Anomaly detection device, anomaly detection system, and anomaly detection method
The abnormality detection system uses frequency analysis of sensor data to distinguish transient peak patterns, enabling precise identification of abnormal movements and preventing potential accidents.
Patent Information
- Application Number
- JP2021030151
- Authority / Receiving Office
- JP · JP
- Patent Type
- Patents
- Current Assignee / Owner
- Filing Date
- 2021-02-26
- Publication Date
- 2024-11-07
- Estimated Expiration
- 2041-02-26
AI Technical Summary
Existing methods struggle to accurately distinguish abnormal movements from daily activities using acceleration sensors, making it difficult to detect potential accidents like falls or near-misses.
An abnormality detection system that analyzes time-series data from sensors, converting it into frequency components and comparing peak patterns across different periods to identify transient abnormalities, distinguishing them from continuous daily behaviors.
The system effectively detects abnormal movements by identifying peak patterns in frequency distributions, allowing for accurate differentiation from daily activities and timely notification of potential hazards.
Smart Images

Figure 0007579177000001 
Figure 0007579177000002 
Figure 0007579177000003
Abstract
Description
[Technical field]
[0001] The disclosure of this specification relates to an anomaly detection device, an anomaly detection system, and an anomaly detection method. [Background technology]
[0002] In recent years, attempts to prevent serious accidents from occurring by monitoring the behavior of a subject and detecting abnormalities or signs thereof at an early stage have been actively studied. Technologies related to monitoring the behavior of a subject are described in, for example, Non-Patent Document 1.
[0003] In the above-mentioned research, in addition to detecting the subject's everyday actions, such as walking, running, and climbing and descending stairs, from the measurements of sensors attached to the subject, it is also important to detect actions that the subject performs unintentionally or against his or her will (i.e., abnormal actions of the subject; hereafter, simply referred to as abnormalities in the subject).
[0004] Typical examples of abnormalities of a subject are the subject's movements that occur together with so-called near misses, and specifically include, for example, tripping, slipping, falling, contact, collision, falling, etc. Note that here, a near miss generally refers to an incident that did not result in a serious accident but could have caused a serious accident, or the recognition of such an incident. [Prior art documents] [Non-patent literature]
[0005] [Non-Patent Document 1] Yoshimitsu Shinagawa et al., "Detection of human walking and falls using acceleration sensors," Journal of Kawasaki Medical Welfare Society, 1999, vol.9 no.2 243-250 Summary of the Invention [Problem to be solved by the invention]
[0006] Non-Patent Document 1 discusses a method for detecting a subject's walking (everyday behavior) and falls (abnormal behavior) using an acceleration sensor, and describes the results of the study. However, as concluded in Non-Patent Document 1, it is not easy to detect a subject's abnormality, such as a fall, from only the measurement results of an acceleration sensor. In particular, it is difficult with conventional technology to distinguish between the subject's everyday behavior and detect the subject's abnormality, and further technological improvements based on new ideas are expected.
[0007] In light of the above-mentioned circumstances, an object of one aspect of the present invention is to provide a technique for detecting an abnormality in a subject by distinguishing it from daily behavior. [Means for solving the problem]
[0008] The abnormality detection device according to one aspect of the present invention detects time-series data of measurement values of a sensor that measures information on the condition of a subject. The frequency component distribution for each period of the time series data obtained by converting of the time series data covering different periods The above Frequency Components Peak patterns appearing in the distribution A detection unit that detects an abnormality in the subject based on the comparison The detection unit detects an abnormality in the subject based on the transience of the peak pattern identified by the comparison of the peak patterns. do. Effect of the Invention
[0009] According to the above aspect, abnormalities in a subject can be detected by distinguishing them from daily behavior. [Brief description of the drawings]
[0010] [Figure 1] FIG. 1 is a diagram illustrating a configuration of a system according to an embodiment. [Diagram 2] FIG. 1 is a diagram illustrating a hardware configuration of a measurement device according to an embodiment. [Diagram 3] FIG. 2 is a diagram illustrating a functional configuration of an abnormality detection device according to an embodiment; [Figure 4] FIG. 2 is an example of a sequence diagram illustrating interactions between components of a system according to an embodiment. [Diagram 5]1 is an example of a flowchart illustrating a process performed in a system according to an embodiment. [Figure 6] 1 is an example of data measured by a measurement device according to an embodiment. [Figure 7] 1 is an example of data showing a frequency distribution of data measured by a measurement device according to an embodiment. [Figure 8] 4 is a diagram illustrating a time-frequency analysis performed by the anomaly detection device according to an embodiment. FIG. [Figure 9] FIG. 11 is a diagram illustrating an example of classification criteria information. [Figure 10] 11 is a flowchart showing a specific example of processing performed in a system according to an embodiment. [Figure 11] FIG. 2 is a diagram illustrating the relationship between the functional configuration and the device configuration of a system according to an embodiment. [Figure 12] FIG. 11 is a diagram illustrating the relationship between the functional configuration and the device configuration of a system according to another embodiment. [Figure 13] FIG. 11 is a diagram showing the relationship between the functional configuration and the device configuration of a system according to yet another embodiment. [Figure 14] FIG. 11 is a diagram showing the relationship between the functional configuration and the device configuration of a system according to yet another embodiment. [Figure 15] FIG. 11 is a diagram showing the relationship between the functional configuration and the device configuration of a system according to yet another embodiment. [Figure 16] FIG. 1 is a diagram illustrating a hardware configuration of a computer for realizing the abnormality detection device according to the embodiment described above. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0011] Fig. 1 is a diagram illustrating an example of a system configuration according to an embodiment. The system 100 illustrated in Fig. 1 is an example of an abnormality detection system that detects an abnormality in a subject 1. First, the configuration of the system 100 will be described with reference to Fig. 1.
[0012] In the following, an example will be described in which the subject 1 is a worker working in a factory or a construction site where an abnormality is likely to lead to a serious accident. However, the subject 1 is not limited to the above-mentioned worker. The subject 1 may be, for example, an inpatient in a hospital, a person in need of care in a nursing home, or an elderly person who needs to be watched over. That is, the system 100 may be constructed for any person for whom early detection of an abnormality and warning are effective in preventing the abnormality from becoming serious or causing a serious accident. In addition, by analyzing the operating environment under which the abnormality of the subject 1 occurred, the system may be useful for improving the working environment, living environment, and the like.
[0013] In this specification, the abnormality of the subject mainly refers to an abnormal movement of the subject, such as tripping, slipping, falling, bumping, colliding, dropping, or other movements that the subject makes unintentionally or against his / her will. However, the abnormality of the subject may also include physical abnormalities of the subject, such as fever, cold sweat, palpitations, shortness of breath, and trembling. In other words, the abnormality of the subject may broadly include abnormal conditions of the subject.
[0014] The system 100 includes a measuring device 10 attached to a subject 1, and an abnormality detection device 40 that analyzes information acquired by the measuring device 10. The system 100 may further include an alarm device 20 carried by the subject 1.
[0015] The devices constituting the system 100 are connected to each other so as to be able to communicate with each other via a network 30. The communication between the devices may be performed via another device (not shown). The communication between the devices may also be performed directly in an ad-hoc communication mode. Here, the communication may be wireless communication or wired communication. The communication may also be a combination of wireless communication and wired communication.
[0016] In the system 100, the abnormality detection device 40 detects an abnormality in the subject 1 based on the information acquired by the measurement device 10. The abnormality detection device 40 may further analyze the detected abnormality and classify it into several categories. The abnormality detection device 40 may also record the detected abnormality in association with other information of the subject 1, such as location information. The abnormality detection device 40 may also notify the alarm device 20 of the detected abnormality, and the alarm device 20 may notify the subject 1 of information according to the classification of the abnormality.
[0017] 1 illustrates an example of only one subject 1 wearing the measuring device 10, but the system 100 may include multiple measuring devices 10 worn by multiple subjects 1, and the abnormality detection device 40 may process information from the multiple measuring devices 10. Also, multiple measuring devices 10 may be worn by one subject 1.
[0018] Fig. 2 is a diagram illustrating a hardware configuration of a measurement device according to an embodiment. Fig. 3 is a diagram illustrating a functional configuration of an anomaly detection device according to an embodiment. The configuration of the system 100 will be described in further detail with reference to Figs. 2 and 3.
[0019] 2, the measurement device 10 includes a sensor 11 that measures information on the state of the subject 1. The measurement device 10 may be, for example, an inertial measurement unit (IMU), and may include an inertial sensor (acceleration sensor, gyro sensor) that measures acceleration, angular velocity, etc. as the sensor 11. The sensor 11 may further include an inclination sensor that measures an inclination angle, a Global Positioning System (GPS) or an indoor positioning sensor that measures a position, a pressure sensor that measures pressure, a biosensor that measures vital signs and other bioinformation, etc.
[0020] In addition to the sensor 11, the measuring device 10 includes a microprocessor 12 that controls the operation of the measuring device 10, a memory 13 that stores data such as measurement values measured by the sensor 11, a communication circuit 14 that communicates with external devices (such as anomaly detection device 40 and alarm device 20), and a battery 15 that supplies power.
[0021] In the measuring device 10, under the control of the microprocessor 12, the sensor 11 measures information on the condition of the subject 1, and the memory 13 stores the measured values of the sensor 11. Furthermore, the communication circuit 14 transmits the time series data of the measured values stored in the memory 13 to the abnormality detection device 40.
[0022] 1, the measurement device 10 is attached to the waist of the subject 1 by fastening it to a belt or the like worn by the subject 1. By attaching the measurement device 10 to the waist, which is close to the center of the body, it is possible to measure the movement of the entire body while suppressing the influence of local movements of parts of the body such as the arms and legs. For this reason, this is an ideal attachment position for the measurement device 10 as an inertial measurement device.
[0023] However, the mounting position of the measuring device 10 is not limited to the above example. For example, the measuring device 10 may be fixed to a helmet and mounted on the head. Also, the measuring device 10 may be mounted at an appropriate position depending on the type of the sensor 11.
[0024] Furthermore, the measurement device 10 may be incorporated into a wearable device such as a smartwatch or smart glasses. The measurement device 10 may be held by the subject 1, for example, may be incorporated into a mobile terminal such as a smartphone or tablet held by the subject 1. The measurement device 10 may be embedded in the subject 1.
[0025] The notification device 20 is a device that notifies the subject 1 of information. The information notified by the notification device 20 may be presented to the subject 1 visually, audibly, or tactilely. The notification device 20 may be, for example, a display device equipped with a display, and may notify the subject 1 of information by displaying information. The notification device 20 may be, for example, a light-emitting device equipped with an LED or the like, and may notify the subject 1 of information by turning on or blinking a light. The notification device 20 may be, for example, an audio output device equipped with a speaker, and may notify the subject 1 of information by outputting a voice or a warning sound. The notification device 20 may be, for example, a vibrator, and may notify the subject 1 of information by vibration.
[0026] When the notification device 20 receives a notification of abnormality detection from the abnormality detection device 40, it notifies the subject 1 of different information depending on the classification of the abnormality included in the notification, for example.
[0027] The notification device 20 does not necessarily need to be fixed to the subject 1, and may be held by the subject 1 in his / her hand, as shown in Fig. 1. The subject 1 may also carry the notification device 20 in his / her pocket or bag. Alternatively, the notification device 20 may be fixed to a belt or the like, similar to the measurement device 10. The notification device 20 may be incorporated in a mobile terminal such as a smartphone or tablet held by the subject 1, or may be incorporated in a wearable device worn by the subject 1.
[0028] The abnormality detection device 40 is a device that detects an abnormality in the subject 1. The abnormality detection device 40 is not particularly limited, and may be, for example, a server computer (tower server, rack-mounted server, blade server) or a personal computer (desktop computer, laptop computer, tablet computer). The abnormality detection device 40 may also be other computers, such as a smartphone or mobile phone, as long as it has sufficient processing power for abnormality detection.
[0029] 3, the abnormality detection device 40 includes a control unit 41, a communication unit 45, and a recording unit 46. The control unit 41 includes a conversion unit 42, a detection unit 43, and a classification unit 44, and may further include other functional units.
[0030] The control unit 41 performs time-frequency analysis on the time series data of the measurement value of the sensor 11 output from the measuring device 10, and detects abnormality of the subject 1 based on the analysis result. The conversion unit 42 converts the time series data of the measurement value of the sensor 11 into a frequency component distribution for each period of the time series data. The detection unit 43 detects abnormality of the subject 1 based on a comparison of the frequency component distribution of the time series data for different periods generated by the conversion unit 42. The classification unit 44 classifies the detection result of the detection unit 43. The communication unit 45 communicates with other devices such as the measuring device 10 and the notification device 20 according to an instruction from the control unit 41. The recording unit 46 stores abnormality of the subject 1 detected by the detection unit 43. The recording unit 46 may also store information that the classification unit 44 refers to when classifying the detection result.
[0031] The abnormality detection process performed by the system 100 will now be described.
[0032] 4 is an example of a sequence diagram showing interactions between components of a system according to an embodiment. The anomaly detection process performed in the system 100 may be composed of a process (measurement process) performed by the measuring device 10, a process (detection process) performed by the anomaly detection device 40, and a process (notification process) performed by the notification device 20, as shown in FIG.
[0033] First, the measuring device 10 starts the measurement process. In the measurement process, time series data of the measurement values measured by the measuring device 10 (measurement data in FIG. 4) is transmitted from the measuring device 10 to the abnormality detection device 40, for example, periodically. The abnormality detection device 40 starts the detection process when it receives the measurement data. In the detection process, the measurement data is analyzed. When an abnormality in the subject 1 is detected, the abnormality detection device 40 notifies the alarm device 20 of the abnormality detection. When the alarm device 20 receives the notification of the abnormality detection, it starts the alarm process. In the alarm process, information corresponding to the received notification is notified to the subject 1.
[0034] FIG. 5 is an example of a flowchart showing processing performed in a system according to an embodiment. FIG. 6 is an example of data measured by a measuring device according to an embodiment. FIG. 7 is an example of data showing a frequency distribution of data measured by a measuring device according to an embodiment. FIG. 8 is a diagram for explaining a time-frequency analysis performed in an anomaly detection device according to an embodiment. FIG. 9 is a diagram showing an example of classification criterion information. The anomaly detection processing performed in the system 100 will be described in further detail with reference to FIGS. 5 to 9.
[0035] The anomaly detection process performed by the system 100 may include, for example, six processes (measurement process, conversion process, comparison / detection process, classification process, recording process, and notification process) as shown in Fig. 5. The anomaly detection process shown in Fig. 5 may be started by the system 100 when the measuring device 10 is started, for example.
[0036] When the abnormality detection process is started, the measurement device 10 performs a measurement process (step S1). In step S1, the sensor 11 measures information on the condition of the subject 1. The information on the condition of the subject 1 measured by the sensor 11 is sampled, for example, at a constant period, and the memory 13 stores the data obtained by sampling.
[0037] 6 is an example of time-series data of the measurement values of the sensor 11 obtained by sampling the measurement values of the sensor 11 while the subject 1 is walking when the sensor 11 is an acceleration sensor. The time-series data stored in the memory 13 is transmitted from the measuring device 10 to the abnormality detection device 40 using the communication circuit 14 at a predetermined timing under the control of the microprocessor 12. Note that the measurement process is repeated, for example, after the measuring device 10 is started up, until the measuring device 10 is stopped.
[0038] Upon receiving the time series data of the measurement values, the abnormality detection device 40 performs a time-frequency analysis on the time series data of the measurement values to detect an abnormality in the subject 1 (steps S2 to S4). That is, the abnormality detection device 40 analyzes how the frequency components contained in the time series data change depending on the target period of the time series data of the measurement values, and detects an abnormality in the subject 1 based on the analysis results.
[0039] The abnormality detection device 40 first converts the time series data of the measurement values (step S2). Here, the control unit 41 operates as the conversion unit 42. The control unit 41 performs, for example, a Fourier transform on the time series data of the measurement values to generate data in which frequency components contained in the time series data are extracted. Data 2000 shown in Fig. 7 is an example of a frequency distribution obtained by Fourier transforming the data 1000 shown in Fig. 6.
[0040] More specifically, in step S2, the converter 42 repeatedly converts the time series data while gradually shifting the target period of the time series data of the measurement values. The converter 42 may, for example, extract the time series data using a window function of a certain size and perform a Fourier transform on the extracted data. Fig. 8 shows an example in which time series data (data 1001, data 1002, data 1003, data 1004) of different target periods are extracted from the time series data of the measurement values, and a Fourier transform is performed on each of them to generate a frequency component distribution (data 2001, data 2002, data 2003, data 2004) of the time series data for each target period.
[0041] When the frequency component distributions are generated, the abnormality detection device 40 compares the generated frequency component distributions to detect abnormality in the subject 1 (step S3). Here, the control unit 41 operates as the detection unit 43. The control unit 41 detects abnormality in the subject 1 based on a comparison of frequency components of time series data covering different time periods. The control unit 41 may detect abnormality in the subject 1 based on, for example, a comparison of peak patterns appearing in the frequency component distributions of time series data covering different time periods.
[0042] In general, abnormal human behaviors such as tripping, slipping, falling, bumping, collisions, and drops occur instantaneously, and in this respect they differ from everyday human behaviors such as walking, running, climbing, etc. In other words, everyday behaviors and abnormal behaviors differ in terms of the presence or absence of continuity in the frequency component distribution.
[0043] The control unit 41 may detect an abnormality in the subject 1 by focusing on the presence or absence of continuity in the frequency component distribution. Specifically, the control unit 41 may, for example, compare peak patterns of frequency components of time series data covering different periods, and detect the abnormality of the subject 1, distinguishing it from daily behavior, based on the transience of the peak pattern identified by the comparison of the peak patterns. Furthermore, the control unit 41 may detect the daily behavior of the subject 1 in addition to the abnormality of the subject 1, and may detect the behavior of the subject 1, for example, based on the continuity of the peak pattern identified by the comparison of the peak patterns.
[0044] Daily actions include various frequency components depending on the type of action. Therefore, daily actions may include a wide range of frequency components, from relatively low frequency components (e.g., walking) to relatively high frequency components (e.g., running). In contrast, abnormal actions are actions that involve instantaneous large changes in acceleration, and are dominated by high frequency components. In other words, abnormal actions can be distinguished from at least some daily actions by the frequency of the peaks included in the frequency component distribution.
[0045] The control unit 41 may focus on the frequency range of this peak to detect the abnormality of the subject 1 with higher accuracy. Specifically, the control unit 41 may compare peak patterns of frequency components of time series data covering different periods, for example, and detect the abnormality of the subject 1 based on the frequency of a peak that occurs temporarily. In this case, the control unit 41 detects the abnormality of the subject 1 when the frequency of the peak is high, but the method of determining whether the frequency of the peak is high or not is not particularly limited. It may be determined by a predetermined threshold value, or may be determined by comparing with the frequency of other peaks such as walking. Furthermore, the control unit 41 may detect the behavior of the subject 1 based on the frequency of a peak that occurs continuously, for example.
[0046] With reference to FIG. 8, a case where the subject 1 stumbles while walking will be described. As shown in FIG. 8, the peak pattern appearing in the frequency component distribution while walking includes only peaks of relatively low frequency indicating walking (data 2001, data 2002). The height and frequency of the peaks vary to some extent depending on the walking speed, etc., but peaks always occur at relatively low frequencies while walking. Therefore, from the frequency and continuity of the peaks, the control unit 41 can detect that the subject 1 is walking. Next, when the subject 1 stumbles while walking, the peak pattern appearing in the frequency component distribution for the period including the timing of the stumble includes a peak having a higher frequency than walking in addition to the peak indicating walking (data 2003). This peak is not included in the peak pattern appearing in the frequency component distribution for the next period (data 2004). Therefore, from the frequency and transience of the peaks, the control unit 41 can detect that the subject 1 stumbles while walking.
[0047] Although an example has been shown here in which a peak is determined to be transient if it disappears within less than two cycles of the target period, whether or not a peak is transient may be changed as appropriate depending on the length of the target period. For example, a peak that disappears within less than three cycles of the target period may be determined to be a transient peak, and a peak that is detected continuously for three or more cycles may be determined to be a continuous peak.
[0048] When an abnormality in the subject 1 is detected in step S3, the abnormality detection device 40 classifies the detected abnormality (step S4). Here, the control unit 41 operates as a classification unit 44. The control unit 41 may classify the detected abnormality based on, for example, the characteristics of a peak indicating the detected abnormality that is included in the frequency component distribution.
[0049] For example, classification standard information 3000 stored in the recording unit 46 may be used to classify abnormalities based on the characteristics of the peaks. For example, as shown in FIG. 9, the classification standard information 3000 may include the characteristics of "continuous / transient", "frequency band", and "other" of the peaks for each classification of abnormality. "Other" may include, for example, characteristics such as the intensity and width of the peak. As shown in FIG. 9, the classification standard information 3000 may include information for classifying daily actions in addition to information for classifying abnormalities.
[0050] In step S4, classification may be performed using a trained model instead of the classification standard information 3000. For example, the detection results may be classified using a trained model that has previously learned the relationship between the features of the peaks converted into vector data and abnormalities and actions.
[0051] In step S4, in addition to classifying the detected abnormality, the detected behavior may also be classified. That is, the control unit 41 may classify the detection results detected by the detection unit 43 regardless of whether they are abnormal actions or everyday behavior. Furthermore, the control unit 41 may classify the detection results using other information in addition to the peak features. For example, the control unit 41 may classify the detection results using information about the sensor 11 (e.g., type, installation direction), or may classify the detection results using data from another sensor.
[0052] When the abnormality is classified in step S4, the abnormality detection device 40 records the detected abnormality (step S5). Here, under the control of the control unit 41, the recording unit 46 records the abnormality of the subject 1 detected in step S3 as an abnormality in the category (classification result) classified in step S4. The recording unit 46 may record various information together with the abnormality category. For example, the recording unit 46 may record date and time information of the target period in which the abnormality was detected, location information of the subject 1 during the target period, information on the behavior of the subject 1 before and after the occurrence of the abnormality, and the like together with the abnormality category. Note that in step S5, the abnormality detection device 40 records the detected abnormality and transmits an abnormality detection notification including information on the abnormality category to the alarm device 20.
[0053] Upon receiving the notification from the abnormality detection device 40, the notification device 20 notifies the subject 1 of information corresponding to the category (step S6). The information corresponding to the category is, for example, a warning that alerts the subject 1 to the content corresponding to the category.
[0054] As described above, the abnormality detection device 40 can detect an abnormality of the subject 1 by distinguishing it from daily behavior by performing all or part of the abnormality detection process shown in Fig. 4 and Fig. 5. In particular, the abnormality detection device 40 can convert an abnormal feature that appears as a complex pattern in the time series data into a relatively simple pattern (for example, a peak in the frequency component distribution) by using the frequency components of the time series data of the measurement value of the sensor. In addition, the abnormality detection device 40 can compare the frequency components of the time series data covering different periods by performing a time-frequency analysis on the time series data, and detect an abnormal behavior by distinguishing it from daily behavior based on the presence or absence of continuity of the simplified pattern. Therefore, the abnormality detection device 40 can easily and accurately detect an abnormality of the subject 1 by distinguishing it from daily behavior.
[0055] 4 and 5, the system 100 can notify the subject 1 of information corresponding to the classification of the abnormality detected by the abnormality detection device 40. Therefore, according to the system 100, the subject 1 can be alerted by notifying the subject 1 of information corresponding to the classification of the detected abnormality, and therefore the occurrence of a serious accident can be prevented.
[0056] Fig. 10 is a flowchart showing a specific example of processing performed in a system according to an embodiment. Hereinafter, with reference to Fig. 10, an example will be described in which the processing shown in Fig. 5 is implemented in a system 100 including a measuring device 10 including at least an acceleration sensor and an inclination sensor as sensors 11.
[0057] In the system 100, first, the measuring device 10 attached to the subject 1 measures acceleration with the sensor 11 (step S101). Next, the abnormality detection device 40 extracts data for a target period from the time series data of acceleration measured by the sensor 11 and performs a Fourier transform (step S102). After that, the abnormality detection device 40 determines whether or not there is a peak indicating something other than walking in the frequency component distribution obtained by the Fourier transform (step S103). The method of determining whether or not there is a peak indicating something other than walking is not particularly limited, but may be performed by, for example, regarding the most frequently occurring relatively low peak as the peak indicating walking.
[0058] If there is no peak other than walking (step S103 NO), the system 100 ends the process shown in Fig. 10. On the other hand, if there is a peak other than walking (step S103 YES), the abnormality detection device 40 judges whether the peak other than walking continues to occur (step S104). In step S104, for example, the continuity of the peak may be evaluated based on whether the peak other than walking also occurred in the frequency component distribution in the preceding target period.
[0059] When the abnormality detection device 40 determines that the peak is not continuing (step S104 NO), the abnormality detection device 40 further determines whether the subject 1 has been in a continuously inclined state since the target period in which the peak other than walking was detected, based on the measurement value (tilt angle) of the tilt sensor (step S105). If the measurement value of the tilt sensor indicates that the subject 1 is continuously inclined (step S105 YES), it is assumed that the subject 1 has fallen, so the abnormality detection device 40 classifies the abnormality of the subject 1 as a fall and notifies the alarm device 20, and the alarm device 20 issues an alarm for when the subject 1 falls (step S106). On the other hand, if the measurement value of the inclination sensor does not indicate a continuous inclination (step S105 NO), it is assumed that the subject 1 has tripped, so the abnormality detection device 40 classifies the abnormality of the subject 1 as a tripping and notifies the alarm device 20, and the alarm device 20 issues a warning for tripping (for example, a message such as "Watch your step") (step S107).
[0060] When the abnormality detection device 40 determines that the peak continues (step S104 YES), it determines whether the width of the peak is less than a threshold value (step S108). When the width of the peak is not less than the threshold value (step S108 NO), the peak is assumed to be caused by the behavior of the subject 1, so the abnormality detection device 40 further determines whether the acceleration axis is horizontal or vertical (step S109). Here, for example, the acceleration axis may be determined based on the measurement value of a three-axis acceleration sensor during the target period in which the peak was detected. When the abnormality detection device 40 determines that the acceleration axis is horizontal, it is assumed that the peak is caused by running, so the abnormality detection device 40 classifies the behavior of the subject 1 as running and notifies the notification device 20, and the notification device 20 issues a warning for running (for example, a message such as "It is dangerous, so please do not run") (step S110). On the other hand, when the abnormality detection device 40 determines that the axis of acceleration is vertical, it is considered that the peak is due to the movement of ascending or descending, so the abnormality detection device 40 classifies the movement of the subject 1 as ascending or descending and notifies the notification device 20, and the notification device 20 issues a warning for ascending or descending (e.g., a message such as "Be careful on the stairs") (step S111). Also, if the width of the peak is less than the threshold in step S108 (YES in step S108), it is assumed that the peak is due to mechanical vibration, so the abnormality detection device 40 notifies the notification device 20 that it is assumed that heavy machinery or the like that generates mechanical vibration is present around the subject 1, and the notification device 20 issues a warning for vigilance against the surroundings (e.g., a message such as "Please keep a close eye on your surroundings before acting") (step S112).
[0061] As described above, the system 100 can distinguish between abnormalities and everyday behavior of the subject 1 based on information from the measurement device 10 including an acceleration sensor and an inclination sensor. It can also detect the condition of the subject 1 that requires vigilance against the surroundings.
[0062] Fig. 11 is a diagram showing the relationship between the functional configuration and the device configuration of the system according to the embodiment described above. The above-mentioned system 100 adopts a configuration in which, among the functional configurations that perform the above-mentioned anomaly detection processing, the measurement device (measurement device 10) serves as the sensor unit, the notification device (notification device 20) serves as the alarm unit, and the anomaly detection device (anomaly detection device 40) serves as the other functional units (conversion unit, detection unit, classification unit, recording unit), but the relationship between the functional configuration and the device configuration is not limited to this example, and for example, the relationships shown in Figs. 12 to 15 may be adopted. Note that the configuration shown in Fig. 11 is a configuration in which the functional units other than those required on the subject 1 side are consolidated on the server side, and is therefore desirable in that similar functions can be efficiently provided to multiple subjects 1.
[0063] Fig. 12 is a diagram showing the relationship between the functional configuration and device configuration of a system according to another embodiment. System 200 shown in Fig. 12 differs from system 100 in that the sensor unit and the notification unit are realized by the same device (measurement notification device). The other points are the same. The configuration shown in Fig. 12 is desirable in that it can minimize the number of devices required on the subject side.
[0064] Fig. 13 is a diagram showing the relationship between the functional configuration and the device configuration of a system according to yet another embodiment. The system 300 shown in Fig. 13 differs from the system 100 in that the measurement device also serves as a conversion unit. The other points are the same. The configuration shown in Fig. 13 may be adopted depending on the performance of the measurement device.
[0065] Fig. 14 is a diagram showing the relationship between the functional configuration and device configuration of a system according to yet another embodiment. In the system 400 shown in Fig. 14, the functional units other than the alarm unit are handled by an anomaly detection device. In this configuration, the anomaly detection device is provided on the subject side, which is desirable in that it is less susceptible to the effects of communication delays and communication failures, for example, compared to a case where the anomaly detection device is placed on the Internet. In addition, by having a separate device (alarm device) handle only the alarm unit, it is possible to effectively utilize existing terminals (for example, a smartphone owned by the subject 1).
[0066] Fig. 15 is a diagram showing the relationship between the functional configuration and the device configuration of a system according to yet another embodiment. In the system 500 shown in Fig. 15, all the functional units are handled by an anomaly detection device. This configuration is desirable in that it can avoid the effects of communication delays and communication failures while minimizing the number of devices required on the subject side, similar to the system 200 shown in Fig. 12.
[0067] Fig. 16 is a diagram illustrating a hardware configuration of a computer 50 for realizing the anomaly detection device according to the embodiment described above. The hardware configuration illustrated in Fig. 16 includes, for example, a processor 51, a memory 52, a storage device 53, a reading device 54, a communication interface 56, and an input / output interface 57. The processor 51, the memory 52, the storage device 53, the reading device 54, the communication interface 56, and the input / output interface 57 are connected to each other, for example, via a bus 58.
[0068] The processor 51 may be, for example, a single processor, or a multi-processor or a multi-core processor. The processor 51 provides some or all of the functions of the control unit 41 by using the memory 52 to execute a program in which, for example, all or part of the procedures of the above-mentioned operation flow are described. The processor 51 reads out and executes a program stored in the storage device 53 to operate, for example, as the conversion unit 42, the detection unit 43, and the classification unit 44.
[0069] The memory 52 is, for example, a semiconductor memory and may include a RAM area and a ROM area. The storage device 53 is, for example, a semiconductor memory such as a hard disk or a flash memory, or an external storage device.
[0070] The reader 54 accesses the removable storage medium 55 according to, for example, an instruction from the processor 51. The removable storage medium 55 is realized by, for example, a semiconductor device, a medium in which information is input and output by magnetic action, a medium in which information is input and output by optical action, or the like. The semiconductor device is, for example, a Universal Serial Bus (USB) memory. The medium in which information is input and output by magnetic action is, for example, a magnetic disk. The medium in which information is input and output by optical action is, for example, a Compact Disc (CD)-ROM, a Digital Versatile Disk (DVD), a Blu-ray Disc, etc. (Blu-ray is a registered trademark).
[0071] The above-mentioned recording unit 46 may include, for example, a memory 52, a storage device 53, and a removable storage medium 55. The communication interface 56 communicates with other devices, for example, according to instructions from the processor 51. For example, the computer 50 may collect measurement values from the sensor 11 via the communication interface 56. The communication interface 56 is an example of the above-mentioned communication unit 45.
[0072] The input / output interface 57 is, for example, an interface between an input device and an output device. The input device is, for example, a device such as a keyboard, a mouse, or a touch panel that receives instructions from a user. The output device is, for example, a display device such as a display, and an audio device such as a speaker.
[0073] Each program according to the embodiment is provided to the computer 50 in the following form, for example. (1) Pre-installed in storage device 53. (2) Provided by a removable storage medium 55. (3) Provided from a server such as a program server.
[0074] Note that the hardware configuration of the computer 50 for realizing the anomaly detection device described with reference to Fig. 16 is an example, and the embodiment is not limited to this. For example, a part of the above-mentioned configuration may be deleted, or a new configuration may be added. In another embodiment, for example, a part or all of the functions of the above-mentioned control unit 41 may be implemented as hardware using a field programmable gate array (FPGA), a system-on-a-chip (SoC), an application specific integrated circuit (ASIC), a programmable logic device (PLD), or the like.
[0075] In the above, several embodiments are described. However, the embodiments are not limited to the above embodiments, and should be understood as including various modified and alternative forms of the above-mentioned embodiments. For example, it will be understood that the various embodiments can be embodied by modifying the components without departing from the spirit and scope of the embodiments. It will also be understood that various embodiments can be implemented by appropriately combining multiple components disclosed in the above-mentioned embodiments. Furthermore, it will be understood by those skilled in the art that various embodiments can be implemented by deleting some components from all the components shown in the embodiments, or by adding some components to the components shown in the embodiments.
[0076] For example, in the above-described embodiment, an example was shown in which the abnormality of the subject 1 was notified to the subject 1 himself, but the subject to be notified is not limited to the subject 1. If the subject 1 encounters an accident, the abnormality of the subject 1 may be notified to people around the subject 1 rather than to the subject 1 himself. Also, if the subject 1 is working in a factory or the like, the site supervisor may be notified. Also, the system administrator may be notified, or a predetermined arbitrary person may be notified to notify the abnormality of the subject 1.
[0077] In the above embodiment, an example was shown in which the measurement data acquired by the measuring device 10 provided on the subject 1 side was transmitted to the abnormality detection device 40 via a network, but data exchange is not limited to a method using network communication. For example, the measurement data acquired by the measuring device 10 may be input to the abnormality detection device 40 by exchanging a recording medium such as a memory card.
[0078] In the above embodiment, when an abnormality in the subject 1 is detected, information corresponding to the abnormality is notified, but the detected abnormality does not necessarily have to be notified. By collecting abnormality information and constructing a near-miss database, information requiring attention may be shared among all related parties including the subject 1. [Explanation of symbols]
[0079] 1. Target Audience 10. Measurement Equipment 11 Sensors 20. Notification device 40 Anomaly detection device 41 Control section 42 Conversion section 43 Comparison Section 44 Classification Department 45 Communications Department 46 Recording Section 50 Computer 51 Processors 100, 200, 300, 400, 500 Systems
Claims
1. a detection unit that detects an abnormality in the subject based on a comparison of peak patterns that appear in the frequency component distribution of the time series data for different periods, the frequency component distribution being obtained by converting time series data of a measurement value of a sensor that measures information on the subject's condition, The detection unit detects an abnormality in the subject based on the transience of the peak pattern identified by the comparison of the peak patterns.
2. An abnormality detection device comprising:
2. The abnormality detection device according to claim 1 , further comprising: A conversion unit converts the time series data into the frequency component distribution for each period of the time series data.
2. An abnormality detection device comprising:
3. 3. The abnormality detection device according to claim 1, The detection unit detects the behavior of the subject based on the continuity of the peak pattern identified by the comparison of the peak patterns.
2. An abnormality detection device comprising:
4. The abnormality detection device according to any one of claims 1 to 3, further comprising: The sensor measures information on the condition of the subject.
2. An abnormality detection device comprising:
5. The abnormality detection device according to any one of claims 1 to 4, further comprising: a notification unit that notifies the subject of information according to a classification result obtained by classifying the detection result by the detection unit; 2. An abnormality detection device comprising:
6. A sensor that measures information on the state of a subject; An abnormality detection device, The abnormality detection device includes: a detection unit that detects an abnormality in the subject based on a comparison of peak patterns that appear in the frequency component distribution of the time series data for different periods, the frequency component distribution of the time series data being obtained by converting the time series data of the measurement values of the sensor; The detection unit detects an abnormality in the subject based on the transience of the peak pattern identified by the comparison of the peak patterns. An anomaly detection system comprising:
7. The anomaly detection system according to claim 6, further comprising: a notification device that notifies the subject of information according to a classification result obtained by classifying the detection result by the detection unit; An anomaly detection system comprising:
8. comparing peak patterns that appear in frequency component distributions of time-series data obtained by converting time-series data of measured values of a sensor that measures information on the subject's condition, the frequency component distributions of the time-series data being for different periods; Detecting an abnormality in the subject based on the transience of the peak pattern identified by the comparison of the peak patterns.
4. An anomaly detection method comprising:
Citation Information
Patent Citations
Method, device and system for recognizing action and behavior
JP1998113343A
Gait analysis
JP2009525107A
Driver's physical condition detecting device and method
JP2018032337A
Systems and methods for monitoring and determining patient parameters from sensed venous waveforms - Patents.com
JP2020501707A
