JP7362575B2 - Method, device and program for detecting abnormality based on output of adjacent sensor device - Google Patents

Description

The present invention relates to a maintenance technique for a group of sensor devices installed at a site.

The technology of installing a large number of sensor devices in a predetermined area, detecting the occurrence of a predetermined phenomenon in the area or its vicinity, and analyzing the details of the occurrence is well known, and in fact is widely used in various fields. ing.

As an example of the use of such technology, for example, Patent Document 1 discloses a vibration detection device installed in a mesh fence to prevent pests from entering. This vibration detection device has (a) an acceleration sensor for detecting a collision, contact, or vandalism of a vermin against a mesh fence as a vibration, and detects the occurrence of vibration based on acceleration data output from this sensor. (b) a register that stores acceleration data output from this sensor; and (c) a vibration information detection unit that acquires predetermined data from among the acceleration data stored in this register, and (b) a register that stores acceleration data output from the sensor; It also includes an arithmetic unit that performs processing.

Here, the vibration information detection section (a) above sends a first notification to the calculation section when the occurrence of vibration is detected, and the calculation section (c) above sends this first notification. , if it is received while in the first state (sleep state), it moves to the second state (operating state) which consumes more power than the first state, and furthermore, it moves to the second state (operating state) which has higher power consumption than the first state, and furthermore, it moves to the second state (operating state) which has higher power consumption than the first state. The point in time at which the acceleration data is obtained from the register in (b) above is the first point in time. With such a device, the present vibration detection device can acquire sufficient data while reducing power consumption.

Further, Non-Patent Document 1 discloses a method of impact detection in a polyethylene vermin protection fence in which the vibration detection device disclosed in Patent Document 1 is installed. Specifically, with the aim of reducing the burden of inspection on polyethylene mesh fences, where there is a concern that vermin may enter through damaged areas, the acceleration sensor described above is used to detect impacts that could destroy the fence. Additionally, the machine learning method XGBoost is used to estimate the causes of detected shocks and vibrations.

Furthermore, Patent Document 2 discloses that the vibration of an optical fiber divided into multiple sections is detected for each section, and false alarms that may occur due to external noise vibrations caused by external factors such as weather are discriminated and detected. A vibration sensing sensor system for level adjustment is disclosed.

Specifically, this vibration detection system includes a detection device that is connected to an optical fiber divided into multiple sections and detects the vibrations of this optical fiber, and processes the vibration of each section detected by this detection device as section data. Then, a section data processing section that stores the section data in the section data storage device, and a section data extraction section that extracts the section data from the section data storage device, compare the extracted section data and determine whether the section data is due to external factors. an interval data comparison unit that determines whether external noise vibration is large; an alarm level control unit that controls an alarm level for detecting an abnormality when it is determined that external noise vibration is large; and an abnormality determination unit that determines whether or not there is an abnormality detection signal exceeding the alarm level.

JP2020-122755A Japanese Patent Application Publication No. 2007-233643

Takuya Kato, Uchide Nogaki, Eiji Utsunomiya, “Impact detection in polyethylene pest protection fence”, IEICE Technical Report, Vol. 119, No. 53, SeMI2019-13, pp. 177-181, 2019.

For example, a group of sensor devices installed in a predetermined area to detect the occurrence of a predetermined phenomenon, such as the vibration detection device installed in a pest protection fence described in Patent Document 1 and Non-Patent Document 1, In many cases, these systems exist in remote locations from the administrator's perspective, and are often deployed in locations that are inconvenient to access, such as mountainous areas.

On the other hand, maintenance and inspection of such a group of sensor devices is usually carried out by a manager who goes to the site where the group of sensor devices is installed, and in most cases it is carried out periodically in order to maintain the group of devices. It is. Therefore, for maintenance and inspection of sensor devices installed in remote and inconvenient areas, it is possible to reduce the number of on-site inspections, reduce the time required for on-site inspections, and reduce human costs. It is highly desirable to reduce travel costs and transportation costs.

In this regard, for example, the vibration detection devices described in Patent Document 1 and Non-Patent Document 1 are often installed in areas where it is difficult to secure a power source, so the acceleration sensor, calculation section, etc. are operated by built-in batteries. . Therefore, in order to reduce power consumption, the arithmetic unit and the like are normally in a sleep state, and are set to start sensor data processing using detection of vibrations caused by contact with an animal or the like as a trigger.

Therefore, realistically, maintenance and inspection of such a vibration detection device must be carried out periodically and on-site, taking a considerable amount of time. For example, maintenance could be performed by requesting information on the output status (alive status) of the vibration detection device via a wireless communication network, but this would require the cost of adding a timer to the vibration detection device and battery life due to the power consumption of wireless communication. Considering the increase in the number of replacements, etc., the cost effectiveness is not worth it.

Furthermore, the maintenance and inspection of the optical fiber system described in Patent Document 2 also has to be carried out periodically and on-site, taking a considerable amount of time. Here again, for example, maintenance could be carried out via a wireless communication network, but considering the costs of on-site power supply equipment, optical fiber, etc., it is important to be certain that it is equally cost-effective. It is difficult.

SUMMARY OF THE INVENTION Therefore, an object of the present invention is to provide a sensor abnormality detection method, device, and program that can reduce the burden of checking the status or maintaining a group of installed sensor devices.

According to the present invention, when a certain phenomenon to be detected occurs among a plurality of sensor devices installed within a predetermined installation range and equipped with sensors of the same type , at least two sensor devices adjacent to each other The interlocking probability, which is the probability of outputting a signal related to the detection of the phenomenon in conjunction with the influence of the phenomenon , is statistically determined as a non-zero value. A sensor abnormality detection device that detects an abnormality in a sensor device to be detected,
The state of the sensor device to be detected in a situation where at least one sensor device adjacent to the sensor device to be detected has output the signal, and the sensor device to be detected is not outputting the signal. a sensor state analysis means for identifying a certain uncoupled state ;
For the sensor device to be detected, if the uncoupled state is continuously identified under the situation , and the sensor device to be detected has not output the signal even once during that period, It is determined whether the specified number of times is a large value before satisfying a predetermined condition, and if it is determined that the specified number of times is a large value before satisfying a predetermined condition, the detection target sensor device is A sensor abnormality detection device is provided that includes a sensor abnormality determination means for determining an abnormality.

In the sensor abnormality detection device according to the present invention, the predetermined condition is that the number of times exceeds the predetermined threshold, and the sensor abnormality determination means determines whether the number of times exceeds the predetermined threshold . It is also preferable to determine that the sensor device to be detected is abnormal if the number of times exceeds the predetermined threshold value and therefore the value is large enough to satisfy the predetermined condition. . Furthermore, it is preferable that the present sensor abnormality detection device statistically determines the interlocking probability for the plurality of sensor devices, and sets the predetermined threshold value to a smaller value as the interlocking probability increases. Furthermore, it is also preferable to set the predetermined threshold value such that the value obtained by subtracting the value of the linked probability from 1 to the power of the predetermined threshold value is equal to the value of the linked probability.

Further, as an embodiment of the sensor abnormality detection device according to the present invention, the sensor abnormality detection device is configured to detect a signal from the time when one sensor device among the plurality of sensor devices outputs the signal to the time when the next signal is output. setting an output interval threshold based on a statistical output interval , which is a statistically determined time across the plurality of sensor devices ;
The sensor abnormality determination means determines that the sensor device to be detected is abnormal even if the sensor device to be detected does not output the signal for a period of time corresponding to the output interval threshold from a predetermined reference point in time. It is also preferable to determine.

Furthermore, as another embodiment of the sensor abnormality detection device according to the present invention, the present sensor abnormality detection device is configured to detect abnormality in one sensor device when each of the plurality of sensor devices is targeted for abnormality detection. An anomaly occurrence interval threshold is set based on a statistical anomaly occurrence interval, which is the time between when an anomaly is determined for another sensor device and which is statistically determined across the plurality of sensor devices. ,
The sensor abnormality determination means determines whether the sensor device to be detected does not output the signal for a period of time equal to the abnormality occurrence interval threshold from a predetermined reference point in time. It is also preferred to determine anomalies.

Furthermore, as yet another embodiment of the sensor abnormality detection device according to the present invention, the plurality of sensor devices described above are powered by a battery,
The sensor abnormality determination means uses a battery life threshold that is set based on the lifespan set for the battery or based on the statistics of the actually measured lifespan, and uses a battery lifespan threshold that is set based on the lifespan set for the battery or based on the statistical value of the actually measured lifespan. It is also preferable to determine the abnormality of the sensor device to be detected even when the signal is not output until the time equal to the battery life threshold has elapsed.

Further, as still another embodiment of the sensor abnormality detection device according to the present invention, the plurality of sensor devices described above are normally in a sleep state with respect to at least the function of outputting the signal, and the plurality of sensor devices described above are normally in a sleep state. It is also preferable that the sensor device that has detected the phenomenon cancels the sleep state and outputs the signal.

Furthermore, as another embodiment regarding the setting of the predetermined threshold value in the sensor abnormality detection device according to the present invention,
This sensor abnormality detection device classifies the phenomenon into multiple types based on the signal detected by the sensor device , determines the interlocking probability for each type of phenomenon, and sets the predetermined threshold value for the phenomenon. Set for each type,
The sensor state analysis means determines the state of the sensor device to be detected in a situation where at least one sensor device adjacent to the sensor device to be detected has output a signal related to the detection of a certain type of phenomenon, and Identifying an uncoupled state in which the target sensor device is not outputting a signal related to the detection of the relevant type of phenomenon,
It is also preferable that the sensor abnormality determining means determines whether the number of times exceeds a predetermined threshold set for the type.

Further, as a specific application example of the sensor abnormality detection device according to the present invention, the plurality of sensor devices described above are installed side by side on a fence to prevent the intrusion of a predetermined object, and the phenomenon It is also preferable to include the subject's collision with, contact with, or vandalism of this fence.

According to the present invention, in a plurality of sensor devices equipped with the same type of sensors installed within a predetermined installation range, when a certain phenomenon to be detected occurs, at least two sensors adjacent to each other Among multiple sensor devices, the interlocking probability, which is the probability that the device outputs a signal related to the detection of the phenomenon in conjunction with the influence of the phenomenon , is statistically determined as a non-zero value. A computer-implemented method for detecting an abnormality in a sensor device to be detected included in a computer, the method comprising :
The state of the sensor device to be detected in a situation where at least one sensor device adjacent to the sensor device to be detected has output the signal, and the sensor device to be detected is not outputting the signal. identifying an uncoupled condition ;
For the sensor device to be detected, if the uncoupled state is continuously identified under the situation , and the sensor device to be detected has not output the signal even once during that period, It is determined whether the specified number of times is a large value before satisfying a predetermined condition, and if it is determined that the specified number of times is a large value before satisfying a predetermined condition, the detection target sensor device is and determining an anomaly.

According to the present invention, in a plurality of sensor devices installed within a predetermined installation range and equipped with the same type of sensors , when a certain phenomenon to be detected occurs, at least two sensors adjacent to each other Among multiple sensor devices, the interlocking probability, which is the probability that the device outputs a signal related to the detection of the phenomenon in conjunction with the influence of the phenomenon , is statistically determined as a non-zero value. A program that causes a computer to function to detect an abnormality in a sensor device to be detected included in the
hand,
The state of the sensor device to be detected in a situation where at least one sensor device adjacent to the sensor device to be detected has output the signal, and the sensor device to be detected is not outputting the signal. a sensor state analysis means for identifying a certain uncoupled state ;
For the sensor device to be detected, if the uncoupled state is continuously identified under the situation , and the sensor device to be detected has not output the signal even once during that period, It is determined whether the specified number of times is a large value before satisfying a predetermined condition, and if it is determined that the specified number of times is a large value before satisfying a predetermined condition, the detection target sensor device is A sensor abnormality detection program is provided that causes a computer to function as a sensor abnormality determination means for determining an abnormality.

According to the sensor abnormality detection method, device, and program of the present invention, it is possible to reduce the burden of checking the status or maintaining a group of installed sensor devices.

1 is a schematic diagram schematically showing two embodiments of a sensor operation system including a sensor abnormality detection device according to the present invention. FIG. FIG. 1 is a functional block diagram showing a functional configuration of an embodiment of a sensor abnormality detection device according to the present invention, and a functional block diagram showing a functional configuration of an embodiment of a sensor device according to the present invention. 1 is a flowchart schematically showing an embodiment of a sensor abnormality detection method according to the present invention. 1 is a heat map for explaining an embodiment of a sensor abnormality detection method according to the present invention. It is a table for explaining an example of a sensor abnormality detection method according to the present invention.

Hereinafter, embodiments of the present invention will be described in detail using the drawings.

[Sensor abnormality detection device, sensor operation system]
FIG. 1 is a schematic diagram schematically showing two embodiments of a sensor operation system including a sensor abnormality detection device according to the present invention.

First, the sensor operation system of the embodiment shown in FIG. 1(A) will be described. In the field where this sensor operation system is deployed, in order to prevent pests from entering the specified area (for example, farmland or residential area), A net fence is installed around the outer periphery of the area.

Here, this sensor operation system detects the occurrence of collision, contact, or vandalism (predetermined phenomena) by vermin against the mesh fence, and based on the detected data, controls the damage caused by the vermin and the mesh fence. The system aims to maintain the safety of the equipment and also maintains the detection sensor equipment.

Specifically, this sensor operation system:
(A) It is equipped with a sensor 301 (Fig. 2, an acceleration sensor in this embodiment) capable of detecting vibrations caused by the occurrence of the above-mentioned collision, contact, or vandalism (predetermined phenomenon). A plurality of sensor devices 3 installed in each of the
(B) Sensor output data wirelessly transmitted from the sensor device 3 (three sensor devices 3 in Figure 1 (A)) installed in the assigned fence section (fence unit) among all the sensor devices 3 a plurality of relay devices 2 that wirelessly transfer the information to the management device 1 after adding information related to the reception time;
(c) By receiving the sensor output data of each sensor device 3 transmitted via the plurality of relay devices 2 and analyzing the sensor output data, (c) The occurrence status of the act (predetermined phenomenon) (time of occurrence, location of occurrence, extent/size, etc.), and further damage status of the mesh fence (time of occurrence of change in fence shape, location of occurrence, degree of damage, etc.) etc.), and (c) detects the occurrence of abnormalities such as failures and defects in each sensor device 3, and generates information for maintenance and inspection of the sensor device 3, as a sensor abnormality detection device. It has a management device 1.

Among these, in the three groups of sensor devices installed on the net fence in (a) above, when a pest collides with, contacts, or destroys the net fence (predetermined phenomenon), at least two groups of sensor devices adjacent to each other There is a possibility that each of the two sensor devices 3 outputs (transmits) sensor output data (vibration detection data). In other words, the "interlocking probability", which is the probability that at least two mutually adjacent sensor devices 3 output (transmit) sensor output data at approximately the same time, can be statistically determined as a non-zero value.

For example, when a pest collides with a net stretched between two adjacent pillars, at least the two adjacent sensor devices 3 installed on these pillars each output (transmit) sensor output data at approximately the same time. On the other hand, if a vermin simply comes into contact with the net, the possibility is not so high; rather, only one (single) sensor device 3 outputs (transmits) sensor output data. It is thought that the possibility of doing so may arise. Here, a certain non-zero probability of these possibilities is statistically determined depending on the manner in which the net fence is installed as a whole and the type, nature, size, etc. of the pests present.

As will be explained in detail later, the sensor device 3 of this embodiment is of a type that is powered by a battery 30 (FIG. 2) on the premise that it will be installed in an area where it is difficult to secure a power source, and the To save power, all devices except sensor 301 (FIG. 2) are normally in a sleep state. Here, when a vibration of a magnitude exceeding a predetermined threshold is detected, this sleep state is canceled, the state shifts to an operating state, and sensor output data is outputted (transmitted). Therefore, in this embodiment, it is also possible to consider that the sleep state has been canceled normally if the sensor output data has been acquired.

Incidentally, in this embodiment, the relay device 2 described in (a) above and the management device 1 described in (c) above are driven by power from, for example, a solar panel (solar battery)/storage battery system installed nearby. It's okay.

Similarly, in FIG. 1(A), the management device 1 in (c) above has a maintenance and operation unit 113 to perform the maintenance and operation function in (c) above, and this maintenance and operation unit 113 has the following functions:
(A) When at least one sensor device 3 adjacent to the sensor device 3 that is the detection target outputs (transmits) a “signal related to the detection of a phenomenon”, the sensor device 3 that is the detection target a sensor state analysis unit 113a (FIG. 2) that extracts a "non-coupled state", which is a state in which no "relevant signal" is output (transmitted);
(B) If the "non-coupled state" is continuously extracted and the sensor device 3 to be detected has not outputted "a signal related to the detection of a phenomenon" during that period, the continuously extracted state is If it is determined whether the "number of unlinked operations" is a large value until a predetermined condition is satisfied, and it is determined that the "unlinked number of times" is a large value before the predetermined condition is met, It has a sensor abnormality determination section 113b (FIG. 2) that determines an abnormality in the sensor device 3 to be detected.

In this way, the management device 1 (its maintenance operation unit 113) determines the above-mentioned "number of non-couplings" for the sensor device 3 to be detected, and based on this "number of non-couplings", the sensor device 3 to be detected The presence or absence of an abnormality in the device 3 is determined. As described above, in this sensor operation system, at least the "interlocking probability" that two adjacent sensor devices 3 each output (transmit) sensor output data can be statistically determined as a non-zero value. ing.

Therefore, assuming that the sensor device 3 to be detected is in a normal state, the above-mentioned "number of non-couplings" will be a "large value until a predetermined condition is met" determined by the "coupling probability". is extremely unlikely to occur. To put it more simply, in a system where "interlocking" occurs with a non-zero probability, it is almost impossible for "non-interlocking" to occur continuously for a predetermined period or more unless an abnormality occurs in the system. Based on such knowledge, the above-mentioned (B) sensor abnormality determination unit 113b (FIG. 2) detects the It is possible to determine an abnormality in the target sensor device 3.

Thereby, according to the management device 1, an abnormal sensor device 3 can be identified in advance using the output from this device (maintenance operation unit 113). It is also possible to shorten the maintenance work time by narrowing down the target of the maintenance work performed by going to the identified sensor device 3 (the net fence where the group is installed). Further, if there is no sensor device 3 determined to be abnormal, on-site maintenance work can be postponed. In this way, according to the management device 1, it is possible to reduce the burden of checking the status or maintaining the sensor device 3 group on site.

Incidentally, the above-described abnormal sensor device 3 may be identified by visiting the installation location of the management device 1, but in this embodiment, the management device 1 collects information related to the identified abnormal sensor device 3, For example, it is possible to wirelessly transmit a list including installation positions of abnormal sensor devices 3 to the administrator's personal computer (PC) 8 as information for maintenance and inspection. This eliminates the need for the administrator to visit the location where the management device 1 is installed, for example, when formulating a maintenance inspection plan.

In the present embodiment, the "signal related to detection of a phenomenon" in (A) and (B) above may be sensor output data (vibration detection data) generated and output (transmitted) by the sensor device 3. can. In addition, as a modification, when the sensor device 3 outputs (sends) the sleep wake-up signal generated within the device when transitioning from the sleep state to the operating state, the sleep wake-up signal is It is also possible to use "such a signal". Furthermore, when the sensor device 3 transitions to the operating state, when it generates an operating state activation signal indicating that it has entered the operating state and outputs (sends) it to the outside, the operating state activation signal is used to detect the phenomenon. It may also be referred to as "such a signal." That is, as this "signal related to detection of a phenomenon", various signals can be adopted as long as they are signals indicating (suggesting) that the sensor device 3 as the output (transmission) source is normal.

The sensor operation system deployed in a mesh fence for preventing pest intrusion has been described above with reference to FIG. , but is not limited to such embodiments. For example, a sensor operating system as shown in FIG. 1(B), which is deployed on a slope beside a road, may be used to detect the occurrence of a landslide and to identify the location of the occurrence.

Specifically, this sensor operation system, as shown in FIG. 1(B),
(a) It is equipped with an acceleration sensor (sensor 301) for detecting displacement and movement of the ground, and is a battery-powered sensor placed at a predetermined interval on a slope beside the road where there is a risk of landslides. A plurality of sensor devices 3 each having a sleep function,
(b) Out of all the sensor devices 3 placed on the slope, the sensor output data transmitted wirelessly from the sensor device 3 installed in the slope area in charge is added with information regarding the reception time. a plurality of relay devices 2 that wirelessly transfer data to the management device 1;
(c) By receiving sensor output data from each sensor device 3 transmitted via multiple relay devices 2 and analyzing the sensor output data, (c1) determining the time and location of the landslide occurrence, and (c2) As a sensor abnormality detection device that grasps the extent of the landslide that has occurred, and (c2) detects the occurrence of abnormalities such as failures and defects in each sensor device 3 and generates information for maintenance and inspection of the sensor device 3. It has a management device 1.

In particular, the management device 1 (its maintenance operation unit 113) performs the same functions and produces the same effects as in the case of FIG. 1(A), and as a result, the same as in the case of FIG. This makes it possible to reduce the burden of checking the status or maintaining the three groups of sensor devices at the site (on the slope beside the road).

Furthermore, of course, the sensor operation system including the sensor abnormality detection device (management device 1) according to the present invention can take various other embodiments. For example, the sensor 301 (FIG. 2) provided in the sensor device 3 is not limited to an acceleration sensor, and may include a visible light sensor, an infrared sensor, a temperature sensor, a humidity sensor, a voice sensor, an ultrasonic sensor, a gas sensor, etc. Various sensors can be used, such as an atmospheric pressure sensor and a tilt sensor.

Furthermore, the "phenomena" whose occurrence should be detected are not limited to those described above. In other words, when a phenomenon occurs in a group of 3 installed sensor devices, there is a possibility that at least two adjacent sensor devices 3 each output (send) a "signal related to detection of the phenomenon". As long as it is a phenomenon, a wide variety of phenomena can be adopted as the "phenomenon". For example, the "phenomena" may include various natural disasters, natural phenomena, changes in various environmental conditions, and even human-made and social phenomena such as the formation of crowds and changes in the operation status of transportation means. It also becomes possible.

Furthermore, the sensor devices 3 are arranged linearly (one-dimensionally) in the embodiment shown in FIG. 1(A), while in the embodiment shown in FIG. ), but it is also possible to adopt an embodiment in which they are arranged (three-dimensionally) within a predetermined space as a predetermined area. For example, it is also possible to adopt an embodiment in which a large number of sensor devices 3 equipped with temperature (infrared) sensors are arranged at high density in various parts of a certain huge stadium to accurately detect the appearance, behavior, and movement of people. It is.

Furthermore, the sensor device 3 may be a device that does not have a sleep function and is always in operation, and the power source of the sensor device 3, the relay device 2, and the management device 1 may not be a battery or a solar panel. For example, it may be a commercial power source. Furthermore, the sensor device 3 may be equipped with multiple types of sensors, and in this case, for example, sensor output data from each type of sensor may be handled individually to prevent the occurrence of sensor abnormality as described above. may be detected. Of course, communication between the sensor device 3 and the relay device 2 and/or between the relay device 2 and the management device 1 may be performed by wire. It is also possible to adopt a mode in which data from all sensor devices 3 is directly received.

[Device functional configuration]
FIG. 2 is a functional block diagram showing a functional configuration of an embodiment of a sensor abnormality detection device according to the present invention, and a functional block diagram showing a functional configuration of an embodiment of a sensor device according to the present invention. Note that the sensor device of this embodiment is installed in the mesh fence for preventing pest intrusion shown in FIG. 1(A).

First, according to the functional block diagram of the sensor device 3 in FIG. 2, the sensor device 3 of this embodiment includes a sensor 301 including a phenomenon detection and determination section 301a, a register 302, a calculation section 311, and a communication control section. 312, a communication interface 303, and a battery 30. Here, the arithmetic unit 311 and the communication control unit 312 may be functional components realized by a processor/memory loaded with corresponding programs, or may be incorporated into the sensor device 3 as dedicated modules respectively. There may be.

Furthermore, in this embodiment, the sensor 301, the register 302, the calculation section 311, the communication control section 312, and the communication interface 303 are all designed on the assumption that the sensor device 3 will be installed in a place where it is difficult to secure a commercial power source. It is powered by a built-in battery 30, for example a dry battery. Here, for example, after completing the initial settings, the sensor device 3 shifts the arithmetic unit 311, communication control unit 312, and communication interface 303, which consume large amounts of power, to a sleep state that is a low power consumption state to reduce power consumption. It is.

Similarly, in the functional block diagram of the sensor device 3 in FIG. 2, the sensor 301 of this embodiment detects vibrations generated in the mesh fence, for example, vibrations caused by collision or contact of pests with the mesh fence, or vandalism, and detects the vibrations. It is an acceleration sensor that outputs acceleration data related to It can be an acceleration sensor (capable of outputting data).

In this embodiment, the sensor 301 measures acceleration every predetermined period of time and outputs acceleration data to the register 302. The register 302 sequentially stores the received acceleration data in a predetermined storage area, and when the storage area runs out of space, deletes the oldest stored acceleration data. As a result, the register 302 always stores acceleration data for a certain period of time in the past.

In addition, the phenomenon detection and determination section 301a included in this sensor 301 (or may be provided outside the sensor 301) is a vibration detection and determination section in this embodiment, and is used at the time of reset (for example, when the mesh fence is stationary). Using the acceleration data at a reference value as a reference value, it is determined whether a change from the reference value in acceleration data measured and output every predetermined time period exceeds a predetermined threshold value. Here, it may be determined whether any one of the changes in the X component, Y component, and Z component of the acceleration exceeds a predetermined threshold, or a combination of the X component, Y component, and Z component may be determined. It may be determined whether the value (for example, the magnitude of the acceleration vector) exceeds a predetermined threshold value.

Further, the phenomenon detection/determination unit 301a performs the above determination on the acceleration data obtained every predetermined time period, and if it is determined that the acceleration data exceeds the predetermined threshold value consecutively for a predetermined number of times or more, the phenomenon detection/judgment unit 301a detects the target vibration. Upon detection (detection of a predetermined phenomenon), an interrupt notification is sent to the arithmetic unit 311 in the sleep state using a sleep release signal.

Similarly, in the functional block diagram of the sensor device 3 in FIG. 2, the calculation unit 311 that receives this interrupt notification (sleep release signal) shifts from the sleep state to the operating state,
(a) Read a predetermined number of samples of acceleration data from the register 302 that corresponds to the past immediately before that point in time (the point in time when it was determined that the target vibration was detected), and further,
(b) Sequentially read the acceleration data after that point from the register 302,
A predetermined process is performed on the acceleration data read out in this way. Here, this predetermined process may be, for example, a predetermined data analysis process, but in this embodiment, in order to simplify the explanation, it is assumed that the predetermined process is a process that only compiles the acceleration data.

Furthermore, when the processed acceleration data reaches a predetermined number of samples, the calculation unit 311 notifies the communication control unit 312 and the communication interface 303 in the sleep state of an interrupt using a sleep release signal, and The data is output to the communication control unit 312 that is in an operating state. Next, the communication control unit 312 appropriately transmits the received acceleration data as sensor output data to the management device 1 via the relay device 2 using the communication interface 303 which is also in an operating state.

In this embodiment, the calculation unit 311, the communication control unit 312, and the communication interface 303 start operating at the stage when a predetermined number of samples of acceleration data have been transmitted from that point in time (the point in time when it is determined that the target vibration has been detected). The device shifts from the state to the sleep state and waits until the next target vibration detection/judgment is performed.

As explained in detail above, each sensor device 3 installed on the mesh fence is normally in a sleep state, at least with respect to the function of outputting and transmitting "signals related to phenomenon detection" (sensor output data in this embodiment). When vibration of the mesh fence exceeding a predetermined level (predetermined phenomenon) is detected, this sleep state is canceled and a "signal related to the detection of the phenomenon" (sensor output data) is sent to the management device 1. and then send it. Next, the functional configuration of this management device 1 will be explained.

According to the functional block diagram of the management device 1 in FIG. 2, the management device 1 as an embodiment of the sensor abnormality detection device according to the present invention includes a communication interface 101, a sensor data storage unit 102, and a processor memory. have

Here, this processor memory stores an embodiment of the sensor abnormality detection program according to the present invention, further has a computer function, and executes the sensor abnormality detection process by executing the sensor abnormality detection program. implement. Also, from this, the management device 1 (sensor anomaly detection device) can be a dedicated device for the sensor anomaly detection process, but it can also be a general-purpose cloud server or a non-cloud server equipped with the sensor anomaly detection program according to the present invention. It may be a server, and furthermore, it may be a personal computer (PC), a notebook or tablet computer, a smartphone, or the like.

Furthermore, the above processor memory includes a sensor data management section 111, a phenomenon monitoring and detection section 112, a maintenance operation section 113 including a sensor state analysis section 113a and a sensor abnormality determination section 113b, and a communication control section 114. have Here, these functional components can be regarded as functions of a sensor abnormality detection program stored in the processor memory. Further, the process flow shown by connecting the functional components of the management device 1 (sensor abnormality detection device) with arrows in FIG. 1 can be understood as an embodiment of the sensor abnormality detection method according to the present invention.

Similarly, in the functional block diagram of the management device 1 in FIG. 2, the sensor data management section 111 receives sensor output data (acceleration data, The sensor output data is stored and managed in the sensor data storage unit 102.

Here, in this embodiment, the sensor data storage unit 102 stores data for each sensor ID, which is an identifier of the sensor device 3.
(a) The installation position of the sensor device 3 of the sensor ID (entered in advance by the administrator) (for example, position data of the installed support),
(b) Table data recorded in association with sensor output data received from the sensor device 3 having the sensor ID may be stored and managed.

The phenomenon monitoring detection unit 112 appropriately imports table data including the above sensor output data from the sensor data storage unit 102 via the sensor data management unit 111, and based on the table data for a predetermined period,
(a) Information on the occurrence of collisions, etc. (phenomena), which is information on the occurrence status (time of occurrence, location of occurrence, extent/size, etc.) of collisions, contacts, and destructive acts (predetermined phenomena) by vermin against net fences; or,
(b) Generate net fence damage information, which is information on the damage status of the net fence (time of occurrence of change in fence shape, location, degree of change, etc.).

Here, regarding (a) above, the phenomenon monitoring detection unit 112 detects, for example, the position (range) of the sensor device 3 (single or multiple adjacent to each other) that outputs (transmits) sensor output data during the predetermined period, The occurrence of the above phenomenon is determined based on the statistical values (average value, variance, etc.) of the direction and magnitude of acceleration (vibration) related to the sensor output data, its pattern (waveform), and the feature quantities generated from them. It is also preferred to generate information.

Regarding (b) above, the phenomenon monitoring detection unit 112 may detect, for example, the first predetermined number of samples of sensor output data output (transmitted) from one sensor device 3 that exists continuously during the predetermined period. The acceleration data of is taken as the data immediately before the vibration (phenomenon) occurs, and the acceleration data for a predetermined number of samples at the end is taken as the data immediately after the end of the vibration (phenomenon).Based on these data, the slope before and after the vibration (phenomenon) is generated. A difference may be calculated, and if the slope difference exceeds a predetermined threshold value, a change in the shape of the net fence may occur. Furthermore, the phenomenon monitoring detection unit 112 can generate the above-mentioned mesh fence damage information based on the sensor output data in which it is determined that a change in the shape of the mesh fence has occurred.

In a further preferred embodiment, the phenomenon monitoring and detection unit 112 generates the detected vibration (phenomenon) using a vibration cause estimation model learned by XGBoost or the like based on the table data for a predetermined period. The cause may be estimated and vibration (phenomenon) occurrence cause information, which is information specifying this cause, may be generated. For example, the correct labels for the causes of vibrations in the net fence include "collision with vermin," "contact with vermin," "collision/contact with objects smaller than vermin," "wind," and "other." It is also preferable to prepare in advance ", etc., and to presume one of these as the cause.

In addition, the phenomenon monitoring and detection unit 112 transmits, for example, information on the occurrence of collisions, etc., information on damage to mesh fences, and information on the cause of vibrations as described above to the administrator who has visited the installation location of the management device 1 ( The information may also be displayed on a display (not shown). However, in this embodiment, this information can be transmitted to the administrator PC 8 via the communication control unit 114 and the communication interface 101. This allows the administrator to check this information without going to the location where the management device 1 is installed.

Similarly, in the functional block diagram of the management device 1 in FIG. 2, the maintenance operation section 113 is a main functional section that implements life-and-death management of the sensor device 3 as preventive maintenance and predictive maintenance. Specifically, the sensor state analysis unit 113a of the maintenance operation unit 113 sequentially targets all the sensor devices 3 that are maintenance targets, one by one. When at least one sensor device 3 adjacent to the sensor device 3 transmits sensor output data (as a “signal related to detection of a phenomenon”), the sensor device 3 to be detected transmits sensor output data (as a “signal related to detection of a phenomenon”). Extract the "non-interlocking state" which is a state in which no sensor output data is being transmitted.

In addition, regarding the extraction of such a "non-coupled state", for example, as in the embodiment shown later, the output of sensor output data (( A heat map showing the transmission) status may be generated, and from this heat map, the "unlinked state" of the sensor device 3 to be detected may be identified and extracted.

Next, in this embodiment, the sensor abnormality determination unit 113b of the maintenance operation unit 113 detects that the “non-coupled state” is continuously extracted, and during that time the sensor device 3 to be detected has not outputted sensor output data even once. In this case, it is determined whether the number of uncoupling times K, which is the number of consecutive extractions, exceeds the threshold value K_th, and when it is determined that the number of uncouplings K exceeds the threshold value K_th. An abnormality in the sensor device 3 to be detected is determined. Further, such abnormality determination processing is performed for all sensor devices 3 that are maintenance targets (each one is sequentially treated as a detection target).

Here, the threshold value K_th used in the above abnormality determination process can be set to a predetermined value (for example, 3 times) based on experience. However, it is preferable that it is determined as appropriate depending on the actual situation of all the sensor devices 3 that are subject to maintenance. for example,
(a) When one sensor device 3 outputs (transmits) sensor output data (as a “signal related to detection of a phenomenon”), at least one sensor device 3 adjacent to this one sensor device 3 also becomes a sensor. The interlocking probability PWPN (P-sensor-wake-previous-next), which is the probability of outputting (transmitting) output data, is statistically determined for the system consisting of all the sensor devices 3 to be maintained,
(b) It is also preferable that the larger the PWPN is, the smaller the threshold K_th is set.

In this regard, in this embodiment, the following formula (1) K_th = log(PWPN)/log(1-PWPN)
The threshold value K_th is determined by: In other words, the threshold value K_th is calculated using the following formula (2) (1-PWPN) ^K_th = PWPN so that the value obtained by subtracting PWPN from 1 and raising the threshold value K_th to the power is equal to this PWPN.
It is decided to satisfy the following. Here, the number of uncoupling times K exceeding the threshold K_th is calculated using the following formula (3) (1-PWPN) ^K < PWPN
This will satisfy the following.

Here, the contents indicated by the above equation (3) will be explained. First, if it is assumed that the sensor device 3 to be detected is in a normal state, this normal sensor device 3 to be detected will output (transmit) despite the output (transmission) of the adjacent sensor device 3. Naturally, as the K value increases, the event in which "non-coordination" in which "non-coupling" occurs consecutively K times of discoupling (the probability is (1-PWPN) ^K ) becomes extremely unlikely to occur. Therefore, when the number of disengagements K, which is a large value that satisfies the above formula (3), is actually counted, the sensor device 3 to be detected outputs (in a situation where it should originally output (transmit) sensor output data). It is determined that the device is in an abnormal state (in the sense that it does not transmit data).

Note that the above-mentioned interlocking probability PWPN is obtained when the sensor devices 3 are installed linearly (one-dimensionally) side by side in a net fence for preventing pest intrusion (as in the present embodiment), and the front and back relationship between them is determined. In cases where
(a) When one sensor device 3 outputs (transmits) sensor output data, the interlocking probability PWP ( P-sensor-wake-previous) and
(b) When one sensor device 3 outputs (transmits) sensor output data, the interlocking probability PWN ( It is known that it can be calculated as the average of P-sensor-wake-next). Here, both the PWP in (a) above and the PWN in (b) above can be calculated as statistical values over a predetermined period (for example, half a year) for all sensor devices 3 that are subject to maintenance.

Furthermore, as another preferred embodiment regarding the threshold value K_th setting, as described above, the phenomenon monitoring and detection unit 112 classifies the cause of vibration (predetermined phenomenon) in the mesh fence into multiple types (the (categorizing the phenomenon into multiple types),
(a) The interlocking probability PWPN is statistically calculated for each vibration cause type (the relevant phenomenon type), and therefore the threshold value K_th is also set for each vibration cause type (the relevant phenomenon type),
(b) When at least one sensor device 3 adjacent to the sensor device 3 to be detected outputs sensor output data related to vibration caused by a type of cause (signal related to detection of a phenomenon by type), the detection target Extracting a "non-coupled state" in which the sensor device 3 is not outputting sensor output data (signal related to detection of phenomenon of the type) related to vibration caused by the type of cause,
(c) An abnormality in the sensor device 3 to be detected may be determined when it is determined that the number of times K of uncoupling in the extracted "non-coupling state" exceeds the threshold value K_th set for the relevant type. preferable.

In this case, for example, the threshold value K_th is set for each of the vibration causes of "collision with vermin," "contact with vermin," "collision/contact with objects smaller than vermin," "wind," and "others." For example, for the number of disengagements K for sensor output data related to vibration caused by "contact with vermin," the threshold value K_th set for "contact with vermin" is used to detect an abnormality. We will carry out the following. Thereby, it is possible to perform a threshold value determination that is suitable for the phenomenon that is actually occurring, and as a result, it becomes possible to detect an abnormality in the sensor device 3 more accurately.

Furthermore, in this embodiment, the sensor abnormality determination unit 113b (of the maintenance operation unit 113)
(a1) Average (statistical) output interval AWP: The time from when one sensor device 3 outputs (transmits) sensor output data to when it outputs (transmits) sensor output data the next time, for all sensor devices the average value (or other statistical value) over a given period in 3;
(b1) Average (statistical) abnormality occurrence interval MTBF: The time from when an abnormality is determined for one sensor device 3 to when an abnormality is determined for another sensor device 3 for all sensor devices 3. Average value (or other statistical value) over a predetermined period, and (c1) (statistical) battery life ABL: average value (or other statistics)
It is also preferable to calculate at least one of these, and use the calculated index to determine whether or not the sensor device to be detected is abnormal. Here, in this embodiment, this determination is performed in parallel with the threshold value determination of the number of uncoupled times K described above.

Specifically, the sensor abnormality determination unit 113b
(a2) Using the output interval threshold T_th_AWP (for example, AWP/3) set based on the average (statistical) output interval AWP, the sensor device 3 to be detected is If the sensor output data is not output until the time equal to the output interval threshold T_th_AWP (=AWP/3) has elapsed, it is also preferable to determine an abnormality in the sensor device to be detected.
(b2) Using the abnormality occurrence interval threshold T_th_MTBF (for example, MTBF/3) set based on the average (statistical) abnormality occurrence interval MTBF, the sensor device 3 to be detected is detected at a predetermined reference time (for example, at the start of maintenance operation) ) until the time equal to the abnormality occurrence interval threshold T_th_MTBF (=MTBF/3) has elapsed, when the sensor output data is not output, it is also preferable to determine the abnormality of the sensor device 3 to be detected,
(c2) (Statistics) Using the battery life threshold T_th_ABL (for example, ABL/2) that is set based on the battery life ABL, the sensor device 3 to be detected is If the sensor output data is not output until the time equal to the life threshold T_th_ABL (=ABL/2) has elapsed, it is also preferable to determine an abnormality in the sensor device 3 to be detected.

Furthermore, among the output interval threshold T_th_AWP, the abnormality occurrence interval threshold T_th_MTBF, and the battery life threshold T_th_ABL, the smallest threshold value is T_th, and the sensor device 3 to be detected is detected at a predetermined reference point in time (for example, at the start of maintenance operation ) It is also preferable to determine the abnormality of the sensor device 3 to be detected when the sensor output data is not output until the time corresponding to the threshold value T_th has elapsed from ).

Here, the output interval threshold T_th_AWP, the abnormality occurrence interval threshold T_th_MTBF, and the battery life threshold T_th_ABL are the calculated average (statistical) output interval AWP, average (statistical) abnormality occurrence interval MTBF, and (statistical) battery life ABL, respectively. On the other hand, it is also preferable to use a value multiplied by a coefficient determined from actual maintenance and operation data. Further, the sensor abnormality determination unit 113b may perform abnormality determination using time interval indicators other than those described above. For example, if three groups of sensor devices using sensors that can be affected by the weather are installed outdoors, it is possible to determine whether the sensor device 3 is abnormal using (annual average sunshine hours)/3 as the determination threshold. .

Above, as a method of abnormality determination for the sensor device 3, in addition to the threshold value determination process for the number of disengagements K, determination processes using the output interval threshold T_th_AWP, the abnormality occurrence interval threshold T_th_MTBF, and the battery life threshold T_th_ABL have been described. In both cases, the determination process is based on the knowledge that "a sensor device 3 that does not output (does not start) for a long period of time longer than a predetermined period is likely to be in an abnormal state such as failure or defect." .

In this embodiment, the maintenance operation department 113, which has been explained in detail above, further includes:
(a) Identify the sensor device 3 that is in an abnormal state by setting each of the sensor devices 3 that are maintenance targets as the detection target,
(b) Information related to the identified abnormal sensor device 3, for example, a list including the installation position of the abnormal sensor device 3, is wirelessly transmitted to the administrator PC 8 via the communication control unit 114 and the communication interface 101 as information for maintenance and inspection. Send.

This eliminates the need for the administrator to go to the location where the management device 1 is installed, for example, when formulating a maintenance inspection plan. However, of course, it is also possible to present information such as a list of abnormal sensor devices 3 on a display (not shown) to the administrator who has visited the installation location of the management device 1, for example. be. As a modification, when the maintenance operation unit 113 determines the presence of an abnormal sensor device 3, it is also preferable that the maintenance operation unit 113 notifies the administrator PC 8 of an alarm including information that specifies the sensor device 3.

[Sensor abnormality detection method]
FIG. 3 is a flowchart schematically illustrating an embodiment of the sensor abnormality detection method according to the present invention. It goes without saying that the sensor abnormality detection method according to the present invention is not limited to the form shown in this flowchart.

(S101) Maintenance operation values such as interlock probability PWPN, average (statistical) output interval AWP, average (statistical) abnormality occurrence interval MTBF, and (statistical) battery life ABL are set. Specifically, these values that have been statistically calculated in advance are input and set as maintenance operation values.
(S102) Using the set maintenance operation value, determine and set the determination threshold K_th and T_th (threshold of the smallest value among T_th_AWP, T_th_MTBF, and T_th_ABL).

Thereafter, an abnormality determination loop is entered in which steps S201 to S205) are executed for each time (time interval) t (=1, 2, 3, . . . ) constituting a predetermined determination period. Incidentally, in this embodiment, it is assumed that all sensor devices 3 are normal (no abnormal sensor device 3 exists) at the start of the predetermined determination period.

(S201) For each of all the sensor devices 3 that are subject to maintenance, the output (transmission) status of sensor output data at the current time (time interval) t is analyzed, and the number of disengagements K is calculated. in particular,
(a) After setting the initial value to zero for K_n, which is the number of disengagements of each sensor device 3_n (n = 1, 2, ..., N (total number of sensor devices)),
(b) At the current time (time interval) t, the sensor device is not outputting (transmitting) sensor output data even though the adjacent (immediately or immediately following) sensor device 3 is outputting sensor output data. For 3_i, add 1 to K_i, which is the number of unlinked operations,
(c) For the sensor device 3_W that outputs (transmits) sensor output data at the current time (time interval) t, the number of non-interlocking times K_W is set to zero (reset).

(S202, S203) For each sensor device 3, it is determined whether or not the number of disengagements K exceeds a threshold value K_th (K>K_th) at the current time (time interval) t, and the number of disengagements K exceeds the threshold value K_th. The sensor device 3 determined to exceed K_th is determined to be abnormal.
(S204, S205) If the current time (time interval) t exceeds the threshold t_th (t>t_th), each sensor device 3 has never outputted (transmitted) sensor output data so far. The sensor device 3 determined to have never outputted (transmitted) sensor output data is determined to be abnormal.

(S301) After the above abnormality determination loop ends (after a predetermined determination period has elapsed), the sensor devices 3 determined to be abnormal are summarized (for example, the sensor ID and the installation position in the mesh fence are associated with each other). ) Generate an abnormal sensor device list.

[Example]
Examples of the sensor abnormality detection method according to the present invention will be shown below. This example is an example of abnormality determination processing performed during the determination period from March 25th to June 11th in the management device 1 of the sensor operation system as shown in FIG. 1(A). . By the way, in this sensor operation system, a battery-powered sensor device 3 (30 in total) with a sleep function is attached to each of the 30 pillars of a mesh fence installed in mountainous areas to prevent pests from entering. PL001 to PL030) are installed, and the sensor output data sent from these devices is transmitted wirelessly to a solar battery-powered management device 1 via 10 relay devices 2 that are also solar battery powered. It receives the data and analyzes it.

Further, in this embodiment, the interlocking probability PWPN is set to 0.3, and as a result, K exceeding the threshold value K_th is (an integer of) 4 or more. In addition, the threshold T_th_AWP (=AWP/3), the threshold T_th_MTBF (=MTBF/3), and the threshold T_th_ABL (=ABL/2) are 84 days, 608 days, and 182 days, respectively, and as a result, the threshold T_th is set to 84 days, which is the shortest number of days.

FIG. 4 shows a heat map generated by the phenomenon monitoring detection unit 112 of the management device 1 in this embodiment, which shows the heat map of the sensor devices PL001 to PL030 during a part of the judgment period (June 1st to June 15th). A heat map showing the output (transmission) status of sensor output data is shown. Note that the sensor devices PL001 to PL030 are arranged in a line in descending order of their numbers (1 to 30) on the mesh fence.

The heat map in Figure 4 shows each day from June 1st to June 15th as 0:00-6:00, 6:00-12:00, 12:00-18:00, and 18:00- After dividing into four time intervals of 24:00, check whether sensor output data is output (transmitted) from each sensor device PL001 to PL030 in each of the 60 time intervals in gray squares. (with sensor output data) or with white squares (without sensor output data). Note that the degree of brightness in the gray squares indicates the intensity of the sensor output data, but in this example, the gray squares are used to indicate that "sensor output data is present" regardless of the degree of brightness. I'm handling it.

Next, FIG. 5 shows an output status table, which is a table generated from the heat map generated in this embodiment, and represents the output (transmission) status of sensor output data in the sensor devices PL001 to PL030. Here, in the same figure, only the table portion related to abnormality determination processing of the output status table is shown. In addition, in the table section, the output (transmission) status is summarized in units of one day that constitute the judgment period, and "0" in the table section indicates that there was no output (transmission) of sensor output data. On the other hand, "1" indicates that sensor output data was output (transmitted).

According to the table part shown in FIG. 5, sensor devices PL001, PL002, PL023, and PL024 to PL030 are not outputting (transmitting) sensor output data on June 11, which is the last day of the determination period. . The following describes the process and results of checking the output (transmission) status of sensor output data for these sensor devices that may be abnormal.

(a) According to the table part of FIG. 5, the number of disengagements K in the sensor device PL001 is 4 (>K_th) on April 2nd, so it is determined that the sensor device PL001 is abnormal.
(b) According to the table part of FIG. 5, sensor device PL002 has a non-cooperation count K of 4 (>K_th) on June 11th, so it is determined that the sensor device PL002 is abnormal.
(c) Sensor device PL023 does not have a non-interlocking count K of 4 or more, and has output (transmitted) sensor output data at least on June 8, so the non-output (transmission) period is the threshold T_th (=84 days) is not exceeded, and as a result, it is judged and determined to be normal (not abnormal).
(d) In any of the sensor devices PL024 to PL030, the number of non-interlocking times K never exceeds 4, and the non-output (transmission) period never exceeds the threshold T_th (=84 days). , is determined to be normal (not abnormal).

In this way, in this example, two sensor devices 3 (PL001 and PL002) are identified as abnormal sensor devices 3 among the 30 sensor devices 3 (PL001 to PL030) installed on the mesh fence. ing. As a result, the administrator who receives this information can go to the site (the mesh fence where 30 sensor devices 3 are installed) and limit the target of maintenance inspection to these two devices. Compared to the conventional case in which all machines have to be inspected, the burden of maintenance work can be significantly reduced (for example, maintenance work time is significantly shortened).

As explained in detail above, in the present invention, the number of times K of non-interlocking as described above is determined for the sensor device to be detected among the sensor devices installed at the site, and the number of times K of non-interlocking as described above is determined. , determines whether there is an abnormality in the sensor device to be detected. This makes it possible, for example, to identify in advance a sensor device that is abnormal, and, for example, to reduce maintenance work time by targeting on-site maintenance work to the identified sensor device. . Furthermore, if there is no sensor device determined to be abnormal, on-site maintenance work can be postponed. As described above, according to the present invention, it is possible to reduce the burden of checking the status or maintaining the sensor device group on site.

Furthermore, the present invention is applicable to cases in which there is a possibility that adjacent sensor devices in a group of installed sensor devices perform sensor outputs at approximately the same time, and there is a need or desire to maintain and manage the group of sensor devices, It can be applied to cases in a wide variety of fields. In other words, the present invention is highly versatile.

Regarding the various embodiments of the present invention described above, various changes, modifications, and omissions within the scope of the technical idea and viewpoint of the present invention can be easily made by those skilled in the art. The above description is merely an example and is not intended to be a restriction in any way. The invention is limited only by the claims and their equivalents.

1 Management device (sensor abnormality detection device)
101 Communication interface 102 Sensor data storage unit 111 Sensor data management unit 112 Phenomenon monitoring detection unit 113 Maintenance operation unit 113a Sensor state analysis unit 113b Sensor abnormality determination unit 114 Communication control unit 2 Relay device 3 Sensor device 30 Battery 301 Sensor 301a Phenomenon detection determination Section 302 Register 303 Communication interface 311 Arithmetic section 312 Communication control section 8 Administrator personal computer (PC)

Claims (12)

When a certain phenomenon to be detected occurs among a plurality of sensor devices equipped with the same type of sensor installed within a predetermined installation range, at least two sensor devices adjacent to each other are capable of detecting the phenomenon. For multiple sensor devices whose interlocking probability, which is the probability of outputting such signals in a linked manner under the influence of the phenomenon, is statistically determined as a non-zero value , the sensor device to be detected included therein. A sensor abnormality detection device for detecting an abnormality in,
The state of the sensor device to be detected in a situation where at least one sensor device adjacent to the sensor device to be detected has output the signal, and the sensor device to be detected is not outputting the signal. a sensor state analysis means for identifying a certain uncoupled state ;
For the sensor device to be detected, if the uncoupled state is continuously identified under the situation , and the sensor device to be detected has not output the signal even once during that period, It is determined whether the specified number of times is a large value before satisfying a predetermined condition, and if it is determined that the specified number of times is a large value before satisfying a predetermined condition, the detection target sensor device is 1. A sensor abnormality detection device comprising: sensor abnormality determination means for determining abnormality. The predetermined condition is that the number of times exceeds a predetermined threshold;
The sensor abnormality determining means determines whether the number of times exceeds the predetermined threshold, and determines that the number of times exceeds the predetermined threshold and is therefore a large value that satisfies the predetermined condition. 2. The sensor abnormality detection device according to claim 1, wherein the sensor abnormality detection device determines an abnormality in the sensor device to be detected when the above-mentioned detection is performed. The sensor abnormality detection according to claim 2, characterized in that the interlocking probability is statistically determined for the plurality of sensor devices, and as the interlocking probability becomes larger, the predetermined threshold value is set to a smaller value. equipment . Claim 3, characterized in that the predetermined threshold value is set such that the value obtained by subtracting the value of the linked probability from 1 to the power of the predetermined threshold value is equal to the value of the linked probability. The sensor abnormality detection device described in . Statistics , which is the time from when one sensor device of the plurality of sensor devices outputs the signal until the next output of the signal, which is the time statistically determined across the plurality of sensor devices; Set the output interval threshold based on the output interval ,
The sensor abnormality determining means determines whether the sensor device to be detected does not output the signal for a period of time corresponding to the output interval threshold from a predetermined reference point in time. determine abnormality
The sensor abnormality detection device according to any one of claims 1 to 4. When each of the plurality of sensor devices is targeted for abnormality detection , the time from when an abnormality is determined for one sensor device until the next time when an abnormality is determined for another sensor device, setting an anomaly occurrence interval threshold based on a statistical anomaly occurrence interval that is a statistically determined time across the sensor device ;
The sensor abnormality determining means may also detect the sensor device as the detection target even if the sensor device as the detection target does not output the signal for a period of time corresponding to the abnormality occurrence interval threshold from a predetermined reference time. determine the abnormality of
The sensor abnormality detection device according to any one of claims 1 to 5. The plurality of sensor devices are powered by batteries,
The sensor abnormality determination means uses a battery life threshold value that is set based on the lifespan set for the battery or based on the statistical value of the actually measured lifespan, and uses a battery lifespan threshold that is set based on the lifespan set for the battery or based on the statistical value of the actually measured lifespan. Even if the signal is not output until the time corresponding to the battery life threshold has elapsed, the abnormality of the sensor device to be detected is determined.
The sensor abnormality detection device according to any one of claims 1 to 6. The plurality of sensor devices are normally in a sleep state at least regarding the function of outputting the signal , and among the plurality of sensor devices, the sensor device that detects the phenomenon cancels the sleep state and performs the corresponding signal. The sensor abnormality detection device according to claim 1, wherein the sensor abnormality detection device outputs a signal. Classifying the phenomenon into multiple types based on signals detected by the sensor device,
determining the linked probability for each type of phenomenon, and setting the predetermined threshold for each type of phenomenon;
The sensor state analysis means determines the state of the sensor device to be detected in a situation where at least one sensor device adjacent to the sensor device to be detected has output a signal related to the detection of a certain type of phenomenon, and Identifying an uncoupled state in which the target sensor device is not outputting a signal related to the detection of the relevant type of phenomenon,
5. Sensor abnormality detection according to claim 3, wherein the sensor abnormality determining means determines whether the number of times exceeds a predetermined threshold set for the type. equipment . The plurality of sensor devices are installed side by side on a fence for preventing a predetermined object from entering, and the phenomenon includes a collision, contact, or vandalism of the predetermined object with the fence. The sensor abnormality detection device according to any one of claims 1 to 9. When a certain phenomenon to be detected occurs among a plurality of sensor devices equipped with the same type of sensor installed within a predetermined installation range, at least two sensor devices adjacent to each other are capable of detecting the phenomenon. For multiple sensor devices whose interlocking probability, which is the probability of outputting such signals in a linked manner under the influence of the phenomenon, is statistically determined as a non-zero value , the sensor device to be detected included therein. A computer-implemented method for detecting an anomaly in a
The state of the sensor device to be detected in a situation where at least one sensor device adjacent to the sensor device to be detected has output the signal, and the sensor device to be detected is not outputting the signal. identifying an uncoupled condition ;
For the sensor device to be detected, if the uncoupled state is continuously identified under the situation , and the sensor device to be detected has not output the signal even once during that period, It is determined whether the specified number of times is a large value before satisfying a predetermined condition, and if it is determined that the specified number of times is a large value before satisfying a predetermined condition, the detection target sensor device is A sensor abnormality detection method comprising the step of determining an abnormality. When a certain phenomenon to be detected occurs among a plurality of sensor devices equipped with the same type of sensor installed within a predetermined installation range, at least two sensor devices adjacent to each other are capable of detecting the phenomenon. For multiple sensor devices whose interlocking probability, which is the probability of outputting such signals in a linked manner under the influence of the phenomenon, is statistically determined as a non-zero value , the sensor device to be detected included therein. A program that makes a computer function to detect abnormalities in,
The state of the sensor device to be detected in a situation where at least one sensor device adjacent to the sensor device to be detected has output the signal, and the sensor device to be detected is not outputting the signal. a sensor state analysis means for identifying a certain uncoupled state ;
For the sensor device to be detected, if the uncoupled state is continuously identified under the situation , and the sensor device to be detected has not output the signal even once during that period, It is determined whether the specified number of times is a large value before satisfying a predetermined condition, and if it is determined that the specified number of times is a large value before satisfying a predetermined condition, the detection target sensor device is A sensor abnormality detection program characterized by causing a computer to function as a sensor abnormality determination means for determining an abnormality.

Images