estimation system 1 is equipped with a seismic intensity estimation device 20. The seismic intensity estimation device 20 collects acceleration waveforms from each of the multiple smart devices 10 and estimates the seismic intensity class based on the acceleration waveforms. Figure 8 is an explanatory diagram of a smart device. The smart device 10 is, for example, a smartphone or a tablet. The smart device 10 is formed in a flat shape and has a screen 10a on one of its surfaces. As shown in Figure 7, the smart device 10 incorporates an earthquake information acquisition sensor 11, a transmission unit 12, and a storage unit 13. The earthquake information acquisition sensor 11 is equipped with an acceleration sensor 11a. The acceleration sensor 11a acquires data in at least two mutually orthogonal components in a first and second direction within a plane parallel to the screen 10a of the smart device 10. In this embodiment, the acceleration sensor 11a also acquires acceleration in an orthogonal direction perpendicular to the screen 10a. In this embodiment, the width direction of the smart device 10 is the first direction X, the length direction is the second direction Y, and the thickness direction is the orthogonal direction Z. When the smart device 10 is placed on a horizontal surface such as the surface of a stand T, the surface forming the screen 10a of the smart device 10, or the surface on the opposite side, i.e., the back side, is often positioned facing the installation surface. In this case, the screen 10a of the smart device 10 is positioned horizontally. Therefore, when an earthquake force acts on the smart device 10 placed on the installation surface, the acceleration sensor 11a observes and acquires the components of the acceleration acting on the smart device 10 in a first direction X along the horizontal plane, a second direction Y perpendicular to the first direction X in the horizontal plane, and a component in the perpendicular direction Z, i.e., the vertical direction. The observed acceleration waveform is stored in the memory unit 13.

[0021] The acceleration data acquired by the acceleration sensor 11a is stored in the memory unit 13 at predetermined short time intervals or continuously. If the smart device 10 does not have a function to store acceleration data in the memory unit 13 at predetermined time intervals or continuously, an application that implements this function may be created and added to the smart device 10. In this way, when an earthquake occurs, the smart device 10 acquires the components of the acceleration waveform, i.e., the time history waveform of acceleration, in the first direction X, the second direction Y, and the orthogonal direction Z, as observation results from the acceleration sensor 11a.

[0022] The transmitting unit 12 transmits the acceleration waveform stored as a time history waveform in the storage unit 13 to the seismic intensity estimation device 20, which will be described next, via the network 100. The network 100 is, for example, a public communication network that enables wireless or wired communication between the transmitting unit 12 of the smart device 10 and the acceleration waveform collection unit 23 of the seismic intensity estimation device 20.

[0023] The seismic intensity estimation device 20 includes a building designation unit 21, a smart device search unit 22, an acceleration waveform collection unit 23, an acceleration waveform adjustment unit 24, a frame vibration characteristic derivation unit 25, a floor seismic intensity scale estimation unit 26, and a storage unit 27. The building designation unit 21 accepts the designation of a building 110 for which the seismic intensity scale will be estimated by the seismic intensity estimation system 1. More specifically, for example, the user inputs information such as an identification number uniquely assigned to the building 110 and the latitude and longitude of the building 110, using an input device (not shown) to identify the building 110 for which the seismic intensity scale will be estimated. The building designation unit 21 also accepts information regarding the height of the floor 111 of the building 110 for which the seismic intensity scale will be estimated.

[0024] The memory unit 27 stores information about all smart devices 10 that are configured to communicate with the seismic intensity estimation device 20. For example, the memory unit 27 stores location information such as the latitude, longitude, and altitude of each smart device 10 when the earthquake occurred. Such location information can be obtained, for example, by known functions of the smart device 10, such as GPS (Global Positioning System). The smart device search unit 22 searches for multiple smart devices 10 that were located near the specified height of the building 110 designated by the building designation unit 21 at the time of the earthquake, based on the information stored in the memory unit 27.

[0025] The acceleration waveform collection unit 23 collects acceleration waveforms acquired by the acceleration sensor 11a during an earthquake from each of the multiple smart devices 10. The acceleration waveform collection unit 23 communicates with the transmission unit 12 of each of the multiple smart devices 10 that have been searched by the smart device search unit 22, and acquires the acceleration waveforms stored in the storage unit 13 of each smart device 10 when an earthquake occurs, and stores them in the storage unit 27.

[0026] Here, when an earthquake occurs, the smart device 10 may be being carried by a user, or it may be stationary on the support frame T. When the smart device 10 is being carried by a user, the acquired acceleration waveform contains acceleration caused by transportation other than seismic motion, making it unsuitable for use in estimating seismic intensity. On the other hand, when the smart device 10 is stationary on the support frame T, it is generally assumed that only seismic motion is reflected in the acceleration waveform. Therefore, the acceleration waveform collection unit 23 determines whether or not the smart device 10 was placed on the support frame T when the earthquake occurred. Specifically, the acceleration waveform collection unit 23 determines whether or not the acceleration of the smart device 10 in the orthogonal direction Z matches the acceleration due to gravity. If the acceleration in the orthogonal direction Z matches the acceleration due to gravity, the unit determines that the smart device 10 was placed on the support frame T and that only seismic motion is reflected in the acceleration waveform, and the acceleration waveform acquired by the smart device 10 is included in the subsequent processing. If the acceleration in the orthogonal direction Z does not match the acceleration due to gravity, the unit determines that the smart device 10 was not placed on the support frame T, and the acceleration waveform acquired by the smart device 10 is excluded from the subsequent processing.

[0027] In the following explanation, the process will be described based on the premise that multiple racks T are provided on the same floor 111, and that each of the multiple smart devices 10 found by the smart device search unit 22 is placed on one of the racks T. Furthermore, the processing of each component in the situation where n smart devices 10, from smart device 10-1 to smart device 10-n, have been found by the smart device search unit 22 will be explained below.

[0028] Figure 9 is an explanatory diagram showing a state in which multiple smart devices are mounted facing different directions. Figure 10 is an explanatory diagram showing a state in which the directions of the acceleration waveforms acquired by each of the smart devices in Figure 9 have been adjusted to match each other. Hereafter, as shown in Figures 9 and 10, when it is necessary to distinguish the mounting base T on which the smart devices 10-i (i=1 to n) are mounted, the mounting base T on which the smart devices 10-i are mounted will be referred to as mounting base Ti. As shown in Figure 9, it is assumed that when an earthquake occurs, the orientations of each smart device 10 will not be aligned. For example, the first direction X, which is x1, x2, ..., xn, for each smart device 10-1, 10-2, ..., 10-n may be different. In order to perform subsequent processing appropriately, the acceleration waveform adjustment unit 24 rotates the acceleration waveforms of the multiple smart devices 10 so that their orientations in the horizontal plane coincide, as shown in Figure 10, for example, to generate orientation-adjusted acceleration waveforms.

[0029] When the smart device 10 is placed on a support frame T, if an earthquake is strong, the horizontal force acting on the smart device 10 may exceed the static friction force between the surface of the support frame T and the smart device 10, potentially causing the smart device 10 to move relative to the support frame T. In such a case, the orientation of the smart device 10 may change during the relative movement. Therefore, the acceleration waveform adjustment unit 24 first extracts portions from each of the time-series acceleration waveforms acquired by each of the smart devices 10 that are considered not to have moved relative to the support frame T and whose orientation has not changed. For example, the magnetic flux density time history waveform is obtained by the magnetic sensor provided in the smart device 10, and in the magnetic flux density time history waveform, when the magnetic flux density changes greatly, it is determined that the orientation of the smart device 10 has changed, so that the time zone in which the smart device 10 has moved relative to the pedestal T can be grasped. Therefore, in this case, by using the portion of the acceleration waveform of the time history other than the time zone in which the smart device 10 has moved relative to the pedestal T, the portion that is considered to have no change in orientation can be extracted.

[0030] In this way, the acceleration waveform adjustment unit 24 first obtains, for each of the smart devices 10-i (i = 1 to n), the acceleration waveform acc of the time history in which the orientation has not changed in each of the first direction X and the second direction Y. xi i (t), acc yi i (t). Here, the subscripts (xi, yi) indicate the coordinate system of the smart device 10-i. Also, the superscript (i) indicates the number of the smart device 10-i.

[0031] These acceleration waveforms acc in the horizontal direction xi i (t), acc yi i (t) reflect the responses of each of the pedestals Ti on which the smart devices 10-i are placed. However, for the frequency components lower than the natural vibration frequency of the pedestal Ti, the pedestal Ti is in a state of swaying in the same manner as the floor 111 on which it is provided. Therefore, by comparing the long-period components with low frequencies of the acceleration waveforms acc in the horizontal direction xi i (t), acc yi i (t), it is possible to compare and adjust the orientations of each of the smart devices 10-i. Based on this idea, the acceleration waveform adjustment unit 24 uses the acceleration waveforms acc xi i (t), acc yi iFrom (t), the acceleration waveform accl is obtained by extracting only the frequency components that are lower than a predetermined frequency threshold that is set sufficiently low. xi i (t), accl yi i (t) is generated. The above frequency threshold should be set to a value greater than the lower limit of the frequency that can be acquired by the accelerometer sensor 11a of the smart device 10.

[0032] Then, the acceleration waveform adjustment unit 24 adjusts the acceleration waveform accl from which the low-frequency components have been extracted. xi i (t), accl yi i Based on (t), the acceleration waveform acc of each smart device 10-i xi i (t), acc yi i Determine the rotation angle to rotate (t), and the acceleration waveform acc xi i (t), acc yi i (t) is the acceleration waveform acc xi i (t), acc yi i By rotating (t) by the rotation angle determined, the acceleration waveform acc xi i (t), acc yi i Align the direction of (t). Specifically, the acceleration waveform adjustment unit 24 first arbitrarily selects one smart device 10. The acceleration waveform adjustment unit 24 then rotates the acceleration waveforms of other smart devices 10 to match the coordinate system of the selected smart device 10, thereby aligning the orientation of the acceleration waveforms. In other words, this selected smart device 10 becomes the reference smart device 10 when aligning the orientation of the acceleration waveforms. For the sake of simplicity, we will now assume that smart device 10-1 is arbitrarily selected, and explain the case where the orientations of other smart devices 10-i (i=2~n) are aligned with respect to this smart device 10-1.

[0033] Next, the acceleration waveform adjustment unit 24 extracts the acceleration waveform accl for each of the smart devices 10-i (i=2~n) with low frequencies. xi i (t), accl yi i Rotate (t) between 1 degree and 360 degrees, for example in 1-degree increments, and obtain the acceleration waveform after rotation (accl). xi´ i (t), accl yi´ i (t) and the acceleration waveform with low frequencies extracted from smart device 10-1 accl x1 1 (t), accl y1 1 A correlation analysis of the waveforms is performed between (t) and (t), and the angle that yields the strongest correlation is determined as the rotation angle. The above-mentioned waveform rotation can be performed, for example, by multiplying the pre-rotation coordinate system (x, y) by a rotation matrix that rotates it by an angle θ, as shown in equation (1), to generate the rotated coordinate system (x', y').

number

number

[0034] The acceleration waveform adjustment unit 24 adjusts the horizontal acceleration waveform of each smart device 10-i (i=2~n) by the rotation angle determined for that smart device 10-i. xi i (t), acc yi i Rotate (t). In this way, the acceleration waveform adjustment unit 24 adjusts the azimuth, i.e., the coordinate system, of the acceleration waveform acc of smart device 10-1 for each of the smart devices 10-i (i=2~n). x1 1 (t), acc y1 1 Acceleration waveform after azimuth adjustment, aligned to coordinate system x1, y1 of (t) x1 i (t), acc y1 i Generate (t). In the smart device 10-1, which serves as the reference for aligning orientations, the rotation operation described above is not actually performed. However, from here on, the acceleration waveform acc will be used for the smart device 10-1. x1 1 (t), acc y1 1 Acceleration waveform acc after adjusting the orientation of (t) x1 1 (t), acc y1 1 Let us consider this to be (t) and continue with the following explanation.

[0035] In this way, the acceleration waveform adjustment unit 24 adjusts the acceleration waveform (with low frequencies extracted) of the second smart device 10-i (i=2~n) for each of the multiple angles θ (for example, in increments of 1 degree in the range of 1 to 360 degrees) with respect to the first smart device 10-1 and the second smart device 10-i (i=2~n) among the multiple smart devices 10. xi i (t), accl yi i Rotating (t) by the angle θ in the horizontal plane, the acceleration waveform (with low frequencies extracted) of the first smart device 10-1 is obtained. x1 1 (t), accl y1 1 A correlation analysis is performed between (t) and (t), and the angle that shows the strongest correlation among multiple angles θ is determined as the rotation angle, and the acceleration waveform acc of the second smart device 10-i (i=2~n) is analyzed. xi i (t), accyi i The acceleration waveform resulting from rotating (t) by the rotation angle within the horizontal plane is the post-orientation adjustment acceleration waveform acc of the second smart device 10-i (i = 2 to n). x1 i (t), acc y1 i (t).

[0036] The pedestal vibration characteristic derivation unit 25 derives the vibration characteristics of the pedestal T based on the post-orientation adjustment acceleration waveforms acc x1 i (t), acc y1 i (t) (i = 1 to n). In the present embodiment, as will be described later, the vibration characteristics are derived for one pedestal Tj on which one of the plurality of smart devices 10-i (i = 1 to n), specifically, a certain smart device 10-j, is placed. In the present embodiment, the derived vibration characteristics of the pedestal Tj are the natural vibration frequency of the pedestal Tj. The pedestal vibration characteristic derivation unit 25 first performs Fourier transform on each of the post-orientation adjustment acceleration waveforms acc x1 i (t), acc y1 i (t) (i = 1 to n) to generate the Fourier spectra F x1 i (f), F y1 i (f) (i = 1 to n) of each of the smart devices 10-i (i = 1 to n). Since these Fourier spectra are generated based on the post-orientation adjustment acceleration waveforms, the orientation is in a state aligned with the coordinate system x1, y1 of the smart device 10-1.

[0037] Here, regarding an arbitrary smart device 10-j among the smart devices 10-i (i = 1 to n), considering that the Fourier spectra F x1 j (f), F y1 j (f) obtained by the smart device 10-j are the result of inputting the response of the floor 111 to the pedestal Tj on which the smart device 10-j is placed, they are expressed as in the following equation (3). F x1 j (f) = F x1 floor (f) × tf x1 Tj (f) ···(3) In the above equation, F x1 floor (f) is the Fourier spectrum of the response of floor 111, tf x1 Tj (f) is the transfer function of the mount Tj. y1 j (f) can also be expressed in the same way as above. Similarly, for any other smart device 10-k besides smart device 10-j, the Fourier spectrum F x1 k (f) can be expressed as shown in equation (4). F x1 k (f) = F x1 floor (f) × tf x1 Tk (f) ···(4) In the above equation, tf x1 Tk (f) is the transfer function of the mount Tk on which the smart device 10-k is mounted. y1 k (f) can also be expressed in the same way as above.

[0038] By dividing the left and right sides of equation (3) by the left and right sides of equation (4), we obtain the Fourier spectrum F relating to the floor response, as shown in equation (5). x1 floor (f) is canceled out, and the ratio of the transfer functions R between the pedestal Tj on which smart device 10-j is mounted and the pedestal Tk on which smart device 10-k is mounted is x1 j、k (f) is calculated.

number

[0039] Thus, the frame vibration characteristic derivation unit 25 calculates the azimuth-adjusted acceleration waveform acc of each of the smart devices 10 in the pair of smart devices 10 (one smart device 10-j and the other smart device 10-k). x1 j (t), acc y1 j Based on (t), the vibration characteristics of at least one support structure Tj are derived. More specifically, the frame vibration characteristic derivation unit 25 calculates the azimuth-adjusted acceleration waveform acc of one of the smart devices 10-j among the multiple smart devices 10. x1 j (t), acc y1 j (t) and the acceleration waveform after orientation adjustment of other smart devices 10-k acc x1 k (t), acc y1 k Based on (t), the natural frequency of the mount Tj on which the smart device 10-j is mounted is derived. Furthermore, the frame vibration characteristic derivation unit 25 calculates the acceleration waveform acc of a smart device 10-j after orientation adjustment. x1 j (t), acc y1 j Fourier spectrum F of (t) x1 j (f), F y1 j (f) Acceleration waveform after orientation adjustment of other smart devices 10-k acc x1 k (t), acc y1 k Fourier spectrum F of (t) x1 k (f), F y1k By dividing by (f), we obtain the ratio R of the transfer functions between the rig Tj on which one smart device 10-j is mounted and the rig Tk on which the other smart device 10-k is mounted. x1 j、k (f), R y1 j、k (f) is calculated, and the ratio of the transfer functions R x1 j、k (f), R y1 j、k In (f), the frequency that results in the peak value is defined as the natural frequency of the mount Tj on which one smart device 10-j is mounted.

[0040] Here, depending on the shape of the Fourier spectrum, for example, when the natural frequencies of mount Tj and mount Tk are close, the value of the peak in equation (5) may not be clearly expressed. Therefore, the ratio R of the transfer function between one smart device 10-j and each of the other smart devices 10-k is x1 j、k (f), R y1 j、k (f) is calculated, and the ratio of the calculated transfer functions R for multiple other smart devices 10-k x1 j、k (f), R y1 j、k (f) Average R x1 j (f), R y1 j (f) is calculated, and the average R x1 j (f), R y1 j In (f), it is more desirable to set the frequency that results in the peak value to the natural frequency of the mount Tj on which one smart device 10-j is mounted. For example, the average R x1 j (f) can be calculated as shown in equation (6).

number

[0041] Furthermore, if, for example, the mount Tj on which smart device 10-j is mounted and the mount Tk on which smart device 10-k is mounted are the same mount T, or if the type of mount T is the same, then when calculating the ratio of the transfer functions, the peaks will be almost completely canceled out. Therefore, in order to eliminate such cases, the acceleration waveform acc of one smart device 10-j after orientation adjustment is calculated. x1 j (t), acc y1 j Fourier spectrum F of (t) x1 j (f), F y1 j (f) and the azimuthal-adjusted acceleration waveform acc of each of the other smart devices 10-k x1 k (t), acc y1 k Fourier spectrum F of (t) x1 k (f), F y1 k It is desirable to perform a correlation analysis between (f) and and, if the result is above the upper threshold, to exclude the other smart device 10-k from the calculation of the average. In this case, the upper threshold can be set to a value such as 0.95.

[0042] Furthermore, in the above description, the frame vibration characteristic derivation unit 25 is described as deriving the natural frequency of the frame Tj on which a smart device 10-j, which is arbitrarily selected from among the multiple smart devices 10, is mounted. However, in practice, it is desirable for the frame vibration characteristic derivation unit 25 to derive the natural frequency of the frame T on which each of the multiple smart devices 10 is mounted, and then select one smart device 10 as smart device 10-j from among them, for which the natural frequency of the frame T is considered to be an appropriate value. For the smart device 10-j thus selected, the seismic intensity class is calculated by the floor seismic intensity class estimation unit 26, which will be explained later. If the acceleration waveform of a smart device 10 mounted on a support frame T, which has a small natural frequency close to the frequency of the seismic intensity filter, is used in calculating the seismic intensity scale, the response of the support frame T may have a significant impact on the seismic intensity scale. Therefore, it is conceivable to select a smart device 10 mounted on a support frame T with a sufficiently large calculated natural frequency and a small impact on the seismic intensity calculation as one smart device 10-j. Alternatively, for example, it is conceivable to select a smart device 10 with the smallest relative movement to the support frame T as one smart device 10-j. Or, if it is possible to identify a smart device 10 with a high coefficient of friction with the support frame T, such as one equipped with an anti-slip cover, it is conceivable to select such a smart device 10 as one smart device 10-j.

[0043] Here, we will explain a verification example regarding the processing of the frame vibration characteristic derivation unit 25 described above. Figure 11 is a table showing the natural frequencies of the mounting structure used for verification and the orientation of the smart device placed on the mounting structure. As shown in Figure 11, smart devices 10 were placed on each of the five rigging structures T, each having a different natural frequency, with different orientations. Then, the transfer function was calculated by treating each rigging structure T as a single mass system, and seismic waves corresponding to the response of the floor 111 were input to create seismic waves recorded by the smart devices 10 placed on the rigging structures T. Based on these seismic waves, the natural frequency of rigging structure T1 on which smart device 10-1 was placed was derived in the manner described above. Figure 12 is a graph showing the result of calculating the ratio of transfer functions by dividing the Fourier spectrum of the azimuth-adjusted acceleration waveform of smart device 10-1 in Figure 11 by the Fourier spectrum of the azimuth-adjusted acceleration waveform of other smart devices, and then calculating the average of these ratios among the other smart devices. According to Figure 11, the natural frequencies of the mount T1 on which the smart device 10-1 is mounted are 12 Hz and 25 Hz. In Figure 12, it can be seen that the values ​​are peaked at both 12 Hz and 25 Hz. In this way, the natural frequencies of the mount T are appropriately derived.

[0044] The floor seismic intensity scale estimation unit 26 calculates the acceleration waveform acc of one smart device 10-j, which is a smart device 10 placed on a frame Tj from which the vibration characteristics have been derived, as selected as described above. xj j (t), acc yj j Based on (t) and the vibration characteristics of the support frame Tj on which the smart device 10-j is installed, i.e., its natural frequency, the Fourier spectrum of the floor 111 is derived, and the seismic intensity class of the floor 111 is estimated. Conceptually, the floor seismic intensity class estimation unit 26 treats the support frame T as a vibration system of a single point mass having the derived natural frequency, and then calculates the shaking of the floor 111 from the response of the support frame T.

[0045] For this purpose, the floor seismic intensity scale estimation unit 26 first uses the acceleration waveform acc of a smart device 10-j. xj j (t), acc yj j The rotated Fourier spectrum is obtained by rotating (t) at each of several angles (for example, in increments of 1 degree in the range of 1 to 360 degrees) and performing a Fourier transform. The floor seismic intensity scale estimation unit 26 selects the Fourier spectrum that has the maximum peak value from among the multiple Fourier spectra generated as described above. Furthermore, the floor seismic intensity scale estimation unit 26 calculates the transfer function of a single mass system with respect to the frame Tj based on the natural frequency of the frame Tj on which the smart device 10-j is mounted. The floor seismic intensity scale estimation unit 26 then derives the Fourier spectrum of the floor 111 by dividing the Fourier spectrum of the acceleration waveform of the smart device 10-j, which was selected as described above, by the transfer function of a single-mass system.

[0046] Figure 13 is a graph showing the results of applying the above embodiment to the Fourier spectrum of the acceleration waveform acquired on the support structure for a Tsukidate wave normalized at 200 gal. In Figure 13, the Fourier spectrum derived as described above is shown with the sign SPC. The Fourier spectrum SPC calculated by the floor seismic intensity scale estimation unit 26 in Figure 13 has a shape closer to the Fourier spectrum SPF of the acceleration waveform acquired on the floor 111 than the Fourier spectrum SPT generated by performing a Fourier transform on the acceleration waveform acquired on the support structure T, as shown in Figure 5. Thus, the Fourier spectrum SPC calculated by the floor seismic intensity scale estimation unit 26 has a significantly reduced influence from the response of the support structure T.

[0047] By using the Fourier spectrum of floor 111 derived as described above, the seismic intensity class of floor 111 can be estimated. Specifically, the floor seismic intensity class estimation unit 26 calculates the acceleration waveforms in the floor 111 in the first and second horizontal floor directions, which are mutually orthogonal in the horizontal plane, by performing an inverse Fourier transform on the Fourier spectrum of the floor 111. The floor seismic intensity class estimation unit 26 further combines the acceleration waveform in the first horizontal floor direction, the acceleration waveform in the second horizontal floor direction, and the acceleration waveform in the orthogonal direction Z of a smart device 10-j to calculate the composite acceleration. Based on this composite acceleration, the floor seismic intensity class estimation unit 26 estimates the seismic intensity class of the floor 111.

[0048] Figure 14 is a table added to Figure 2, showing the seismic intensity class estimated by applying each of the processes of this embodiment to the acceleration waveforms acquired on the support structure for each case shown in Figure 2. In Figure 14, the seismic intensity class estimated by applying each of the processes of this embodiment is indicated as "After support structure seismic intensity correction". As can be seen from Figure 14, the seismic intensity scale estimated by applying each process of this embodiment is sufficiently close to the seismic intensity scale estimated based on the acceleration waveform acquired at the floor.

[0049] Next, the seismic intensity estimation method using the seismic intensity estimation system 1 described above will be explained using Figures 1 to 14 and Figure 15. Figure 15 is a flowchart of the seismic intensity estimation method. First, the building designation unit 21 receives designation of a building for which the seismic intensity class will be estimated by the seismic intensity estimation system 1. The building designation unit 21 also receives information regarding the height of the floor 111 of the building for which the seismic intensity class will be estimated (step S1). The smart device search unit 22 searches for multiple smart devices 10 located near a specified height in a building designated by the building designation unit 21 when an earthquake occurs (step S2). The acceleration waveform acquisition unit 23 collects acceleration waveforms acquired by the acceleration sensor 11a during an earthquake from each of the multiple smart devices 10 (Step S3: Acceleration waveform acquisition step). The acceleration waveform adjustment unit 24 adjusts the acceleration waveform of each smart device 10-i. xi i (t), acc yi i By rotating (t) and aligning the orientation, the acceleration waveform after orientation adjustment can be obtained. x1 i (t), acc y1 i (t) is generated (Step S4: Acceleration waveform adjustment step).

[0050] The frame vibration characteristic derivation unit 25 calculates the acceleration waveform after orientation adjustment. x1 i (t), acc y1 i Each of (t)(i=1~n) is subjected to a Fourier transform, and the Fourier spectrum F of each smart device 10-i(i=1~n) is obtained. x1 i (f), F y1 i (f) Generate (i=1~n) (Step S5). The frame vibration characteristic derivation unit 25 calculates the acceleration waveform of a smart device 10-j after orientation adjustment. x1 j (t), acc y1 j Fourier spectrum F of (t) x1 j (f), F y1 j (f) Acceleration waveform after orientation adjustment of other smart devices 10-k acc x1 k (t), acc y1 k Fourier spectrum F of (t) x1 k (f), F y1 k By dividing by (f), we obtain the ratio R of the transfer functions between the rig Tj on which one smart device 10-j is mounted and the rig Tk on which the other smart device 10-k is mounted. x1 j、k (f), R y1 j、k (f) is calculated, and the ratio of the transfer functions R x1 j、k (f), R y1 j、k In (f), the frequency that results in the peak value is taken as the natural frequency of the mount Tj on which one smart device 10-j is mounted (Step S6: Mounting Vibration Characteristic Derivation Process).

[0051] The floor seismic intensity class estimation unit 26 uses the acceleration waveform acc of one smart device 10-j selected as described above. xj j (t), acc yj j (t) and the vibration characteristics of the mounting frame Tj on which the smart device 10-j is installed, i.e., the natural frequency, are used to derive the Fourier spectrum of the floor 111 (step S7). The floor seismic intensity scale estimation unit 26 estimates the seismic intensity scale of the floor (Step S8: Floor seismic intensity scale estimation process).

[0052] The seismic intensity estimation system 1 described above is a seismic intensity estimation system 1 that uses a smart device 10 with a built-in earthquake information acquisition sensor 11, wherein multiple frames T are provided on the same floor 111, and each of the multiple smart devices 10 is placed on one of the frames T, the earthquake information acquisition sensor 11 is equipped with an acceleration sensor 11a, and an acceleration waveform collection unit 23 collects acceleration waveforms acquired by the acceleration sensor 11a during an earthquake from each of the multiple smart devices 10, and rotates the acceleration waveforms of the multiple smart devices 10 so that their orientations in the horizontal plane coincide with each other. The system includes an acceleration waveform adjustment unit 24 that generates an acceleration waveform after orientation adjustment, a base vibration characteristic derivation unit 25 that derives the vibration characteristics of at least one base Tj based on the acceleration waveforms after orientation adjustment of at least one pair of smart devices 10-j and 10-k among the multiple smart devices 10, and a floor seismic intensity class estimation unit 26 that derives the Fourier spectrum of the floor 111 based on the acceleration waveform of the smart device 10-j placed on the base Tj from which the vibration characteristics have been derived, and the vibration characteristics of the base Tj, and estimates the seismic intensity class of the floor 111. With the above configuration, multiple mounting frames T are provided on the same floor 111, and each of the multiple smart devices 10 is placed on one of the mounting frames T. Acceleration waveforms acquired by the acceleration sensor 11a during an earthquake are collected from each of the multiple smart devices 10. Next, the acceleration waveforms of the multiple smart devices 10 are rotated so that their orientations in the horizontal plane coincide with each other to generate orientation-adjusted acceleration waveforms. Then, for at least one pair of smart devices 10-j and 10-k among the multiple smart devices 10, the vibration characteristics of at least one mounting frame Tj are derived based on the orientation-adjusted acceleration waveforms of each of the pair of smart devices 10-j and 10-k. For example, if a pair of smart devices 10-j and 10-k includes one smart device 10-j and the other smart device 10-k, the Fourier spectrum of the azimuthal-adjusted acceleration waveform of one smart device 10-j can be expressed as the product of the Fourier spectrum of the acceleration waveform of the floor 111 and the transfer function of the pedestal Tj on which the one smart device 10-j is mounted. Similarly, the Fourier spectrum of the azimuthal-adjusted acceleration waveform of the other smart device 10-k can be expressed as the product of the Fourier spectrum of the acceleration waveform of the floor 111 and the transfer function of the pedestal Tk on which the other smart device 10-k is mounted. Therefore, by dividing the Fourier spectrum of the azimuthal-adjusted acceleration waveform of one smart device 10-j by the Fourier spectrum of the azimuthal-adjusted acceleration waveform of the other smart device 10-k, the ratio of the transfer functions between the pedestal Tj on which the one smart device 10-j is mounted and the pedestal Tk on which the other smart device 10-k is mounted can be calculated. By extracting the peak frequency within this ratio of transfer functions, it is possible to derive the natural frequency of the mount Tj on which one smart device 10-j is mounted, as the vibration characteristic of the mount Tj on which one smart device 10-j is mounted. As described above, if the vibration characteristics such as the natural frequency of at least one support structure (for example, support structure Tj on which one smart device 10-j is placed) are known, then, for example, by considering the response by support structure Tj as the response by a single-mass system, calculating the transfer function of the single-mass system based on the vibration characteristics of support structure Tj, and then dividing the Fourier spectrum of the acceleration waveform of the smart device 10-j by the transfer function of the single-mass system, the Fourier spectrum of floor 111 can be derived. The Fourier spectrum of floor 111 derived in this way excludes the influence of the vibration of support structure Tj. Therefore, by using such a Fourier spectrum of floor 111, the seismic intensity class of floor 111 can be estimated with high accuracy. In this way, when the smart device 10 is placed on a stand T on the floor 111, it becomes possible to further improve the estimation accuracy when estimating the seismic intensity class based on the acceleration waveform obtained by the acceleration sensor 11a of the smart device 10.

[0053] In particular, the frame vibration characteristic derivation unit 25 derives the vibration characteristics of frame Tj on which one smart device 10-j is placed, based on the azimuth-adjusted acceleration waveform of one smart device 10-j among the multiple smart devices 10 and the azimuth-adjusted acceleration waveform of another smart device 10-k. The floor seismic intensity class estimation unit 26 derives the Fourier spectrum of floor 111 based on the acceleration waveform of one smart device 10-j and the vibration characteristics of frame Tj, and estimates the seismic intensity class of floor 111. With the configuration described above, the seismic intensity estimation system 1 can be properly implemented.

[0054] Furthermore, in the acceleration sensor 11a, the acceleration waveform is acquired in at least the first direction X and second direction Y components which are orthogonal to each other in a plane parallel to the screen 10a of the smart device 10. The acceleration waveform adjustment unit 24, with respect to the first smart device 10-1 and the second smart device 10-i (i=2~n) among the multiple smart devices 10, rotates the acceleration waveform of the second smart device 10-i by that angle in the horizontal plane for each of a plurality of angles, performs correlation analysis with the acceleration waveform of the first smart device 10-1, determines the angle that yields the best correlation among the plurality of angles as the rotation angle, and the acceleration waveform resulting from rotating the acceleration waveform of the second smart device 10-i by the rotation angle in the horizontal plane is taken as the orientation-adjusted acceleration waveform of the second smart device 10-i. With the above configuration, the acceleration waveforms of multiple smart devices 10 can be rotated so that their orientations in the horizontal plane coincide with each other, thereby appropriately generating orientation-adjusted acceleration waveforms.

[0055] Furthermore, the vibration characteristics are natural frequencies, and the mount vibration characteristic derivation unit 25 calculates the ratio of the transfer functions between the mount Tj on which one smart device 10-j is mounted and the mount Tk on which the other smart device 10-k is mounted by dividing the Fourier spectrum of the acceleration waveform of one smart device 10-j after orientation adjustment by the Fourier spectrum of the acceleration waveform of the other smart device 10-k, for one smart device 10-j and the other smart device 10-k among the pair of smart devices 10-j and 10-k. The frequency that is the peak value in the ratio of the transfer functions is taken as the natural frequency of the mount Tj on which one smart device 10-j is mounted. With the configuration described above, the ratio of the transfer functions between the pedestal Tj on which the first smart device 10-j is mounted and the pedestal Tk on which the other smart device 10-k is mounted is calculated by dividing the Fourier spectrum of the pedestal acceleration waveform of the other smart device 10-k by the Fourier spectrum of the pedestal acceleration waveform of the other smart device 10-k. By extracting the peak frequency from this ratio of transfer functions, the natural frequency of the pedestal Tj on which the first smart device 10-j is mounted can be derived. Using the natural frequency of the pedestal Tj derived in this way, the seismic intensity class of the floor 111 can be estimated with high accuracy, as already explained.

[0056] Furthermore, the mount vibration characteristic derivation unit 25 calculates the transfer function ratio for each of the multiple other smart devices 10-k by dividing the Fourier spectrum of the azimuth-adjusted acceleration waveform of one smart device 10-j by each of the Fourier spectra of the azimuth-adjusted acceleration waveforms of multiple other smart devices 10-k. It then calculates the average of the transfer function ratios among the multiple other smart devices 10-k, and sets the frequency that is the peak value among the average of the transfer function ratios as the natural frequency of the mount on which the one smart device 10-j is mounted. Furthermore, the frame vibration characteristic derivation unit 25 performs a correlation analysis between the Fourier spectrum of the azimuth-adjusted acceleration waveform of one smart device 10-j and the Fourier spectrum of the azimuth-adjusted acceleration waveform of another smart device 10-k. If the result exceeds an upper threshold, the other smart device 10-k is excluded from the average calculation. With the above configuration, the natural frequency of the stand on which a single smart device 10-j is mounted can be reliably extracted.

[0057] Furthermore, the vibration characteristics are natural frequencies, and the floor seismic intensity class estimation unit 26 considers the response by the support structure Tj to be the response by a single-mass system. Based on the natural frequency of the support structure Tj, it calculates the transfer function of the single-mass system, and derives the Fourier spectrum of the floor 111 by dividing the Fourier spectrum of the acceleration waveform of the smart device 10-j placed on the support structure Tj, from which the vibration characteristics have been derived, by the transfer function of the single-mass system. With the above configuration, if the vibration characteristics such as the natural frequency of the support structure Tj are known, the response by the support structure Tj can be considered as the response of a single-mass system. Based on the vibration characteristics of the support structure Tj, the transfer function of the single-mass system can be calculated. By performing processes such as dividing the Fourier spectrum of the acceleration waveform of the smart device 10-j placed on the support structure Tj (from which the vibration characteristics have been derived) by the transfer function of the single-mass system, the Fourier spectrum of the floor 111 can be derived. The Fourier spectrum of the floor 111 derived in this way excludes the influence of the vibration of the support structure Tj. Therefore, by using such a Fourier spectrum of the floor 111, the seismic intensity class of the floor 111 can be estimated with high accuracy.

[0058] Furthermore, the above-described seismic intensity estimation method is a seismic intensity estimation method using a smart device 10 with a built-in acceleration sensor 11a, wherein multiple frames T are provided on the same floor 111, and each of the multiple smart devices 10 is placed on one of the frames T, and the acceleration waveform acquired by the acceleration sensor 11a during an earthquake is collected from each of the multiple smart devices 10 in an acceleration waveform collection step, and the acceleration waveforms of the multiple smart devices 10 are rotated so that their orientations in the horizontal plane coincide with each other. The system includes: an acceleration waveform adjustment step for generating an acceleration waveform after orientation adjustment; a frame vibration characteristic derivation step for deriving the vibration characteristics of a frame Tj on which one smart device 10-j is mounted, based on the acceleration waveform after orientation adjustment of one smart device 10-j and the acceleration waveforms after orientation adjustment of another smart device 10-k; and a floor seismic intensity class estimation step for deriving the Fourier spectrum of the floor 111 based on the acceleration waveform of one smart device 10-j and the vibration characteristics of the frame Tj, and estimating the seismic intensity class of the floor 111. With the above configuration, as explained with respect to the seismic intensity estimation system 1, when the smart device 10 is placed on a stand T on the floor 111, it becomes possible to further improve the estimation accuracy when estimating the seismic intensity class based on the acceleration waveform obtained by the acceleration sensor 11a of the smart device 10.

[0059] (Modified version of the embodiment) Let's consider the case where an earthquake occurs while the smart device 10 is placed on a horizontal surface, such as the surface of a mounting frame T. If the earthquake is small, the smart device 10 will move in accordance with the mounting surface due to the frictional force between the smart device 10 and the mounting surface, and will not move relative to the mounting surface. In this case, the acceleration waveform observed by the acceleration sensor 11a, i.e., the time history waveform of acceleration, is considered to be an accurate observation of the earthquake. However, if the earthquake is strong enough that a force exceeding the frictional force between the smart device 10 and the installation surface acts on the smart device 10, the smart device 10 may slide on the installation surface, bounce up and down on the installation surface, or fall off the installation surface, thus moving relative to the installation surface.

[0060] Figure 16(a) shows the horizontal component of the acceleration waveform acquired by a smart device fixed on a pedestal; Figure 16(b) is an illustrative diagram of the horizontal component of the acceleration waveform that may be acquired when a smart device not fixed on a pedestal slides relative to the pedestal; and Figure 16(c) is an example of the horizontal component of the acceleration waveform actually acquired when a smart device not fixed on a pedestal slides relative to the pedestal. In Figure 16(a), the smart device 10 is fixed to the mounting base T in a way that prevents relative movement, and therefore the acceleration waveform is correctly acquired by the smart device 10.

[0061] In contrast, if the smart device 10 is not fixed to the mounting base T, and the smart device 10 moves relative to the mounting base T, the acceleration waveform observed by the acceleration sensor 11a may not accurately reflect the earthquake. For example, in the acceleration waveform shown in Figure 16(b), the value is approximately 150 cm / s². 2 At a certain level of acceleration, a region PA occurs where the amplitude becomes constant and plateaus, resulting in a loss of acceleration data in this region. In such cases, at what magnitude of acceleration does the acceleration plateau and become constant? (e.g., 150 cm / s in Figure 16(b)) 2It is relatively easy to determine the value of ( ). Therefore, by supplementing the portion where the absolute value of the amplitude is larger than the acceleration using a sinusoidal curve fitting process (curve fitting), as will be explained later using Figure 25, it is possible to estimate the acceleration waveform when the acceleration has not plateaued, assuming that the smart device 10 is fixed to the mount T. However, in reality, as shown in Figure 16(c), even though the smart device 10 is sliding, the portion PB that is thought to be sliding does not reach a constant amplitude and plateau, and large accelerations are sometimes observed in certain areas. In such acceleration waveforms, it is not easy to automatically determine, at least using information processing equipment, which parts of the acceleration are plateauing and which parts are not. Therefore, it is not easy to automatically interpolate such acceleration waveforms and estimate the acceleration waveform assuming that the smart device 10 is fixed to the mount T.

[0062] For example, when estimating seismic intensity levels for many smart devices 10 covering a wide area and creating a seismic intensity distribution for that area, it is necessary to perform interpolation processing on many acceleration waveforms. As mentioned above, it is not easy to perform this interpolation processing automatically. Furthermore, even if the interpolation processing is performed manually, it becomes considerably more time-consuming and impractical when the number of acceleration waveforms is large. In contrast, the seismic intensity estimation system of this modified version determines whether the smart device 10 has moved relative to the installation surface. If it is determined that it has moved, it adjusts and corrects the acceleration waveform at the time the movement was determined to have occurred, and estimates the seismic intensity class based on the result.

[0063] Figure 17 is a block diagram of the seismic intensity estimation system related to this modified example. For the processing described above, the smart device 10 of this modified example includes a magnetic sensor 11b in addition to an acceleration sensor 11a in the earthquake information acquisition sensor 11. Furthermore, the seismic intensity estimation device 20A of the seismic intensity estimation system 1A of this modified example includes a building designation unit 21, a smart device search unit 22, an acceleration waveform collection unit 23, an acceleration waveform adjustment unit 24, a frame vibration characteristic derivation unit 25, a floor seismic intensity class estimation unit 26, and a storage unit 27, as well as a sliding section identification unit 31, a sliding acceleration threshold calculation unit 32, and an acceleration waveform correction unit 33.

[0064] The magnetic sensor 11b, like the acceleration sensor 11a, acquires magnetic flux density in the first direction X, the second direction Y, and the orthogonal direction Z. Magnetic flux density is an indicator that represents the orientation of the smart device 10, indicating which direction it is facing. When an earthquake force acts on the smart device 10 placed on the installation surface, the magnetic sensor 11b observes and acquires the component of the magnetic flux density in a first direction X along the horizontal plane, a component in a second direction Y perpendicular to the first direction X in the horizontal plane, and a component in the perpendicular direction Z, i.e., the vertical direction.

[0065] In such a smart device 10, when an earthquake occurs, the components of the time history waveform of the magnetic flux density (hereinafter referred to as the magnetic flux density waveform) in the first direction X, the second direction Y, and the orthogonal direction Z are acquired as measurement results from the magnetic sensor 11b and stored in the memory unit 13. The transmitting unit 12 transmits the acceleration waveform, i.e., the time history waveform of acceleration, as well as the magnetic flux density waveform stored as a time history waveform in the storage unit 13, to the seismic intensity estimation device 20A via the network 100. The acceleration waveform collection unit 23 collects magnetic flux density waveforms acquired by the magnetic sensor 11b during an earthquake from each of the multiple smart devices 10, in addition to the acceleration waveforms. The acceleration waveform collection unit 23 communicates with the transmission unit 12 of each of the multiple smart devices 10 that have been searched by the smart device search unit 22, and retrieves the magnetic flux density waveforms stored in the storage unit 13 of each smart device 10 when an earthquake occurs, and stores them together with the acceleration waveforms in the storage unit 27.

[0066] The sliding section identification unit 31 identifies the sliding section, which is the time interval when the smart device 10 slides on the mounting frame T, based on the measurement results from the earthquake information acquisition sensor 11. In this modified seismic intensity estimation system 1A, the change in the orientation of the smart device 10 is observed to determine whether the smart device 10 is moving relative to the installation surface or the installation surface. In this modified example, the orientation of the smart device 10 is determined by the magnetic flux density measured by the magnetic sensor 11b. That is, in this modified example, the sliding section identification unit 31 identifies the sliding section based on the magnetic flux density waveform obtained from the measurement result of the magnetic sensor 11b.

[0067] Figure 18(a) shows the horizontal component of the acceleration waveform during an earthquake, and Figure 18(b) shows the horizontal component of the time-history waveform of the magnetic flux density acquired by a magnetic sensor. Figure 19(a) shows the vertical component of the acceleration waveform during an earthquake, and Figure 19(b) shows the vertical component of the time-history waveform of the magnetic flux density acquired by a magnetic sensor. In the earthquakes shown in Figures 18 and 19, as can be seen from the vertical component of the acceleration waveform shown in Figure 19(a), the vertical component of acceleration is 1G (980 cm / s²). 2 ) does not exceed this value. Therefore, it is considered that the smart device 10 is basically in contact with the installation surface. However, since the magnetic flux density waveform is changing in the vertical direction, it is considered that the smart device 10 is bouncing by bouncing slightly in the vertical direction while sliding horizontally on the installation surface.

[0068] More specifically, in the cases shown in Figures 18 and 19, the magnetic flux density remains almost constant until time TR1, then begins to change significantly after time TR1, and this large change ends a little before time TR2, after which it returns to an almost constant value. Thus, because the magnetic flux density continues to change significantly from time TR1 to time TR2, the orientation of the smart device 10 is continuously changing during this period, and therefore, it can be considered that the smart device 10 is moving relative to the installation surface. Thus, if the orientation of the smart device 10 changes, the magnetic flux density value also changes accordingly. Therefore, based on the magnetic flux density waveform, it is possible to determine whether the smart device 10 has moved relative to the installation surface from a stationary state on the installation surface due to the earthquake, and if it is determined that it has moved, it is possible to identify the movement period, which is the time interval during which the movement occurred.

[0069] Here, as shown in Figures 18 and 19 for the time before time TR1 and for the time after time TR2, the magnetic flux density value measured by the magnetic sensor 11b changes slightly even when no earthquake is occurring. For this reason, it is not advisable to determine whether or not the smart device 10 has moved by simply comparing the magnetic flux density values. In this modified example, the sliding section identification unit 31 identifies the sliding section, which is the time interval when the smart device 10 slides on the mounting base T, by calculating the STA / LTA ratio based on either the horizontal component or the vertical component or both of the magnetic flux density waveform as a measurement result from the earthquake information acquisition sensor 11. LTA (Long Term Average) is a long-term moving average of the magnetic flux density waveform, while STA (Short Term Average) is a short-term moving average of the magnetic flux density waveform. The STA / LTA ratio is calculated by dividing STA by LTA.

[0070] For example, in the case of a magnetic flux density waveform where the value changes significantly with time, such as from time TR1 to time TR2 as shown in Figures 18 and 19, the LTA cannot keep up with this change, while the STA clearly reflects this change, causing a large change in the STA / LTA ratio. Therefore, by comparing the value of the STA / LTA ratio for each time point included in the magnetic flux density waveform with a threshold, and extracting the times when the value is greater than the threshold, the sliding interval can be identified by setting the earliest time as the time when the smart device 10 started sliding and the latest time as the time when the smart device 10 ended sliding. For example, the time period for calculating STA can be set to 0.5 seconds, the time period for calculating LTA to 2 seconds, and the threshold to a value such as 0.01.

[0071] In the acceleration waveform correction unit 33, which will be explained later, the waveform corresponding to the sliding section of the acceleration waveform is corrected to generate a corrected acceleration waveform. Here, if the vertical acceleration is 1G or more, it is considered that the smart device 10 is falling downward from the surface of the support base T, and in such a case, it is difficult to properly perform the waveform correction in the acceleration waveform correction unit 33. Therefore, in this modified example, the sliding section identification unit 31 identifies a time interval in which the STA / LTA ratio is above a certain threshold and the vertical acceleration is less than 1G as the sliding section.

[0072] Once the sliding interval, which is the time interval when the smart device 10 slides on the mount T, is identified as described above, the value of the amplitude that appears most frequently in the sliding interval waveform, which is the waveform corresponding to the sliding interval of the acceleration waveform obtained by extracting only the sliding interval portion from the acceleration waveform, is determined. This gives the acceleration value of the portion where the amplitude becomes constant and plateaus (150 cm / s in Figure 16(b)). 2 It may be possible to determine the value of ). Figure 20 is a graph showing an example of the waveform u(t). Figure 21 is a graph showing the absolute value |u(t)| of the waveform u(t) in Figure 20. For example, with respect to the waveform u(t) shown in Figure 20, it is possible to calculate the absolute value |u(t)| as shown in Figure 21, and then, by finding the amplitude value that appears most frequently on the graph of this absolute value |u(t)|, it is possible to determine the acceleration value of a peak in the waveform u(t). However, in the graph of the absolute value |u(t)| shown in Figure 21, the waveform value changes rapidly, and even if the frequency is calculated from this graph of the absolute value |u(t)|, it may not be possible to determine the acceleration value of the peak with sufficient accuracy. Therefore, in this modified example, the sliding acceleration threshold calculation unit 32 generates an envelope for the sliding interval waveform and calculates the mode of the values ​​on the envelope as the sliding acceleration threshold, which is the threshold of the acceleration at which the smart device 10 slides on the support frame T.

[0073] The sliding acceleration threshold calculation unit 32 generates the envelope described above by, for example, extracting only the portion corresponding to the sliding section from the acceleration waveform, thereby extracting the sliding section waveform, which is the waveform corresponding to the sliding section of the acceleration waveform. The sliding acceleration threshold calculation unit 32 generates the envelope of the sliding section waveform by applying a Hilbert transform to the sliding section waveform. Figure 22 is an explanatory diagram of the Hilbert transformation. The Hilbert transform is a type of signal analysis. The Fourier spectrum U(ω) in the frequency domain, generated by applying the Fourier transform to the input waveform u(t), is multiplied by -j × sgn(ω) (where j is the imaginary unit and sgn(ω) is the sign determined based on the sign of the frequency) to obtain the function U H (ω) is obtained. This function U H Applying the inverse Fourier transform to (ω) gives the Hilbert transform pair of the input waveform u(t) u H (t) is obtained. Figure 23 shows the Hilbert transform of the waveform u(t) compared to Figure 20. H This is a graph showing (t) superimposed. As shown in Figure 23, Hilbert transform pair u H (t) is a waveform in which the phase of the input waveform u(t) has been shifted by 90 degrees.

[0074] Here, as shown in equation (7), the input waveform u(t) is taken as the real part, and the Hilbert transform pair u H A waveform a(t), generally referred to as the analytical signal, is calculated with (t) as the imaginary part.

number

[0075] The sliding acceleration threshold calculation unit 32 calculates the mode of the values ​​on the envelope HL of the sliding interval waveform generated as described above. Specifically, the sliding acceleration threshold calculation unit 32 acquires the value of the sliding section waveform (acceleration value) for each of several time points set at constant minute time intervals within the sliding section, and creates a frequency distribution from the acquired multiple acceleration values. The sliding acceleration threshold calculation unit 32 acquires the mode, that is, the acceleration value that appears most frequently, from the created frequency distribution and sets this as the sliding acceleration threshold. In this modified example, the sliding acceleration threshold calculation unit 32 extracts sliding interval waveforms for each of the acceleration waveforms in the first direction X, the second direction Y, and the orthogonal direction Z, and calculates the sliding acceleration threshold.

[0076] The acceleration waveform correction unit 33 corrects the portion of the acceleration waveform whose absolute value is greater than the sliding acceleration threshold, and generates a corrected acceleration waveform. Figure 25 is an explanatory diagram showing an example of the acceleration waveform when an earthquake occurs and a smart device is in a sliding state, and the corrected acceleration waveform after applying corrections to it. As already explained, in the sliding section, as shown in Figure 25, with respect to the components W1 of the acceleration waveform in the first direction X and the second direction Y, the acceleration waveform WR that should be observed when the smart device 10 is fixed to the mount T is not observed. Instead, the amplitude becomes constant, resulting in a plateau portion WP, and in this portion, acceleration data is missing.

[0077] Therefore, the acceleration waveform correction unit 33 compensates for the missing parts of the acceleration data. Specifically, the acceleration waveform correction unit 33 generates a corrected sliding section waveform WC by interpolating the portion WP of the sliding section waveform where the amplitude of component W1 is constant, and generating the portion WR, which should be observed but whose absolute value is greater than the sliding acceleration threshold TH, by a sinusoidal curve fitting process. The acceleration waveform correction unit 33 generates a corrected acceleration waveform by replacing the portion of the acceleration waveform corresponding to the sliding section waveform with the corrected sliding section waveform WC. In this modified example, the acceleration waveform correction unit 33 generates a corrected sliding section waveform WC by interpolating the acceleration data from two points before and two points after the point where the acceleration plateaus, as shown as point P in Figure 25, using a sinusoidal curve fitting process (curve fitting). For curve fitting, for example, a nonlinear least squares method can be used. In this way, the acceleration waveform correction unit 33 generates a corrected acceleration waveform based on the acceleration waveform. In this modified example, the acceleration waveform correction unit 33 generates a corrected acceleration waveform for each of the acceleration waveforms in the first direction X, the second direction Y, and the orthogonal direction Z.

[0078] The acceleration waveform acquisition unit 23 generates an azimuth-adjusted acceleration waveform by using the corrected acceleration waveform generated as described above from an acceleration waveform obtained from a certain smart device 10 as the acceleration waveform obtained from the smart device 10. The subsequent processing is as described in the embodiment.

[0079] Next, we will explain the seismic intensity estimation method using the seismic intensity estimation system 1A described above, using Figure 26. Figure 26 is a flowchart of the seismic intensity estimation method in this modified example. In this modified example, steps S1 to S3 are performed as described with reference to Figure 15 in the above embodiment. Specifically, first, the building designation unit 21 receives designation of a building for which the seismic intensity class will be estimated by the seismic intensity estimation system 1. The building designation unit 21 also receives information regarding the height of the floor 111 of the building for which the seismic intensity class will be estimated (step S1). The smart device search unit 22 searches for multiple smart devices 10 located near a specified height in a building designated by the building designation unit 21 when an earthquake occurs (step S2). The acceleration waveform acquisition unit 23 collects acceleration waveforms acquired by the acceleration sensor 11a during an earthquake from each of the multiple smart devices 10 (Step S3: Acceleration waveform acquisition step). In this modified example, in step S3, the acceleration waveform collection unit 23 collects acceleration waveforms from each of the multiple smart devices 10, and at the same time collects magnetic flux density waveforms acquired by the magnetic sensor 11b during an earthquake.

[0080] Next, the sliding section identification unit 31 identifies a sliding section, which is a time interval when the smart device 10 slides on the support frame T, based on the measurement results from the earthquake information acquisition sensor 11 (step S11). Next, the sliding acceleration threshold calculation unit 32 generates an envelope for the sliding interval waveform and calculates the mode of the values ​​on the envelope as the sliding acceleration threshold, which is the threshold of the acceleration at which the smart device 10 slides on the support frame T (step S12). Then, the acceleration waveform correction unit 33 corrects the portion of the acceleration waveform WP whose absolute value is greater than the sliding acceleration threshold, and generates a corrected acceleration waveform (step S13). Subsequently, the process proceeds to step S4 as described in the above embodiment, and the acceleration waveform adjustment unit 24 generates the azimuth-adjusted acceleration waveform. In this case, the acceleration waveform adjustment unit 24 uses the corrected acceleration waveform generated as described above as the acceleration waveform obtained from the smart device 10 to generate the azimuth-adjusted acceleration waveform.

[0081] In this modified seismic intensity estimation system 1A, a sliding section identification unit 31 identifies a sliding section, which is a time interval when the smart device 10 slides on the support base T, based on the measurement results from the earthquake information acquisition sensor 11; a sliding acceleration threshold calculation unit 32 generates an envelope HL for the sliding section waveform, which is the waveform corresponding to the sliding section of the acceleration waveform of the smart device 10, and calculates the mode of the value on the envelope HL as the sliding acceleration threshold TH, which is the threshold of the acceleration when the smart device 10 slides on the support base T; and an acceleration waveform correction unit 33 corrects the portion WP of the sliding section waveform whose absolute value is greater than the sliding acceleration threshold TH to generate a corrected acceleration waveform. The acceleration waveform adjustment unit 24 uses the corrected acceleration waveform generated for the smart device 10 as the acceleration waveform of the smart device 10 to generate an azimuth-adjusted acceleration waveform. During an earthquake, if the smart device 10, which was stationary and placed on a support frame T, is large enough that a force exceeding the static frictional force between the smart device 10 and the support frame T on which it is installed acts on the smart device 10, it may slide relative to the support frame. In this state, a portion WP occurs in the acceleration waveform where the amplitude becomes constant and plateaus, and in this portion, acceleration data is lost. In contrast, with the above configuration, the sliding section identification unit 31 first identifies the sliding section, which is the time interval when the smart device 10 slides on the support frame T, based on the measurement results from the earthquake information acquisition sensor 11. Next, the sliding acceleration threshold calculation unit 32 generates an envelope HL for the sliding section waveform, which is the waveform of the acceleration waveform of the smart device 10 that corresponds to the sliding section identified as described above. Such an envelope HL generally has a shape that connects the parts of the sliding section waveform that have maximum values. Here, when the smart device 10 slides on the support frame T and a part occurs in the acceleration waveform where the amplitude becomes constant and plateaus, it is considered highly likely that the acceleration value of this plateauing part will be the mode of the frequency distribution when values ​​are acquired from this envelope HL at regular time intervals and a frequency distribution is created from the acquired values. Therefore, if the mode of the values ​​on the envelope HL generated as described above is calculated, that value can be considered to be the acceleration value of the part WP where sliding occurs and the amplitude of the acceleration waveform becomes constant and plateaus. Based on this idea, the sliding acceleration threshold calculation unit 32 calculates the mode of the values ​​on the envelope HL as the sliding acceleration threshold TH, which is the threshold for the acceleration at which the smart device 10 slides on the support base T. Then, the acceleration waveform correction unit 33 corrects the portion WP of the acceleration waveform whose absolute value is greater than the sliding acceleration threshold TH, and generates a corrected acceleration waveform. With the above configuration, the sliding acceleration threshold TH, which is the threshold for the acceleration at which the smart device 10 slides on the support base T, is set with high accuracy, so that the acceleration waveform correction unit 33 can accurately estimate and generate the acceleration waveform of a non-sliding state, even when the smart device 10 slides. In this way, the accurately estimated and generated corrected acceleration waveform is used as the acceleration waveform of the smart device 10, and subsequent processing, including the generation of the azimuth-adjusted acceleration waveform, is performed. Therefore, even when the smart device 10 slides, it becomes possible to further improve the accuracy of the seismic intensity scale estimation.

[0082] (Example of calculating the sliding acceleration threshold using a modified seismic intensity estimation system) Next, we will explain an example of calculating the sliding acceleration threshold TH using the seismic intensity estimation system 1A of the above modified example. Figure 27 shows a first example of calculating the sliding acceleration threshold. In the first example, the Tsukidate wave (M7.2 earthquake) normalized to 200 gal was used. Figure 27(a) shows the horizontal acceleration waveform obtained from the shaking table experiment. Figure 27(b) shows the vertical magnetic flux density waveform obtained from the magnetic sensor 11b. Figure 27(c) shows the STA / LTA ratio calculated by the sliding section identification unit 31. The sliding sections calculated based on the STA / LTA ratio are indicated with the symbol SA. Furthermore, Figure 27(a) shows the envelope HL obtained by calculating the absolute value |a(t)| of the analyzed signal a(t). Figure 27(d) shows the frequency distribution calculated for this envelope HL. The frequency distribution was created by dividing the range from 0 to 1000 gal into 40 segments of 25 gal each. In this frequency distribution, 150 to 175 gal was obtained as the mode.

[0083] Figure 28 shows a second example of calculating the sliding acceleration threshold. In the second example, the Tsukidate wave (M7.2 earthquake) normalized at 400 gal was used. In the second example, as shown in Figure 28(d), 125-150 gal was obtained as the most frequent value.

[0084] Figure 29 shows a third example in which the sliding acceleration threshold was calculated. In the third example, the Kumamoto wave (M6.5 earthquake) normalized to 400 gal was used. In the third example, as shown in Figure 29(d), 125-150 gal was obtained as the most frequent value.

[0085] It should be noted that the seismic intensity estimation system of the present invention is not limited to the embodiments and modifications described above with reference to the drawings, and various other modifications are conceivable within its technical scope. For example, in the above modified example, the sliding section identification unit 31 identified the sliding section based on the time history waveform of the magnetic flux density acquired by the magnetic sensor 11b, but instead, the sliding section identification unit 31 may identify the sliding section based on the acceleration waveform. In addition to the above, it is possible to select or discard the configurations listed in the above embodiments and their respective modifications, or to change them to other configurations as appropriate. [Explanation of Symbols]

[0086] 1. 1A Seismic Intensity Estimation System 25 Frame Vibration Characteristic Derivation Section 10, 10-1~10-n Smart Device 26 Floor Seismic Intensity Scale Estimation Unit 11 Earthquake information acquisition sensor 111 floor 11a Accelerometer T, T1~Tn Mount 11b Magnetic sensor HL envelope 23 Acceleration waveform acquisition unit TH Sliding acceleration threshold 24 Acceleration waveform adjustment section

Claims

1. A seismic intensity estimation system using a smart device with a built-in earthquake information acquisition sensor, Multiple stands are provided on the same floor, and each of the multiple smart devices is placed on one of the stands. The aforementioned earthquake information acquisition sensor is equipped with an acceleration sensor. An acceleration waveform collection unit collects acceleration waveforms acquired by the acceleration sensor during an earthquake from each of the multiple smart devices, An acceleration waveform adjustment unit rotates the acceleration waveforms of multiple smart devices so that their orientations coincide in the horizontal plane, thereby generating an orientation-adjusted acceleration waveform. A frame vibration characteristic derivation unit that derives the vibration characteristics of at least one frame based on the orientation-adjusted acceleration waveform of each of the smart devices in the pair of smart devices, among the multiple smart devices, A floor seismic intensity class estimation unit estimates the seismic intensity class of the floor by deriving the Fourier spectrum of the floor based on the acceleration waveform of the smart device placed on the frame from which the vibration characteristics have been derived, and the vibration characteristics of the frame, A seismic intensity estimation system characterized by having the following features.

2. The aforementioned vibration characteristics are natural frequencies. The frame vibration characteristic derivation unit determines the relationship between one of the pair of smart devices and the other smart device. By dividing the Fourier spectrum of the azimuth-adjusted acceleration waveform of the first smart device by the Fourier spectrum of the azimuth-adjusted acceleration waveform of the other smart device, the ratio of the transfer functions between the stand on which the first smart device is mounted and the stand on which the other smart device is mounted is calculated. The frequency that has the peak value within the ratio of the transfer function is defined as the natural frequency of the frame on which the first smart device is mounted. The seismic intensity estimation system according to feature 1.

3. The aforementioned vibration characteristics are natural frequencies. The floor seismic intensity scale estimation unit is, Assuming that the response of the aforementioned mounting structure is the response of a single-mass system, the transfer function of the single-mass system is calculated based on the natural frequency of the aforementioned mounting structure. The Fourier spectrum of the floor is derived by dividing the Fourier spectrum of the acceleration waveform of the smart device placed on the stand from which the vibration characteristics were derived by the transfer function of the single-mass system. The seismic intensity estimation system according to feature 1.

4. Based on the measurement results from the earthquake information acquisition sensor, a sliding section identification unit identifies a sliding section, which is a time interval when the smart device slides on the mounting base. A sliding acceleration threshold calculation unit generates an envelope for the sliding interval waveform, which is the waveform corresponding to the sliding interval of the acceleration waveform of the smart device, and calculates the mode of the values ​​on the envelope as the sliding acceleration threshold, which is the threshold of the acceleration at which the smart device slides on the stand. An acceleration waveform correction unit corrects the portion of the acceleration waveform whose absolute value is greater than the sliding acceleration threshold to generate a corrected acceleration waveform. Equipped with, The acceleration waveform adjustment unit uses the corrected acceleration waveform generated for the smart device as the acceleration waveform of the smart device to generate the azimuth-adjusted acceleration waveform. The seismic intensity estimation system according to any one of claims 1 to 3.

Images