Earthquake sensitive instrument, earthquake sensitive method and program

Abstract

To enable accurate determination of whether detected vibration is caused by an earthquake.SOLUTION: An earthquake sensitive instrument 1 comprises a detection unit 20 that detects vibration of a building (structure) and outputs a detection signal indicating the detection result as its function, a waveform output unit 41 that outputs a frequency distribution waveform or time-series acceleration waveform (output waveform) based on the detection signal, and a determination unit 42 that determines whether an earthquake has occurred in the vicinity of the building (structure) based on the output waveform. The determination unit 42 determines that an earthquake has occurred if the output waveform is in a predetermined frequency domain. Additionally, the determination unit 42 uses an inference model to infer whether the input time-series acceleration waveform of the output waveform is caused by an earthquake and outputs an inference result, thereby determining whether an earthquake has occurred.SELECTED DRAWING: Figure 2

Description

Technical Field

[0001] This invention relates to a technique for detecting the occurrence of an earthquake.

Background Art

[0002] Conventionally, earthquake detectors for detecting the occurrence of an earthquake have been known. For example, when an earthquake occurs, an earthquake detector detects the vibration of the earthquake and outputs a detection signal, and causes a predetermined operation for dealing with the earthquake to be performed.

[0003] For example, in a cooking appliance (such as a gas stove or an electromagnetic induction heating device) that performs cooking, when an earthquake occurs, a heating device that stops heating at the time of the earthquake and ensures safety by detecting the vibration of the earthquake and outputting a detection signal is known (see Patent Document 1).

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] By the way, depending on the environment where the earthquake detector is installed, vibrations other than earthquakes may occur regularly, so the earthquake detector may malfunction. Therefore, in the technique described in Patent Document 1, it is determined whether the detected vibration is due to an earthquake according to the vibration level (acceleration). However, when the vibration level other than the earthquake is not much different from the vibration level of the earthquake, there is a problem that it is difficult to determine whether the detected vibration is due to the earthquake.

[0006] Therefore, an object of the present invention is to provide an earthquake detection device, an earthquake detection method, and a program that can accurately determine whether the detected vibration is due to an earthquake.

Means for Solving the Problem

[0007] In order to solve the above problems, the earthquake detection device according to the present invention includes a detection unit that detects vibrations of a structure and outputs a detection signal indicating the detection result, an output unit that outputs frequency distribution data based on the detection signal, and a determination unit that determines whether an earthquake has occurred in the vicinity of the structure based on the frequency distribution data. The frequency distribution data includes the frequency distribution data due to an earthquake and the frequency distribution data due to other than an earthquake. The determination unit determines that an earthquake has occurred in the vicinity of the structure when the frequency of the frequency distribution data is in a substantially predetermined frequency region. This is the gist of the invention.

[0008] The earthquake detection device according to the present invention may be configured such that the determination unit determines whether an earthquake has occurred in the vicinity of the structure based on an inference result obtained by inputting the frequency distribution data into a machine learning model that has performed machine learning based on past frequency distribution data.

[0009] The earthquake detection device according to the present invention includes a detection unit that detects vibrations of a structure and outputs a detection signal indicating the detection result, an output unit that outputs time-series acceleration data based on the detection signal, and a determination unit that determines whether an earthquake has occurred in the vicinity of the structure based on the time-series acceleration data. The determination unit determines whether an earthquake has occurred in the vicinity of the structure based on an inference result obtained by inputting the time-series acceleration data into a machine learning model that has performed machine learning based on past time-series acceleration data. This is the gist of the invention.

[0010] The earthquake detection device according to the present invention may be configured such that the determination unit determines whether an earthquake has occurred in the vicinity of the structure based on both the determination result based on the time-series acceleration data and the inference result by the machine learning model.

[0011] In the earthquake detection device according to the present invention, the determination unit may determine whether an earthquake has occurred in the vicinity of the structure based on the difference between the time-series acceleration data and the output data from the machine learning model.

[0012] In the earthquake detection device according to the present invention, when the difference between the time-series acceleration data and the output data from the machine learning model is equal to or greater than a predetermined threshold value, the determination unit may determine that an earthquake has occurred in the vicinity of the structure.

[0013] In the earthquake detection device according to the present invention, when the determination result based on the time-series acceleration data is determined to be an earthquake occurrence and the inference result from the machine learning model is determined that no earthquake has occurred, the determination unit may determine that no earthquake has occurred in the vicinity of the structure.

[0014] In the earthquake detection device according to the present invention, the machine learning model may be configured by a recurrent neural network that performs prediction of time-series data.

[0015] In the earthquake detection device according to the present invention, the machine learning model may be configured by an LSTM including a forget gate.

[0016] An earthquake detection method according to the present invention is an earthquake detection method to be executed by a computer including a processor and a memory. The processor executes a detection step of detecting vibrations of a structure and outputting a detection signal indicating the detection result, an output step of outputting frequency distribution data based on the detection signal, and a determination step of determining whether an earthquake has occurred in the vicinity of the structure based on the frequency distribution data. The frequency distribution data includes the frequency distribution data due to an earthquake and the frequency distribution data due to other than an earthquake. In the determination step, when the frequency of the frequency distribution data is in a substantially predetermined frequency region, it is determined that an earthquake has occurred in the vicinity of the structure.

[0017] The earthquake detection method according to the present invention is an earthquake detection method to be executed by a computer including a processor and a memory, wherein the processor executes a detection step of detecting vibrations of a structure and outputting a detection signal indicating the detection result, an output step of outputting time-series acceleration data based on the detection signal, and a determination step of determining whether an earthquake has occurred in the vicinity of the structure based on the time-series acceleration data. In the determination step, it is characterized in that whether an earthquake has occurred in the vicinity of the structure is determined based on an inference result obtained by inputting the time-series acceleration data into a machine learning model trained based on past time-series acceleration data.

[0018] The program according to the present invention is a program for causing a computer including a processor and a memory to execute, wherein the processor is caused to execute a detection step of detecting vibrations of a structure and outputting a detection signal indicating the detection result, an output step of outputting frequency distribution data based on the detection signal, and a determination step of determining whether an earthquake has occurred in the vicinity of the structure based on the frequency distribution data. The frequency distribution data includes the frequency distribution data due to an earthquake and the frequency distribution data due to other than an earthquake. In the determination step, it is characterized in that when the frequency of the frequency distribution data is in a substantially predetermined frequency region, it is determined that an earthquake has occurred in the vicinity of the structure.

[0019] Further, the program according to the present invention is a program for causing a computer including a processor and a memory to execute, wherein the processor is caused to execute a detection step of detecting vibrations of a structure and outputting a detection signal indicating the detection result, an output step of outputting time-series acceleration data based on the detection signal, and a determination step of determining whether an earthquake has occurred in the vicinity of the structure based on the time-series acceleration data. In the determination step, it is characterized in that whether an earthquake has occurred in the vicinity of the structure is determined based on an inference result obtained by inputting the time-series acceleration data into a machine learning model trained based on past time-series acceleration data.

Advantages of the Invention

[0020] According to the earthquake detection device, earthquake detection method, and program of the present invention, the frequency distribution data based on the detection signal detecting vibration includes frequency distribution data due to an earthquake and frequency distribution data due to other than an earthquake. When the frequency of the frequency distribution data is in a substantially predetermined frequency range, it is determined that an earthquake has occurred. Therefore, it becomes possible to detect an earthquake based on the frequency distribution data detecting vibration. Thereby, it is possible to accurately determine whether the detected vibration is due to an earthquake or due to other than an earthquake.

[0021] According to the earthquake detection device of the present invention, based on the inference result of inputting the frequency distribution data into the machine learning model that has performed machine learning based on the past frequency distribution data, it is determined whether an earthquake has occurred. Thereby, it is possible to more accurately determine whether the detected vibration is due to an earthquake or due to other than an earthquake.

[0022] According to the earthquake detection device, earthquake detection method, and program of the present invention, based on the inference result of inputting the time series acceleration data based on the detection signal detecting vibration into the machine learning model that has performed machine learning based on the past time series acceleration data, it is determined whether an earthquake has occurred. Therefore, it becomes possible to detect an earthquake based on the time series acceleration data detecting vibration. Thereby, it is possible to accurately determine whether the detected vibration is due to an earthquake or due to other than an earthquake.

[0023] According to the earthquake detection device of the present invention, it is determined whether an earthquake has occurred based on the determination result based on the time-series acceleration data and the inference result by the machine learning model. At this time, for example, it is determined whether an earthquake has occurred based on the difference between the time-series acceleration data and the output data by the machine learning model. Further, for example, when the difference between the time-series acceleration data and the output data by the machine learning model is equal to or greater than a predetermined threshold value, it is determined that an earthquake has occurred. Thereby, it is possible to more accurately determine whether the detected vibration is due to an earthquake or due to other than an earthquake.

[0024] According to the earthquake detection device of the present invention, when the determination result based on the time-series acceleration data is determined to be an earthquake occurrence and the inference result by the machine learning model is determined that no earthquake has occurred, it is determined that no earthquake has occurred. Thereby, it is possible to more accurately determine whether the detected vibration is due to an earthquake or due to other than an earthquake.

[0025] According to the earthquake detection device of the present invention, the machine learning model is constituted by a recurrent neural network that performs prediction of time-series data. Or, the machine learning model is constituted by an LSTM including a forgetting gate. Thereby, it is possible to determine the occurrence of an earthquake based on the time-series acceleration data and to prevent a decrease in accuracy.

Brief Description of Drawings

[0026]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Best Mode for Carrying Out the Invention

[0027] Hereinafter, the present invention will be described based on the illustrated embodiments. The following description and drawings are examples for explaining the present invention, and for the sake of clarity of explanation, appropriate omissions and simplifications are made. The present invention can also be implemented in various other forms. Unless otherwise particularly limited, each component may be singular or plural. In the following description, the same components are denoted by the same reference numerals. Their names and functions are also the same. Therefore, detailed descriptions thereof will not be repeated.

[0028] (Embodiment 1) <1 Overall Configuration of Earthquake Detection System 100> FIG. 1 is a schematic diagram showing the overall configuration of an earthquake detection system 100 according to Embodiment 1.

[0029] As shown in FIG. 1, the earthquake detection system 100 includes an earthquake detection device 1 and an elevator EV, which is an example where the earthquake detection device 1 is installed. The earthquake detection device 1 and the elevator EV are connected to each other via a communication cable (not shown). The communication cable is a wired communication line for transmitting and receiving control signals, and is constituted by a wired interface such as various cables. Note that the earthquake detection device 1 may be installed in a structure other than the elevator EV, for example, the building itself in which the elevator EV is installed. However, in the present embodiment, it will be described as an example of being installed in the elevator EV.

[0030] Further, the earthquake detection system 100 may be configured as a device in which the earthquake detection device 1 and the elevator EV are integrated.

[0031] The earthquake detection device 1 is a device that detects the occurrence of an earthquake in the vicinity of a building or other structure (structure) where the elevator EV is installed and controls the operation of the elevator EV. Specifically, the earthquake detection device 1 detects the shaking caused by the occurrence of an earthquake as acceleration, and when the detected acceleration, that is, the degree of shaking due to the earthquake, is equal to or greater than a predetermined value, it performs operation control associated with the occurrence of the earthquake, for example, stops the operation of the elevator EV.

[0032] A device for controlling the elevator during an earthquake is required to be installed in accordance with the Building Standards Act in order to ensure the safety of the building during disasters such as earthquakes. Therefore, in order to comply with the standards, the elevator EV is equipped with an earthquake detection device 1 for detecting the occurrence of an earthquake.

[0033] The elevator EV is installed in various buildings and, as shown in FIG. 1, is a device that moves a person H up and down like an arrow L with the person H on board, thereby moving the person H to different floors in the building. The elevator EV is provided, for example, with a car for carrying the person H and a rope for moving the car, and is configured such that the car moves up and down in the hoistway of the building by winding the rope with a hoisting machine.

[0034] Here, the structures such as buildings where the elevator EV is installed are various types of buildings, etc., and there are cases where buildings generate vibrations regularly. A typical example is an elevator installed in the precincts of a station, which is installed to move between the floor of the moving passage in the precincts of the station and the floor of the platform where trains arrive and depart. In such a case, since trains arrive and depart regularly on the platform, the vibrations generated by the movement of the trains are also detected by the elevator EV. Therefore, in the earthquake detection device 1 installed in the elevator EV, when vibrations generated by the movement of the trains are detected, it is determined that they are vibrations due to an earthquake, and the elevator EV has been unnecessarily stopped.

[0035] Therefore, in the present embodiment, when the frequency distribution data of the detection signal detecting the vibration of a structure such as a building, for example, the frequency distribution waveform, is in a substantially predetermined frequency range, it is determined that an earthquake has occurred. Further, in the present embodiment, the time-series acceleration data of the detection signal detecting the vibration of a structure such as a building, for example, the time-series acceleration waveform, is input into a machine learning model obtained by performing machine learning based on past time-series acceleration waveforms, and it is determined whether an earthquake has occurred based on the inference result. With such a configuration, it is possible to more accurately determine whether the vibration is caused by an earthquake or by other factors than an earthquake.

[0036] <1.1 Configuration of Earthquake Detection Device 1> FIG. 2 is a functional block diagram showing the earthquake detection device 1 of FIG. 1. As shown in FIG. 2, the earthquake detection device 1 functions as a communication unit 10, a detection unit 20, a storage unit 30, and a control unit 40.

[0037] The communication unit 10 is a communication interface for performing wired communication with the elevator EV via a communication cable, and any communication protocol may be used as long as mutual communication can be executed.

[0038] The detection unit 20 has a function of detecting the vibration of a building or the like (structure) in which the earthquake detection device 1 and the elevator EV are installed, and outputting a detection signal indicating the detection result. Specifically, the detection unit 20 is composed of a vibration sensor that detects the vibration of a building or the like (structure), that is, displacement, an acceleration sensor that detects the acceleration of a building or the like (structure), and the like. When the detection unit 20 detects the displacement or acceleration of a building or the like (structure), it outputs displacement data, acceleration data, etc. as a detection signal.

[0039] The storage unit 30 stores programs, input data, etc. for executing various control processes and each function in the control unit 40, and is composed of a memory including a RAM (Random Access Memory), a ROM (Read Only Memory), etc., and a storage including an HDD (Hard Disk Drive), an SSD (Solid State Drive), a DRAM (Dynamic Random Access Memory), etc. Further, the storage unit 30 stores a detection signal database 31 and a machine learning model 32. Furthermore, the storage unit 30 temporarily stores the data generated in each process described later.

[0040] The detection signal database 31 stores the detection signals detected by the detection unit 20. In the detection signal database 31, for example, displacement data, acceleration data, etc., which are detection signals, are stored in time series together with the time data at the time of detection.

[0041] Note that each element stored in the detection signal database 31 does not necessarily need to be stored in the same physical database, as long as the information of each element is identified by its own unique identification information and each element is stored in association with the identification information.

[0042] The machine learning model 32 stores a machine learning model obtained by performing machine learning based on past detection signals such as displacement data and acceleration data by the detection unit 20. The machine learning model 32 is, for example, a machine learning model in which machine learning is performed together with result data indicating whether or not it was vibration due to an earthquake, for example, frequency distribution data shown from displacement data detecting the vibration of a building (structure), for example, a frequency distribution waveform. When the frequency distribution data is input to this machine learning model 32, an inference is made as to whether or not it is vibration due to an earthquake, and the inference result is output.

[0043] Further, the machine learning model 32 is a machine learning model in which machine learning is performed together with result data indicating whether or not vibration is due to an earthquake for a time-series acceleration waveform shown from acceleration data detecting the acceleration of a building or the like (structure). When the time-series acceleration waveform is input, this machine learning model 32 makes an inference as to whether or not the vibration is due to an earthquake and outputs the inference result.

[0044] The machine learning model 32 is constituted by, for example, a deep learning model (deep neural network model). A deep learning model is a method that learns rules and patterns based on a large amount of learning data and enables data analysis without human intervention, and is a model used in fields such as image recognition, natural language processing, and anomaly detection. Further, the machine learning model 32 may be constituted by a Recurrent Neural Network (RNN) or LSTM (Long Short-Term Memory). A recurrent neural network is a method that recursively uses calculation results for time-series data that changes over time and can give time dependence. Further, LSTM is provided with an input gate, an output gate, and a forget gate, and has a function of forgetting time-series data from long-term memory, and thus is a method capable of making predictions associated with time dependence from short term to long term.

[0045] The control unit 40 controls the overall operation of the earthquake detection device 1 by executing a program stored in the storage unit 30. As functions of the control unit 40, a waveform output unit (output unit) 41, a determination unit 42, and an elevator operation control unit 43 are provided. The waveform output unit 41, the determination unit 42, and the elevator operation control unit 43 are activated by a program stored in the storage unit 30 and executed by the earthquake detection device 1.

[0046] The waveform output unit 41 controls the process of outputting a frequency distribution waveform, which is an example of frequency distribution data based on the detection signal detected by the detection unit 20. For example, the waveform output unit 41 classifies displacement data, which is the detection signal of the vibration sensor stored in the detection signal database 31, by frequency bands and outputs it as frequency distribution data, for example, a frequency distribution waveform. Note that the frequency distribution data at this time includes that caused by earthquake vibrations and that caused by vibrations other than earthquake vibrations (for example, vibrations caused by the movement of trains).

[0047] Also, the waveform output unit 41 controls the process of outputting a time-series acceleration waveform, which is an example of time-series acceleration data based on the detection signal detected by the detection unit 20. For example, the waveform output unit 41 uses the acceleration data, which is the detection signal of the acceleration sensor stored in the detection signal database 31, as time-series data and outputs it as a time-series acceleration waveform. Note that the time-series acceleration waveform at this time includes that caused by earthquake vibrations and that caused by vibrations other than earthquake vibrations (for example, vibrations caused by the movement of trains).

[0048] The determination unit 42 controls the process of determining whether an earthquake has occurred in the vicinity of a building or other structure based on the frequency distribution data output by the waveform output unit 41. Specifically, the determination unit 42 determines that an earthquake has occurred when the frequency of the frequency distribution waveform, which is an example of the frequency distribution data, is in a substantially predetermined frequency region. Note that the determination unit 42 may use an inference model, which is an example of the machine learning model 32, to infer whether the frequency distribution data is vibration caused by an earthquake when the frequency distribution data is input and output the inference result to determine whether an earthquake has occurred.

[0049] FIG. 3 is a diagram showing an example of the determination of earthquake occurrence by the determination unit 42 in FIG. 2. The frequency distribution waveform, which is an example of the frequency distribution data shown in FIG. 3(a), is an example of the frequency distribution waveform of the detection signal detected by the detection unit 20, and it is shown that the vibration waveforms are distributed in various frequency bands. Here, it is known that the frequency band region of the vibration due to an earthquake is generally from about 1 Hz to 10 Hz. For example, as in the frequency distribution waveform shown in FIG. 3(b), when the frequency band region of the vibration is within the range of 1 Hz to 10 Hz, it can be determined as the vibration due to an earthquake. On the other hand, as in the frequency distribution waveform shown in FIG. 3(c), when the frequency band region of the vibration is outside the range of 1 Hz to 10 Hz, it can be determined as the vibration due to an earthquake. However, as in the frequency distribution waveform shown in FIG. 3(a), there may be a case where the frequency band region of the vibration is mixed within and outside the range of 1 Hz to 10 Hz. In particular, when the elevator is installed inside the station building as described above, since the building itself vibrates, there may be a case where it is mixed within and outside the range of 1 Hz to 10 Hz.

[0050] Therefore, the determination unit 42 makes a determination as follows, for example. · When the frequency band region is substantially (mainly) within the range of 1 Hz to 10 Hz: Determine that it is during an earthquake (earthquake occurrence). · When the frequency band region is within both the range of 1 Hz to 10 Hz and outside the range: Determine that it is normal times (no earthquake occurrence). · When the frequency band region is substantially (mainly) outside the range of 1 Hz to 10 Hz: Determine that it is normal times (no earthquake occurrence). · When there is no vibration: Determine that it is normal times (no earthquake occurrence). Thus, the vibration when an earthquake occurs is accurately determined.

[0051] In addition, the determination unit 42 controls a process of determining whether an earthquake has occurred in the vicinity of a building or the like (structure) based on the time-series acceleration waveform, which is an example of the time-series acceleration data output by the waveform output unit 41. Specifically, when the time-series acceleration waveform, which is an example of the time-series acceleration data and is an example of the machine learning model 32, is input, the inference model that makes an inference as to whether it is vibration due to an earthquake and outputs the inference result is used to determine whether an earthquake has occurred.

[0052] For example, the determination unit 42 determines whether an earthquake has occurred based on the determination result based on the time-series acceleration waveform output by the waveform output unit 41 and the inference result obtained by inputting the time-series acceleration waveform into the machine learning model 32. Specifically, the determination unit 42 makes determinations as follows, for example. · When the determination result of the time-series acceleration waveform is normal time (no earthquake occurrence) and the inference result of the machine learning model is normal time (no earthquake occurrence): Determine normal time (no earthquake occurrence) At this time, since the determination results of both are the same, it is considered that no earthquake has occurred. · When the determination result of the time-series acceleration waveform is normal time (no earthquake occurrence) and the inference result of the machine learning model is earthquake time (earthquake occurrence): Determine normal time (no earthquake occurrence) At this time, it is considered to be a misjudgment of the machine learning model 32. · When the determination result of the time-series acceleration waveform is earthquake time (earthquake occurrence) and the inference result of the machine learning model is normal time (no earthquake occurrence): Determine normal time (no earthquake occurrence) At this time, it is considered to be an event that the machine learning model 32 should learn. · When the determination result of the time-series acceleration waveform is earthquake time (earthquake occurrence) and the inference result of the machine learning model is earthquake time (earthquake occurrence): Determine earthquake time (earthquake occurrence) At this time, since the determination results of both are the same, it is considered that an earthquake has occurred. Thereby, the vibration when an earthquake occurs is accurately determined.

[0053] Also, for example, the determination unit 42 determines whether an earthquake has occurred based on the difference between the time-series acceleration waveform output by the waveform output unit 41 and the output waveform of the inference result obtained by inputting the time-series acceleration waveform into the machine learning model 32. At this time, when the difference is equal to or greater than a predetermined threshold value, the determination unit 42 determines that an earthquake has occurred. This is because an earthquake is not an event that occurs frequently, so the opportunity for the machine learning model 32 to learn data during an earthquake is rare. Therefore, considering that the machine learning model 32 has not learned data during an earthquake, it focuses on the error between normal time (no earthquake occurrence) and earthquake time (earthquake occurrence).

[0054] The elevator operation control unit 43 controls the process of controlling the operation of the elevator EV. Specifically, when the determination result by the determination unit 42 indicates an earthquake (earthquake occurrence) and the seismic intensity of the earthquake is equal to or greater than a predetermined level, the elevator operation control unit 43 stops the operation of the elevator EV. This makes it possible to ensure the safety of buildings and the like during an earthquake.

[0055] <2. Flow of processing> With reference to FIG. 4, an example of the earthquake detection process (earthquake detection method) executed by the earthquake detection system 100 and the flow of the operation control of the elevator EV will be described. FIG. 4 is a flowchart showing the operation control process by the earthquake detection device 1 in FIG. 1.

[0056] As the process of step S101, the detection unit 20 of the earthquake detection device 1 detects the vibration of the building (structure) where the earthquake detection device 1 and the elevator EV are installed.

[0057] As the process of step S102, the detection unit 20 of the earthquake detection device 1 outputs a detection signal indicating the detection result for the vibration of the building (structure) detected in step S101 (detection step).

[0058] As the process of step S103, the waveform output unit 41 of the earthquake detection device 1 outputs the output waveform of the detection signal detected in step S102 (output step). In step S103, a frequency distribution waveform, which is an example of frequency distribution data based on the detection signal, is output from the displacement data which is the detection signal. Or, in step S103, a time-series acceleration waveform, which is an example of time-series acceleration data based on the detection signal, is output from the acceleration data which is the detection signal.

[0059] As the process of step S104, the determination unit 42 of the earthquake detection device 1 determines whether an earthquake has occurred in the vicinity of a building or other structure (structure) based on the frequency distribution waveform output in step S103 (determination step). At this time, if the frequency of the frequency distribution waveform is approximately (mainly) in a predetermined frequency range (for example, 1 Hz to 10 Hz), it is determined that an earthquake has occurred. If the frequency range includes values outside the range of 1 Hz to 10 Hz, it is determined that no earthquake has occurred.

[0060] Also, as the process of step S104, the determination unit 42 of the earthquake detection device 1 determines whether an earthquake has occurred in the vicinity of a building or other structure (structure) based on the time-series acceleration waveform output in step S103. At this time, using an inference model that, in the example of the machine learning model 32, makes an inference as to whether the time-series acceleration waveform is vibration due to an earthquake when the time-series acceleration waveform is input and outputs the inference result, it is determined whether an earthquake has occurred.

[0061] As the process of step S105, the earthquake detection device 1 performs a process branch based on the result determined in step S104. If the result determined in step S104 is normal time (no earthquake occurrence) ("Y" shown in FIG. 4), the process is terminated. If it is earthquake time (earthquake occurrence) ("N" shown in FIG. 4), the process proceeds to step S106.

[0062] As the process of step S106, the elevator operation control unit 43 of the earthquake detection device 1 controls the operation of the elevator EV and stops the operation of the elevator EV when it is earthquake time (earthquake occurrence) and the seismic intensity of the earthquake is a predetermined value or more.

[0063] <3. Effects> According to the earthquake detection device 1 and the earthquake detection method according to Embodiment 1, based on the frequency distribution waveform which is an example of the frequency distribution data based on the detection signal detecting the vibration of a building or the like (structure), when the frequency of the frequency distribution waveform is substantially (mainly) in a predetermined frequency range (1 Hz to 10 Hz), it is determined that an earthquake has occurred. When the frequency of the frequency distribution waveform includes outside the predetermined frequency range (1 Hz to 10 Hz), it is determined that an earthquake has not occurred. Therefore, it becomes possible to detect an earthquake based on the frequency distribution waveform detecting the vibration. Thereby, it becomes possible to accurately determine whether the detected vibration is due to an earthquake or due to vibration other than an earthquake.

[0064] Further, according to the earthquake detection device 1, when a time series acceleration waveform which is an example of the time series acceleration data based on the detection signal detecting the vibration of a building or the like (structure) is input, an inference as to whether the vibration is due to an earthquake is performed, and using an inference model which outputs the inference result, it is determined whether an earthquake has occurred. For example, based on the determination result based on the time series acceleration waveform and the inference result by a machine learning model, it is determined whether an earthquake has occurred. Also, for example, based on the difference between the time series acceleration waveform and the output waveform by the machine learning model, it is determined whether an earthquake has occurred. Therefore, it becomes possible to detect an earthquake based on the time series acceleration waveform detecting the vibration. Thereby, it becomes possible to accurately determine whether the detected vibration is due to an earthquake or due to vibration other than an earthquake.

[0065] (Embodiment 2 (Program)) FIG. 5 is a functional block configuration diagram showing an example of the configuration of a computer (electronic computer) 700 according to Embodiment 2. In the present embodiment, an example is shown in which the earthquake detection device 1 according to Embodiment 1 is an existing earthquake detection device, and the functions of the waveform output unit 41, the determination unit 42, and the elevator operation control unit 43 are realized by an external computer 700. The computer 700 includes a CPU 701, a main storage device 702, an auxiliary storage device 703, and an interface 704.

[0066] Here, the details of the program for realizing each function constituting the waveform output unit 41, the determination unit 42, and the elevator operation control unit 43 according to the first embodiment will be described. These functional blocks are implemented in the computer 700. And the operations of these respective components are stored in the auxiliary storage device 703 in the form of a program. The CPU 701 reads the program from the auxiliary storage device 703, expands it in the main storage device 702, and executes the above-described processing according to the program. Also, the CPU 701 secures a storage area corresponding to the above-described storage unit in the main storage device 702 according to the program.

[0067] Specifically, the program is a program that, in the computer 700, realizes a detection step of detecting the vibration of a structure and outputting a detection signal indicating the detection result, an output step of outputting frequency distribution data based on the detection signal or time series acceleration data based on the detection signal, and a determination step of determining whether an earthquake has occurred in the vicinity of the structure based on the frequency distribution data or the time series acceleration data, by the computer.

[0068] The CPU 701 is hardware for executing the instruction set described in the program, and is composed of an arithmetic unit, registers, peripheral circuits, etc. The CPU 701 is composed of at least one or more processors, and is typically a microprocessor, but may also be other types of processors including an MPU (Micro Processing Unit), a GPU (Graphics Processing Unit), a microprocessor, a processor core, and a multiprocessor. At least one processor may be a single core or a multi-core. Also, at least one processor may be a processor in a broad sense such as a hardware circuit (for example, an FPGA (Field-Programmable Gate Array) or an ASIC (Application Specific Integrated Circuit)) that performs part or all of the processing.

[0069] The main memory device 702 is for temporarily storing programs and data processed by the programs, etc., and is, for example, a memory such as a DRAM (Dynamic Random Access Memory). The main memory device 702 is at least one or more memories, and typically may be composed of a main memory device. The one or more memories may be volatile memories or non-volatile memories.

[0070] The auxiliary storage device 703 is an example of a non-temporary tangible medium. Other examples of non-temporary tangible media include magnetic disks, magneto-optical disks, CD-ROMs, DVD-ROMs, semiconductor memories, etc. connected via the interface 704. Also, when this program is distributed to the computer 700 via a network, the receiving computer 700 may expand the program in the main memory device 702 and execute the above-described processing.

[0071] Also, the program may be for realizing a part of the above-described functions. Furthermore, the program may be what realizes the above-described functions in combination with other programs already stored in the auxiliary storage device 703, i.e., a so-called difference file (difference program).

[0072] As described above, the embodiments of the present invention have been explained, but the specific configuration is not limited to the above-described embodiments, and even if there are design changes, etc. within the scope not departing from the gist of the present invention, they are included in the present invention.

Explanation of Reference Numerals

[0073] 1: Earthquake detection device 10: Communication unit 20: Detection unit 30: Storage unit 31: Detection signal database 32: Machine learning model 40: Control unit 41: Output unit 42: Judgment unit 43: Elevator operation control unit 100: Earthquake detection system 700: Computer 701: CPU 702: Main memory device 703: Auxiliary storage device 704: Interface

Claims

1. A detection unit that detects vibrations of a structure and outputs a detection signal indicating the detection result; An output unit that outputs frequency distribution data based on the detection signal; A determination unit that determines whether an earthquake has occurred in the vicinity of the structure based on the frequency distribution data, and comprises: The frequency distribution data includes frequency distribution data due to an earthquake and frequency distribution data due to other than an earthquake; When the frequency of the frequency distribution data is in a substantially predetermined frequency region, the determination unit determines that an earthquake has occurred in the vicinity of the structure. An earthquake sensing device characterized by the above.

2. The determination unit determines whether an earthquake has occurred in the vicinity of the structure based on an inference result obtained by inputting the frequency distribution data into a machine learning model that has performed machine learning based on past frequency distribution data. The earthquake sensing device according to claim 1, characterized by the above.

3. A detection unit that detects vibrations of a structure and outputs a detection signal indicating the detection result; An output unit that outputs time-series acceleration data based on the detection signal; A determination unit that determines whether an earthquake has occurred in the vicinity of the structure based on the time-series acceleration data, and comprises: The determination unit determines whether an earthquake has occurred in the vicinity of the structure based on an inference result obtained by inputting the time-series acceleration data into a machine learning model that has performed machine learning based on past time-series acceleration data. An earthquake sensing device characterized by the above.

4. The determination unit determines whether an earthquake has occurred in the vicinity of the structure based on a determination result based on the time-series acceleration data and an inference result by the machine learning model. The earthquake sensing device according to claim 3, characterized by the above.

5. The determination unit determines whether an earthquake has occurred in the vicinity of the structure based on a difference between the time-series acceleration data and output data by the machine learning model. The earthquake sensing device according to claim 4, characterized by the above.

6. When the difference between the time-series acceleration data and the output data by the machine learning model is equal to or greater than a predetermined threshold value, the determination unit determines that an earthquake has occurred in the vicinity of the structure. The earthquake sensing device according to claim 5, characterized by the above.

7. When the determination unit determines that an earthquake has occurred based on the time-series acceleration data and the inference result by the machine learning model determines that no earthquake has occurred, it determines that no earthquake has occurred in the vicinity of the structure. The earthquake detection device according to claim 4, characterized in that.

8. The machine learning model is composed of a recurrent neural network that predicts time-series data. The earthquake detection device according to any one of claims 3 to 6, characterized in that.

9. The machine learning model is composed of an LSTM including a forget gate. The earthquake detection device according to any one of claims 3 to 6, characterized in that.

10. An earthquake detection method to be executed by a computer including a processor and a memory, wherein the processor a detection step of detecting the vibration of the structure and outputting a detection signal indicating the detection result; an output step of outputting frequency distribution data based on the detection signal; a determination step of determining whether an earthquake has occurred in the vicinity of the structure based on the frequency distribution data, and the frequency distribution data includes the frequency distribution data due to an earthquake and the frequency distribution data due to other than an earthquake, in the determination step, when the frequency of the frequency distribution data is in a substantially predetermined frequency region, it is determined that an earthquake has occurred in the vicinity of the structure. An earthquake detection method characterized by that.

11. An earthquake detection method to be executed by a computer including a processor and a memory, wherein the processor a detection step of detecting the vibration of the structure and outputting a detection signal indicating the detection result; an output step of outputting time-series acceleration data based on the detection signal; a determination step of determining whether an earthquake has occurred in the vicinity of the structure based on the time-series acceleration data, and in the determination step, based on the inference result obtained by inputting the time-series acceleration data into a machine learning model that has performed machine learning based on past time-series acceleration data, it is determined whether an earthquake has occurred in the vicinity of the structure. An earthquake detection method characterized by that.

12. A program for causing a computer including a processor and a memory to execute, wherein the processor a detection step of detecting the vibration of the structure and outputting a detection signal indicating the detection result; an output step of outputting frequency distribution data based on the detection signal; a determination step of determining whether an earthquake has occurred in the vicinity of the structure based on the frequency distribution data, and causing the steps to be executed; the frequency distribution data includes the frequency distribution data due to an earthquake and the frequency distribution data due to other than an earthquake; in the determination step, when the frequency of the frequency distribution data is in a substantially predetermined frequency region, it is determined that an earthquake has occurred in the vicinity of the structure; A program characterized by the above.

13. A program for causing a computer including a processor and a memory to execute, to the processor, a detection step of detecting vibrations of a structure and outputting a detection signal indicating the detection result; an output step of outputting time-series acceleration data based on the detection signal; a determination step of determining whether an earthquake has occurred in the vicinity of the structure based on the time-series acceleration data, and causing the steps to be executed; in the determination step, based on an inference result obtained by inputting the time-series acceleration data into a machine learning model that has performed machine learning based on past time-series acceleration data, it is determined whether an earthquake has occurred in the vicinity of the structure; A program characterized by the above.

Images