JP5547124B2 - Program and measurement data management device - Google Patents

Description

  The present invention relates to a measurement data management device that manages measurement data of each sensor node of a sensor network.

  The sensor network system is a system for collecting measurement data (sensor data) from sensor nodes installed in various places and monitoring the state of an object. In a wireless sensor network that is one form of this network system, measurement data from sensor nodes is repeatedly wirelessly transferred via other sensor nodes and relay nodes, thereby being collected in a data aggregation device. The measurement data aggregated in the data aggregating apparatus is transmitted to an external apparatus via, for example, an existing communication cable (for example, see Patent Document 1).

JP-A-2005-328230

  When the wireless sensor network is applied to, for example, a monitoring system for structures such as railway bridges and tunnels, a large number of measurement data are collected and accumulated because the monitoring target is large and the number of sensor nodes is large. When analyzing aggregated and accumulated measurement data, changes in measurement data take into account what external events (train passage, periodic repairs and adjustment work, etc.) occurred during measurement. It is necessary to determine whether or not indicates an abnormality of the object. When there are a large number of measurement data, it is very troublesome to identify and associate an occurrence event and its occurrence period for each measurement data.

  The present invention has been made in view of the above circumstances, and an object of the present invention is to automatically associate a large number of measurement data obtained by a sensor network with an external event that has occurred. .

The first form for solving the above problem is
A program for causing a computer to manage measurement data of each sensor node of a sensor network configured by a large number of sensor nodes arranged at various locations of a structure related to the passage of a moving body,
For each external event, refer to the case database storing the measurement data of each sensor node when the external event occurs as case data, and generate an external event that matches the measurement data of each sensor node that constitutes the sensor network. Individual estimation means to estimate as an event,
Comprehensive determination means for comprehensively determining the occurrence events based on the occurrence events of the individual sensor nodes estimated by the individual estimation means and the measurement time difference based on the arrangement position relationship of the sensor nodes
An occurrence event that is comprehensively determined by the overall determination means, and that is stored in a storage unit in association with measurement data of each sensor node that constitutes the sensor network;
As a program for causing the computer to function.

As another form,
A measurement data management device that manages measurement data of each sensor node of a sensor network configured by a large number of sensor nodes arranged at various locations of a structure related to the passage of a mobile body,
For each external event, refer to the case database storing the measurement data of each sensor node when the external event occurs as case data, and generate an external event that matches the measurement data of each sensor node that constitutes the sensor network. Individual estimation means for estimating as an event;
Comprehensive determination means for comprehensively determining occurrence events based on sensor node individual occurrence events estimated by the individual estimation means and a measurement time difference based on the arrangement position relationship of the sensor nodes;
An occurrence event that is comprehensively determined by the overall determination means, and that is associated with the measurement data of each sensor node constituting the sensor network and stored in a storage unit; and
You may comprise the measurement data management apparatus provided with.

  According to the first aspect and the like, referring to the case database, an external event that matches the measurement data of each sensor node is individually estimated as an occurrence event, and then the individual occurrence event and sensor of each sensor node estimated individually An occurrence event is comprehensively determined based on a measurement time difference based on the arrangement position relationship of the nodes. Then, the comprehensively determined occurrence event is stored in the storage unit in association with the measurement data of each sensor node. As a result, the occurrence event is automatically associated with the measurement data of each sensor node.

  When some external event occurs, some change occurs in the measurement data of not only one sensor node but also a plurality of surrounding sensor nodes. For this reason, after referring to the case database storing the measurement data of each sensor node due to the occurrence of the external event as case data for each external event, individually estimating the occurrence events for each sensor node, a plurality of measurement data are collected. By comprehensively determining the occurrence event as a group, the accuracy of the occurrence event determination can be improved.

As a second form, a program of the first form,
The occurrence event association means includes
Based on the sensor node individual occurrence event estimated by the individual estimation means and the measurement time difference based on the arrangement position relationship of the sensor nodes, the data portion related to the occurrence event comprehensively determined from the measurement data of each sensor node Extraction means for extracting
Storing the comprehensively determined occurrence event in the data portion extracted by the extraction means in the storage unit;
As described above, a program for causing the computer to function may be configured.

  According to the second embodiment, the data portion related to the comprehensively determined occurrence event is extracted from the measurement data of each sensor node, and the comprehensively determined occurrence event is associated with the extracted data portion. Stored in the storage unit.

As a third form, a program according to the first or second form,
The individual estimating means is
An occurrence event is estimated for measurement data of a part of sensor nodes constituting the sensor network, and the case data corresponding to the occurrence event is compared with measurement data of other sensor nodes, and Having a validity judging means for judging validity,
When it is determined to be appropriate by the validity determination means, the same occurrence event is estimated for the measurement data of the other sensor nodes.
As described above, a program for causing the computer to function may be configured.

  According to the third embodiment, as individual estimation of individual occurrence events of sensor nodes, first, occurrence events are individually estimated for measurement data of some sensor nodes constituting the sensor network. Next, the case data corresponding to the occurrence event is compared with the measurement data of the other sensor nodes, and the validity of the occurrence event estimated earlier is determined. Presumed to be the same event.

A fourth form is a program of the third form,
The validity determination means extracts the validity determination target portion from the measurement data of the other sensor nodes according to a measurement time difference based on a positional relationship between the some sensor nodes and the other sensor nodes. And determining the validity,
As described above, a program for causing the computer to function may be configured.

  According to the fourth embodiment, the validity of the occurrence event for other sensor nodes is determined according to the measurement time difference based on the positional relationship between some sensor nodes and other sensor nodes. The validity judgment target part is extracted from the inside, and the validity judgment is performed. Thereby, since the validity is determined not for the entire measurement data but for the extracted determination target portion, it is possible to efficiently determine the occurrence event.

As a fifth form, the program according to any one of the first to fourth forms,
In the case database, for each external event, measurement data for each sensor node is stored in association with each other, and a measurement time difference based on the arrangement positional relationship between the sensor nodes is stored in association with each other.
Claims for causing the computer to function as case update means for additionally storing an occurrence event stored in the storage unit and measurement data of each sensor node constituting the sensor network as new case data in the case database. You may comprise the program as described in any one of 1-4.

  According to the fifth embodiment, in the case database, the measurement time difference based on the arrangement positional relationship between the sensor nodes is stored in association with the measurement data for each sensor node for each external event. Furthermore, the stored occurrence event and the measurement data of each sensor node are additionally stored in the case database as new case data. That is, every time an occurrence event is comprehensively determined and stored in association with measurement data of each sensor node, new case data is additionally stored and the case database is updated. For this reason, it is possible to improve the accuracy of the individual estimation of the occurrence event of each sensor node performed with reference to the case data.

As a sixth form, the program according to any one of the first to fifth forms,
You may comprise the program for functioning the said computer as a reliability calculation means to calculate the reliability with respect to the occurrence event comprehensively determined by the said comprehensive determination means using the estimation result by the said individual estimation means.

  According to the sixth aspect, the reliability for the occurrence event comprehensively determined is calculated using the individual estimation result of the occurrence event for each sensor node.

The schematic diagram of a sensor network. An example of sensor node installation and measurement data. Explanatory drawing of similarity determination with case data of measurement data. Explanatory drawing of determination of a similar event. Explanatory drawing of the similarity determination about a proximity sensor. Explanatory drawing of determination of an occurrence event. Explanatory drawing of estimation of an event occurrence period. The function block diagram of a data management apparatus. The data structural example of measurement value DB. The data structural example of measurement value management DB. Data structure example of case DB. The flowchart of a data management process.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. However, the applicable embodiment of the present invention is not limited to this.

[Overview]
The data management apparatus according to the present embodiment manages a large number of measurement data (sensor data) obtained by a sensor network. Specifically, event metadata that is data related to a situation at the time of measurement is included in the measurement data. Are associated and managed. A sensor network is a system that monitors the state of an object by aggregating measurement data from sensor nodes installed in various places.

  FIG. 1 shows a schematic configuration of the sensor network. As shown in FIG. 1, the sensor network 1 includes a sensor node S, a relay node R, a gateway G, and a management server 3.

  The sensor node S is a wireless terminal on which a microprocessor, a sensor, a wireless chip, a memory, and the like are mounted, and is installed at various locations on the monitoring target. And the measurement data by the sensor mounted is transmitted to the relay node R or gateway G predetermined as a transmission destination. The relay node R has a wireless communication function, and transmits received data from the sensor node S and other relay nodes R to another relay node R or gateway G that is predetermined as a transmission destination. The gateway G is connected to the external management server 3 through the communication line N by wired communication via an existing external cable, or by a wireless communication network such as a mobile phone network or mobile WiMAX via a wireless LAN access point AC or router. Has been. As described above, in the sensor network 1, measurement data from the sensor node S is repeatedly transmitted to the management server 3 by being transferred via the relay node R.

  FIG. 2 shows an installation example of the sensor node S and an example of measurement data. FIG. 2 shows an example in which a railway bridge (structure) is a monitoring object. As shown in the upper side of FIG. 2, a plurality (10 in FIG. 2) of sensor nodes S (S1 to S10) are installed on the steel frame portion of the bridge. The sensor node S is equipped with a vibration sensor, and outputs the degree of vibration (swing) of the installation part as a measured value.

  Then, when an external event (event) that “a train passes” occurs through this bridge, the measurement data of each sensor node S changes as shown in the lower side of FIG. That is, the measured value of each sensor node S fluctuates (vibrates) with a time difference in the order along the train traveling direction (in FIG. 2, the order of sensor nodes S1 to S10).

  In the data management device according to the present embodiment, the measurement data from the plurality of sensor nodes S is compared with case data that is measurement data at the time of occurrence of a past external event (event). The type or the like of the event that has occurred is judged (analogue), and event metadata indicating that the data is related to the event that has occurred is added to the corresponding portion of each measurement data.

  Specifically, first, as shown in FIG. 3, similarity determination between the measurement data 10 by the target sensor node S (target sensor) and the case data 20 stored in the case DB 240 is performed. The case data 20 is measurement data when an event has occurred in the past, and the case data 240 associated with the combination of the event and the sensor node S is stored in the case DB 240.

  In FIG. 3, the measurement data 10-1 of the sensor node S1 is targeted. In this case, the measurement data 10-1 of the sensor node S1 (target sensor) and each of the case data 20 of the sensor node S1 associated with each event among the case data 20 stored in the case DB 240. Similarity determination (individual estimation) is performed.

  Similarity determination between the measurement data 10 and the case data 20 is performed by correlation calculation. That is, as shown in FIG. 4, the data portion having the same time width tw as the time width tw (twa, twb) of the case data 20 is shifted from the measurement data 10 in the time direction as the data portion 12 to be compared. A so-called phase direction (time axis direction) search is performed in which the data is cut out and correlated with the case data 20. This is performed for each case data 20. If the maximum correlation value (maximum correlation value) is greater than or equal to a predetermined threshold (for example, “0.9”) among the calculated correlation values for each case data 20, it is determined that there is a “similar” event, The event corresponding to the case data 20 having the maximum correlation value is determined as the “similar event” of the measurement data 10. In addition, the data portion 12 having the maximum correlation value with the case data 20 determined as a similar event is extracted from the measurement data 10 as the similar measurement data portion. The time width tw of the similar measurement data portion is the same as the time width tw of the case data 20 of the determined similar event.

  For example, in FIG. 4, among the correlation values between the measurement data 10 and the case data 20, the correlation value between the data portion 12-1 of the measurement data 10-1 and the case data 20a-1-2 of “event A”. Exceeded the predetermined threshold. Therefore, the similar event is determined as “event A”. In the measurement data 10-1, the data portion 12-1 having the maximum correlation value with the case data 20a-1-2 of “event A” becomes the “similar measurement data portion”.

  When the similar event is determined, as shown in FIG. 5, the similar determination (similar to the measurement data 10 and each of the case data 20 associated with the similar event) is performed for each of the other sensor nodes S in the vicinity. Individual estimation) is performed to determine whether a similar event is valid as an occurrence event.

  The “neighbor sensor” is a sensor node S in which a change occurs in measurement data due to a similar event, and is determined based on past measurement data. In the example of FIG. 2, the event is “passage of train”. In the measurement data based on this event, a change appears in the measurement data of all sensor nodes S installed in the steel frame portion of the bridge. All sensor nodes S2 to SN other than S1 are set as proximity sensors.

  Then, the similarity determination between the measurement data 10-2 to 10-N of each of the (N-1) sensor nodes S2 to SN that are proximity sensors and the case data 20 of “event A” that is a similar event is performed. .

  Thus, when the similarity determination about each measurement data 10 of each sensor node S is performed, the comprehensive determination of the occurrence event is performed based on the determination result. That is, as shown in FIG. 6, the ratio of the number of neighboring sensors determined as “similar” by similarity judgment (individual estimation) with the similar event case data 20 to the total number of neighboring sensors is a predetermined threshold (for example, 0.7) In the case of the above, “similar event” is determined as “occurrence event”.

  When a similar event is determined as an occurrence event, event metadata related to the occurrence event is added to the measurement data 10 of each of the target sensor and the proximity sensor. Specifically, an event occurrence period is specified for each measurement data 10, and event metadata is added to the measurement data portion 12 of the specified event occurrence period.

  For specifying the event occurrence period, for the measurement data 10 of the target sensor, the data portion having the maximum correlation value with the case data 20 is set as the event occurrence period. Among the measurement data 10 of the proximity sensors, for the measurement data 10 determined as “similar” by the similarity determination, the data portion 12 having the maximum correlation value with the case data 20 is referred to as an “event occurrence period”. To do. Further, among the measurement data 10 of the proximity sensors, the measurement data determined as “dissimilar” by the similarity determination, as shown in FIG. 7, the time relationship with the other measurement data 10 determined as “similar” From the above, the event occurrence period is estimated.

  FIG. 7 shows measurement data 10 and case data 20 for each of the sensor nodes S1 to S3. In FIG. 7, the measurement data 10 of each of the sensor nodes S <b> 1 to S <b> 3 is shown on the upper side, and the case data 20 of the occurrence event of each of the sensor nodes S <b> 1 to S <b> 3 stored in the case DB 240 is shown on the lower side. Among these, the measurement data 10-2 of the sensor node S2 is determined as “dissimilar”, and the measurement data 10-1 and 10-3 of the sensor nodes S1 and S3 are determined as “similar”.

In this case, the “event occurrence period” of the measurement data 10-2 of the sensor node S2 determined as “dissimilar” is estimated from the case data 20-1 of the similar event of the sensor node S1 determined as “similar”. . That is, the time difference Δu ₁ , ₂ between the start times u ₁ , u ₂ of the case data 20-1, 20-2 of the sensor nodes S1, S2 is calculated. Next, the time t ₂ = (t ₁ + Δu _1,2 ) delayed by the time difference Δu _1,2 from the start time t ₁ of the event occurrence period of the measurement data 10-1 of the sensor node S1 is set to the measurement data 10-2. and start time t ₂ of the event occurrence period. Then, the same period as the event occurrence period U ₂ of the case data 20-2, that is, the period of “time t ₂ (= t ₁ + Δu _1,2 ) to t ₂ + U ₂ ” is generated in the measurement data 10-2. Period.

When the occurrence event is confirmed, the measurement data 10 is associated with the occurrence event and additionally registered in the case DB 240 as new case data 20. That is, the similar measurement data portion 12 of the measurement data 10 of the target sensor is additionally registered in the case DB 240 as new case data 20 in association with the confirmed occurrence event. Further, for the measurement data 10 of the proximity sensor, the data portion (similar measurement data portion) 12 corresponding to each event occurrence period is additionally registered in the case DB 240 as new case data 20 in association with the confirmed occurrence event. To do.
At this time, the time width tw of the similar measurement data portion 12 of the measurement data 10 is the time width tw of the case data 20 of the confirmed occurrence event.

  Further, when the occurrence event is determined, the measurement data 10 of each sensor node S is associated with the “reliability” of determination of the occurrence event in the event metadata.

  The reliability of the determination of the occurrence event is the ratio of the number of sensor nodes S determined as “similar” by the similarity determination out of the total number N of sensor nodes S. For example, in FIG. 6, the similar event of the sensor node S1 that is the target sensor is “event A”. Then, it was determined that M (≦ N−1) sensor nodes “similar” to the event A among the (N−1) sensor nodes S2 to SN that are neighboring sensors. Therefore, in this case, the reliability of event A, which is an occurrence event, is “(1 + M) / N”.

[Constitution]
FIG. 8 is a functional configuration diagram of the data management apparatus 100. According to FIG. 8, the data management apparatus 100 is functionally configured as a computer system including a processing unit 110, an operation input unit 120, a display unit 130, a communication unit 140, and a storage unit 200. The

  The processing unit 110 is a control device realized by a processor such as a CPU, for example. The processing unit 110 controls the data management device 100 based on data input from the operation input unit 120, programs and data stored in the storage unit 200, and the like. Instructions and data transfer to each constituent part are performed, and overall control of the data management apparatus 100 is performed.

  In the present embodiment, the processing unit 110 performs data management processing according to the data management program 210. In this data management process, measurement value management data is generated by adding “event metadata” that is data related to an event (event) that occurred during measurement of the measurement data to the measurement data acquired from the sensor node. Here, the measurement data acquired from the sensor node is accumulated and stored in the measurement value DB 220. The measurement value management data generated by adding event metadata to the measurement data is accumulated and stored in the measurement value management DB 230.

  FIG. 9 is a diagram illustrating an example of a data configuration of the measurement value DB 220. As shown in FIG. 9, the measurement value DB 220 is a set of sensor-specific measurement data groups 221 for each sensor node, and each sensor-specific measurement data group 221 includes the measurement time t and the measurement along with the sensor ID 222 of the corresponding sensor node. Measurement data 223 (10) in which values and normal metadata are associated in time series is stored.

  FIG. 10 is a diagram illustrating an example of a data configuration of the measurement value management DB 230. As shown in FIG. 10, the measurement value management DB 230 is a set of sensor-specific measurement management data groups 231 for each sensor node, and each sensor-specific measurement management data group 231 includes a measurement time along with the sensor ID 232 of the corresponding sensor node. Measurement management data 233 in which t, measurement values, normal metadata, and event metadata are associated in time series is stored. The measurement management data 233 has a configuration in which event metadata is added to the measurement data 223 of the sensor-specific measurement data group 221 of the corresponding sensor node.

  The processing unit 110 performs similarity determination between the measurement data 223 stored in the measurement value DB 220 and the case data 252 stored in the case DB 240 for each sensor node, and determines a similar event.

  FIG. 11 is a diagram illustrating an example of a data configuration of the case DB 240. As shown in FIG. 11A, the case DB 240 is a set of event-specific case data groups 241 for each event, and each event-specific case data group 241 has a corresponding sensor ID 242 and a nearby sensor ID 243 and its vicinity. The sensor-specific case data group 250 for each sensor node specified by the sensor ID 243 is stored. As shown in FIG. 11B, each sensor-specific case data group 250 includes case data 252 (20) in which the time, measurement value, and normal metadata are associated in time series together with the sensor ID 251 of the corresponding sensor node. Is stored. The case data 252 is partial data of the measurement data 223 corresponding to the corresponding event and sensor node in the measurement value DB 220.

  The processing unit 110 calculates a correlation value by performing a correlation operation between the measurement data 223 of the target sensor node (target sensor) and each of the case data 252 associated with each event of the target sensor. Then, among the calculated correlation values for each case data 252, the event corresponding to the case data 252 having the maximum correlation value is determined as a “similar event”.

  Subsequently, the processing unit 110 identifies the proximity sensor of the target sensor with reference to the event-specific case data group 241 of the similar event, and the similarity between the measurement data 223 and the similar event case data 252 for each of the identified nearby sensors. A determination is made to determine “similar” or “dissimilar”. Then, the ratio of the number of neighboring sensors that have determined “similarity” with the case data 252 to the total number N of sensor nodes is calculated. If this ratio is equal to or greater than a predetermined threshold, a similar event is determined as an occurrence event.

  When the occurrence event is determined, event metadata related to the occurrence event is added to each measurement data 223 and stored as measurement management data 233 in the measurement value management DB 230. At this time, for the measurement data 223 of the target sensor, a data portion (similar data portion) having a maximum correlation value with the case data 252 determined to be similar is set as an event occurrence period, and an event related to the generated event is included in the similar data portion. Add metadata.

  Among the measurement data 223 of the proximity sensors, for the measurement data determined to be “similar” to the case data 252, a data portion (similar data portion) having a maximum correlation value with the case data 252 determined to be similar is used. The event occurrence period is set, and event metadata related to the occurrence event is added to the similar data portion.

  Of the measurement data 223 of the proximity sensor, the measurement data 223 determined as “dissimilar” with the case data 252 is the event occurrence period of the measurement data 223 of the proximity sensor determined as “similar” with the case data 252. The event occurrence period is estimated based on the time difference between and event metadata related to the occurrence event is added to the data portion corresponding to the event occurrence period.

  In addition, when the generated event is determined, the processing unit 110 additionally registers the new case data 252 based on the measurement data 223 in the case DB 240. That is, among the measurement data 223 of the target sensor, the data portion (similar data portion) corresponding to the event occurrence period is additionally registered in the case DB 240 as new case data 252. In addition, the data portion (similar data portion) corresponding to the event occurrence period in the measurement data 223 of each neighboring sensor is additionally registered in the case DB 240 as new case data 252.

  Returning to FIG. 8, the operation input unit 120 is an input device realized by, for example, a keyboard, a mouse, a touch panel, various switches, and the like, and outputs an input signal corresponding to the operation input unit to the processing unit 110.

  The display unit 130 is a display device realized by an LCD or the like, for example, and displays various screens based on display signals input from the processing unit 110.

  The communication unit 140 is a communication device realized by, for example, a wireless communication module, a router, a modem, a TA, a wired cable jack, a control circuit, and the like, and performs data communication with an external device.

  The storage unit 200 stores a system program for realizing various functions for the processing unit 110 to control the data management apparatus 100 in an integrated manner, a program and data for executing the data management processing of the present embodiment, and the like. At the same time, it is used as a work area of the processing unit 110 and temporarily stores calculation results executed by the processing unit 110 according to various programs and input signals from the operation input unit 120. In the present embodiment, the storage unit 200 stores a data management program 210 as a program, and stores a measurement value DB 220, a measurement value management DB 230, and a case DB 240 as data.

[Process flow]
FIG. 12 is a flowchart for explaining data management processing. According to FIG. 12, the processing unit 110 performs an iterative process (loop A) for each sensor node. In this loop A, similarity determination is performed by correlation calculation between measurement data 223 of the target sensor node (target sensor) and each case data 252 of each event (step A1).

  As a result, if it is determined that there is a similar event (step A3: YES), then a proximity sensor of the target sensor is specified (step A5). Then, for each specified proximity sensor, similarity determination is performed by correlation calculation between the measurement data 223 of the proximity sensor and the similar event case data 252 (step A7).

  As a result, it is determined whether or not the similar event is regarded as an occurrence event based on the number of sensor nodes that have determined “similarity” to the similar event among neighboring sensors (step A9). (Step A11: YES), the reliability of determination of similar cases is calculated based on the number of sensor nodes that have determined “similarity” to similar cases among all sensor nodes. (Step A13). Next, similar events are presented as candidates for occurrence events to the user together with the calculated reliability by displaying a predetermined display screen or the like (step A15).

  If the occurrence event candidate is confirmed by the user (step A17: YES), the similar event is determined as the occurrence event (step A19). Next, the event occurrence period in the measurement data 223 of the target sensor and the measurement data 223 of each neighboring sensor is specified (steps A21 to A23). Then, event metadata relating to the occurrence event is added to the data portion (similar data portion) corresponding to the specified event occurrence period in the measurement data 223 of each of the target sensor and the proximity sensor (step A25). Further, the data portion corresponding to the event occurrence period in the measurement data 223 of each of the target sensor and the proximity sensor is additionally stored in the case DB 240 as new case data 252 regarding the generated event (step A27). The process of loop A is performed in this way.

  When the process of Loop A for all sensor nodes is performed, the processing unit 110 ends the data management process.

[Action / Effect]
As described above, according to the present embodiment, the measurement data 10 obtained from each sensor node S configuring the sensor network system 1 is subjected to similarity determination with the case data 20, and the occurrence event is estimated. Event metadata related to the estimated occurrence event is added to the data portion of the event occurrence period of each measurement data 10 of the node S.

[Modification]
Note that embodiments to which the present invention can be applied are not limited to the above-described embodiments, and it is needless to say that the embodiments can be appropriately changed without departing from the spirit of the present invention.

(A) Estimation of Occurrence Event For example, in the above-described embodiment, first, the number of sensor nodes that have determined a similar event of the target sensor and have determined “similarity” to the case data 20 of this similar event among neighboring sensors. Based on the above, it was decided to determine similar events as occurrences. For each measurement data 10 of each sensor node, correlation calculation with the case data 20 is performed to individually determine similar events, and among the similar events for each individually determined sensor node, the most similar events are defined as occurrence events. You may do that.

(B) Reliability of Occurring Event In the above-described embodiment, the reliability of determining an occurrence event is a value based on the number of sensor nodes that have determined “similarity” to the occurrence event among the total number N of sensor nodes. However, this may be a value based on the correlation value between the measurement data 10 and the case data 20. Specifically, for example, the average value of the correlation values between the measurement data 10 and the case data 20 of each sensor node determined to be “similar”, or the measurement data 10 and the case data of all the sensor nodes, respectively. The average value of the correlation values with 20.

(C) Similarity determination Further, when performing a correlation calculation between the measurement data 10 of the sensor node and the case data 20, a correlation calculation with the case data 20 is performed only for a data portion of the measurement data 10 that is estimated to have a high correlation. It is also good to do. Specifically, for example, when the similarity determination for the event “train passing” is performed, a predetermined time range (for example, ± 2 minutes) before and after the train passing time obtained from the train schedule is used. The data portion is subjected to similarity determination with the case data 20 corresponding to the event “train passing”.

(D) Multiple events that occur in relation to multiple events that occur in relation to each other, once the occurrence of one event is determined, has any other related events occurred in the same way? May be estimated. For example, in the example shown in FIG. 2, after the external event “train passage”, an event (related event) “free vibration” resulting from this “train passage” occurs.

  Specifically, event-related data in which a plurality of events that are related to each other in a temporal context are associated in advance. In the data management process (see FIG. 12), after the occurrence of the target sensor is confirmed, the event metadata is added, and the case sensor 240 is additionally registered as new case data (steps A1 to A27). Subsequently, it is determined whether or not a related event has occurred. That is, in the event-related data, other events (related events) that are defined as related to the previously determined occurrence event are specified, and the measurement data of the target sensor and the nearby sensor and the related event case data Similarity determination is performed to determine whether a related event has occurred. If it is determined that the related event has occurred, event metadata related to the related event is added to the corresponding data portion in the measurement data of each of the target sensor and the proximity sensor, and the corresponding data portion is It is additionally registered in the case DB 240 as new case data. In this way, by establishing the relationship between related events, it is more efficient to estimate occurrence events and add event metadata.

(E) External Event In the above-described embodiment, “train passage” is described as an example of the external event. However, the present invention is not limited to this, and the same applies to external events such as construction, maintenance inspection, and repair work. It is.

(F) Monitoring Object In the above-described embodiment, the case where “bridge” is taken as an example of the monitoring object of the sensor network has been described. However, the present invention is not limited to this, and the same applies to an object such as “tunnel”. It is. Moreover, although the sensor mounted in the sensor node is a vibration sensor, it can be a sensor of a type corresponding to the monitoring object, such as an illuminance sensor, a temperature sensor, or a strain sensor. Further, the present invention is not limited to railroads, and can be similarly applied to structures such as tunnels and bridges of automobile roads.

DESCRIPTION OF SYMBOLS 1 Sensor network system S Sensor node, R relay node, G gateway 3 Management server 10 Measurement data, 12 Similar data part 20 Case data 100 Data management apparatus 110 Processing part, 120 Operation input part, 130 Display part, 140 Communication part 200 Storage Unit 210 Data management program 220 Measurement value DB, 230 Measurement value management DB, 240 Case DB

Claims (5)

A program for causing a computer to manage measurement data of each sensor node of a sensor network configured by a large number of sensor nodes arranged at various locations of a structure related to the passage of a moving body,
For each external event, refer to the case database storing the measurement data of each sensor node when the external event occurs as case data, and generate an external event that matches the measurement data of each sensor node that constitutes the sensor network. Individual estimation means to estimate as an event,
Comprehensive determination means for comprehensively determining the occurrence events based on the occurrence events of the individual sensor nodes estimated by the individual estimation means and the measurement time difference based on the arrangement position relationship of the sensor nodes
An occurrence event that is comprehensively determined by the overall determination means, and that is stored in a storage unit in association with measurement data of each sensor node that constitutes the sensor network;
It makes the computer function as,
The individual estimating means is
An occurrence event is estimated for measurement data of a part of sensor nodes constituting the sensor network, and the case data corresponding to the occurrence event is compared with measurement data of other sensor nodes, and Having a validity judging means for judging validity,
When it is determined to be appropriate by the validity determination means, the same occurrence event is estimated for the measurement data of the other sensor nodes.
program. The occurrence event association means includes
Based on the sensor node individual occurrence event estimated by the individual estimation means and the measurement time difference based on the arrangement position relationship of the sensor nodes, the data portion related to the occurrence event comprehensively determined from the measurement data of each sensor node Extraction means for extracting
Storing the comprehensively determined occurrence event in the data portion extracted by the extraction means in the storage unit;
The program according to claim 1. The validity determination means extracts the validity determination target portion from the measurement data of the other sensor nodes according to a measurement time difference based on a positional relationship between the some sensor nodes and the other sensor nodes. And determining the validity,
The program according to claim 1 or 2 . The computer according to any one of claims 1 to 3 , for causing the computer to function as a reliability calculation unit that calculates a reliability for the occurrence event comprehensively determined by the comprehensive determination unit using an estimation result by the individual estimation unit. The program described in. A measurement data management device that manages measurement data of each sensor node of a sensor network configured by a large number of sensor nodes arranged at various locations of a structure related to the passage of a mobile body,
For each external event, refer to the case database storing the measurement data of each sensor node when the external event occurs as case data, and generate an external event that matches the measurement data of each sensor node that constitutes the sensor network. Individual estimation means for estimating as an event;
Comprehensive determination means for comprehensively determining occurrence events based on sensor node individual occurrence events estimated by the individual estimation means and a measurement time difference based on the arrangement position relationship of the sensor nodes;
An occurrence event that is comprehensively determined by the overall determination means, and that is associated with the measurement data of each sensor node constituting the sensor network and stored in a storage unit; and
Equipped with a,
The individual estimating means is
An occurrence event is estimated for measurement data of a part of sensor nodes constituting the sensor network, and the case data corresponding to the occurrence event is compared with measurement data of other sensor nodes, and Having a validity judging means for judging validity,
When it is determined to be appropriate by the validity determination means, the same occurrence event is estimated for the measurement data of the other sensor nodes.
Measurement data management device.