Monitoring system, information processing device, and program

Abstract

A monitoring system according to the present invention comprises: n (n is an integer of not less than 3) oscillation sensors that are disposed so as not to be aligned on one straight line; and an information processing device configured to execute information processing based on oscillation waveform data which has been acquired by the n oscillation sensors. The information processing device includes a position estimation unit that estimates positional information of an oscillation source of the oscillation waveform data on the basis of respective pieces of the oscillation waveform data which have been acquired by the n oscillation sensors and respective pieces of positional information of the n oscillation sensors.

Description

Monitoring system, information processing device, and program

[0001] This invention relates to a monitoring system, an information processing device, and a program.

[0002] Conventionally, technologies have been known that use cameras and vibration sensors to detect intruders in order to prevent unauthorized entry into restricted areas (see Patent Document 1).

[0003] U.S. Patent Application Publication No. 2021 / 0104150

[0004] However, the inventors have come to recognize the following problems: Cameras have blind spots, and deploying a large number of cameras to fill these blind spots increases costs. The technology described in Patent Document 1 uses vibration sensors in addition to cameras, but it does not adequately compensate for blind spots, and the inventors realized that there is room to more accurately estimate the location of vibration sources such as intruders.

[0005] This invention has been made in view of these circumstances, and one of its exemplary objectives is to provide a technique for easily and accurately estimating the location of a vibration source.

[0006] One aspect of the present invention is a monitoring system. The monitoring system comprises n vibration sensors (where n is an integer of 3 or more) arranged so as not to be in a straight line, and an information processing device configured to perform information processing based on vibration waveform data acquired by the n vibration sensors. The information processing device has a position estimation unit that estimates the position information of the vibration source of the vibration waveform data based on the vibration waveform data acquired by each of the n vibration sensors and the position information of each of the n vibration sensors.

[0007] Furthermore, any combination of the above components, as well as conversions of the expression of the present invention between methods, apparatus, systems, recording media, computer programs, etc., are also valid embodiments of the present invention.

[0008] According to the present invention, it is possible to easily and accurately estimate the location of the vibration source.

[0009] This is a block diagram of a monitoring system according to one embodiment of the present invention. This is a functional block diagram of an information processing device according to the same embodiment. This is a diagram showing an example of a vibration waveform shown by a vibration waveform model according to the same embodiment. This is a block diagram showing an example of the hardware configuration of an information processing device according to the same embodiment. This is a diagram showing an example of the arrangement of vibration sensors according to the same embodiment. This is a diagram showing the time change in the intensity of vibration waves detected by three vibration sensors according to the same embodiment. This is a conceptual diagram for explaining a first example of identifying the type of vibration source based on vibration waveform data. This is a conceptual diagram for explaining a second example of identifying the type of vibration source based on vibration waveform data. This is a flowchart showing an example of the operation of an information processing device according to one embodiment of the present invention. Figure 10(a) is a diagram showing an example of a vibration waveform of an automobile observed by a seismograph, and Figure 10(b) is a spectrogram of the vibration waveform shown in Figure 10(a). Figure 11(a) is a diagram showing an example of a vibration waveform of a person observed by a seismograph, and Figure 11(b) is a spectrogram of the vibration waveform shown in Figure 11(a). Figure 12(a) is a diagram showing an example of a vibration waveform of noise observed by a seismograph, and Figure 12(b) is a spectrogram of the vibration waveform shown in Figure 12(a). Figure 13(a) shows an example of a human vibration waveform from test data, and Figure 13(b) shows the result of classifying the vibration source of the vibration waveform in Figure 13(a) using a trained model. Figure 14(a) shows a car vibration waveform from unknown vibration waveform data, Figure 14(b) shows the result of classifying the vibration source of the vibration waveform in Figure 14(a) using a trained model, and Figure 14(c) shows the result of classifying the vibration source of the vibration waveform in Figure 14(a) after transformation processing. Figure 15(a) shows the classification result when the car is in the range of 0 to 20 m from the seismometer, and Figure 15(b) shows the classification result when the car is in the range of 20 to 40 m from the seismometer. Figure 16(a) shows the classification result when the person is in the range of 0 to 20 m from the seismometer, and Figure 16(b) shows the classification result when the person is in the range of 20 to 40 m from the seismometer. This figure shows an example of a heatmap of RMS (Root Mean Square) error. Figures 18(a) to 18(j) show examples of the results of estimating the position of a moving vibration source, respectively.

[0010] (Embodiments) Hereinafter, embodiments for carrying out the present invention will be described in detail with reference to the drawings. In the description of the drawings, the same elements will be denoted by the same reference numerals, and redundant explanations will be omitted as appropriate.

[0011] Figure 1 is a block diagram of a monitoring system 1 according to one embodiment of the present invention. The monitoring system 1 may be used, for example, to prevent the theft of items within a monitoring area (for example, a premises). Specifically, the monitoring system 1 may be used to detect vehicles approaching a predetermined premises or to detect people entering the premises.

[0012] The monitoring system 1 according to this embodiment comprises an information processing device 10, n vibration sensors (first vibration sensor 20_1 to nth vibration sensor 20_n) arranged so as not to be in a straight line (n is an integer of 3 or more), and a mobile body 30. Hereinafter, when the first vibration sensor 20_1 to nth vibration sensor 20_n are not distinguished from each other, they will be collectively referred to simply as "vibration sensor 20". Furthermore, the n vibration sensors 20 refer to all of the first vibration sensor 20_1 to nth vibration sensor 20_n.

[0013] The information processing device 10, n vibration sensors 20, and mobile body 30 are connected to each other via a network 15 so that they can communicate with one another. This connection may be wireless or wired. The network 15 includes, for example, the Internet and a LAN (Local Area Network).

[0014] The vibration sensor 20 is a sensor capable of detecting various vibration waves, such as seismic waves and sound waves. The vibration sensor 20 according to this embodiment is a seismic sensor. The vibration sensor 20 detects vibration waves generated by a vibration source and acquires vibration waveform data. The type of vibration source may be, for example, a bus, a light truck, a van (minivan, large van), a passenger car (small car, medium car, large car), a motorcycle, or a person.

[0015] The vibration sensor 20 may be a one-dimensional sensor that detects only the z-axis (vertical) component of the vibration wave, or it may be a three-dimensional sensor that can detect the three-dimensional components (vertical and horizontal components) of the vibration wave. Using a one-dimensional sensor is less expensive, while using a three-dimensional sensor results in higher accuracy information processing in the information processing device 10. For this reason, the vibration sensor 20 can be used as either a one-dimensional or a three-dimensional sensor as needed.

[0016] The vibration sensor 20 according to this embodiment has a wireless communication function that transmits acquired vibration waveform data to the information processing device 10 in real time. The vibration waveform data acquired by the vibration sensor 20 is transmitted to the information processing device 10 in real time via the network 15. Even if there is a delay in the transmission of vibration waveform data, if it is sequential processing, this processing is considered to be in real time.

[0017] The information processing device 10 is configured to perform information processing based on vibration waveform data acquired by n vibration sensors 20. In this embodiment, the information processing device 10 performs information processing based on vibration waveform data transmitted from the vibration sensors 20 in real time (for example, estimation of the location information of the vibration source and identification of the type of vibration source). If the information processing based on the vibration waveform data transmitted from the vibration sensors 20 is sequential, then this processing is considered to be in real time.

[0018] The information processing device 10 may collaborate with security companies and any external services and cloud services. These services may include, for example, an AI (Artificial Intelligence) chatbot that provides answers to any question.

[0019] The mobile body 30 may be, for example, an aircraft, a vehicle, or a ship. An aircraft may be, for example, a drone. The mobile body 30 may move or fly in response to operations by an operator, or it may move or fly unmanned by an autonomous driving function. The mobile body 30 may have wireless communication functions for sending and receiving various types of data via the network 15.

[0020] The mobile body 30 according to this embodiment has an imaging device 32. The imaging device 32 captures an image including a vibration source identified as meeting threat conditions described later during the movement of the mobile body 30. The imaging device 32 (or the mobile body 30 equipped with it) may be arranged in the central region of the n vibration sensors 20. For example, when there are three vibration sensors, it may be arranged inside the triangle formed by connecting the three vibration sensors in a straight line. When the direction of the vibration source is known in advance by the estimation of the position estimation unit of the information processing device 10 described later, the imaging device 32 may be directed in that direction. Thereby, it becomes possible to more reliably capture an image including the vibration source.

[0021] FIG. 2 is a functional block diagram of the information processing device 10 according to this embodiment. The information processing device 10 according to this embodiment includes a communication unit 100, a storage unit 120, and a processing unit 140.

[0022] The communication unit 100 transmits and receives various information via the network 15. For example, it receives the received data D １ or transmits the transmission data D ２ .

[0023] The received data D １ is, for example, vibration waveform data acquired by the vibration sensor 20. The received data D １ is transmitted to the storage unit 120 and the processing unit 140.

[0024] The transmission data D ２ includes data generated by the processing of the processing unit 140 and may include, for example, data indicating alarm information. The alarm information may include that a vibration source satisfying the threat conditions described later has been detected, information indicating the time when such a vibration source was detected, and information regarding the vibration source (for example, the type and position information of the vibration source). Alternatively, the transmission data D ２ may include, for example, a signal for operating a security device arranged within the site (for example, outputting a warning sound). The transmission data D ２ may be transmitted via the network 15 to, for example, the terminal of the administrator of the monitoring system 1, the mobile body 30, the security device arranged within the site, the terminal of an external security company, etc.

[0025] The storage unit 120 stores various types of data, for example, learning data, a learned model, data indicating pre-determined threat conditions, a database of vibration waveform models, received data D １ and a processing program, etc. The processing program is a program for the processing unit 140 to execute various information processes (for example, machine learning, estimation of the position information of the vibration source, identification of the type of the vibration source, identification of the vibration source satisfying the threat conditions, etc.).

[0026] The learning data includes data based on vibration waves. Specifically, it includes vibration waveform data in the time domain or data obtained by converting this data into the frequency domain (for example, data indicating a spectrogram). These data may be data of only one component (vertical component) of the vibration wave, or may be data of three components (vertical component and horizontal components) of the vibration wave. Using the data of three components can improve the accuracy of the estimation of the position information and the identification of the type of the vibration source.

[0027] The learning data may include various types of information associated with the vibration waveform data, for example, information such as the type of the vibration source, the position and speed of the vibration source, information regarding the weather (for example, clear, rain, and wind speed, etc.), and environmental conditions such as the terrain and geology of the location where the vibration sensor for acquiring the vibration waveform data is arranged.

[0028] The learned model may be machine-learned to output information regarding the type of the vibration source of the input vibration waveform data with the vibration waveform data as the input. Specifically, it may be machine-learned to classify the vibration source of the vibration waveform data. Also, a plurality of learned models may be stored in the storage unit 120, and these plurality of learned models may be constructed by machine learning according to the environmental conditions of the location where the vibration sensor is arranged respectively. Also, the learned model may be machine-learned to output information indicating the position information of the vibration source (the position of the vibration source and the moving direction of the vibration source) in addition to the type of the vibration source. The machine learning of these learned models will be described later.

[0029] A threat condition is a condition that defines an increased risk of a threat occurring or that a threat has occurred. A threat condition may be, for example, that a specified vibration source (e.g., a light truck) has stopped within a specified distance from the monitoring area, or that a specified vibration source (e.g., a person) has entered the monitoring area.

[0030] A vibration waveform model database is a database that stores vibration waveform models for multiple types of vibration sources. A vibration waveform model is data that shows a typical vibration waveform of that vibration source.

[0031] Figure 3 shows an example of a vibration waveform represented by the vibration waveform model according to this embodiment. Figure 3(a) shows an example of a bus vibration waveform 122, Figure 3(b) shows an example of a light truck vibration waveform 124, and Figure 3(c) shows an example of a minivan vibration waveform 126. As shown in Figures 3(a) to (c), different vibration waveforms are shown for each type of vibration source (bus, light truck, minivan). For example, by comparing the vibration waveform model showing these vibration waveforms with the vibration waveform data acquired by the vibration sensor 20, it is possible to identify the type of vibration source of the vibration waveform data.

[0032] Returning to Figure 2, the functions of the processing unit 140 will be explained. The processing unit 140 processes various data and generates various data. The processing unit 140 according to this embodiment includes a learning unit 142, a location estimation unit 144, a type identification unit 146, and a threat identification unit 148.

[0033] The learning unit 142 performs machine learning using the training data and generates a trained model. The trained model generated by the learning unit 142 is stored in the storage unit 120. The trained model may include various known models, such as models using neural networks like deep neural networks, convolutional neural networks, and recurrent neural networks.

[0034] The learning unit 142 may perform machine learning so that the trained model takes vibration waveform data as input and outputs information about the type of vibration source of the input vibration waveform data. The training data used in this machine learning is vibration waveform data for which the type of vibration source is known. By inputting vibration waveform data acquired by the vibration sensor 20 into this trained model, information about the type of vibration source of the input vibration waveform data is output.

[0035] The training data preferably includes information about weather. Furthermore, the training data preferably includes environmental conditions such as the topography and geology of the location where the vibration sensor that acquired the vibration waveform data is positioned.

[0036] A trained model may output probabilities for each type of vibration source. For example, suppose vibration waveform data from one type of vibration source is input to a trained model. In this case, the trained model may output information such as: there is a 90% probability that the vibration source is a light truck, a 5% probability that the vibration source is a minivan, and a 5% probability that the vibration source is something else.

[0037] The way vibration waves generated by a vibration source propagate varies depending on environmental conditions such as topography and geology. Therefore, even if the same type of vibration source generates vibration waves, the detected vibration waveform may differ depending on the location. For this reason, it is sometimes preferable to build a trained model that is tailored to the environmental conditions.

[0038] The learning unit 142 may perform machine learning for each environmental condition to generate multiple trained models. For example, suppose the learning unit 142 generates a first trained model using vibration waveform data acquired by a vibration sensor placed at a first location as training data. The learning unit 142 may then generate a second trained model using vibration waveform data acquired by a vibration sensor placed at a second location under different environmental conditions than the first location as training data.

[0039] By creating different environmental conditions at the first and second locations, it is possible to generate trained models suited to two different environmental conditions. Preferably, the first and second locations are the actual locations where the vibration sensor 20 of the monitoring system 1 is placed, but they may also be different locations similar to where the vibration sensor 20 is actually placed.

[0040] Furthermore, the learning unit 142 can generate three or more trained models suited to environmental conditions by performing machine learning using training data under three or more different environmental conditions. These trained models may be stored in the storage unit 120 in association with the environmental conditions.

[0041] The learning unit 142 may generate a trained model that outputs the position and velocity (speed and direction) of the vibration source in addition to the type of vibration source. In this case, the training data may be vibration waveform data in which the position and velocity of the vibration source are known in addition to the type of vibration source.

[0042] The learning unit 142 may perform additional machine learning as needed after generating a trained model. For example, after placing the vibration sensors 20 in predetermined locations, the learning unit 142 may combine surveillance cameras and micro-weather information to further train the trained model on the relationship between the vibration source and the vibration waveforms observed by each vibration sensor 20. This can improve the accuracy of the output of the trained model.

[0043] The trained model estimates the combination of baseline and vibration waveforms from multiple sources that make up the complex vibration waveform observed by the vibration sensor 20. For example, it becomes possible to instantly identify the combination of vibration sources with the highest probability.

[0044] The processing flow using the trained model is as follows. Assume that vibration waveform data acquired by each vibration sensor 20 is input to the trained model. This vibration waveform data may include vibration waveforms generated by multiple vibration sources. The trained model performs, for example, the following: identification of combinations of vibration sources based on the training data, identification of vibration waveforms identical to vibration waveform data observed by multiple vibration sensors 20, identification of differences in detection times, identification of differences in vibration wave intensity, identification of the location of the vibration source, and identification of the direction of movement of the vibration source based on changes from the vibration waveform data at the previous time. As a result, the trained model can output data including the type of vibration source, the location of the vibration source, and the direction of movement of the vibration source.

[0045] The position estimation unit 144 estimates the position information of the vibration source based on the vibration waveform data acquired by each of the n vibration sensors 20 and the position information of each of the n vibration sensors 20. Since the n vibration sensors 20 are placed in different locations, the position estimation unit 144 estimates the position information of the vibration source using the vibration waveforms detected at the n locations.

[0046] The vibration source location information may include various types of information relating to the vibration source's position, such as the vibration source's position and information indicating the time evolution of the vibration source's position (specifically, the direction of movement of the vibration source and the speed of movement of the vibration source). Alternatively, the vibration source location information may be information indicating the probabilistic distribution of the vibration source's position (for example, a two-dimensional distribution).

[0047] The position estimation unit 144 may estimate the position information of the vibration source based on the intensity information of the vibration waveform data. The intensity information may be, for example, information indicating the peak intensity of the vibration waveform. The position estimation unit 144 can obtain intensity information from each of the n vibration waveform data acquired by the n vibration sensors 20, and estimate the position of the vibration source based on the acquired n intensity information and the position information of each of the n vibration sensors 20.

[0048] For example, the position estimation unit 144 may estimate the position of the vibration source by calculating the difference between any two peak intensities selected from the n acquired peak intensities. Furthermore, if a three-dimensional sensor (three-component seismometer) is available, the position estimation unit 144 may also estimate the position information by utilizing information on the vibration intensity (particle motion) of the horizontal component of the vibration wave. The magnitude of attenuation during the propagation of vibration waves varies depending on the environment. For this reason, it is preferable to use machine learning to obtain an appropriate difference in peak intensity. Furthermore, by pre-estimating the speed at which vibration waves propagate in or around the monitoring area 50 (elastic wave velocity) and utilizing this velocity information, the accuracy of position information estimation based on the difference can be improved.

[0049] An example of a method for estimating the specific location of a vibration source is described below. Based on vibration waveform data acquired at a certain time, it is estimated that the vibration source is located in a donut-shaped region (a circle with a certain width) centered on the vibration sensor 20 that acquired the data. By estimating a donut-shaped region for each of three or more vibration sensors 20, the location of the vibration source is uniquely identified as the overlapping location of these regions. Furthermore, by tracking the position information while considering changes over time, it becomes possible to determine the location of the vibration source with greater accuracy.

[0050] The position estimation unit 144 may estimate the position information of the vibration source based on the time information of the vibration waveform data. The time information may be, for example, information indicating the time when the intensity of the vibration wave reaches its peak. The position estimation unit 144 can obtain time information from each of the n vibration waveform data acquired by the n vibration sensors 20, and estimate the position of the vibration source based on the acquired n time information and the position information of each of the n vibration sensors 20. Since the speed at which vibration waves propagate (speed of sound) is constant, using time information allows for a more accurate estimation of the distance from the vibration sensor 20 to the vibration source than using the difference in peak intensity as described above.

[0051] The position estimation unit 144 may estimate the position information of the vibration source based on the intensity information and time information of the vibration waveform data. As described above, it is possible to estimate the position information of the vibration source using only one of the intensity information or time information, but by using both the intensity information and time information, it is possible to estimate the position information of the vibration source with greater accuracy.

[0052] Furthermore, the position estimation unit 144 may estimate the position information of the vibration source using a trained model that has been machine-trained to take vibration waveform data as input and output position information of the vibration source, as described above. The position estimation unit 144 may input vibration waveform data acquired by each vibration sensor 20 into the trained model and estimate the position information of the vibration source based on the output from the trained model. In this case, the trained model may output information about the type of vibration source in addition to the position information of the vibration source.

[0053] The position estimation unit 144 may estimate the position information of vibration sources located outside the monitoring area. Specifically, the position estimation unit 144 may estimate position information indicating that a vibration source is approaching the monitoring area, for example, it may estimate the position and speed of a light truck approaching the monitoring area. Alternatively, the position estimation unit 144 may estimate the position information of vibration sources within the monitoring area, for example, it may estimate the position of a person who has entered the monitoring area.

[0054] The type identification unit 146 identifies the type of vibration source of the vibration waveform data based on the vibration waveform data acquired by the vibration sensor 20.

[0055] The type identification unit 146 may identify the type of vibration source of the vibration waveform data by comparing the vibration waveform data acquired by the vibration sensor 20 with the vibration waveform model of the vibration source. For example, the type identification unit 146 may evaluate the similarity between each of the multiple types of vibration source vibration waveform models stored in the storage unit 120 and the vibration waveform data acquired by the vibration sensor 20. The type identification unit 146 may identify the type of vibration source of the vibration waveform model whose similarity exceeds a threshold as the type of vibration source of the vibration waveform data. Alternatively, the type identification unit 146 may evaluate the similarity between each of the multiple types of vibration waveform models and the vibration waveform data, and identify the type of vibration source of the vibration waveform model with the highest similarity as the type of vibration source of the vibration waveform data.

[0056] The type identification unit 146 may identify the type of vibration source in the vibration waveform data using a trained model that has been machine-learned as described above. Specifically, the type identification unit 146 may identify the type of vibration source in the vibration waveform data based on the output of the trained model when the vibration waveform data acquired by the vibration sensor is taken as input to the trained model. For example, if the trained model outputs a probability for each type of vibration source, the type identification unit 146 may identify the type of vibration source with the highest probability as the type of vibration source in the vibration waveform data.

[0057] Furthermore, if there are multiple trained models, each constructed using machine learning tailored to the environmental conditions of the location where the vibration sensor is placed, the type identification unit 146 may use a trained model that matches the environmental conditions of the vibration waveform data to identify the type of vibration source. This makes it possible to identify the type of vibration source more accurately.

[0058] The threat identification unit 148 can identify vibration sources that match the threat conditions based on the type of vibration source identified by the type identification unit 146 and the location information of the vibration source estimated by the location estimation unit 144. The threat identification unit 148 may identify vibration sources that match the threat conditions by referring to data indicating the threat conditions stored in the storage unit 120.

[0059] For example, suppose the threat condition is that a light truck has stopped within a predetermined distance from the monitoring area. In this case, the threat identification unit 148 can determine that the vibration source meets the threat condition if the type identification unit 146 identifies the type of vibration source as a light truck, and the position information of the vibration source estimated by the position estimation unit 144 indicates that the vibration source has stopped within a predetermined distance from the monitoring area. The threat identification unit 148 may determine that the vibration source does not meet the threat condition if the light truck has entered within a predetermined distance from the monitoring area but has moved away from the monitoring area without stopping.

[0060] Let's assume the threat condition is that a person has entered the monitored area. In this case, the threat identification unit 148 can determine that the vibration source meets the threat condition if the type identification unit 146 identifies the type of vibration source as a person, and the location information of the vibration source estimated by the location estimation unit 144 indicates that the vibration source is located inside the monitored area.

[0061] The communication unit 100 may transmit alarm information in response to the identification of a vibration source that matches the threat conditions by the threat identification unit 148. The alarm information may include signals that trigger the implementation of various security measures. For example, the mobile unit 30 may start moving or flying in response to the alarm information, and the imaging device 32 may photograph the vibration source that matches the threat conditions. This makes it possible to collect more detailed information about the vibration source that matches the threat conditions. In addition, security devices placed in the monitoring area may flash warning lights, emit warning sounds, throw color balls, etc. in response to the alarm information. Alternatively, the alarm information may be transmitted to a terminal of an external security company, and security guards or other personnel may be dispatched from that company to the monitoring area.

[0062] Figure 4 is a block diagram showing an example of the hardware configuration of the information processing device 10 according to this embodiment. The information processing device 10 comprises a processing device 160, a storage device 162, a communication device 164, an input device 166, and an output device 168. The processing device 160 may include a processor such as a CPU (Central Processing Unit) and implement the functions of the processing unit 140. The storage device 162 may include volatile / non-volatile storage devices such as RAM (Random Access Memory), ROM (Read Only Memory), and HDD (Hard Disk Drive) and implement the functions of the storage unit 120. The communication device 164 is an interface for communicating with an external device via a network 15 and may implement the functions of the communication unit 100. The input device 166 may include a mouse, keyboard, and touch panel display. The output device 168 may include various known display devices and audio output devices.

[0063] Figure 5 shows an example of the arrangement of the vibration sensors 20 according to this embodiment. Figure 5 shows an example in which three vibration sensors 20 are arranged (i.e., an example where n=3). The first vibration sensor 20_1, the second vibration sensor 20_2, and the third vibration sensor 20_3 are arranged at points A, B, and C within the monitoring area 50, respectively, along the boundary 52 between the inside and outside of the monitoring area 50. More specifically, the monitoring area 50 has a substantially triangular shape, and the first vibration sensor 20_1, the second vibration sensor 20_2, and the third vibration sensor 20_3 are each arranged near the vertices of this triangle. By arranging the vibration sensors 20 near the boundary 52 in this way, vibration sources near the monitoring area 50 can be detected with greater accuracy.

[0064] Incidentally, there may be height differences among Point A, Point B, and Point C. FIG. 5 shows an example in which three vibration sensors 20 are arranged, but four or more vibration sensors 20 may be arranged. FIG. 5 shows an example in which the vibration sensors 20 are arranged inside the monitoring area 50, but the vibration sensors 20 may be arranged outside the monitoring area 50 or on the boundary 52. The shape of the monitoring area 50 is not particularly limited, and it may have any arbitrary polygon such as a quadrilateral, or any arbitrary shape such as a circle or an ellipse. When the monitoring area 50 is a polygon (for example, a quadrilateral), the vibration sensors 20 may be arranged near each vertex of the polygon. Also, the position where the vibration sensors 20 are arranged is not limited to the vicinity of the boundary 52, and may be, for example, the center of the monitoring area 50 or its vicinity. Furthermore, the monitoring area 50 as shown in FIG. 5 does not necessarily have to be set.

[0065] The vibration source 40 is located at Point X outside the monitoring area 50. Assume that the vibration source 40 is a light truck that generates vibration waves 42 (seismic waves) and moves westward while approaching the monitoring area 50. The position of Point X changes as the vibration source 40 moves. In the example shown in FIG. 5, the monitoring area 50 is partially surrounded by a forest 54, making it difficult to see the vibration source 40 from inside the monitoring area 50.

[0066] Assume that Point A, Point B, and Point C are in this order, closer to Point X. The vibration waves 42 are detected by the first vibration sensor 20_1, the second vibration sensor 20_2, and the third vibration sensor 20_3 respectively, and the vibration waveform data acquired by each of the first vibration sensor 20_1, the second vibration sensor 20_2, and the third vibration sensor 20_3 is transmitted to the information processing device 10 in real time.

[0067] FIG. 6 is a diagram showing the time change of the intensity of the vibration waves detected by the three vibration sensors 20 according to the present embodiment. The horizontal axis represents time, and the vertical axis represents the intensity of the vibration waves. FIG. 6 shows the intensities 60_1, 60_2, and 60_3 of the vibration waves detected by the first vibration sensor 20_1, the second vibration sensor 20_2, and the third vibration sensor 20_3 respectively.

[0068] The intensity 60_1 of the vibration wave is at time t ＡAt peak intensity I ＰＥＡＫ＿Ａ The vibration wave intensity 60_2 is taken at time t Ｂ At peak intensity I ＰＥＡＫ＿Ｂ The vibration wave intensity 60_3 is taken at time t Ｃ At peak intensity I ＰＥＡＫ＿Ｃ As shown above, points A, B, and C are closest to point X in this order, so I ＰＥＡＫ＿Ａ > I ＰＥＡＫ＿Ｂ > I ＰＥＡＫ＿Ｃ And so, at time t Ａ ,t Ｂ ,t Ｃ This is the order in which they are fastest.

[0069] The position estimation unit 144 of the information processing device 10 can estimate the position of point X as described above. Specifically, the position estimation unit 144 can estimate the peak intensity I ＰＥＡＫ＿Ａ , I ＰＥＡＫ＿Ｂ , I ＰＥＡＫ＿Ｃ The difference (for example, I ＰＥＡＫ＿Ａ -I ＰＥＡＫ＿Ｂ , I ＰＥＡＫ＿Ｂ -I ＰＥＡＫ＿Ｃ The position of point X can be estimated using ). Alternatively, the position estimation unit 144 can estimate the position of point X at time t Ａ ,t Ｂ ,t Ｃ Based on this, the distance from point A, point B, and point C to point X may be estimated, and the position of point X may be estimated. Furthermore, the position estimation unit 144 uses the peak intensity I ＰＥＡＫ＿Ａ , I ＰＥＡＫ＿Ｂ , I ＰＥＡＫ＿Ｃ and time t Ａ ,t Ｂ ,t Ｃ Based on this, the position of point X may be estimated.

[0070] For example, the position estimation unit 144 determines that point X is d east of point A. １ It may be estimated that the location is at a distance of m. Furthermore, the position estimation unit 144 may estimate the time change in the position of point X, for example, if point X is moving at a speed of v per hour. １ It can be estimated that the object is moving westward at a speed of km / h. In this case, it can be estimated that point X is approaching the monitoring area 50.

[0071] Furthermore, the peak intensity of the vibration wave 42 detected at each of the vibration sources 40 at points A, B, and C can be difficult to accurately estimate because the propagation of the vibration wave 42 differs depending on the surrounding environment at each point, and the degree of attenuation differs depending on the frequency of the vibration wave 42. By using a trained model that achieves the following (1) and (2) through machine learning, it becomes possible to estimate the position of the vibration source 40 more accurately. (1) The vibration waveforms to be compared at points A, B, and C can be identified, and the difference in the time when the intensity of the vibration wave reaches its peak can be appropriately compared. (2) The difference in intensity can be compared not only based on the distance between points A, B, and C and point X, but also taking into account the propagation of the vibration wave and the surrounding environment.

[0072] As described above, according to the monitoring system 1 of this embodiment, the location information of the vibration source 40 can be estimated using n vibration sensors 20. As described above, the monitoring area 50 is partially surrounded by a forest 54, and even if cameras are installed at points A, B, and C, it may not be possible to monitor the vibration source 40 located outside the monitoring area 50. Even in such cases, the location information of the vibration source 40 can be estimated by using the vibration waveform data acquired by each of the n vibration sensors 20.

[0073] Figure 7 is a conceptual diagram illustrating a first example of identifying the type of vibration source 40 based on vibration waveform data. The upper part of Figure 7 shows the vibration waveform 70 at point A, as indicated by the vibration waveform data acquired by the first vibration sensor 20_1. This vibration waveform 70 is a combination of the vibration waveform 72 of the light truck and the noise waveform 74 shown at the bottom of Figure 7.

[0074] The type identification unit 146 of the information processing device 10 may identify that the type of vibration source 40 is a light truck by comparing the vibration waveform data with a vibration waveform model of a light truck. Alternatively, the type identification unit 146 may input the vibration waveform data into a trained model and identify that the type of vibration source 40 is a light truck based on the output of the trained model.

[0075] In this explanation, we have described a method for identifying the type of vibration source 40 using vibration waves detected at point A. However, the type of vibration source 40 may also be identified using vibration waves detected at point B or point C. Furthermore, the type identification unit 146 may use a common pre-trained model for each of points A, B, and C, or it may use a pre-trained model that is suited to the environmental conditions of each location.

[0076] Figure 8 is a conceptual diagram illustrating a second example of identifying the type of vibration source 40 based on vibration waveform data. The upper part of Figure 8 shows the vibration waveform 80 at point A, as indicated by the vibration waveform data acquired by the first vibration sensor 20_1. This vibration waveform 80 is a combination of the vibration waveform 82 from a minivan, the vibration waveform 84 from a bus, the vibration waveform 86 from a light truck, and the noise waveform 88, as shown in the lower part of Figure 8. Therefore, although not shown in Figure 5, there are vibration sources from minivans and buses in addition to the light truck. The type identification unit 146 may, for example, input vibration waveform data into a trained model and, based on the output of the trained model, identify that the vibration source of the vibration waveform 80 at point A includes minivans, buses, and light trucks. By using a trained model that has been machine-learned in this way, it becomes possible to easily identify the vibration source even in the case of complex vibration waveforms that include vibration waveforms from multiple vibration sources.

[0077] Figure 9 is a flowchart showing an example of the operation of an information processing device 10 according to one embodiment of the present invention. The operation flow of the information processing device 10 will be described below in accordance with the flowchart shown in Figure 9.

[0078] First, the communication unit 100 receives vibration waveform data acquired by each of the n vibration sensors 20 (S101). As a result, vibration waveform data acquired at n locations is received.

[0079] Next, the position estimation unit 144 estimates the position information of the vibration source based on the vibration waveform data acquired by each of the n vibration sensors 20 received in S101 and the position information of each of the n vibration sensors (S103). Then, the type identification unit 146 identifies the type of vibration source of the vibration waveform data based on the vibration waveform data received in S101 (S105).

[0080] Next, the threat identification unit 148 determines whether the type of vibration source identified in S105 and the location information of the vibration source estimated in S103 satisfy the threat conditions (S107). If it is determined that the threat conditions are not satisfied (S107: NO), the process ends. On the other hand, if it is determined that the threat conditions are satisfied (S107: YES), the process proceeds to S109.

[0081] If it is determined in S107 that the threat conditions are met, the communication unit 100 transmits the alarm information, including the transmission data D. ２ Send (S109). Transmission data D ２ This may be transmitted, for example, to the administrator's terminal of the monitoring system 1 or to the mobile device 30. In response, the mobile device 30 may start moving, and the imaging device 32 may photograph the vibration source that matches the threat conditions. Transmitted data D ２ The process ends when the message is sent.

[0082] The configuration and operation of the monitoring system 1 according to this embodiment have been described above. The monitoring system 1 according to this embodiment comprises n vibration sensors 20 (where n is an integer of 3 or more) arranged so as not to be located in a straight line, and an information processing device 10 configured to perform information processing based on vibration waveform data acquired by the n vibration sensors 20. The information processing device 10 has a position estimation unit 144 that estimates the position information of the vibration source of the vibration waveform data based on the vibration waveform data acquired by each of the n vibration sensors 20 and the position information of each of the n vibration sensors 20.

[0083] This configuration eliminates the need to prepare numerous cameras to cover blind spots, and by using three or more vibration waveform data, the location of the vibration source can be accurately estimated. Therefore, the monitoring system 1 or its information processing device 10 according to this embodiment can easily and accurately estimate the location of the vibration source.

[0084] (Examples) The following will be explained in more detail using examples, but the following descriptions of methods for classifying vibration sources and methods for estimating the location of vibration sources do not limit the embodiments of the monitoring system of the present invention in any way.

[0085] 1. Classification of Vibration Sources In this embodiment, vibration waveform data of automobiles and people was collected using a seismometer, and a trained model was constructed to classify vibration sources using the collected vibration waveform data. The accuracy of vibration source classification using the trained model was also verified. The details of this embodiment are described below.

[0086] First, vibration waveform data from vehicles and people were collected. Specifically, twelve three-component seismometers were placed in mainly flat areas within the mountains, and vibration waves generated from moving vehicles and walking people were detected to collect vibration waveform data. Vehicles traveled along predetermined routes at predetermined speeds (e.g., 20 km / h), and people walked along predetermined or random routes. The vehicle speed was changed as needed. Low levels of noise were also detected through mountains and other sources. In this way, 148 patterns of vibration waveform data were obtained.

[0087] Figure 10(a) shows an example of a vibration waveform of an automobile observed by a seismograph, and Figure 10(b) is the spectrogram of the vibration waveform shown in Figure 10(a). Figure 11(a) shows an example of a vibration waveform of a person observed by a seismograph, and Figure 11(b) is the spectrogram of the vibration waveform shown in Figure 11(a). Figure 12(a) shows an example of a vibration waveform of noise observed by a seismograph, and Figure 12(b) is the spectrogram of the vibration waveform shown in Figure 12(a).

[0088] In Figures 10(a), 11(a), and 12(a), the horizontal axis represents time [s] and the vertical axis represents the intensity of the vibration wave [m / s]. ２In Figures 10(b), 11(b), and 12(b), the horizontal axis represents time [s] and the vertical axis represents the frequency [Hz] of the vibration wave. The shades of gray indicate intensity; specifically, a particular frequency component at a given time is shown brighter the stronger its intensity.

[0089] As shown in Figure 10(b), the vibration waves from an automobile have strong frequency components in the range of 10 to 40 Hz. As shown in Figures 11(a) and 11(b), the intensity of human vibration waves increases with each step. The vibration waves from noise are weaker in intensity compared to those from automobiles and people, and as shown in Figure 12(b), there were hardly any frequency components above 30 Hz.

[0090] Training and validation data were created using 148 patterns of vibration waveform data acquired from a seismometer. Specifically, noise was added to these 148 vibration waveform data to create 740 composite data, resulting in a total of 888 vibration waveform data. Of these 888 vibration waveform data, 666 were used as training data, and the remaining 222 were used as validation data. The original 148 vibration waveform data patterns were also used as test data.

[0091] Using the created training and validation data, a convolutional neural network (CNN) model was trained to classify vibration sources in vibration waveform data into three types (car, person, and noise), and a trained model was constructed.

[0092] When 148 patterns of test data were classified into a trained model, the classification accuracy was 92.5%. For example, all 49 vibration sources whose correct label was "automobile" were classified as automobiles. Of the 47 vibration sources whose correct label was "person," 39 were classified as "person." Of the remaining 8 vibration sources, 2 were classified as automobiles and 6 were classified as noise. Furthermore, of the 52 vibration sources whose correct label was noise, 49 were classified as noise and the remaining 3 were classified as "person."

[0093] Figure 13(a) shows an example of a human vibration waveform from test data, and Figure 13(b) shows the result of classifying the vibration source of the vibration waveform in Figure 13(a) using a trained model. In Figure 13(a), the horizontal axis represents time [s], and the vertical axis represents the intensity of the vibration wave [m / s]. ２ This represents [ ]. In Figure 13(b), the horizontal axis represents time [s], and the shades represent the classification results of the vibration source. The light gray area 900 represents that the classification result is noise, and the dark gray area 902 represents that the classification result is a person.

[0094] In Figure 13(a), human vibration waves are generated at the times indicated by the six arrows. As shown in Figure 13(b), the vibration source is generally classified as a human during the times when human vibration waves are generated.

[0095] In this example, the classification accuracy of the trained model was verified not only using test data but also using unknown vibration waveform data, which represents a longer time period compared to the training data.

[0096] Figure 14(a) shows the vibration waveform of an automobile for unknown vibration waveform data, Figure 14(b) shows the result of classifying the vibration source of the vibration waveform in Figure 14(a) using a trained model, and Figure 14(c) shows the result of classifying the vibration source of the vibration waveform in Figure 14(a) after transformation processing. In Figure 14(a), the horizontal axis represents time [s], and the vertical axis represents the intensity of the vibration wave [m / s]. ２ This represents [ ]. In Figures 14(b) and 14(c), the horizontal axis represents time [s], and the shades represent the classification results of the vibration source. The light gray area 900 represents that the classification result is noise, the dark gray area 902 represents that the classification result is a person, and the black area 904 represents that the classification result is an automobile.

[0097] As shown in Figure 14(a), vibration waves from the automobile are generated at the times indicated by the eight arrows. As shown in Figure 14(b), the source of the vibration waves at the times indicated by the arrows is classified as the automobile. However, the source of the vibration is partially misclassified. For example, because the intensity of the vibration wave weakens at the beginning and end of the automobile vibration waveform, the source of the vibration in those parts is incorrectly classified as a person.

[0098] Therefore, a majority vote was taken among five classification results, including the classification result for the target time and the classification results for adjacent times. For example, if automobiles accounted for the majority of the five classification results, which included the classification result for a certain time and four temporally adjacent classification results, then the classification result for that time was set to automobile. In addition, for the classification target, other classification results obtained for other temporally close times were integrated into the classification result of the classification target. Specifically, if the vibration source was classified as a person at a time close to the time when the vibration source was classified as an automobile, that classification result was integrated into automobile. This process of taking a majority vote and integrating classification results is also called "conversion processing." As a result of the conversion processing, eight clear classification results for automobiles were obtained, as shown in Figure 14(c).

[0099] Figure 15 shows the results of classifying unknown vibration waveform data from 16 automobiles. Figure 15(a) shows the classification results when the automobile is within the range of 0 to 20 m from the seismometer, and Figure 15(b) shows the classification results when the automobile is within the range of 20 to 40 m from the seismometer. In Figures 15(a) and 15(b), the row labeled "CNN Classification" shows the results of classifying the vibration source using the trained model without any conversion processing, and the row labeled "Conversion Result" shows the results of classification after conversion processing.

[0100] As shown in Figure 15(a), when the vehicle was located within 0 to 20 m of the seismometer, the trained model classified the vibration waves solely as vehicles (100%, 64 times) without any conversion processing, and never misclassified them as people or noise (0%, 0 times, respectively). Similarly, when conversion processing was performed, the vibration source was classified solely as vehicles.

[0101] As shown in Figure 15(b), when the vehicle was located within 20-40m of the seismometer, the trained model classified the majority of vibration sources as vehicles (56.3%, 45 times), but also misclassified some vibration sources as people (43.7%, 35 times). However, after conversion processing, vibration sources were no longer misclassified as people (0%, 0 times), and although some vibration sources were classified as noise (12.5%, 10 times), vibration sources were mainly classified as vehicles (87.5%, 70 times).

[0102] As described above, when the vehicle was located less than 20 meters from the seismometer, the vibration source was completely classified as the vehicle, regardless of whether or not conversion processing was performed. Furthermore, when the vehicle was located between 20 and 40 meters from the seismometer, the conversion processing was effective.

[0103] Figure 16 shows the results of classifying four unknown vibration waveform data points from people. Figure 16(a) shows the classification results when the person is within 0 to 20 m of the seismograph, and Figure 16(b) shows the classification results when the person is within 20 to 40 m of the seismograph. In Figures 16(a) and 16(b), the "CNN Classification" row shows the results of classifying the vibration source using the trained model without any conversion processing, and the "Conversion Result" row shows the results of classification after conversion processing.

[0104] As shown in Figure 16(a), when a person was within 0 to 20 m of the seismometer, the trained model correctly identified most vibration sources as people (79.7%), although it did misclassify some vibration sources as noise (20.3%, 13 times). The same result was obtained when conversion processing was performed. On the other hand, as shown in Figure 16(b), when a person was within 20 to 40 m of the seismometer, the intensity of the vibration waves from the person was weak, so the vibration source was mostly misclassified as noise. Thus, the trained model was able to appropriately classify vibration sources when a person was walking at a distance of less than 20 m from the seismometer.

[0105] 2. Estimation of the Vibration Source Location In this embodiment, vibration waves from a vibration source were detected using three seismometers, and the location of the vibration source was estimated using the detection results. Specifically, the three seismometers were positioned so as not to be in a straight line, and vibration waveform data was acquired from these seismometers. Using this vibration waveform data, the location of the vibration source was estimated using the following procedure (steps 1 to 3). Step 1: The peak of cross-correlation was searched for, and the difference in the time it takes for the same vibration wave to reach the three seismometers was determined. Specifically, when the three seismometers are designated as seismometers A to C, the time difference was calculated from the cross-correlation for each of the three pairs (seismometer A-B, seismometer B-C, seismometer C-A). Step 2: The region to be analyzed was divided into grids, and the difference in the time it takes for vibration waves to reach the three seismometers, assuming that there is a vibration source in each grid, was determined. At this time, the time difference was determined for each of the three pairs, similar to step 1. Step 3: Calculate the RMS error for each grid between the three time differences calculated based on the cross-correlation determined in Step 1 and the three time differences determined in Step 2, and display the results as a heatmap.

[0106] Figure 17 shows an example of a heatmap of RMS error. In Figure 17, three seismometers 920, 922, and 924 are shown as circles on the XY plane, and the estimated position of the vibration source 926 is shown as a star. The darkness of the grid represents the magnitude of the RMS error, with darker grids indicating a smaller RMS error. Therefore, the vibration source is estimated to be located in the darkest grid. In the example shown in Figure 17, the vibration source is estimated to be located between the two seismometers 920 and 924.

[0107] Figures 18(a) to 18(j) show examples of the results of estimating the position of a moving vibration source (automobile). In Figures 18(a) to 18(j), three seismometers 920, 922, and 924 are shown as circles on the XY plane, the estimated position of the vibration source 926 is shown as a white star, and the correct position of the vibration source 928 is shown as a gray star. The grayscale of the grid represents the magnitude of the RMS error, with darker grids indicating a smaller RMS error. Time progresses in the order of Figures 18(a) to 18(j). Therefore, the correct position of the vibration source 928 moves linearly from the bottom right to the top left.

[0108] The RMS error shown in Figure 18(a) is particularly small in the region 940 enclosed by the dashed line. Region 940 includes the central region 942 of the three seismometers 920, 922, and 924, and extends downward and to the right from the central region 942. The estimated position 926 and the correct position 928 of the vibration source are located inside region 940. Although the estimated position 926 is far from the correct position 928, it can be said that the direction from the positions of seismometers 920, 922, and 924 (specifically the central region 942) to the correct position 928 could be estimated. In Figures 18(b) to 18(j), it can also be said that the direction from the positions of seismometers 920, 922, and 924 to the correct position 928 could be estimated to a reasonable extent.

[0109] Regarding Figures 18(a) to 18(j), position estimation was performed at three points slightly staggered in time, and the average value was taken to estimate the location of the vibration source, but no improvement in the accuracy of position estimation was observed. Furthermore, position estimation was performed with four seismometers, but no improvement in the accuracy of position estimation was observed.

[0110] (First Application Example) The monitoring system 1 according to the above embodiment can also be used to monitor various animals (for example, bears). The vibration wave modes generated by quadrupedal animals such as bears are easily distinguishable from the vibration wave modes of bipedal animals such as humans, and can therefore be easily classified using a trained model.

[0111] (Second Application Example) The monitoring system 1 according to the above embodiment can be used to confirm the movement of people in places where it is difficult to place cameras. For example, in a nursing home, a seismometer can be placed to detect vibration waves from users in the bathroom. Based on the vibration waves detected by the seismometer, the movement of the users can be estimated, and an alert may be issued if, for example, it is estimated that the users have not moved for a predetermined period of time.

[0112] The present invention has been described above based on embodiments. These embodiments are illustrative, and it will be understood by those skilled in the art that various modifications are possible in combinations of these components and processing processes, and that such modifications also fall within the scope of the present invention.

[0113] (First Modified Example) In the above embodiment, an example was described in which the monitoring system is configured to estimate the location information of vibration sources located on land or to identify the type of vibration source located on land. However, the system is not limited to this, and for example, the monitoring system may be configured to identify the type of vibration source located in the ocean or to estimate the location information of vibration sources located in the ocean. For example, n vibration sensors may be placed on the seabed, and an information processing device may be configured to identify the type of vibration source (for example, a vessel attempting to illegally enter a predetermined area) or to estimate the location information of that vibration source. In this case, the vibration sensors are configured to detect vibration waves transmitted through seawater and acquire vibration waveform data.

[0114] (Second Modification) In the above embodiment, an example was described in which one information processing device 10 has all the functions of the learning unit 142, the position estimation unit 144, the type identification unit 146, and the threat identification unit 148. However, the invention is not limited to this, and these functions may be realized using multiple devices. For example, a first device having the function of the learning unit 142 and a second device having functions such as the position estimation unit 144 and the type identification unit 146 may be provided separately. The second device may use the trained model generated by the first device to estimate the position information of the vibration source and identify the type of vibration source.

[0115] This invention relates to a monitoring system, an information processing device, and a program.

[0116] 10 Information processing device, 20 Vibration sensor, 30 Moving body, 32 Imaging device, 100 Communication unit, 120 Storage unit, 140 Processing unit, 142 Learning unit, 144 Position estimation unit, 146 Type identification unit, 148 Threat identification unit.

Claims

1. A monitoring system comprising: n vibration sensors (where n is an integer of 3 or more) arranged so as not to be located in a straight line; and an information processing device configured to perform information processing based on vibration waveform data acquired by the n vibration sensors, wherein the information processing device has a position estimation unit that estimates the position information of the vibration source of the vibration waveform data based on the vibration waveform data acquired by each of the n vibration sensors and the position information of each of the n vibration sensors.

2. The monitoring system according to claim 1, wherein the information processing device further comprises a type identification unit that identifies the type of vibration source of the vibration waveform data based on the vibration waveform data acquired by the vibration sensor.

3. The monitoring system according to claim 2, wherein the information processing device further comprises a storage unit that stores a trained model, the trained model is machine-trained to take vibration waveform data as input and output information regarding the type of vibration source of the input vibration waveform data, and the type identification unit identifies the type of vibration source of the vibration waveform data based on the output of the trained model when the vibration waveform data acquired by the vibration sensor is taken as input to the trained model.

4. The monitoring system according to claim 3, wherein the memory unit stores a plurality of trained models, and each of the plurality of trained models is constructed by machine learning according to the environmental conditions of the location where the vibration sensor is placed.

5. The monitoring system according to claim 2, wherein the information processing device further has a storage unit that stores vibration waveform models of a plurality of types of vibration sources, and the type identification unit identifies the type of vibration source of the vibration waveform data by comparing the vibration waveform data acquired by the vibration sensor with the vibration waveform model of the vibration source.

6. The monitoring system according to claim 2, wherein the information processing device further comprises a threat identification unit that identifies vibration sources that match predetermined threat conditions, and the threat identification unit identifies vibration sources that match the threat conditions based on the type of vibration source identified by the type identification unit and the location information of the vibration source estimated by the location estimation unit.

7. The surveillance system according to claim 6, further comprising a mobile body having an imaging device, wherein the imaging device captures an image including a vibration source identified by the threat identification unit as matching the threat conditions while the mobile body is moving.

8. The monitoring system according to claim 6, wherein the information processing device further comprises a communication unit that transmits alarm information in response to the identification of a vibration source that matches the threat conditions by the threat identification unit.

9. The monitoring system according to claim 1, wherein the location information of the vibration source includes information indicating the change in the position of the vibration source over time.

10. The monitoring system according to claim 1, wherein the position estimation unit estimates the position information of the vibration source of the vibration waveform data based on the intensity information and time information of the vibration waveform data.

11. The monitoring system according to claim 1, wherein each of the n vibration sensors is arranged along the boundary between the inside and outside of the monitoring area, the position estimation unit estimates the position information of a vibration source located outside the monitoring area, and the position information includes information indicating the approach of the vibration source to the monitoring area.

12. The monitoring system according to claim 1, wherein the vibration sensor has a wireless communication function for transmitting acquired vibration waveform data to the information processing device in real time, and the information processing device performs information processing based on the vibration waveform data transmitted from the vibration sensor in real time.

13. Information processing device for a monitoring system, wherein the information processing device is configured to perform information processing based on vibration waveform data acquired by n vibration sensors (where n is an integer of 3 or more) arranged so as not to be located in a straight line, and the information processing device has a position estimation unit that estimates the position information of the vibration source of the vibration waveform data based on the vibration waveform data acquired by each of the n vibration sensors and the position information of each of the n vibration sensors.

14. A program for causing a computer for a monitoring system to perform information processing based on vibration waveform data acquired by n vibration sensors (where n is an integer of 3 or more) arranged so as not to be in a straight line, wherein the information processing includes estimating the position information of the vibration source of the vibration waveform data based on the vibration waveform data acquired by each of the n vibration sensors and the position information of each of the n vibration sensors.

Images