JP2017215870A - Sensor control apparatus, sensor system, sensor control method, and sensor control program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a sensor control apparatus which saves power of a sensor device without reducing measurement accuracy, a sensor system, a sensor control method and a sensor control program.SOLUTION: A sensor control apparatus is configured to control multiple sensor devices each including a battery and comprises: a collection part for collecting a measurement and remaining power of the battery from each of the sensor devices; a calculation part for performing regression analysis on the measurement to calculate an index value indicating importance when estimating a measurement of the other sensor device from the measurement of each of the sensor devices; and a selection part for selecting a sensor device successively from most remaining battery power or from a highest evaluation value that is calculated based on the index value and the remaining battery power, in such a manner that a total of index values becomes equal to or greater than a threshold value. After the selection part selects a sensor device, the collection part collects a measurement from the selected sensor device and estimates a measurement of the other sensor device from the collected measurement.SELECTED DRAWING: Figure 2

Description

  This case relates to a sensor control device, a sensor system, a sensor control method, and a sensor control program.

  IoT (Internet Of Things) solutions that utilize information from sensors installed in various environments and are used for detecting abnormalities and controlling devices are becoming widespread. Examples include solutions for monitoring river and sewer flooding and solutions for detecting road and bridge anomalies.

  In this type of solution, the sensor cannot be connected to a power source to be used outdoors, and operates with power from a battery. Since the sensor is installed in a wide range or in a dangerous environment, the cost of battery replacement or sensor replacement increases. Therefore, power saving of the sensor is important in operating the sensor system.

  Regarding power saving, for example, Patent Documents 1 and 2 disclose techniques related to power saving of nodes in a wireless network. Patent Document 3 discloses that sensor information of an unusable sensor is estimated from sensor information of another sensor related to the sensor.

International Publication No. 2006/067922 JP 2014-204194 A International Publication 2012/165496

  For example, as in the technique disclosed in Patent Document 3, if the measurement value of a sensor is estimated from the measurement values of other sensors, the number of sensor measurements can be reduced. However, the sensors that perform measurement are partially biased. Since the power consumption of the sensor is not averaged, the entire sensor cannot be effectively saved. On the other hand, if a sensor that performs measurement is simply selected based on the remaining battery level, the measurement value of another sensor may be estimated from a measurement value with low relevance, which may reduce the measurement accuracy.

  It is an object of the present invention to provide a sensor control device, a sensor system, a sensor control method, and a sensor control program that save power in the sensor device without reducing measurement accuracy.

  In one aspect, the sensor control device controls a plurality of sensor devices including a battery, the collection unit collecting the measurement values and the remaining amount of the battery from the plurality of sensor devices, and the measurement A calculation unit for calculating an index value indicating importance in estimating a measurement value of another sensor device from the measurement value for each of the plurality of sensor devices by performing regression analysis of the value, and a total of the index values Select one or more sensor devices from the plurality of sensor devices in descending order of the remaining amount of the battery or in descending order of the evaluation value calculated based on the index value and the remaining amount of the battery. A selection unit; and after the selection unit has selected one or more sensor devices, the collection unit is arranged in front of the one or more selected sensor devices among the plurality of sensor devices. The measurements were collected, to infer the measurement values of the other sensor device from the measured values the collection.

  In one aspect, a sensor system has a plurality of sensor devices provided with a battery, and a sensor control device that controls the plurality of sensor devices, and the sensor control device receives measured values from the plurality of sensor devices, and The importance of estimating the measured values of the other sensor devices from the measured values for each of the plurality of sensor devices by performing a regression analysis on the measured values and the collection unit for collecting the remaining amount of the battery. A calculation unit that calculates an index value to be shown, and a calculation based on the index value and the remaining amount of the battery from the plurality of sensor devices in order of increasing the remaining amount of the battery so that the sum of the index values is equal to or greater than a threshold value A selection unit that selects one or more sensor devices in descending order of evaluation value, and the collection unit selects the one or more sensor devices after the selection unit selects one or more sensor devices. Of, and collecting the measurements from one or more sensor devices that are the selected infer a measurement of another sensor device from the measured values the collection.

  In one aspect, the sensor control method includes a step of collecting a measured value and a remaining amount of the battery from the plurality of sensor devices, respectively, in the sensor control method for controlling a plurality of sensor devices including a battery, and the measured value. And calculating the index value indicating the importance in estimating the measured value of the other sensor device from the measured value for each of the plurality of sensor devices, and the sum of the index values is a threshold value Selecting one or more sensor devices from the plurality of sensor devices in descending order of the remaining amount of the battery or in descending order of the evaluation value calculated based on the index value and the remaining amount of the battery; Collecting the measurement values from the selected one or more sensor devices out of the plurality of sensor devices; and from the collected measurement values, the plurality of sensor devices. Of, and a step to estimate the measure of the other sensor device, a method executed by a computer.

  In one aspect, the sensor control program collects a measured value and a remaining amount of the battery from each of the plurality of sensor devices, and returns the measured value in the sensor control program for controlling the plurality of sensor devices including a battery. By analyzing, for each of the plurality of sensor devices, an index value indicating the importance in estimating the measured value of another sensor device from the measured value is calculated, and the total of the index values is equal to or greater than a threshold value. And selecting one or more sensor devices from the plurality of sensor devices in descending order of the remaining amount of the battery or in descending order of the evaluation value calculated based on the index value and the remaining amount of the battery. The measured values are collected from the selected one or more sensor devices, and the other sensors of the plurality of sensor devices are collected from the collected measured values. I guess measurements of location, and a process is a program for causing a computer to execute.

  As one aspect, it is possible to save power in the sensor device without reducing measurement accuracy.

It is a lineblock diagram showing an example of a sensor system. It is a sequence diagram which shows an example of operation | movement of a sensor system. It is a block diagram which shows an example of a sensor apparatus. It is a block diagram which shows an example of a sensor control apparatus. It is a figure which shows an example of a sensor information table. It is a figure which shows an example of a measured value table. It is a flowchart which shows an example of a regression analysis process. It is a figure which shows an example of a regression coefficient table and an importance table. It is a flowchart which shows an example of a threshold value setting process. It is a figure which shows an example of a regression coefficient. It is a flowchart which shows an example of a sensor selection process. It is a flowchart which shows the operation example at the time of execution of an accuracy monitoring process. It is a figure which shows an example of the measured value and regression coefficient which are used for accuracy determination. It is a figure which shows the other example of the measured value and regression coefficient which are used for accuracy determination. It is a figure which shows an example of the operation time of the sensor system of a comparative example and an Example. It is a flowchart which shows the other example of a sensor selection process.

  FIG. 1 is a configuration diagram illustrating an example of a sensor system. The sensor system includes a control server 1, a plurality of gateway devices (GW devices) 3, and a plurality of sensor devices (# 1 to # 9) 2. “# 1” to “# 9” are sensor IDs for identifying the sensor device 2. Further, although nine sensor devices 2 are provided in this example, the number is not limited.

  The control server 1 is an example of a sensor control device and controls each sensor device 2. The control server 1 is connected to a network such as the Internet. The control server 1 wirelessly communicates with some sensor devices (# 1, # 2, # 4, # 5, # 7, # 8) 2 via the GW device 3. The GW device 3 has a routing function, and relays wireless communication between the control server 1 and each sensor device 2. Further, the control server 1 communicates wirelessly with other sensor devices (# 3, # 6, # 9) 2 without passing through the GW device 3. The connection form and connection form between the control server 1 and the sensor device 2 are not limited to this example.

  Each of the plurality of sensor devices 2 includes a battery, and is disposed, for example, in a predetermined outdoor range FD. The plurality of sensor devices 2 are installed in, for example, rivers, sewers, roads, bridges, and the like, and detect abnormalities by executing measurement processing. For this reason, for example, in the case of rivers and sewers, flooding is detected, and in the case of roads and bridges, defects due to aging are detected.

  The plurality of sensor devices 2 measure the measured values and transmit them to the control server 1. Examples of the measured value include temperature and humidity. Each sensor device 2 transmits a measurement value to the control server 1 for each measurement, and the control server 1 detects an abnormality based on the assumed value received from each sensor device 2.

  In general, not all of the plurality of sensor devices 2 are in an operating state. The control server 1 selects a sensor device 2 (hereinafter referred to as “operating sensor device”) to be in an operating state from a plurality of sensor devices 2 by a sensor selection process to be described later, and performs measurement processing only on the operating sensor device 2. Let it be done. On the other hand, the other sensor devices 2 not selected by the control server 1 (hereinafter referred to as “non-operating sensor devices”) are in the sleep state, and thus do not perform the measurement process.

  The control server 1 estimates the measurement value of the non-operation sensor device 2 from the measurement value of the operation sensor device 2. That is, the control server 1 acquires the estimated value instead of the measured value of the non-operating sensor device 2. Accordingly, since it is not necessary for all the sensor devices 2 to perform measurement processing in parallel, power saving can be realized by reducing the number of measurements of the sensor device 2.

  However, if the operating sensor device 2 is biased to a part of the sensor device 2, the power consumption of each sensor device 2 is not averaged, so that the entire sensor device 2 cannot be effectively saved. On the other hand, when the operating sensor device 2 is simply selected based on the remaining battery level, the measurement value of the non-operating sensor device 2 may be estimated from the measurement values with low relevance, and thus the measurement accuracy may be reduced.

  Therefore, the control server 1 according to the present embodiment is important in estimating the measurement values of the other sensor devices 2 from the measurement values of the sensor devices 2 by performing regression analysis processing on the measurement values of the sensor devices 2. The index value shown (hereinafter referred to as “importance”) is calculated. The control server 1 selects one or more operation sensor devices 2 in descending order of the remaining battery power from the plurality of sensor devices 2 such that the total importance is equal to or greater than the threshold value.

  For this reason, the control server 1 can select the sensor device 2 having a sufficient remaining battery level as the operation sensor device 2 while considering the accuracy of estimation of the measurement value according to the importance. Therefore, according to the present embodiment, it is possible to save power in the sensor device 2 without reducing the measurement accuracy. An example of operation of the sensor system will be described below.

  FIG. 2 is a sequence diagram illustrating an example of the operation of the sensor system. First, the control server 1 sets a measurement value collection frequency H and a measurement value estimation accuracy K from a terminal such as a computer via a network. The frequency H is, for example, a measurement value collection period, and the accuracy K is, for example, an error of an estimated value with respect to the measurement value.

  The control server 1 executes the entire collection process SQ1 with the frequency H. In the overall collection process SQ1, the control server 1 sequentially transmits a transmission request for the measured value and the remaining battery level to each sensor device 2. Upon receiving the transmission request, each sensor device 2 executes a measurement process and transmits the measurement value and the battery remaining amount to the control server 1.

  Symbol X indicates an example of a transmission request. The transmission request X includes, for example, “sensorID” indicating the sensor ID of the requesting sensor device 2, “interval” indicating the frequency H, and “attribute” indicating the collection target information. In “attribute”, “temperature” and “battery_charge” are set. “Temperature” indicates a temperature that is an example of a measured value, and “battery_charge” indicates a remaining battery level.

  The entire collection process SQ1 is repeatedly executed a plurality of times at a period corresponding to “interval” in the transmission request, that is, the frequency H.

  In this way, the control server 1 collects the measured values and the remaining battery capacity of each sensor device 2. In this example, the case where the measurement value and the battery remaining amount are collected from each sensor device 2 to the control server 1 by transmitting a transmission request X from the control server 1 to each sensor device 2 is not limited to this. . For example, each sensor device 2 may actively transmit the measurement value and the remaining battery level to the control server 1 at a frequency H without receiving a transmission request. Measurements and battery level are collected.

  Next, the control server 1 executes a regression analysis process SQ2 of the measurement values of each sensor device 2. More specifically, the control server 1 calculates a regression coefficient indicating the degree of correlation between the measured values by performing regression analysis on the measured values of each sensor device 2, and each sensor device 2 uses the regression coefficient to calculate each sensor device. 2 is calculated. As described above, the importance is an index value indicating the importance in estimating the measured value of another sensor device 2 from the measured value of each sensor device 2. As regression analysis, for example, LASSO regression can be used.

  Next, the control server 1 executes a threshold value setting process SQ3 in order to set a threshold value of importance used for selection of the operation sensor device 2. For example, the control server 1 selects the sensor devices 2 in descending order of importance, and estimates the measured values of the other sensor devices 2 from the measured values of the selected sensor devices 2. Here, the regression coefficient obtained by the regression analysis is used to estimate the measured value.

  The control server 1 determines the accuracy of the estimated value. When the estimated value satisfies the predetermined accuracy with respect to the measured value, the control server 1 uses the sum of the importance levels of all the sensor devices 2 selected so far as a threshold value. Set. Here, as a determination condition for the estimation accuracy, a preset accuracy K is used. More specifically, when the difference between the estimated value and the measured value is equal to or less than K, the control server 1 determines that the estimated value satisfies a predetermined accuracy with respect to the measured value.

  As described above, the control server 1 selects the sensor device 2 in order of importance until the estimated value satisfies the predetermined accuracy with respect to the measured value, and sets the sum of the importance of all the selected sensor devices 2 as a threshold value. To do. For this reason, the control server 1 can set a threshold value so that the estimated value satisfies a predetermined accuracy with respect to the measured value.

  Next, the control server 1 executes a sensor selection process SQ4 in order to select the operating sensor device 2. The control server 1 selects one or more sensor devices 2 as the active sensor devices 2 in descending order of the remaining amount of the battery so that the total importance is equal to or greater than the threshold value. For this reason, the control server 1 can select the sensor device 2 having a sufficient remaining battery level as the operation sensor device 2 while considering the accuracy of estimation of the measurement value according to the importance. The sensor selection process SQ4 is executed at a predetermined cycle, for example, because the remaining battery capacity of each sensor device 2 decreases with time.

  The method of the sensor selection process SQ4 is not limited to the above. For example, the control server 1 selects one or more sensor devices 2 as the active sensor device 2 in descending order of evaluation values calculated based on the importance level and the remaining battery level so that the sum of the importance levels is equal to or greater than the threshold value. Also good. Examples of the evaluation value include, but are not limited to, the product of the importance and the remaining battery level. Also in this case, the control server 1 can select the sensor device 2 having a sufficient remaining battery capacity as the operation sensor device 2 while considering the accuracy of estimation of the measurement value according to the importance.

  Next, the control server 1 executes the normal collection process SQ5 by the selected operation sensor device 2. At this time, the transmission request for collection is as described above. The normal collection process SQ5 is repeatedly executed a plurality of times at a period corresponding to “interval” in the transmission request, that is, the frequency H.

  The control server 1 collects measurement values only from the operation sensor device 2 and does not collect measurement values from the non-operation sensor device 2, unlike the whole collection process SQ 1 after selecting the operation sensor device 2. For example, when the sensor device (# 1) 2 is the non-operating sensor device 2 and the other sensor devices (# 2 to # 9) 2 are the operating sensor devices 2, the control server 1 includes the sensor devices (# 2 to # 2). # 9) Collect measurements from 2 only.

  For this reason, the non-operating sensor device 2 does not execute the measurement process, enters the sleep state, and can suppress power consumption. Note that the remaining battery capacity may be a collection target or a non-collection target in the normal collection process SQ5.

  Next, the control server 1 executes a measurement value estimation process SQ6. The control server 1 estimates the measurement value of the non-operation sensor device 2 from the measurement value of the operation sensor device 2. At this time, since the operation sensor device 2 is selected based on both the estimation accuracy and the remaining battery level, the estimation value satisfies a predetermined accuracy.

  However, for example, when the environmental conditions change, it is considered that the accuracy of the estimated value cannot be maintained. For this reason, the control server 1 periodically executes the monitoring collection process SQ7 and the accuracy monitoring process SQ8. Note that the frequency of the monitoring collection process SQ7 is performed at a frequency H 'lower than the frequency H of the overall collection process SQ1 and the normal collection process.

  The control server 1 collects measurement values only from the non-operating sensor device 2 by executing the monitoring collection process SQ7. In this example, since the non-operating sensor device 2 is the sensor device (# 1) 2, the control server 1 collects measurement values only from the sensor device (# 1) 2.

  Next, the control server 1 executes an accuracy monitoring process SQ8. The control server 1 compares the measured value collected in the monitoring collection process SQ7 with the estimated value, determines whether or not the estimated value satisfies a predetermined accuracy with respect to the measured value, and as indicated by a dotted arrow, Processing according to the determination result is performed.

  When the estimated value satisfies the predetermined accuracy with respect to the measured value, the control server 1 executes the normal collection process SQ5 again because the selection of the operation sensor device 2 is appropriate. Further, when the estimated value does not satisfy the predetermined accuracy with respect to the measured value, the control server 1 executes the threshold setting process SQ3 or the entire collection process SQ1.

  More specifically, when the control server 1 determines that the estimated value does not satisfy the predetermined accuracy in the first accuracy monitoring processing SQ8 after the regression analysis processing SQ2, the control server 1 changes the importance threshold value by the threshold setting processing SQ3. To do. At this time, the control server 1 selects the sensor device 2 by reversing the order of importance from the previous threshold setting.

  That is, in the threshold setting process SQ3, the control server 1 selects the sensor device 2 in ascending order of importance, and estimates the measured value of the other sensor device 2 from the measured value of the selected sensor device 2. The control server 1 determines the accuracy of the estimated value. When the estimated value satisfies the predetermined accuracy with respect to the measured value, the control server 1 uses the sum of the importance levels of all the sensor devices 2 selected so far as a threshold value. Set.

  Therefore, the control server 1 re-executes the sensor selection process SQ4 using a threshold value different from the threshold value in the first sensor selection process SQ4. Thereby, since the working sensor apparatus 2 is selected again, the estimation accuracy can be improved.

  If the control server 1 determines that the estimated value does not satisfy the predetermined accuracy in the second accuracy monitoring processing SQ8 after the regression analysis processing SQ2, the overall collection processing SQ1, the regression analysis processing SQ2, and the threshold setting processing SQ3. And sensor selection processing SQ4. That is, in this case, the control server 1 calculates the importance of each sensor device 2 again and reselects the operation sensor device 2. Thereby, since the working sensor apparatus 2 is selected again, the estimation accuracy can be improved.

  In this way, whenever the control server 1 determines that the estimated value does not satisfy the predetermined accuracy in the accuracy monitoring process SQ8, the control server 1 re-executes the regression analysis process SQ2 to change the selection of the operation sensor device 2. The selection of the operation sensor device 2 is changed alternately by re-executing the threshold value setting process SQ3 with the order of importance reversed. Next, the configuration of the sensor device 2 will be described.

  FIG. 3 is a configuration diagram illustrating an example of the sensor device 2. The sensor device 2 includes a CPU (Central Processing Unit) 20, a ROM (Read Only Memory) 21, a RAM (Random Access Memory) 22, a nonvolatile memory 23, a communication port 24, a sensor device 25, a battery 27, and a battery interface (battery). INF) portion 26. The CPU 20 is connected to a ROM 21, a RAM 22, a nonvolatile memory 23, a communication port 24, a sensor device 25, and a battery INF unit 26 via a bus 29 so that signals can be input and output.

  The battery 27 supplies power to the sensor device 2. The battery INF unit 26 includes an analog circuit connected to the battery 27 and detects the remaining amount of the battery 27. The sensor device 25 measures the measured value.

  The ROM 21 stores a program for driving the CPU 20. The program includes an OS (Operating System) and a sensor control program for executing the sensor control method of the embodiment.

  The RAM 22 functions as a working memory for the CPU 20. The communication port 24 is, for example, a PHY (Physical Layer) / MAC (Media Access Control) device, and transmits and receives packets to and from the control server 1. In addition, although an IP (Internet Protocol) packet is mentioned as a packet, it is not limited to this.

  When the CPU 20 reads a program from the ROM 21, a control unit 200, a state management unit 201, a measurement unit 202, a battery monitoring unit 203, and a communication processing unit 204 are formed as functions. The nonvolatile memory 23 stores transmission request X setting information 230. As the non-volatile memory 23, for example, an EPROM (Erasable Programmable ROM) is cited.

  The control unit 200 controls the overall operation of the sensor device 2. The state management unit 201, the measurement unit 202, the battery monitoring unit 203, and the communication processing unit 204 execute various processes according to the control of the control unit 200.

  The state management unit 201 manages the state of the sensor device 2. The state management unit 201 sets the state of the sensor device 2 to the sleep state when the transmission request X is not received from the control server 1 for a certain period, and operates the state of the sensor device 2 when the transmission request X is received in the sleep state. State. That is, the state management unit 201 switches the state of the sensor device 2 between the operation state and the sleep state in response to the transmission request X. Since power supply to most parts of the sensor device 2 is controlled to be stopped in the sleep state, the power consumption of the sensor device 2 is reduced as compared with the operation state.

  In the operating state, the measurement unit 202 measures the measurement value by the sensor device 25. More specifically, the measurement unit 202 performs measurement at the frequency H of the setting information 230, that is, the cycle indicated by “interval”, and outputs the measurement value to the communication processing unit 204 via the control unit 200. Further, since power supply to the sensor device 25 is stopped in the sleep state, the measurement unit 202 does not perform measurement.

  The battery monitoring unit 203 monitors the remaining amount of the battery 27 using the battery INF unit 26. The battery monitoring unit 203 acquires the remaining battery level from the battery INF unit 26 and outputs the battery remaining amount to the communication processing unit 204 via the control unit 200.

  The communication processing unit 204 processes communication via the communication port 24. The communication processing unit 204 receives the transmission request X from the control server 1 and transmits the sensor value measured by the measurement unit 202 and the remaining battery level acquired by the battery monitoring unit 203 to the control server 1.

  More specifically, when receiving the transmission request X via the communication port 24, the communication processing unit 204 determines whether “sensorID” included in the transmission request X matches the sensor ID of the own device. The communication processing unit 204 discards the transmission request X when “sensorID” does not match the sensor ID of its own device, and “interval” included in the transmission request X when “sensorID” matches the sensor ID of its own device. And “attribute” are stored in the nonvolatile memory 23 as setting information 230.

  The communication processing unit 204 notifies the control unit 200 of reception of the transmission request X addressed to its own device. The control unit 200 instructs the measurement unit 202 to perform measurement according to the setting information 230 and instructs the battery monitoring unit 203 to acquire the remaining battery level. More specifically, when “temperature” is included in “attribute”, the control unit 200 instructs the measurement unit 202 to perform measurement at a cycle indicated by “interval”. In addition, when “battery_charge” is included in “attribute”, the battery monitoring unit 203 is instructed to acquire the remaining battery capacity at a cycle indicated by “interval”.

  The communication processing unit 204 transmits the measurement value measured by the measurement unit 202 and the remaining battery level acquired by the battery monitoring unit 203 to the control server 1 via the communication port 24. Next, the configuration of the control server 1 will be described.

  FIG. 4 is a configuration diagram illustrating an example of the control server 1. The control server 1 includes a CPU 10, a ROM 11, a RAM 12, an HDD 13, a plurality of communication ports 14, an input device 15, and an output device 16. The CPU 10 is connected to a ROM 11, a RAM 12, an HDD 13, a plurality of communication ports 14, an input device 15, and an output device 16 via a bus 19 so that signals can be input and output with each other. The CPU 10 is an example of a computer.

  The ROM 11 stores a program for driving the CPU 10. The RAM 12 functions as a working memory for the CPU 10. The plurality of communication ports 14 are, for example, wireless LAN (Local Area Network) cards or NICs (Network Interface Cards), and transmit and receive packets between the GW apparatus 3 and the sensor apparatus 2, respectively.

  The input device 15 is a device that inputs information to the control server 1. Examples of the input device 15 include a keyboard, a mouse, and a touch panel. The input device 15 outputs the input information to the CPU 10 via the bus 19.

  The output device 16 is a device that outputs information of the control server 1. Examples of the output device 16 include a display, a touch panel, and a printer. The output device 16 acquires and outputs information from the CPU 10 via the bus 19.

  When the CPU 10 reads a program from the ROM 11, a control unit 100, a measurement processing unit 101, an importance calculation unit 102, a sensor selection unit 103, and an accuracy monitoring unit 104 are formed as functions. The HDD 13 stores setting information 130, a sensor information table 131, a measurement value table 132, a regression coefficient table 133, an importance level table 134, threshold information 135, and operating sensor information 136.

  The control unit 100 controls the overall operation of the control server 1. The control unit 100 instructs the measurement processing unit 101, the importance calculation unit 102, the sensor selection unit 103, and the accuracy monitoring unit 104 to operate according to the sequence shown in FIG.

  The measurement processing unit 101 is an example of a collection unit, and collects measurement values and battery remaining amounts from the plurality of sensor devices 2. More specifically, the measurement processing unit 101 transmits a transmission request X to each sensor device 2 via the communication port 14. At this time, the measurement processing unit 101 acquires the frequency H from the setting information 130 and generates a transmission request X including “interval” based on the frequency H.

  The measurement processing unit 101 receives the measurement value and the remaining battery level from each sensor device 2 via the communication port 14. The measurement processing unit 101 records the measurement value in the measurement value table 132 and records the remaining battery level in the sensor information table 131.

  FIG. 5 is a diagram illustrating an example of the sensor information table 131. The sensor information table 131 records a sensor ID, an IP address, and a remaining battery level (mAh). When transmitting the transmission request X, the measurement processing unit 101 refers to the IP address corresponding to the sensor ID of the transmission target sensor device 2. Further, when receiving the remaining battery level from the sensor device 2, the measurement processing unit 101 records the remaining battery level corresponding to the sensor ID of the receiving sensor device 2. In the following description, four sensors (# 1 to # 4) 2 will be described, but various functions described below do not depend on the number of sensor devices 2.

  FIG. 6 is a diagram illustrating an example of the measurement value table 132. In the measurement value table 132, measurement values for each time and sensor ID are recorded. When the measurement processing unit 101 receives the measurement value, the measurement processing unit 101 records the measurement value for each sensor ID together with the received time. The time is acquired from, for example, a network or a time circuit (not shown).

  The measurement processing unit 101 executes the overall collection process SQ1, the normal collection process SQ5, and the monitoring collection process SQ7 shown in FIG. The measurement processing unit 101 transmits the transmission request X to all the sensor devices 2 in the overall collection processing SQ1, but transmits the transmission request X only to the active sensor device 2 in the normal collection processing SQ5, and is inactive in the monitoring collection processing SQ7. A transmission request X is transmitted only to the sensor device 2. As described above, the overall collection process SQ1 and the normal collection process SQ5 are executed at the frequency H, while the monitoring collection process SQ7 is executed at the frequency H ′ smaller than the frequency H.

  After the sensor selection process SQ4, that is, after the active sensor device 2 is selected, the measurement processing unit 101 collects measurement values from the selected active sensor device 2 in the normal collection process SQ5, and the collected measurement values From the above, the measured value of the other non-operating sensor device 2 is estimated. For this reason, the power consumption of the non-operating sensor device 2 is reduced by omitting the measurement process.

  The measurement processing unit 101 uses the regression coefficient obtained by the regression analysis process SQ2 when estimating the measurement value.

  The importance calculation unit 102 is an example of a calculation unit, and calculates the importance for each sensor device 2 by performing regression analysis on the measurement values collected by the measurement processing unit 101. More specifically, the importance calculation unit 102 calculates the regression coefficient by executing the regression analysis process SQ2, and calculates the importance of each sensor device 2 based on the regression coefficient.

  FIG. 7 is a flowchart showing an example of the regression analysis process SQ2. The importance calculation unit 102 acquires the measurement value of each sensor device 2 from the measurement value table 132 (step St1). Next, the importance calculation unit 102 performs regression analysis on the measured values of the sensor devices 2 (step St2).

The above equation (1) represents an example of a multiple regression model of each measurement value used for regression analysis. Here, the variable y _k indicates an estimated value of the measured value of the sensor device (#k) 2, and the variable β _{k, i} indicates a regression coefficient when obtaining y _k . Further, the variable x _i represents the measured value of the sensor device (#i) 2, the variable n represents the number of all the sensor device 2, the variable ε indicates the error of the estimated value for the measured value.

The importance calculation unit 102 obtains a regression coefficient β _{k, i} that minimizes the error ε based on the equation (1). Here, as a method for calculating the regression coefficient β _{k, i} , for example, statistical analysis means such as R language can be cited, but there is no limitation.

  Next, the importance calculation unit 102 registers the regression coefficient in the regression coefficient table 133 (step St3). Next, the importance calculation unit 102 calculates the importance of each sensor device 2 from the regression coefficient (step St4) and registers it in the importance table 134 (step St5).

  FIG. 8 is a diagram illustrating an example of the regression coefficient table 133 and the importance level table 134. In the regression coefficient table 133, a regression coefficient is registered for each combination of the influence source sensor ID and the influence destination sensor ID. The influence source sensor ID and the influence destination sensor ID correspond to the sensor ID. The sensor device 2 of the influence source sensor ID affects the sensor device 2 of the influence destination sensor ID with respect to the measurement value, and the sensor device 2 of the influence destination sensor ID is affected by the measurement value from the sensor device 2 of the influence source sensor ID.

  The regression coefficient indicates the degree of influence in the above-mentioned relationship of exchange of influence. That is, the larger the absolute value of the regression coefficient, the greater the influence on the measured value.

y ₁ = 0.85x ₁ + 0.16x ₂ (2)

In the measured value estimation process SQ6, an estimated value is calculated using a regression coefficient. For example, when the measured value of the sensor device (# 1) 2 is estimated, the estimated value y ₁ is the regression coefficients of the influence source sensor IDs “# 1” and “# 2” with respect to the influence target sensor ID “# 1”. It is calculated from the above equation (2) using “0.85” and “0.16”. Here, the variable x ₁ is the last collected sensor device (# 1) 2 measurements, the variable x ₂ is a sensor device was last collected (# 2) 2 measurements. Since each regression coefficient of the influence source sensor ID “# 3” and “# 4” with respect to the influence destination sensor ID “# 1” is “0”, each measured value of the sensor device (# 3, # 4) 2 Does not affect the guess.

  In this way, since the regression coefficient represents the degree of influence on the estimated value, the importance calculation unit 102 calculates the number of regression coefficients having an absolute value greater than a certain value as the importance for each influence source sensor ID, and calculates the importance table. 134. For example, if the number of regression coefficients having an absolute value greater than 0 is the importance, the importance of the sensor device (# 1) 2 is “1” as indicated by the dotted line, and the sensor device (# 2) 2 The importance is 2. The importance level of the sensor device (# 3) 2 is “3”, and the importance level of the sensor device (# 4) 2 is 1. In this way, the regression analysis process SQ2 is executed.

  Referring to FIG. 4 again, the sensor selection unit 103 executes a threshold setting process SQ3 and a sensor selection process SQ4. The sensor selection unit 103 is an example of a selection unit. In the sensor selection process SQ4, the sensor selection unit 103 selects one or more operating sensor devices 2 in descending order of the remaining amount of the battery so that the total importance is equal to or greater than a threshold value. The sensor selection unit 103 stores the sensor ID of the selected operation sensor device 2 in the HDD 13 as operation sensor information 136. In the normal collection process SQ5, the measurement processing unit 101 acquires the sensor ID from the operation sensor information 136 to identify the operation sensor device 2 to be collected. Unlike the above, the sensor selection unit 103 selects one or more operation sensor devices 2 in descending order of evaluation values calculated based on the importance and the remaining battery level so that the total importance is equal to or greater than the threshold. May be.

  Further, in the threshold setting process SQ3, the sensor selection unit 103 selects the sensor device 2 from the plurality of sensor devices 2 in descending order of importance, and measures other sensor devices 2 from the measured values of the selected sensor device 2. Guess the value. Then, when the estimated value satisfies a predetermined accuracy with respect to the measurement value, the sensor selection unit 103 sets the total importance of all the selected sensor devices 2 as the threshold information 135 as a threshold value.

  FIG. 9 is a flowchart showing an example of the threshold setting process SQ3. The sensor selection unit 103 sets 0 to the variable J (step St11). Next, the sensor selection unit 103 determines whether or not the flag F is 0 (step St12). The flag F is controlled to alternately become 1 or 0 by the control unit 100 every time the accuracy monitoring process SQ8 is executed.

  When the flag F is 0 (Yes in Step St12), the sensor selection unit 103 rearranges the sensor IDs in the importance table 134 in descending order of importance (Step St13), and when the flag F is 1 (Step St12). No), the sensor IDs in the importance table 134 are rearranged in ascending order of importance (step St14). Taking the importance table 134 of FIG. 8 as an example, when the flag F = 0, the sensor selection unit 103 sets the sensor ID of the importance table 134 to “# 3”, “# 2”, “# 1”, “ Rearrange in the order of “# 4”. When there are a plurality of sensor IDs having the same importance, the sensor selection unit 103 rearranges the sensor IDs in ascending order, for example.

  Next, the sensor selection unit 103 selects the sensor device 2 with the highest sensor ID from the rearranged importance table 134 (step St15). That is, when the sensor selection unit 103 performs rearrangement in descending order of importance, the sensor selecting unit 2 selects the sensor device 2 having the maximum importance, and when rearranging in ascending order of importance, the sensor having the minimum importance. Device 2 is selected.

  Next, the sensor selection unit 103 adds the importance of the selected sensor device 2 to the variable J (step St16). In the case of the above example, the sensor selection unit 103 selects the sensor device (# 3) 2 having the highest importance, and adds the importance “3” to the variable J, so the variable J becomes 3.

  Next, the sensor selection unit 103 performs regression analysis on the measured value of the selected sensor device 2 (step St17), and estimates the measured value of the other sensor device 2 based on the regression coefficient obtained by the regression analysis (step St18). . The regression analysis and the calculation of the estimated value are as described above.

  Next, the sensor selection unit 103 calculates the maximum error of the estimated value with respect to the measurement value of each sensor device 2 (step St19). More specifically, the sensor selection unit 103 calculates the error between the measured value and the estimated value for each sensor device 2 other than the selected sensor device 2, and acquires the maximum value of the error. At this time, the sensor selection unit 103 may acquire an average value of errors instead of the maximum value of errors.

  Next, the sensor selection unit 103 compares the maximum error with the accuracy K (step St20). The accuracy K is obtained from the setting information 130 and is used as a reference value for the error of the estimated value. As described above, the sensor selection unit 103 determines whether or not the estimated value satisfies the predetermined accuracy K with respect to the measurement value.

  Next, when the maximum error ≦ K is satisfied (Yes in Step St20), the sensor selection unit 103 sets a variable J as a threshold (Step St22). In addition, when the maximum error> K is satisfied (No in Step St20), the sensor selection unit 103 determines whether all the sensor devices 2 have been selected (Step St21). If all the sensor devices 2 have been selected (Yes in step St21), the sensor selection unit 103 executes the process in step St22.

  In addition, when there is an unselected sensor device 2 (No in Step St21), the sensor selection unit 103 selects the highest sensor device 2 in the importance level table 134 among the unselected sensor devices 2 (Step S21). St15), the processing after step St16 is executed again. For this reason, the sensor selection unit 103 adds the importance of the selected sensor device 2 as a threshold value until it determines that the estimated value of the selected sensor device 2 satisfies the predetermined accuracy K with respect to the measured value.

  FIG. 10 is a diagram illustrating an example of the regression coefficient obtained by the process of step St17. In this example, it is assumed that the sensor selection unit 103 has selected the sensor device (# 3) 2.

  In the process of step St18, the sensor selection unit 103 determines the regression coefficient “1.02” for each of the affected sensor IDs “# 1”, “# 2”, and “# 4” corresponding to the affected sensor ID “# 3”. ”,“ 1.03 ”, and“ 1.05 ”are used to calculate the estimated values of the other sensor devices (# 1, # 2, # 4) 2. For example, the sensor selection unit 103 sets each of the measurement value “20.0” of the sensor ID “# 3” at the time “12:01:00” to the time “12:02:00” in the measurement value table 132 of FIG. The estimated value of the sensor device (# 1, # 2, # 4) 2 is calculated.

  In step St19, the sensor selection unit 103 calculates an error with respect to the measured value of the estimated value of each sensor device (# 1, # 2, # 4) 2. The estimated values and errors of each sensor device (# 1, # 2, # 4) 2 are shown below.

  The estimated value of the sensor device (# 1) 2 is 20.4 (= 20 × 1.02), and the error is 0.005 (= | 20.5-20.4 | /20.5). The estimated value of the sensor device (# 2) 2 is 20.6 (= 20 × 1.03), and the error is 0.005 (= | 20.5-20.6 | /20.5). The estimated value of the sensor device (# 4) 2 is 21 (= 20 × 1.05), and the error is 0 (= | 21.0-21.0 | /21.0).

  Therefore, the sensor selection unit 103 calculates 0.005 as the maximum error. When the accuracy K is set to 0.05, the sensor selection unit 103 determines that the maximum error ≦ K (0.005 <0.5) in step St20, and therefore, in step St22, the sensor device (# 3) 2 has a threshold value. Set importance "3".

  Thus, the control server 1 selects the sensor device 2 in order of importance until the estimated value satisfies the predetermined accuracy K with respect to the measurement value, and sets the sum of the importance of all the selected sensor devices 2 as a threshold value. Set. For this reason, the control server 1 can set the threshold value so that the estimated value satisfies the predetermined accuracy K with respect to the measured value. The sensor selection unit 103 executes a sensor selection process SQ4 after the threshold setting process SQ3.

  FIG. 11 is a flowchart showing an example of the sensor selection process SQ4. The sensor selection unit 103 sets the variable M to 0 (step St31). Next, the sensor selection unit 103 rearranges the sensor devices 2 in the sensor information table 131 in descending order of the remaining battery capacity (step St32). Next, the sensor selection unit 103 selects the sensor device 2 with the largest remaining battery level from the sensor information table 131 (step St33).

  Next, the sensor selection unit 103 adds the importance of the selected sensor device 2 to the variable M (Step St34). Next, the sensor selection unit 103 acquires a threshold value from the threshold value information 135 and compares it with the variable M (step St35).

  If M <threshold value is satisfied (No in step St35), the sensor selection unit 103 determines whether all the sensor devices 2 have been selected (step St36). When there is an unselected sensor device 2 (No in Step St36), the sensor selection unit 103 selects the sensor device 2 with the largest remaining battery level from among the unselected sensor devices 2 (Step St33). Steps St34 and St35 are executed again.

  For this reason, the sensor selection part 103 selects the sensor apparatus 2 in order with many remaining battery levels until the variable M becomes more than a threshold value. At this time, the variable M is the total importance of all the selected sensor devices 2.

  When M ≧ threshold value is satisfied (Yes in step St35), or when all the sensor devices 2 have been selected (Yes in step St36), the sensor selection unit 103 operates the selected one or more sensor devices 2 as operating sensors. The operation sensor information 136 is set as the device 2 (step St37). Next, the sensor selection unit 103 determines whether or not the number of operating sensor devices 2 is two or more (step St38). If the number of operating sensor devices 2 is one (No in step St38), the sensor selecting unit 103 determines the selection of the operating sensor device 2 and ends the process.

  In addition, when the number of operating sensor devices 2 is two or more (Yes in Step St38), the sensor selection unit 103 sets the operating sensor devices 2 set in Step St37 in the ascending order of the remaining battery capacity in the sensor information table 131. Rearrange (step St39). In this case, the sensor selection unit 103 reduces the number of selections of the active sensor device 2 so as to reduce the difference between the total importance and the threshold by executing the following process.

  Next, the sensor selection unit 103 selects the sensor device 2 with the minimum remaining battery level from the sensor information table 131 (step St40). Next, the sensor selection unit 103 subtracts the importance of the selected sensor device 2 from the variable M (step St41).

  Next, the sensor selection unit 103 acquires a threshold value from the threshold value information 135 and compares it with the variable M (step St42). If M ≧ threshold value is satisfied (Yes in step St42), the sensor selection unit 103 excludes the selected sensor device 2 from the operation sensor device 2 (step St43). That is, the sensor selection unit 103 deletes the sensor ID of the selected sensor device 2 from the operating sensor information 136. Moreover, the sensor selection part 103 does not perform the process of step St43, when M <threshold value is materialized (No of step St42).

  Next, the sensor selection unit 103 determines whether or not all the sensor devices 2 set in the operation sensor device 2 in Step St37 have been selected (Step St44). The sensor selection part 103 complete | finishes a process, when all the sensor apparatuses 2 have been selected (Yes of step St44). In addition, when there is an unselected sensor device 2 (No in Step St44), the sensor selection unit 103 selects the sensor device 2 with the minimum remaining battery level from among the unselected sensor devices 2 (Step St40). ), The process after step St41 is executed again.

  Thereby, the sensor selection part 103 can reduce the number of the operation | movement sensor apparatuses 2 in the range with which the sum total of importance ≥ threshold value is satisfied. As described above, the sensor selection unit 103 further reduces the power consumption of the sensor device 2 by optimizing the number of the active sensor devices 2.

  Next, an example of the sensor selection process SQ4 will be described. In this example, it is assumed that the initial remaining battery level of each sensor device (# 1 to # 4) 2 is 100 (mAh), and the remaining battery level is reduced by 1 (mAh) by one measurement process. Further, it is assumed that the sensor selection process SQ4 is executed every ten normal collection processes SQ5. The importance of each sensor device 2 is as shown in the example of FIG.

  In the first sensor selection process SQ4, the sensor selection unit 103 rearranges the sensor information table 131 in order of sensor ID (# 1 to # 4) in step St32. In addition, the sensor selection part 103 is rearranged in order of sensor ID, when a battery remaining charge is the same between the sensor apparatuses 2. FIG.

  The sensor selection unit 103 selects the sensor device (# 1) 2 in step St33, and adds the importance “1” of the sensor device (# 1) 2 to the variable M in step St34. Since the sensor selection unit 103 determines that M = 1 <3 in Step St35, the sensor selection unit 103 selects the sensor device (# 2) 2 in Step St33, and in Step St34, sets the variable M to the importance “ Add 2 ”.

  As a result, the sensor selection unit 103 determines that M = 3 ≧ 3 in Step St35, and therefore sets the sensor device (# 1, # 2) 2 as the operation sensor device 2 in Step St37. The sensor selection unit 103 determines that the number of the active sensor devices 2 is two or more in step St38, and rearranges the sensor IDs in the sensor information table 131 in the order of sensor IDs (# 1, # 2) in step St39.

  The sensor selection unit 103 selects the sensor device (# 1) 2 in step St40, and subtracts the importance “1” of the sensor device (# 1) 2 from the variable M (= 3) in step St41. As a result, the sensor selection unit 103 determines that M = 3-1 = 2 <3 in Step St42. The sensor selection unit 103 determines that there is an unselected sensor device 2 in step St44, and executes the process of step St40 again.

  Next, the sensor selection unit 103 selects the sensor device (# 2) 2 in step St40, and subtracts the importance “2” of the sensor device (# 2) 2 from the variable M (= 3) in step St41. As a result, the sensor selection unit 103 determines that M = 3−2 = 1 <3 in Step St42.

  Therefore, the sensor device (# 1) 2 and the sensor device (# 2) 2 are finally selected as the operation sensor device 2. The other sensor devices (# 3, # 4) 2 enter the sleep state as the non-operating sensor device 2.

  It is assumed that the second sensor selection process SQ4 is executed after the ten times of the normal collection process SQ5. At this time, each battery remaining amount of the sensor device (# 1) 2 and the sensor device (# 2) 2 is 90 (mAh), and each battery remaining amount of the sensor device (# 3) 2 and the sensor device (# 4) 2 is left. The amount is 100 (mAh).

  In the second sensor selection process SQ4, in step St32, the sensor selection unit 103 rearranges the sensor information table 131 in descending order of remaining battery power, that is, sensor IDs in the order of # 3, # 4, # 1, and # 2. The sensor selection unit 103 selects the sensor device (# 3) 2 in step St33, and adds the importance “3” of the sensor device (# 3) 2 to the variable M in step St34.

  The sensor selection unit 103 sets the sensor device (# 3) 2 as the active sensor device 2 in step St37 in order to determine that M = 3 ≧ 3 in step St35. In step St38, the sensor selection unit 103 determines that the number of active sensor devices 2 is one, and ends the process.

  Therefore, the sensor device (# 3) 2 is finally selected as the operation sensor device 2. The other sensor devices (# 1, # 2, # 4) 2 enter the sleep state as the non-operating sensor device 2.

  It is assumed that the third sensor selection process SQ4 is executed after the ten times of the normal collection process SQ5. At this time, the remaining battery levels of the sensor device (# 1) 2, the sensor device (# 2) 2, and the sensor device (# 3) 2 are 90 (mAh). The amount is 100 (mAh).

  In the third sensor selection process SQ4, in step St32, the sensor selection unit 103 rearranges the sensor information table 131 in descending order of the remaining battery power, that is, sensor IDs in the order of # 4, # 1, # 2, and # 3. The sensor selection unit 103 selects the sensor device (# 3) 2 in step St33, and adds the importance “1” of the sensor device (# 4) 2 to the variable M in step St34.

  Since the sensor selection unit 103 determines that M = 1 <3 in Step St35, the sensor selection unit 103 selects the sensor device (# 1) 2 in Step St33, and in Step St34, sets the variable M to the importance “ Add 1 ”. Since the sensor selection unit 103 determines that M = 1 + 1 = 2 <3 in Step St35, the sensor selection unit 103 selects the sensor device (# 2) 2 in Step St33, and in Step St34, the sensor device (# 2) 2 is assigned to the variable M. Add degree “2”.

  As a result, in order to determine that M = 2 + 2 = 4 ≧ 3 in Step St35, the sensor selection unit 103 sets the sensor device (# 1, # 2, # 4) 2 as the operation sensor device 2 in Step St37. The sensor selection unit 103 determines that the number of the active sensor devices 2 is two or more in Step St38, and arranges the sensor IDs in the sensor information table 131 in ascending order of the remaining battery capacity (# 1, # 2, # 4) in Step St39. Change.

  The sensor selection unit 103 selects the sensor device (# 1) 2 in step St40, and subtracts the importance “1” of the sensor device (# 1) 2 from the variable M (= 4) in step St41. As a result, the sensor selection unit 103 determines that M = 4-1 = 3 ≧ 3 in Step St42, and excludes the currently selected sensor device (# 1) 2 from the active sensor device 2 in Step St43.

  As a result, the sensor device (# 2) 2 and the sensor device (# 4) 2 are finally selected as the operation sensor device 2. The other sensor devices (# 1, # 3) 2 enter the sleep state as the non-operating sensor device 2.

  As described above, the sensor selection unit 103 selects one or more sensor devices 2 as the active sensor devices 2 in descending order of the remaining amount of the battery so that the total importance is equal to or greater than the threshold value. For this reason, the control server 1 can select the sensor device 2 having a sufficient remaining battery level as the operation sensor device 2 while considering the accuracy of estimation of the measurement value according to the importance.

  Referring to FIG. 4 again, the accuracy monitoring unit 104 is an example of a determination unit. By executing the accuracy monitoring process SQ8, the estimated value estimated by the measurement processing unit 101 has a predetermined accuracy K with respect to the measured value. It is determined whether or not it is satisfied. Thereby, the accuracy of the measured value of the non-operating sensor device 2 estimated from the measured value of the operating sensor device 2, that is, the estimated value is determined. The accuracy monitoring unit 104 notifies the control unit 100 of the determination result.

  The control unit 100 instructs the measurement processing unit 101, the importance level calculation unit 102, and the sensor selection unit 103 to operate according to the determination result of the accuracy monitoring unit 104. More specifically, the control unit 100 gives an instruction according to the value of the flag F when the accuracy monitoring unit 104 determines that the estimated value does not satisfy the predetermined accuracy K.

  The control unit 100 initializes the flag F to 0 when the control server 1 is started, and then sets the flag F to 0 and 1 each time the accuracy monitoring unit 104 determines that the estimated value does not satisfy the predetermined accuracy K. Switch alternately. The measurement processing unit 101, the importance calculation unit 102, and the sensor selection unit 103 operate according to the flag F.

  FIG. 12 is a flowchart showing an operation example during execution of the accuracy monitoring process SQ8. The accuracy monitoring unit 104 refers to the sensor information table 131 and the operating sensor information 136, and selects the sensor device 2 with the largest remaining battery capacity from the non-operating sensors 2 (step St51).

  Next, the accuracy monitoring unit 104 acquires the measured value and the estimated value of the selected sensor device 2 (step St52). More specifically, the accuracy monitoring unit 104 acquires a measurement value from the measurement value table 132 and acquires an estimated value from the measurement processing unit 101.

  Next, the accuracy monitoring unit 104 calculates an error of the estimated value with respect to the measured value (step St53). Next, the accuracy monitoring unit 104 compares the error with the accuracy K (step St54). If the error is equal to or lower than the accuracy K (Yes in step St54), the accuracy monitoring unit 104 notifies the control unit 100 that the estimated value satisfies the predetermined accuracy K as a determination result (step St55), and ends the process. To do. In this case, the control unit 100 causes the measurement processing unit 101 to continue the normal collection processing SQ5 without changing the setting.

  If the error is larger than the accuracy (No in step St54), the accuracy monitoring unit 104 notifies the control unit 100 that the estimated value does not satisfy the predetermined accuracy K as a determination result (step St56). Next, the control unit 100 determines the value of the flag F (step St57). When the flag F = 0 (No in Step St57), the control unit 100 causes the sensor selection unit 103 to execute the following Steps St63 to St65.

  When the flag F = 0 (No in Step St57), the sensor selection unit 103 executes the threshold setting process SQ3 shown in FIG. 9 (Step St63). In the threshold setting process SQ3, since the flag F = 0, the sensor selection unit 103 rearranges the sensor IDs in the importance level table 134 in descending order of importance (step St13). Next, the sensor selection unit 103 executes a sensor selection process SQ4 shown in FIG. 11 (step St64). Next, the control unit 100 sets the flag F = 1 (step St65) and ends the process.

  In addition, when the flag F = 1 (Yes in Step St57), the control unit 100 causes the measurement processing unit 101, the importance calculation unit 102, and the sensor selection unit 103 to execute the processes of Steps St58 to St62 below.

  When the flag F = 1 (Yes in step St57), the measurement processing unit 101 executes the entire collection process SQ1 (step St58). Next, the importance degree calculation unit 102 executes the regression analysis process SQ2 shown in FIG. 7 (step St59).

  The sensor selection unit 103 executes the threshold setting process SQ3 shown in FIG. 9 (step St60). In the threshold setting process SQ3, since the flag F = 1, the sensor selection unit 103 rearranges the sensor IDs in the importance table 134 in ascending order of importance (step St14). Next, the sensor selection unit 103 executes a sensor selection process SQ4 shown in FIG. 11 (step St61). Next, the control unit 100 sets the flag F = 0 (step St62) and ends the process.

  As described above, the control unit 100 alternately switches the flag F between 1 and 0 every time the notification of the determination result that the estimated value does not satisfy the predetermined accuracy K is received. Therefore, the sensor ID selection order in the threshold setting process SQ3 is alternately switched between the descending order and the ascending order.

  That is, when the accuracy monitoring unit 104 determines that the estimated value does not satisfy the predetermined accuracy K, the sensor selection unit 103 changes the threshold value, and in the change, the order of importance is the same as when the previous threshold value was set. The sensor device 2 is selected by reversing the above. Thereby, since the working sensor apparatus 2 is selected again by a new threshold value, the estimation accuracy can be improved.

  In addition, when the flag F = 1, the importance is changed by the regression analysis process SQ2, and the threshold is reset, so that the estimation accuracy can also be improved.

  Next, an example of operation at the time of executing the accuracy monitoring process SQ8 will be described. In this example, a state immediately after the above-described three sensor selection processes SQ4 is assumed.

  In the process of step St51, the accuracy monitoring unit 104 selects the sensor device 2 with the largest remaining battery level from the non-operating sensor devices (# 1, # 3) 2. However, since the remaining battery levels of the sensor devices (# 1, # 3) 2 are the same, the accuracy monitoring unit 104 selects the sensor device (# 1) 2 having a small sensor ID. Next, the accuracy monitoring unit 104 calculates an estimated value from the regression coefficient, and calculates an error with respect to the measured value.

  FIG. 13 is a diagram illustrating an example of measured values and regression coefficients used for accuracy determination. In step St51, the accuracy monitoring unit 104 acquires the measurement value “18.0” of the sensor device (# 1) 2 at time t [n + 1]. In addition, the accuracy monitoring unit 104 affects the measured values “20.0” and “21.0” of the sensor device (# 1, # 4) 2 at time t [n] and the affected sensor ID “# 1”. The measured value of the sensor device (# 1) 2 is estimated from the regression coefficient “0.99” of the original sensor ID “# 2” and the regression coefficient “0.00” of the influence source sensor ID “# 4”.

  As a result, the accuracy monitoring unit 104 acquires 19.8 (= 0.99 × 20.0 + 0.00 × 21.0) as the estimated value.

  Next, in step St53, the accuracy monitoring unit 104 calculates an error (error rate) of the estimated value “19.8” with respect to the measured value “18.0” of the sensor device (# 1) 2. The error is 0.1 (= | 18.0-19.8 | /18.0).

  If the accuracy K = 0.05, the accuracy monitoring unit 104 determines that 0.1> 0.05 in Step St54 (No in Step St54). Further, since the flag F = 0 (No in step St57), the sensor selection unit 103 executes the threshold setting process SQ3 in step St63. In step St13 of the threshold setting process SQ3, the sensor selection unit 103 rearranges the sensor IDs in descending order of importance (# 1, # 4, # 2, # 3). As a result, the threshold value is set to 4.

  Next, the accuracy monitoring unit 104 executes the sensor selection process SQ4 in step St64. As a result, the sensor device (# 1, # 2, # 4) 2 is selected as the operation sensor device 2. Next, the control unit 100 sets the flag F = 1 in step St65.

  Next, the accuracy monitoring unit 104 executes the second accuracy monitoring process SQ8. The accuracy monitoring unit 104 selects the sensor device (# 3) 2 that is the non-operating sensor device 2 in the process of step St51. Next, the accuracy monitoring unit 104 calculates an estimated value from the regression coefficient, and calculates an error with respect to the measured value.

  FIG. 14 is a diagram illustrating another example of measurement values and regression coefficients used for accuracy determination. In step St51, the accuracy monitoring unit 104 acquires each measurement value “19.0” of the sensor device (# 3) 2 at time t [m + 1]. In addition, the accuracy monitoring unit 104 has the measured values “18.0”, “21.0”, “21.0” of the sensor device (# 1, # 2, # 4) 2 at time t [m], Regression coefficients “0.00”, “1.01”, “−0.05” of the influence source sensor IDs “# 1”, “# 2”, “# 4” with respect to the influence destination sensor ID “# 3” The measured value of the sensor device (# 3) 2 is estimated from the above.

  As a result, the accuracy monitoring unit 104 acquires 20.16 (= 0.00 × 18.0 + 1.01 × 21.0−0.05 × 21.0) as an estimated value.

  Next, in step St53, the accuracy monitoring unit 104 calculates an error (error rate) of the estimated value “18.0” with respect to the measured value “19.0” of the sensor device (# 1) 2. The error is 0.061 (= | 19.0-20.16 | /19.0).

  If the accuracy K = 0.05, the accuracy monitoring unit 104 determines that 0.061> 0.05 in Step St54 (No in Step St54). Further, since the flag F = 1 (Yes in Step St57), the measurement processing unit 101 executes the entire collection process SQ1 in Step St58. Next, the importance degree calculation unit 102 executes the regression analysis process SQ2 in step St59.

  Next, the sensor selection unit 103 executes a threshold setting process SQ3 in step St60. In step St13 of the threshold setting process SQ3, the sensor selection unit 103 rearranges the sensor IDs in ascending order of importance. Next, the accuracy monitoring unit 104 executes the sensor selection process SQ4 in step St61. Next, the control unit 100 sets the flag F = 0 in step St62.

  As described above, the control server 1 switches the selection method of the operation sensor device 2 by alternately switching the flag F between 1 and 0 for each determination result that the estimated value does not satisfy the predetermined accuracy K. . Thereby, since the working sensor apparatus 2 is newly selected, the measurement accuracy can be improved.

  According to the sensor control method described above, it is possible to extend the operation time by saving the remaining battery power of each sensor device 2 while maintaining the accuracy of the predetermined measurement. An example will be described below.

  FIG. 15 is a diagram illustrating an example of an operation time of the sensor system of the comparative example and the example. In FIG. 15, the horizontal axis indicates time, and times t0 to t15 are measurement processing times at regular time intervals Tu.

  Moreover, the initial value of the battery remaining amount of each sensor device (# 1 to # 4) 2 is 10 (mAh), and the battery remaining amount is consumed by 1 (mAh) for each measurement at each time t0 to t15. Assume that Therefore, each sensor device (# 1 to # 4) 2 can perform 10 measurements within the period from time t0 to time t15. At this time, it is assumed that the battery is not charged.

  The importance levels of the sensor device (# 1) 2 and the sensor device (# 2) 2 are 1, and the importance levels of the sensor device (# 3) 2 and the sensor device (# 4) 2 are 0.3. Assume that there is.

  The operation sensor device 2 is selected every five measurements. That is, the operation sensor device 2 is selected at times t5, t10, and t15. The operation sensor device 2 is selected such that the sum of the importance levels satisfies the threshold value “1.5” (in the case of the embodiment, this corresponds to step St35 in FIG. 11).

  In the comparative example, the remaining battery level of the sensor device 2 is not taken into consideration, and the operation sensor device 2 is selected based only on its importance. The total importance of the sensor devices (# 1, # 2) 2 is 2 (= 1 + 1), which exceeds the threshold (1.5). For this reason, the sensor device (# 1, # 2) 2 is selected as the operation sensor device 2 in the period of time t0 to t10.

  At time t10, the remaining battery levels of the sensor devices (# 1, # 2) 2 become 0 (mAh). However, since the total importance of the other sensor devices (# 3, # 4) 2 is 0.6 (= 0.3 + 0.3), which is below the threshold (1.5), the sensor device (# 3 , # 4) 2 is not selected as the operation sensor device 2.

  Therefore, the operation time in the comparative example is only the period from time t0 to time t10.

  In contrast, in the embodiment, as described above, the operation sensor device 2 is selected based on both the remaining battery level and the importance level of the sensor device 2. For example, the sensor device (# 1, # 2) 2 is selected as the operating sensor device 2 during the period of time t0 to t5, but the sensor device (# 1, # 3, # 4) is selected during the period of time t5 to t10. 2 is selected as the operating sensor device 2. At this time, the total importance of the sensor devices (# 1, # 3, # 4) 2 is 1.6 (= 1 + 0.3 + 0.3), which exceeds the threshold (1.5).

  Further, the sensor device (# 2, # 3, # 4) 2 is selected as the operation sensor device 2 during the period of time t10 to t15. At this time, the total importance of the sensor devices (# 2, # 3, # 4) 2 is 1.6 (= 1 + 0.3 + 0.3), which exceeds the threshold (1.5). As for the sensor device (# 1, # 2) 2, when the sensor device (# 2) 2 is selected during the period from time t5 to t10, the sensor device (# 1) 2 is selected during the period from time t10 to t15. May be.

  Therefore, the operating time in the example is a period from time t0 to t15, which is 1.5 times the operating time in the comparative example. For this reason, the number of measurements (15 times) in the example is 1.5 times the number of measurements (10 times) in the comparative example.

  As described above, the control server 1 can select the sensor device 2 having a sufficient remaining battery level as the operation sensor device 2 while considering the accuracy of estimation of the measurement value according to the importance. Therefore, according to the present embodiment, it is possible to save the power of the sensor device 2 without reducing the measurement accuracy.

  As described above, the sensor selection process SQ4 is not limited to the method shown in FIG. The sensor selection unit 103 may select one or more operation sensor devices 2 in descending order of evaluation values calculated based on the importance level and the remaining battery level instead of the remaining battery level.

  FIG. 16 is a flowchart showing another example of the sensor selection process SQ4. In FIG. 16, processes that are the same as those in FIG.

  The sensor selection unit 103 sets the variable M to 0 (step St31). Next, the sensor selection unit 103 calculates the evaluation value of each sensor device 2 based on the importance and the remaining battery level (step St32a). Examples of the evaluation value include, but are not limited to, the product of the importance and the remaining battery level.

  Next, the sensor selection unit 103 rearranges the sensor devices 2 in the sensor information table 131 in descending order of evaluation values (step St32b). Next, the sensor selection unit 103 selects the sensor device 2 with the highest remaining battery level from the sensor information table 131 (step St33a). Thereafter, the processing described above is executed.

  Also in this example, the control server 1 can select the sensor device 2 having a sufficient remaining battery capacity as the operation sensor device 2 while considering the accuracy of estimation of the measurement value according to the importance. Therefore, the effect described with reference to FIG. 15 is obtained.

  The above processing functions can be realized by a computer. In that case, a program describing the processing contents of the functions that the processing apparatus should have is provided. By executing the program on a computer, the above processing functions are realized on the computer. The program describing the processing contents can be recorded on a computer-readable recording medium (except for a carrier wave).

  When the program is distributed, for example, it is sold in the form of a portable recording medium such as a DVD (Digital Versatile Disc) or a CD-ROM (Compact Disc Read Only Memory) on which the program is recorded. It is also possible to store the program in a storage device of a server computer and transfer the program from the server computer to another computer via a network.

  The computer that executes the program stores, for example, the program recorded on the portable recording medium or the program transferred from the server computer in its own storage device. Then, the computer reads the program from its own storage device and executes processing according to the program. The computer can also read the program directly from the portable recording medium and execute processing according to the program. Further, each time the program is transferred from the server computer, the computer can sequentially execute processing according to the received program.

  The above-described embodiment is an example of a preferred embodiment of the present invention. However, the present invention is not limited to this, and various modifications can be made without departing from the scope of the present invention.

In addition, the following additional notes are disclosed regarding the above description.
(Additional remark 1) In the sensor control apparatus which controls the several sensor apparatus provided with the battery,
A collection unit that collects measurement values and the remaining amount of the battery from the plurality of sensor devices, and
By calculating regression of the measured value, a calculation unit that calculates an index value indicating the importance in estimating the measured value of another sensor device from the measured value for each of the plurality of sensor devices;
In order that the sum of the index values is equal to or greater than a threshold value, one or more of the plurality of sensor devices is in descending order of the remaining amount of the battery or in descending order of evaluation values calculated based on the index value and the remaining amount of the battery. A selection unit for selecting a sensor device;
The collection unit collects the measurement values from the one or more selected sensor devices among the plurality of sensor devices after the selection unit selects one or more sensor devices, and the collected measurement values A sensor control device which estimates a measured value of another sensor device from the above.
(Supplementary Note 2) The selection unit selects a sensor device from the plurality of sensor devices in the order of the index value, estimates a measurement value of another sensor device from the measurement value of the selected sensor device, and performs the estimation. The sensor control device according to appendix 1, wherein when the value satisfies a predetermined accuracy with respect to the measurement value, the sum of the index values of all the selected sensor devices is set as the threshold value.
(Additional remark 3) It has the determination part which determines whether the estimated value estimated by the said collection part satisfy | fills predetermined precision with respect to the said measured value,
The selection unit includes:
When the determination unit determines that the estimated value does not satisfy the predetermined accuracy, the threshold value is changed,
3. The sensor control device according to appendix 2, wherein in the change, the sensor device is selected by reversing the order of the index values from the previous setting of the threshold value.
(Supplementary Note 4) The sensor control device according to any one of Supplementary Notes 1 to 3, wherein the selection unit reduces the number of sensor devices selected so that a difference between a sum of the index values and the threshold value is reduced. .
(Additional remark 5) The some sensor apparatus provided with the battery,
A sensor control device that controls the plurality of sensor devices;
The sensor control device includes:
A collection unit that collects measurement values and the remaining amount of the battery from the plurality of sensor devices, and
By calculating regression of the measured value, a calculation unit that calculates an index value indicating the importance in estimating the measured value of another sensor device from the measured value for each of the plurality of sensor devices;
In order that the sum of the index values is equal to or greater than a threshold value, one or more of the plurality of sensor devices is in descending order of the remaining amount of the battery or in descending order of evaluation values calculated based on the index value and the remaining amount of the battery. A selection unit for selecting a sensor device;
The collection unit collects the measurement values from the one or more selected sensor devices among the plurality of sensor devices after the selection unit selects one or more sensor devices, and the collected measurement values A sensor system characterized by estimating a measured value of another sensor device from the above.
(Additional remark 6) The said selection part selects a sensor apparatus from the said several sensor apparatus in order of the said index value, estimates the measured value of another sensor apparatus from the said measured value of this selected sensor apparatus, and this estimation The sensor system according to appendix 5, wherein when the value satisfies a predetermined accuracy with respect to the measured value, the sum of the index values of all the selected sensor devices is set as the threshold value.
(Additional remark 7) It has the determination part which determines whether the estimated value estimated by the said collection part satisfy | fills predetermined precision with respect to the said measured value,
The selection unit includes:
When the determination unit determines that the estimated value does not satisfy the predetermined accuracy, the threshold value is changed,
The sensor system according to appendix 6, wherein in the change, the sensor device is selected by reversing the order of the index values from the previous threshold value setting time.
(Supplementary note 8) The sensor system according to any one of supplementary notes 5 to 7, wherein the selection unit reduces the number of sensor devices selected so that a difference between a sum of the index values and the threshold value is reduced.
(Supplementary Note 9) In a sensor control method for controlling a plurality of sensor devices including a battery,
Collecting each of the measurement value and the remaining amount of the battery from the plurality of sensor devices;
Calculating an index value indicating the importance in estimating the measured value of another sensor device from the measured value for each of the plurality of sensor devices by performing regression analysis of the measured value;
In order that the sum of the index values is equal to or greater than a threshold value, one or more of the plurality of sensor devices is in descending order of the remaining amount of the battery or in descending order of evaluation values calculated based on the index value and the remaining amount of the battery. Selecting a sensor device;
Collecting the measurement values from the one or more selected sensor devices among the plurality of sensor devices;
A computer executing a step of estimating a measurement value of another sensor device among the plurality of sensor devices from the collected measurement value.
(Supplementary Note 10) A step of selecting a sensor device from the plurality of sensor devices in the order of the index value;
Estimating a measurement value of another sensor device from the measurement value of the selected sensor device;
Appendix 9 wherein when the estimated value satisfies a predetermined accuracy with respect to the measured value, the computer executes a step of setting a sum of the index values of all the selected sensor devices as the threshold value. The sensor control method described in 1.
(Supplementary Note 11) A step of determining whether or not the estimated value estimated in the step of estimating the measured value of the other sensor device satisfies a predetermined accuracy with respect to the measured value;
When it is determined that the estimated value does not satisfy the predetermined accuracy, the computer executes the step of changing the threshold value,
11. The sensor control method according to appendix 10, wherein in the step of changing the threshold value, a sensor device is selected by reversing the order of the index values from the previous threshold value setting time.
(Supplementary note 12) In the step of setting the total of the index values as the threshold value, the number of selected sensor devices is reduced so that the difference between the total of the index values and the threshold value is reduced. The sensor control method according to any one of the above.
(Additional remark 13) In the sensor control program which controls the several sensor apparatus provided with the battery,
Collecting the measured value and the remaining amount of the battery from the plurality of sensor devices,
By performing regression analysis on the measured value, an index value indicating the importance in estimating the measured value of another sensor device from the measured value for each of the plurality of sensor devices is calculated,
In order that the sum of the index values is equal to or greater than a threshold value, one or more of the plurality of sensor devices is in descending order of the remaining amount of the battery or in descending order of evaluation values calculated based on the index value and the remaining amount of the battery. Select the sensor device,
Collecting the measurement values from the one or more selected sensor devices among the plurality of sensor devices;
A sensor control program that causes a computer to execute a process of inferring a measurement value of another sensor device among the plurality of sensor devices from the collected measurement value.
(Supplementary Note 14) Select a sensor device in the order of the index value from the plurality of sensor devices,
Inferring measurement values of other sensor devices from the measurement values of the selected sensor device,
An additional note that, when the estimated value satisfies a predetermined accuracy with respect to the measured value, causes the computer to execute a process of setting the sum of the index values of all the selected sensor devices as the threshold value. 13. A sensor control program according to item 13.
(Additional remark 15) It is determined whether the estimated value estimated in the process which estimates the measured value of the said other sensor apparatus satisfy | fills predetermined precision with respect to the said measured value,
If it is determined that the estimated value does not satisfy the predetermined accuracy, the threshold value is changed.
15. The sensor control program according to appendix 14, wherein, in the process of changing the threshold value, the sensor device is selected by reversing the order of the index values from the previous setting of the threshold value.
(Additional remark 16) In the process which sets the sum total of the said index value to the said threshold value, the number of selections of a sensor apparatus is reduced so that the difference of the total of the said index value and the said threshold value may be reduced. A sensor control program according to any one of the above.

DESCRIPTION OF SYMBOLS 1 Control server 2 Sensor apparatus 10 CPU
27 Battery 101 Measurement Processing Unit 102 Importance Calculation Unit 103 Sensor Selection Unit 104 Accuracy Monitoring Unit

Claims (7)

In a sensor control device that controls a plurality of sensor devices including a battery,
A collection unit that collects measurement values and the remaining amount of the battery from the plurality of sensor devices, and
By calculating regression of the measured value, a calculation unit that calculates an index value indicating the importance in estimating the measured value of another sensor device from the measured value for each of the plurality of sensor devices;
In order that the sum of the index values is equal to or greater than a threshold value, one or more of the plurality of sensor devices is in descending order of the remaining amount of the battery or in descending order of evaluation values calculated based on the index value and the remaining amount of the battery. A selection unit for selecting a sensor device;
The collection unit collects the measurement values from the one or more selected sensor devices among the plurality of sensor devices after the selection unit selects one or more sensor devices, and the collected measurement values A sensor control device which estimates a measured value of another sensor device from the above.   The selection unit selects a sensor device from the plurality of sensor devices in the order of the index value, estimates a measured value of another sensor device from the measured value of the selected sensor device, and the estimated value is the measured value. 2. The sensor control device according to claim 1, wherein when a predetermined accuracy is satisfied with respect to a value, a sum of the index values of all the selected sensor devices is set as the threshold value. A determination unit that determines whether or not the estimated value estimated by the collection unit satisfies a predetermined accuracy with respect to the measurement value;
The selection unit includes:
When the determination unit determines that the estimated value does not satisfy the predetermined accuracy, the threshold value is changed,
3. The sensor control device according to claim 2, wherein in the change, the sensor device is selected by reversing the order of the index values from the previous setting of the threshold value.   The sensor control device according to claim 1, wherein the selection unit reduces the number of sensor devices selected so that a difference between a sum of the index values and the threshold value is reduced. A plurality of sensor devices with batteries;
A sensor control device that controls the plurality of sensor devices;
The sensor control device includes:
A collection unit that collects measurement values and the remaining amount of the battery from the plurality of sensor devices, and
By calculating regression of the measured value, a calculation unit that calculates an index value indicating the importance in estimating the measured value of another sensor device from the measured value for each of the plurality of sensor devices;
In order that the sum of the index values is equal to or greater than a threshold value, one or more of the plurality of sensor devices is in descending order of the remaining amount of the battery or in descending order of evaluation values calculated based on the index value and the remaining amount of the battery. A selection unit for selecting a sensor device;
The collection unit collects the measurement values from the one or more selected sensor devices among the plurality of sensor devices after the selection unit selects one or more sensor devices, and the collected measurement values A sensor system characterized by estimating a measured value of another sensor device from the above. In a sensor control method for controlling a plurality of sensor devices including a battery,
Collecting each of the measurement value and the remaining amount of the battery from the plurality of sensor devices;
Calculating an index value indicating the importance in estimating the measured value of another sensor device from the measured value for each of the plurality of sensor devices by performing regression analysis of the measured value;
In order that the sum of the index values is equal to or greater than a threshold value, one or more of the plurality of sensor devices is in descending order of the remaining amount of the battery or in descending order of evaluation values calculated based on the index value and the remaining amount of the battery. Selecting a sensor device;
Collecting the measurement values from the one or more selected sensor devices among the plurality of sensor devices;
A computer executing a step of estimating a measurement value of another sensor device among the plurality of sensor devices from the collected measurement value. In a sensor control program for controlling a plurality of sensor devices including a battery,
Collecting the measured value and the remaining amount of the battery from the plurality of sensor devices,
By performing regression analysis on the measured value, an index value indicating the importance in estimating the measured value of another sensor device from the measured value for each of the plurality of sensor devices is calculated,
In order that the sum of the index values is equal to or greater than a threshold value, one or more of the plurality of sensor devices is in descending order of the remaining amount of the battery or in descending order of evaluation values calculated based on the index value and the remaining amount of the battery Select the sensor device,
Collecting the measurement values from the one or more selected sensor devices among the plurality of sensor devices;
A sensor control program that causes a computer to execute a process of inferring a measurement value of another sensor device among the plurality of sensor devices from the collected measurement value.

Images