JP2016115289A - Sensor controller, sensor control program and sensor control method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a sensor controller capable of reducing transmission frequency of measured values even when measured values change variously.SOLUTION: The sensor controller includes: a sensor 2 that measures a measurement object and outputs a measured value as a measurement result; a prediction section 3 that calculates a prediction value of the measurement result using a prediction model in plural prediction models for predicting a measurement result; a transmission section 4 that transmits the measured value to a server when a difference between the measured value and the prediction value is larger than a predetermined threshold level; and a changing section 5 that changes the prediction model according to the transmission frequency of the measured data.SELECTED DRAWING: Figure 1

Description

  The present specification relates to a sensor control device, a sensor control program, and a sensor control method.

  There is a sensor network composed of sensor nodes using battery drive and energy harvester technology. In such a sensor network, from the viewpoint of management cost and system reliability, it is required to reduce the power consumption of the sensor node as much as possible to increase the lifetime.

  On the other hand, even when the battery level of the sensor node is low, the accuracy of sensing data required by the application program operating on the server side cannot be reduced. However, if the sensor node performs sensing at a high frequency and continues to transmit the sensed data to the server, the battery will run out immediately and it will not be possible to operate.

  As a technique related to reduction of power consumption of sensing data, for example, there is a first technique for autonomously suppressing power consumption related to device management (for example, Patent Document 1). The communication device transmits a management information request message to another communication device. The communication device receives management information of another communication device by a response message to the request message. The communication device uses the received management information of the other communication device to determine a response condition that determines whether or not the other communication device responds to the request message. The communication device creates a request message that is transmitted at a predetermined timing after the determination, including the determined response condition. The communication apparatus uses a response method that lowers the frequency of response for a sensor device whose absolute value of the difference between actual management information and creation management information for any past X times is smaller than a threshold T that is arbitrarily determined, and conversely, the threshold T or higher The response method for increasing the response frequency is determined for the sensor device.

  However, in the first technique, if the difference between the measured data and the estimated data is less than a certain value, the data transmission interval is extended. Therefore, there is a sudden change in the measured data between transmissions on the sensor node side. In this case, the server cannot supplement the change in the actual measurement data.

  On the other hand, there is a second technique in which the wireless terminal device transmits information to the base station when the predicted measurement result and the measurement result by the sensor are different (for example, Patent Document 2). In the second technique, power consumption is reduced by reducing the frequency of wireless communication based on prediction.

JP 2014-138245 A JP 2006-33674 A JP 2013-115488 A JP 2010-124396 A JP 2013-255193 A JP 2007-30830 A JP 2006-33674 A JP 2010-206596 A

  In the second technique described above, a simple example of the prediction method is shown. However, in actual use scenes, measurement values obtained as a result of measurement often vary in a complicated manner. Therefore, the frequency of transmission of actual measurement values deviates from the prediction every time, and power consumption is not reduced.

  The present invention provides, as one aspect, a sensor control technique for reducing the frequency of transmission of actual measurement values even when the measurement values vary in various ways.

  A sensor control device according to one aspect of the present invention includes a sensor, a prediction unit, a transmission unit, and a change unit. The sensor measures a measurement target and outputs an actual measurement value as a measurement result. The prediction unit calculates a predicted value of the measurement result using any one of a plurality of prediction models for predicting the measurement result. A transmission part transmits an actual value to a server apparatus, when the difference of an actual value and an estimated value is more than a predetermined threshold value. The changing unit changes the prediction model according to the transmission frequency of the actual measurement data.

  According to the technique described in this specification, it is possible to reduce the transmission frequency of the actual measurement value even when the measurement value varies in various ways.

An example of the sensor control apparatus in this embodiment is shown. An example of the sensor management system in this embodiment is shown. An example of the prediction model applied by switching in this embodiment is shown. An example of the sensor node in this embodiment (Example 1) is shown. An example of the server in this embodiment (Example 1) is shown. It is the sequence diagram (the 1) between the sensor node regarding the transmission of the actual measurement value in this embodiment (Example 1). It is the sequence diagram (the 2) between the sensor node regarding the transmission of the actual measurement value in this embodiment (Example 1). It is the sequence diagram (the 1) between the sensor node regarding the transmission of the actual value according to the comparison result of the actual value and the predicted value in this embodiment (Example 1). It is the sequence diagram (the 2) between the sensor node regarding the transmission of the actual value according to the comparison result of the actual value and the predicted value in this embodiment (Example 1). It is the sequence diagram (the 3) between the sensor node regarding the transmission of the actual value according to the comparison result of the actual value and the predicted value in the present embodiment (Example 1). It is a sequence diagram (the 4) between the sensor node regarding the transmission of the actual value according to the comparison result of the actual value and the predicted value in the present embodiment (Example 1). The processing flow of the prediction model determination part (node side and server side) in this embodiment (Example 1) is shown. The other example of the processing flow of the prediction model determination part (node side and server side) in this embodiment (Example 1) is shown. An example of the sensor node in this embodiment (Example 2) is shown. It is the sequence diagram (the 1) of the sensor node in this embodiment (Example 2). It is the sequence diagram (the 2) of the sensor node in this embodiment (Example 2). It is the sequence diagram (the 3) of the sensor node in this mode (example 2). An example of the sensor node in this embodiment (Example 3) is shown. The processing flow of the sensor node in this embodiment (Example 3) is shown. The processing flow of the sensor node in this embodiment (Example 4) is shown. An example of the sensor node in this embodiment (Example 5) is shown. An example of the predicted value-measurement time relationship information in this embodiment (Example 5) is shown. It is a sequence diagram of a sensor node in the present embodiment (Example 5). An example of the hardware constitutions of the sensor node concerning this embodiment is shown. It is an example of a configuration block diagram of a hardware environment of a computer that executes a program in the present embodiment.

  FIG. 1 shows an example of a sensor control device in the present embodiment. The sensor control device 1 includes a sensor 2, a prediction unit 3, a transmission unit 4, and a change unit 5. An example of the sensor control device 1 is a sensor node 11.

  The sensor 2 measures a measurement target and outputs an actual measurement value that is a measurement result. Examples of the sensor 2 include the sensor 31 or the sensor unit 62.

  The prediction unit 3 calculates a predicted value of the measurement result using any one of a plurality of prediction models for predicting the measurement result. An example of the prediction unit 3 is a control unit 61 that functions as the data prediction unit 26.

  The transmission unit 4 transmits the actual measurement value to the server device when the difference between the actual measurement value and the predicted value is greater than or equal to a predetermined threshold value. As an example of the transmission unit 4, a data comparison unit 37 and a control unit 61 that functions as the transmission unit 38 can be cited.

  The changing unit 5 changes the prediction model according to the transmission frequency of the actual measurement data. An example of the changing unit 5 is a control unit 61 that functions as the prediction model determination unit 32.

  By configuring in this way, it is possible to reduce the transmission frequency of the actual measurement value even when the measurement value varies in various ways. In other words, when the difference between the measured value from the sensor and the predicted value predicted by the prediction model is greater than or equal to the threshold value, the measured value is changed to an appropriate prediction model according to the transmission frequency when transmitting the measured value to the server. Even when the frequency fluctuates in various ways, the transmission frequency can be reduced. As a result, power consumption can be reduced.

  The changing unit 5 changes the prediction model stepwise to a prediction model based on a more complicated approximate expression or a prediction model based on a predetermined order as the transmission frequency increases. By configuring in this way, for example, when the sensor control device 1 is operating with a prediction model set as a default, if the transmission frequency increases more than a certain level, the sensor control device 1 is operated with a prediction model based on a more complicated approximate expression. . When the transmission frequency increases again more than a certain value, the sensor control device 1 switches to the next prediction model. As a result, the transmission frequency can be reduced by changing to a more appropriate prediction model in accordance with the actually measured value.

The changing unit 5 changes the prediction model stepwise to a prediction model based on a simpler approximation formula or a prediction model based on a predetermined order when the transmission frequency is equal to or less than a predetermined value for a predetermined time.
With this configuration, when the prediction model is changed step by step to a prediction model based on a more complicated approximate expression or a prediction model based on a predetermined order, the transmission frequency is constant for a certain period of time. If the value is equal to or smaller than the value, the sensor control device 1 performs the following process. That is, the sensor control device 1 operates by returning to the prediction model before the change step by step.

  The prediction unit 3 calculates a prediction value for each of the same events using the first prediction model and the second prediction model corresponding to the relationship before and after the change in the case where the prediction model is changed in stages among the plurality of prediction models. The changing unit 3 changes the first prediction model to the second prediction model when the second frequency is smaller than the first frequency. Here, the first frequency indicates the number of times that the difference between the actual measurement value and the predicted value is equal to or greater than a predetermined threshold value in a predetermined time, when the prediction is performed using the currently applied first prediction model. . The second frequency indicates the number of times that the difference between the actually measured value and the predicted value is equal to or greater than a predetermined threshold value in a predetermined time, when predicted using the second prediction model.

  By configuring in this way, it is possible to employ a prediction model with better prediction results, and therefore it is possible to perform prediction with higher accuracy.

  When the sensor control device 1 is operated using a battery, or when the power generation amount is 0 when the sensor control device 1 is operated using a battery, the changing unit 5 is configured according to the transmission frequency of measured data. To change.

  With this configuration, when there is no need to consider power consumption, it is not necessary to calculate the predicted value because there is no need to consider the transmission frequency, but when there is a limit to the available power, the transmission frequency The number of transmissions can be reduced by calculating a prediction value using an appropriate prediction model to limit the number of transmissions. Thereby, power consumption can be suppressed while maintaining the accuracy of the sensing data.

  The change unit 5 calculates the power balance in the sensor control device when the sensor control device 1 operates using a battery or operates using an environmental power generation device, and consumes more than the available power. When the power is larger, the prediction model is changed according to the transmission frequency of the actual measurement data.

  By configuring in this way, it is possible to calculate a predicted value using an appropriate prediction model in order to limit the transmission frequency when the power balance is larger than the available power in consideration of the power balance. it can. Thereby, power consumption can be suppressed while maintaining the accuracy of the sensing data.

  The sensor control device 1 further includes a measurement time adjustment unit 6. The measurement time adjustment unit 6 adjusts the measurement time of the sensor according to the predicted value. An example of the measurement time adjustment unit 6 is a sensor control unit 51.

With this configuration, the time required for one measurement at a certain measurement accuracy varies depending on the measured value in the sensor. Therefore, by adjusting to the optimal measurement time, control is performed so that unnecessary power is not consumed. FIG. 2 shows an example of a sensor management system in the present embodiment. The server device (hereinafter referred to as “server”) 21 and the sensor node 11 share a plurality of prediction models. Here, the prediction model is, for example, a data prediction algorithm such as prediction based on a moving average, prediction based on a least square method, and the like. Note that the server functions described below may be realized by a base station (access point).

  The sensor node 11 performs sensing at a frequency according to the requirements of the application program, acquires sensing data (actual measurement value) (S1), and stores it in the storage unit (S2). At the same time, the sensor node 11 predicts sensing data (predicted value) using a prediction model (S3). Note that the prediction model (calculation logic / program) may be stored in advance in the sensor node or distributed from the server. The sensor node 11 compares the measured value and the predicted value of the sensing data, and transmits the measured value to the server 21 when the difference between the measured value and the predicted value is equal to or greater than a predetermined threshold (S4).

  Further, the sensor node 11 measures the frequency at which the actual measurement value is transmitted, and determines whether the prediction model currently applied is appropriate / inappropriate from the transmission frequency. The sensor node 11 switches to an appropriate prediction model according to the determination result (S5).

  Prediction using a prediction model is also performed on the server side. When the actual measurement value is not sent from the sensor node 11, that is, when the transmission frequency (reception frequency) is low, the prediction model is appropriate, so the server 21 performs the following process. That is, the server 21 predicts the latest data from the data series in which the actual measurement value sent from the sensor node 11 and the predicted value calculated by the server 21 are mixed, and the predicted latest data. (Predicted value) is substituted for the actually measured value (S6). In this case, the server 21 sends the predicted value as sensing data to the application side.

  On the other hand, if the transmission frequency (reception frequency in the server 21) at which the actual measurement value is transmitted from the sensor node 11 is high, the prediction model is inappropriate and the server 21 performs the following processing. Do. That is, the server 21 determines whether the currently applied prediction model is appropriate or not based on the reception frequency, and switches to an appropriate prediction model according to the determination result (S7).

  FIG. 3 shows an example of a prediction model applied by switching in the present embodiment. In the prediction model graph of FIG. 3, for example, the vertical axis represents a measured value and the horizontal axis represents elapsed time. The prediction model is a model based on an approximate expression (or approximate curve expression) such as linear approximation, polynomial approximation, logarithmic approximation, exponent, and power approximation. In the present embodiment, the prediction model is switched in the order of FIG. 3 (A), FIG. 3 (B), and FIG. 3 (C). Further, as an example, each of the sensor node 11 and the server 21 stores the prediction models of FIGS. 3A, 3B, and 3C in the storage unit in advance.

  FIG. 3A shows a linear approximation model. The linear approximate value model is used when the actual measurement value changes monotonously. When the prediction model currently applied is FIG. 3 (A), the sensor node 11 and the server 21 respectively compare the prediction value predicted by the linear approximation model with the actual measurement value in S4 and S6. In the prediction of data using the linear approximation model, the amount of calculation is small and the prediction is easy.

  FIG. 3B shows a quadratic curve approximation model. The quadratic curve approximate value model is used when the measured value increases or decreases relatively slowly. When the currently applied prediction model is FIG. 3B, the sensor node 11 and the server 21 compare the predicted value predicted by the quadratic curve approximate value model with the actual measurement value in S4 and S6, respectively. In the prediction of data using the quadratic curve approximate value model, the calculation amount is increased as compared with the linear approximate value model, but prediction is possible if the increase / decrease is moderate.

  FIG. 3C shows a moving average model. When the increase / decrease of the actual measurement value is severe and it is difficult to predict even when the calculation amount is increased and the multidimensional curve approximation is applied, the moving average model is applied. That is, the moving average model is used when the actually measured value is complicatedly increased or decreased, that is, randomly increased or decreased. When the prediction model currently applied is FIG. 3C, the sensor node 11 and the server 21 respectively compare the predicted value based on the moving average of the most recent predetermined number of measured values with the measured values in S4 and S6. To do.

  The prediction models shown in FIGS. 3A to 3C are switched in stages according to the transmission frequency of the actual measurement values. Specifically, a more complicated prediction model is applied step by step as the transmission frequency increases. For example, assume that the prediction model set by default in the sensor node 11 and the server 21 is as shown in FIG. If the transmission frequency increases beyond a certain level, the sensor node 11 and the server 21 switch the currently applied prediction model to the prediction model shown in FIG. If the transmission frequency increases again by a certain amount or more, the sensor node 11 and the server 21 switch the current prediction model to the prediction model of FIG.

  When the prediction model is FIG. 3B or FIG. 3C, the sensor node 11 and the server 21 become a simpler prediction model if the transmission frequency falls below a predetermined value during a predetermined time. Move back in stages.

  In the present embodiment, as an example, three prediction models (a linear approximate value model, a quadratic curve approximate value model, and a moving average model) will be described. However, the present invention is not limited to this, and a plurality of approximate curves, etc. The prediction model may be used.

Below, the Example of this embodiment is further explained in full detail.
Example 1
FIG. 4 shows an example of a sensor node in the present embodiment (Example 1). The sensor node 11 includes a sensor 31, a prediction model determination unit 32, an effective data storage unit 33, an error threshold storage unit 34, a reception unit 35, a data prediction unit 36, a data comparison unit 37, and a transmission unit 38.

  The sensor 31 is a detection device that detects the state of the surrounding environment such as temperature, humidity, water level, and pressure, performs sensing, and outputs the result as sensing data (actual measurement value). The sensor node includes one or more sensors 31.

  The receiving unit 35 receives an error threshold from the server 21. The error threshold T1 is a threshold used for determining whether or not the difference between the predicted value and the actual measurement value is equal to or greater than the error threshold T1.

  The error threshold storage unit 34 stores the error threshold T1 transmitted from the server 21. The effective data storage unit 33 stores one or more recent effective measurement data (for example, measurement values or predicted values for the latest n (arbitrary integer)) in a FIFO (First-In-First-Out) format. . The effective measurement data includes an actual measurement value or a predicted value and a time when the actual measurement or prediction is performed. If the reference time and the measurement interval are fixed in advance, the effective measurement data may include the number of measurements instead of the time. When the difference between the predicted value and the actual measurement value is equal to or greater than the error threshold T1, the actual measurement value is stored as an effective measurement value (effective measurement data). When the difference between the predicted value and the actually measured value is less than the error threshold value T1, the predicted value is stored as an effective measured value.

  For example, the data prediction unit 36 previously stores a plurality of pieces of prediction model information about the linear approximate value model in FIG. 3A, the quadratic curve approximate value model in FIG. 3B, and the moving average model in FIG. I remember it. The data prediction unit 36 reads the latest effective measurement value from the effective data storage unit 33. The data prediction unit 36 generates a predicted value of the sensing data based on the read effective measurement value and the prediction model.

  The data comparison unit 37 reads the error threshold T1 from the error threshold storage unit 34. The sensing data (actual measurement value) acquired from the sensor 31 is compared with the prediction value generated by the data prediction unit 36. When the difference between the actual measurement value and the predicted value is equal to or greater than the threshold value T <b> 1, the data comparison unit 37 transmits the actual measurement value to the server 21 via the transmission unit 38. The data comparison unit 37 stores the transmitted actual measurement value in the effective data storage unit 33 as an effective measurement value.

  The prediction model determination unit 32 measures the transmission frequency of the actual measurement value to the server 21 and determines whether the prediction model currently applied is appropriate or inappropriate from the transmission frequency. The prediction model determination unit 32 switches to an appropriate prediction model according to the determination result, and notifies the data prediction unit 36 of the switched prediction model.

  When the difference between the actual measurement value and the predicted value is equal to or greater than the threshold value T1, the transmission unit 38 transmits data including the sensor node ID, the sensor ID, and the actual measurement value based on the address information of the server 21 set in advance.

  FIG. 5 shows an example of a server in the present embodiment (Example 1). The server 21 includes an error threshold setting unit 41, a prediction model determination unit 42, a transmission data generation unit 43, a transmission unit 44, an application data storage unit 45, a data prediction unit 46, an actual measurement value reception determination unit 47, and a reception unit 48.

  The error threshold setting unit 41 has an external interface, and an error threshold T1 for notifying a sensor node through this external interface is set in advance for each sensor. The transmission data generation unit 43 generates transmission data including the error threshold T1 set in the error threshold setting unit 41. The transmission unit 44 transmits the transmission data generated by the transmission data generation unit 43 to the sensor node 11 based on a preset address of the sensor node 11.

  The application 49 is an application program that analyzes sensing data (actually measured values and predicted values).

  The application data storage unit 45 has one or more recent effective measurement data (for example, measurement values or predictions of the latest n (arbitrary integer)) sent to the application 49 in a FIFO (First-In-First-Out) format. Value) is stored for each sensor of each sensor node. When the actual measurement value is received from the sensor node, the actual measurement value is stored in the application data storage unit 45 as an effective measurement value. When the actual measurement value is not received from the sensor node, the predicted value is stored in the application data storage unit 45 as an effective measurement value.

  The receiving unit 48 receives data including a sensor node ID, a sensor ID, and an actual measurement value from the sensor node 11.

  The actual value reception determination unit 47 performs the following process for each sensor of the sensor node 11. That is, the actual value reception determination unit 47 determines whether or not an actual value has been received from the sensor node 11, notifies the determination result to the data prediction unit 46, and transmits the received actual value to the application 49. Further, it is stored in the application data storage unit 45 as an effective measurement value. When it is determined that the actual measurement value is received from the sensor node 11 at a predetermined frequency, the actual measurement value reception determination unit 47 causes the prediction model determination unit 42 to receive the actual data reception frequency (of the sensor node 11). Send frequency).

  When it is determined that the actual measurement value is not sent from the sensor node 11 within the predetermined time, the actual measurement value reception determination unit 47 notifies the data prediction unit 46 to that effect.

  The data prediction unit 46 reads the effective measurement value stored most recently from the application data storage unit 45 for each sensor of the sensor node 11. The data prediction unit 46 generates a predicted value of the sensing data based on the read transmission data and the prediction model. The data prediction unit 46 transmits the generated predicted value to the application 49 and stores it as an effective measurement value in the application data storage unit 45 when the actual measurement value is not received.

  For each sensor of the sensor node 11, the prediction model determination unit 42 measures the transmission frequency of the actual measurement value of the sensor 13 (the reception frequency of the actual measurement value received by the server 21), and the prediction model currently applied from the transmission frequency. Appropriate / unsuitable. The prediction model determination unit 42 switches to an appropriate prediction model according to the determination result, and notifies the data prediction unit 46 of the switched prediction model.

  FIG. 6A and FIG. 6B are sequence diagrams between the sensor node and the server regarding transmission of actual measurement values in the present embodiment (Example 1). First, on the server 21 side, the error threshold value setting unit 41 notifies the transmission data generation unit 43 of the node ID, sensor ID, and threshold value T1 (S11). The transmission data generation unit 43 generates transmission data including the node ID, sensor ID, and threshold value T1, and sends the transmission data to the transmission unit 44 (S12). The transmission unit 44 transmits the transmission data to the address of the sensor node corresponding to the node ID (S13).

  On the sensor node 11 side corresponding to the node ID, the reception unit 35 receives the transmission data from the server 21, extracts the sensor ID and the threshold T1 from the transmission data, and stores them in the error threshold storage unit 34 (S14). The data comparison unit 37 acquires a threshold value from the error threshold value storage unit 34 (S15).

  Next, the sensor 31 performs sensing at a predetermined interval to generate sensing data (actual measurement value) (S16). The data comparison unit 37 compares the actual measurement value with the predicted value, and if the difference between the actual measurement value and the predicted value is equal to or greater than the threshold value T1 as a result of the comparison, notifies the transmission unit 38 of the actual measurement value (S17). As a result of the comparison, when the difference between the actual measurement value and the predicted value is smaller than the threshold value T1, the data comparison unit 37 notifies the transmission unit 38 of nothing. However, in the initial state (when the threshold value T1 is received from the server 21), the data prediction unit 36 selects a default value (for example, 0) as the prediction value because the prediction model to be applied is not selected. 37 may be provided. Further, in the initial state (when the threshold value T1 is received from the server 21), the data comparison unit 37 unconditionally transmits the actual measurement value to the server 21 in order to transmit the first plurality of sampling data to the server. You may do it.

  If the difference between the actual measurement value and the predicted value is equal to or greater than the threshold value T <b> 1 as a result of the comparison, the data comparison unit 37 writes the actual measurement value in the effective data storage unit 33. As a result of the comparison, when the difference between the actual measurement value and the predicted value is less than the threshold value T1, the data comparison unit 37 writes the predicted value as an effective measurement value in the effective data storage unit 33. When there is no predicted value (immediately after the start of the operation, the actual data is not accumulated and therefore cannot be predicted), the data comparison unit 37 uses the sensing data (actual value) as an effective measurement value in the effective data storage unit 33. Write (S19). Note that the data comparison unit 37 may store the sensing data of both the actual measurement value and the predicted value in the effective data storage unit 33 each time.

  When the measured value is notified from the data comparing unit 37, the transmitting unit 38 transmits transmission data including the sensor node ID, the sensor ID, and the measured value to the server 21 (S18).

  On the server 21 side, when receiving the transmission data from the sensor node 11, the receiving unit 48 transmits it to the measured value reception determining unit 47 (S20).

  The actual value reception determination unit 47 stores the actual value included in the transmission data from the sensor node 11 in the application data storage unit 45 (S21) and transfers it to the upper application 49 (S22). The actual value reception determination unit 47 counts the number of transmission data received for each combination of sensor node ID and sensor ID (S23).

  At the start of the operation, the processes of S16 to S23 are repeated for the number of data that can be predicted by the prediction model.

  FIG. 7 (FIG. 7A, FIG. 7B, FIG. 7C, FIG. 7D) shows the sensor node-server transmission related to the transmission of the actual measurement value according to the comparison result between the actual measurement value and the predicted value in this embodiment (Example 1). It is a sequence diagram. The sequence of FIG. 7A and FIG. 7B is a sequence after the prediction value can be calculated based on the prediction model through the sequence of FIG. 6A and FIG. 6B.

  On the sensor node 11 side, the data prediction unit 36 reads one or more effective measurement values from the effective data storage unit 33 from the newer one (S24). The data prediction unit 36 calculates the predicted value of the next sensing data using the one or more effective measurement values and the prediction model currently effective (S25). The data prediction unit 36 sends the calculated predicted value to the data comparison unit 36 (S26).

  The sensor 31 performs sensing at a predetermined interval, and generates sensing data (actual measurement value) (S27). The data comparison unit 37 compares the actual measurement value and the predicted value, and if the difference between the actual measurement value and the predicted value is equal to or greater than the threshold value T1 as a result of the comparison, notifies the transmission unit 38 of the actual measurement value (S28).

  Further, when the difference between the actually measured value and the predicted value is equal to or greater than the threshold value T1 as a result of the comparison, the data comparing unit 37 writes the sensing data (actually measured value) to the effective data storage unit 33 as an effective measured value (S29). In this case, the data comparison unit 37 counts the number of transmitted actual values for each sensor 31 (S30).

  The data comparison part 37 notifies a prediction error to the prediction model determination part 32, when the transmission number (transmission frequency) of the measured value within the predetermined time exceeds the threshold value T2 (S31). When receiving the prediction error, the prediction model determination unit 32 changes the prediction model according to the rules described with reference to FIG. 3 (S32). The prediction model determination unit 32 notifies the data prediction unit 36 of the change of the prediction model (S33).

  When the measured value is notified from the data comparing unit 37, the transmitting unit 38 transmits transmission data including the sensor node ID, the sensor ID, and the measured value to the server 21 (S34).

  On the server 21 side, when receiving the transmission data from the sensor node 11, the receiving unit 48 transmits it to the measured value reception determining unit 47 (S35).

  The actual value reception determination unit 47 stores the actual value included in the transmission data from the sensor node 11 in the application data storage unit 45 (S36) and transfers it to the upper application 49 (S37). Further, the actual measurement value reception determination unit 47 counts the number of transmission data received for each combination of the sensor node 11 ID and the sensor ID (S38).

  The actual measurement value reception determination unit 47 notifies the prediction model determination unit 42 of a prediction error when the number of transmissions of actual measurement values (transmission frequency) within a predetermined time exceeds the threshold T2 (S39). Note that the actual measurement value reception determination unit 47 may notify the prediction model determination unit 42 only that the actual measurement value has been transmitted, and the prediction model determination unit 42 may calculate the transmission frequency per unit time.

  When the prediction model determination unit 42 receives a prediction error, the prediction model determination unit 42 changes the prediction model according to the rules described with reference to FIG. 3 (S40). The prediction model determination unit 42 notifies the data prediction unit 46 of the change of the prediction model (S41).

  On the other hand, when the measured value reception determination unit 47 does not receive the transmission data from the sensor node 11 within a predetermined time or a predetermined time, the actual value reception determination unit 47 issues a prediction instruction to the data prediction unit 46 (S42).

  The data prediction unit 46 reads one or more effective measurement values from the new one from the application data storage unit 45 (S43). The data prediction unit 46 calculates the predicted value of the next sensing data using the one or more effective measurement values and the prediction model currently effective (S44). The data prediction unit 46 stores the transmission data from the sensor node 11 as an effective measurement value in the application data storage unit 45 (S45) and transfers it to the upper application 49 (S46).

  FIG. 8 shows a processing flow of the prediction model determination unit (node side and server side) in the present embodiment (Example 1). The prediction model determination units 32 and 42 initialize the model change time Tm, which is a parameter for holding the time when the prediction model is changed, at the time of activation (S51).

  When receiving the prediction error notification (“YES” in S52), the prediction model determination units 32 and 42 change the prediction model (S53) and instruct the data prediction units 36 and 46 to change the prediction model. The prediction model determination units 32 and 42 record the time when the prediction model is changed as the model change time Tm (S54). Thereafter, the process returns to S52.

  For S53, for example, when the prediction error notification is received when the current prediction model is FIG. 3A, the prediction model determination units 32 and 42 switch the prediction model to the prediction model of FIG. When the prediction error notification is received when the current prediction model is FIG. 3B, the prediction model determination units 32 and 42 switch the prediction model to the prediction model of FIG.

When the prediction error notification is not received (“NO” in S52), the process proceeds to S55.
The prediction model determination units 32 and 42 obtain the current time and calculate the elapsed time ΔT from the model change time Tm (S55).

  When the model change time Tm ≦ the threshold T3 (when the predetermined time has not elapsed) (“No” in S56), the process returns to S52. When model change time Tm> threshold value T3 (when a predetermined time has elapsed) (“Yes” in S56), the prediction model determination units 32 and 42 perform the following processing. That is, the prediction model determination units 32 and 42 change the prediction model (S57), and instruct the data prediction units 36 and 46 to change the prediction model. The prediction model determination units 32 and 42 record the time when the prediction model is changed as the model change time Tm (S58). Thereafter, the process returns to S52.

  For S57, for example, when a prediction error notification is received when the current prediction model is FIG. 3C, the prediction model determination units 32 and 42 switch the prediction model to the prediction model of FIG. When the prediction error notification is received when the current prediction model is FIG. 3B, the prediction model determination units 32 and 42 switch the prediction model to the prediction model of FIG.

  As described above, in FIG. 8, the prediction model determination units 32 and 42 switch to a more complicated prediction model or moving average model each time an error notification is received, and switch to a simpler model when a predetermined time has elapsed.

  In FIG. 8, the prediction model determination units 32 and 42 receive the error notification issued by the data comparison unit 37 or the actual measurement value reception determination unit 47 based on whether or not the transmission frequency exceeds the threshold value T2, Although the prediction model was switched, it is not limited to this. For example, the prediction model determination units 32 and 42 may determine whether or not the transmission frequency exceeds the threshold T2 in addition to the processing of FIG. This will be described with reference to FIG.

  FIG. 9 shows another example of the processing flow of the prediction model determination unit (node side and server side) in the present embodiment (Example 1). The prediction model determination units 32 and 42 initialize the model change time Tm, which is a parameter for holding the time when the prediction model is changed (S61).

  When the data comparison unit 37 notifies the prediction model determination units 32 and 42 of transmission of actual values or reception of transmission data including actual values from the actual value reception determination unit 47 ("YES" in S62), the transmission frequency is determined. Fs is calculated (S63). Here, each time the prediction model determination units 32 and 42 receive notification of transmission of actual values from the data comparison unit 37 or reception of transmission data including actual values from the actual value reception determination unit 47, the prediction model determination units 32 and 42 count the number of notifications. ing. The prediction model determination units 32 and 42 determine whether or not the number of notifications (transmission frequency Fs) within a predetermined time exceeds the threshold T2.

  When transmission frequency Fs> threshold value T2 (“YES” in S64), the prediction model determination units 32 and 42 change the prediction model (S65) and instruct the data prediction units 36 and 46 to change the prediction model. The process of S65 is the same as S53. The prediction model determination units 32 and 42 record the time when the prediction model is changed as the model change time Tm (S66). Thereafter, the process returns to S62.

  When the data comparison unit 37 does not notify the transmission of the actual measurement value or the reception of the transmission data including the actual measurement value from the actual measurement value reception determination unit 47 (“NO” in S62) or when the transmission frequency Fs ≦ the threshold value T2 (“ NO "), the process proceeds to S67.

  The prediction model determination units 32 and 42 obtain the current time and calculate the elapsed time ΔT from the model change time Tm (S67).

  When the model change time Tm ≦ the threshold T3 (when the predetermined time has not elapsed) (“No” in S68), the process returns to S62. When model change time Tm> threshold value T3 (when a predetermined time has elapsed) (“Yes” in S68), the prediction model determination units 32 and 42 perform the following processing. That is, the prediction model determination units 32 and 42 change the prediction model (S69), and instruct the data prediction units 36 and 46 to change the prediction model. The prediction model determination units 32 and 42 record the time when the prediction model is changed as the model change time Tm (S70). Thereafter, the process returns to S62.

  According to the first embodiment, the prediction model can be changed to a more appropriate prediction model according to the transmission frequency of the actual measurement value. Therefore, by changing to the prediction model corresponding to the change in the observation environment, the error between the actual measurement value and the prediction value is reduced, and the frequency of sending the actual measurement value is reduced, so that the power consumption can be reduced.

(Example 2)
In the second embodiment, an example will be described in which prediction using two prediction models is performed at the same time and the prediction model having a better prediction result is changed. For example, when the prediction model currently applied is FIG. 3B or FIG. 3C, the sensor node performs prediction based on the prediction model currently applied and the prediction model in the previous order. At the same time. Here, when the prediction model currently applied is FIG. 3 (B), the prediction model in the previous order is FIG. 3 (A). When the prediction model currently applied is FIG. 3C, the prediction model in the previous order is FIG. 3B. The sensor node predicts when the prediction result of the previous prediction model is better than the prediction model currently applied, that is, when the transmission frequency of the previous prediction model is lower. Change the model to the better one. At this time, the sensor node adds and transmits the prediction model change information regarding the change of the prediction model at the time of transmission of the measurement value that has triggered the change of the prediction model.

  In the second embodiment, the same configurations, processes, and functions as those in the first embodiment are denoted by the same reference numerals as those in the first embodiment, and description thereof is omitted.

  FIG. 10 shows an example of a sensor node in the present embodiment (Example 2). The sensor node 11 in FIG. 10 is obtained by replacing the data prediction unit 36 and the data comparison unit 37 in the sensor node in FIG. 4 with two pairs of data prediction units 36a and 36b and data comparison units 37a and 37b.

  FIG. 11 (FIGS. 11A, 11B, and 11C) is a sequence diagram of sensor nodes in the present embodiment (Example 2). The sequence in FIG. 11 corresponds to a sequence obtained by replacing S24 to S34 in FIG. 7A with S81 to S97.

  On the sensor node 11 side, the data prediction unit A (36a) reads one or more effective measurement values from the effective data storage unit 33 from the newer one (S81). At this time, the data prediction unit B (36b) similarly reads one or more effective measurement values from the effective data storage unit 33 from the newer one (S82).

  The data prediction unit A (36a) calculates the predicted value of the next sensing data using the one or more effective measurement values and the prediction model currently effective, and the calculated prediction value is used as the data comparison unit. Send to A (37a) (S83). The data prediction unit B (36b) calculates the predicted value of the next sensing data using the read effective measurement value and the prediction model that is currently valid, and uses the calculated prediction value as the data comparison unit B ( 37b) (S84).

  The sensor 31 performs sensing at a predetermined interval, generates sensing data (actually measured value), and outputs it to the data comparison unit A (37a) and the data comparison unit B (37b) (S85, S86).

  The data comparison unit A (37a) compares the actual measurement value with the predicted value. As a result of the comparison, if the difference between the actual measurement value and the predicted value is equal to or greater than the threshold value T1, the data comparison unit A (37a) counts the number of transmissions of the actual measurement value and is currently applied by the data prediction unit A (36a). The transmission frequency (prediction result A) when applying the prediction model is calculated (S87).

  Further, the data comparison unit B (37b) compares the actual measurement value with the predicted value. As a result of the comparison, when the difference between the actually measured value and the predicted value is equal to or greater than the threshold value T1, the data comparison unit B (37b) counts the counter. The data comparison unit B (37b) sets the counter value as the provisional transmission number when the prediction model currently applied by the data prediction unit B (36b) is applied, and determines the transmission frequency (from the provisional transmission number within a predetermined time). Predicted results B) are calculated (S88).

  The data comparison unit A (37a) notifies the prediction result A to the prediction model determination unit 32 (S89). The data comparison part B (37b) notifies the prediction result B to the prediction model determination part 32 (S90).

  The prediction model determination unit 32 compares the prediction results A and B, and changes the prediction model to the prediction model corresponding to the prediction result with the lower transmission frequency (S91). The prediction model determination unit 32 notifies the data prediction unit A (36a) of the change of the prediction model (S92). In addition, the prediction model determination unit 32 notifies the data prediction unit B (36b) of the change of the prediction model (the prediction model immediately before the model notified to the data prediction unit A) (S93).

  The prediction model determination unit 32 instructs the data comparison unit A (37a) to transmit the changed prediction model to the server 21 (S94).

  The data comparison unit A (37a) notifies the transmission unit 38 of the changed prediction model together with the actual measurement value (S95). In addition, the data comparison unit A (37a) writes the sensing data (actual measurement value) to the effective data storage unit 33 (S96).

  Based on the notification from the data comparison unit 37, the transmission unit 38 transmits transmission data including the sensor node ID, the sensor ID, the actual measurement value, and the predicted model information after the change to the server 21 (S97).

  According to the second embodiment, since a prediction model with better prediction results can be adopted, more accurate prediction can be performed.

(Example 3)
When the sensor node power supply is a commercial power supply, etc., there is little need to consider the power consumption. Therefore, to maintain the accuracy of the sensing data, send the sensing data (actually measured value) to the server every time it is measured. Also good.

  However, if the sensor node is operating by battery drive, or if it is operating using an environmental power generation device such as a solar cell, and the power generation amount is 0, the accuracy of the sensing data is maintained. However, it is necessary to reduce power consumption.

  Therefore, in the third embodiment, when the sensor node is operated by a battery or when it is operated using an environmental power generation device such as a solar cell (when the battery is limited), the power generation amount becomes zero. If it is, the prediction model is switched.

  In the second embodiment, the same configurations, processes, and functions as those in the first embodiment are denoted by the same reference numerals as those in the first embodiment, and description thereof is omitted.

  FIG. 12 shows an example of a sensor node in the present embodiment (Example 3). The sensor node 11 of FIG. 13 is obtained by adding a power supply monitoring unit 39 to the sensor node of FIG. The function of the power supply monitoring unit 39 will be described with reference to FIG.

  FIG. 13 shows a processing flow of the sensor node in the present embodiment (Example 3). The power monitoring unit 39 generates zero power when the sensor node 11 is operated by a battery or when the sensor node 11 is operated using an environmental power generation device such as a solar cell (when the battery is limited). It is determined whether it is (S101).

  When operating by battery drive, or when operating using an environmental power generation device such as a solar cell, the power generation amount is 0 (“YES” in S101), The prediction model switching process is started (S102). For example, the power supply monitoring unit 39 notifies the data prediction unit 36 to perform the process of S24 in FIG. 7A. The subsequent processes are the same as those in the sequence diagrams of FIGS. 7A and 7B.

  According to the third embodiment, when it is not necessary to consider power consumption, it is not necessary to calculate the transmission frequency because there is no need to consider the transmission frequency, but there is a limit to the available power. By applying the first embodiment to the sensor node, the power consumption can be further suppressed while maintaining the accuracy of the sensing data.

Example 4
The fourth embodiment is a modification of the third embodiment. In the fourth embodiment, when the sensor node is operated using an environmental power generation device such as a battery drive or a solar cell, the power balance is calculated from the remaining battery power, the power generation amount, the transmission frequency, and the power consumption in the sensor module such as the sensor. When the continuous operation is impossible, the prediction model is switched.

  In the fourth embodiment, the same configurations, processes, and functions as those in the third embodiment are denoted by the same reference numerals as those in the third embodiment, and description thereof is omitted.

  FIG. 14 shows a process flow of the sensor node in the present embodiment (Example 4). The flow of FIG. 14 is obtained by adding the processing of S111 to S113 to the flow of FIG.

  When operating using an environmental power generation device such as a battery or a solar battery (“YES” in S101), the power supply monitoring unit 39 calculates the following power balance.

  First, in the evaluation period T, the remaining battery power P0 [w], and the power generation amount GP [w / h], the power supply monitoring unit 39 calculates total power P [w] = P0 + GP × T (S111).

Next, the power monitoring unit 39 measures the power consumption for each communication unit: Pr [w], CPU unit: Pc [w], sensor unit Ps [w], and the like. Here, the power consumption Pr [w] of the communication unit is calculated from the power consumption per transmission and the transmission frequency. The power consumption Pc [w] of the CPU unit is calculated from the power consumption at startup and the operation time. The power consumption Ps [w] of the sensor unit is calculated from the power consumption per measurement and the number of measurements per T. Further, based on these, the power supply monitoring unit 39 calculates total power consumption C = Pr + Pc + Ps [w] (S112).

  Here, when total power P <total power consumption C (“YES” in S113), the power supply monitoring unit 39 starts the prediction model switching process (S102). For example, the power supply monitoring unit 39 notifies the data prediction unit 36 to perform the process of S24 in FIG. 7A. The subsequent processes are the same as those in the sequence diagrams of FIGS. 7A and 7B.

  In the fourth embodiment, when the available power is limited, the power balance of the sensor node 11 is considered, and when the power consumption is larger than the available power, an appropriate prediction is made to limit the transmission frequency. The predicted value can be calculated using the model. Thereby, power consumption can be suppressed while maintaining the accuracy of the sensing data.

(Example 5)
In the fifth embodiment, when the transmission frequency is low and the data prediction can be performed stably, the next sensing data is predicted using the currently applied prediction model, and the sensor is operated once according to the predicted value. Determining the time (measurement time) required for the measurement will be described.

  In order to measure a single sensing data such as temperature, the sensor often performs a process such as measuring (sampling) a plurality of times and taking an average value. Here, the measurement time is represented by sampling interval (10 ms) × sampling count when sampling is performed, for example, every 10 ms (milliseconds) in one measurement by the sensor. For example, in the case of distance measurement using infrared rays, the measurement time is indicated by sampling interval × infrared irradiation frequency.

  In the fifth embodiment, the same configurations, processes, and functions as those in the first embodiment are denoted by the same reference numerals as those in the first embodiment, and the description thereof is omitted.

  FIG. 15 shows an example of a sensor node in the present embodiment (Example 5). A sensor node 11 in FIG. 15 is obtained by adding a sensor control unit 51 and a measurement time storage unit 52 to the sensor node in FIG.

  The measurement time storage unit 52 stores predicted value-measurement time relationship information in which a predicted value and a measurement time (or the number of sampling times) are associated with each other.

  The sensor control unit 51 refers to the predicted value-measurement time relationship information in the measurement time storage unit 52 and sets the sensor measurement time (sensor sampling interval × number of measurements) based on the predicted value.

  FIG. 16 shows an example of predicted value-measurement time relationship information in the present embodiment (Example 5). In the predicted value-measurement time relationship information, the measurement time corresponding to the predicted value is registered in advance. Note that the measurement time may be the number of samplings.

  FIG. 17 is a sequence diagram of sensor nodes in the present embodiment (Example 5). The sequence of FIG. 17 is obtained by adding the processing of S131 to S133 between S26 to S27 of FIG. 7A.

  On the sensor node 11 side, the data prediction unit 36 reads one or more effective measurement values from the effective data storage unit 33 from the newer one (S24). The data prediction unit 36 calculates the predicted value of the next sensing data using the one or more effective measurement values and the prediction model currently effective (S25). The data prediction unit 36 sends the calculated predicted value to the data comparison unit 36 (S26). In addition, the data prediction unit 36 sends the calculated prediction value to the sensor control unit 51 (S131).

  When the sensor control unit 51 receives the predicted value from the data prediction unit 36, the sensor control unit 51 acquires the measurement time corresponding to the received predicted value from the predicted value-measurement time relationship information in the measurement time storage unit 52 (S132). The sensor control unit 51 sets the acquired measurement time in the sensor 31 (S133).

  The sensor 31 performs sensing at predetermined intervals during the measurement time set by the sensor control unit 51, and generates sensing data (actual measurement value) (S27). Since the subsequent steps are the same as those in FIGS. 7A and 7B, description thereof is omitted.

  According to the fifth embodiment, in the sensor, the time taken for one measurement at a certain measurement accuracy varies depending on the measurement value. Therefore, by adjusting the measurement time to the optimum measurement time, control is performed so as not to waste power. be able to.

  FIG. 18 shows an example of the hardware configuration of the sensor node according to the present embodiment. The sensor node 11 includes a control unit 61, a sensor unit 62, a storage unit 63, a wireless communication interface (I / F), a bus 65, and a power source 66. The control unit 61, sensor unit 62, storage unit 63, and wireless communication I / F 64 are connected by a bus 65.

  The sensor unit 62 is a detection device that monitors a monitoring object such as a water level sensor, a temperature sensor, a humidity sensor, a gas sensor, an optical sensor, and a sound sensor. The storage unit 63 is a generic term for a RAM (Random Access Memory), a nonvolatile memory, and the like. The nonvolatile memory is a semiconductor memory such as a ROM (Read Only Memory) or a readable / writable SSD (Solid State Drive). The storage unit 63 functions as an effective data storage unit 33, an error threshold storage unit 34, and a measurement time storage unit 52. The storage unit 63 stores a plurality of prediction models, threshold values, and the like.

  The control unit 61 is a processor that controls the operation of the entire sensor node 61, and is, for example, a CPU (Central Processing Unit). The control unit 61 executes the program according to the present embodiment, whereby the prediction model determination unit 32, the reception unit 35, the data prediction unit 36, the data comparison unit 37, the transmission unit 38, the data prediction units 36a and 36b, the data comparison Functions as the units 37a and 37b, the power supply monitoring unit 39, and the sensor control unit 51.

The wireless communication I / F 64 is hardware that performs processing for performing wireless communication.
The power source 65 is a power source for supplying electric power to electronic parts such as the control unit 61, the sensor unit 62, the storage unit 63, and the wireless communication I / F 64 to operate, such as a small battery, a rechargeable battery, and a solar cell. is there.

  The sensor node may be realized by a computer including a processor that executes a program according to the present embodiment.

  FIG. 19 is an example of a configuration block diagram of a hardware environment of a computer that executes a program according to the present embodiment. The computer 70 functions as the server 21. The computer 70 includes a CPU 72, a ROM 73, a RAM 76, a communication I / F 74, a storage device 77, an output I / F 71, an input I / F 75, a reading device 78, a bus 79, an output device 81, and an input device 82.

  Here, CPU indicates a central processing unit. ROM indicates a read-only memory. RAM indicates random access memory. I / F indicates an interface. A CPU 72, ROM 73, RAM 76, communication I / F 74, storage device 77, output I / F 71, input I / F 75, and reading device 78 are connected to the bus 79. The reading device 78 is a device that reads a portable recording medium. The output device 81 is connected to the output I / F 71. The input device 82 is connected to the input I / F 75.

  As the storage device 77, various types of storage devices such as a hard disk, a flash memory, and a magnetic disk can be used. In the storage device 77 or the ROM 73, the CPU 72 functions as an error threshold setting unit 41, a prediction model determination unit 42, a transmission data generation unit 43, a transmission unit 44, a data prediction unit 46, an actual value reception determination unit 47, and a reception unit 48. A program according to the present embodiment is stored. The storage device 77 functions as the application data storage unit 45. The storage device 77 stores an error threshold value and the like. Information is temporarily stored in the RAM 76.

  The CPU 72 reads the program according to the present embodiment from the storage device 77 or the ROM 73 and executes the program.

  The program for realizing the processing described in the above embodiment may be stored in, for example, the storage device 77 via the communication network 80 and the communication I / F 74 from the program provider side. Moreover, the program which implement | achieves the process demonstrated by the said embodiment may be stored in the portable storage medium marketed and distribute | circulated. In this case, the portable storage medium may be set in the reading device 78 and the program read by the CPU 72 and executed. As the portable storage medium, various types of storage media such as a CD-ROM, a flexible disk, an optical disk, a magneto-optical disk, an IC card, and a USB memory device can be used. The program stored in such a storage medium is read by the reading device 78.

  As the input device 82, a keyboard, a mouse, an electronic camera, a web camera, a microphone, a scanner, a sensor, a tablet, or the like can be used. The output device 81 can be a display, a printer, a speaker, or the like. The network 80 may be a communication network such as the Internet, a LAN, a WAN, a dedicated line, a wired line, and a wireless line.

  The present invention is not limited to the above-described embodiment, and various configurations or embodiments can be taken without departing from the gist of the present invention.

DESCRIPTION OF SYMBOLS 1 Sensor control apparatus 2 Sensor 3 Prediction part 4 Transmission part 5 Change part 6 Measurement time adjustment part 11 Sensor node 31 Sensor 32 Prediction model determination part 33 Effective data storage part 34 Error threshold value storage part 35 Reception part 36 Data prediction part 37 Data comparison Unit 38 transmission unit 41 error threshold setting unit 42 prediction model determination unit 43 transmission data generation unit 44 transmission unit 45 application data storage unit 46 data prediction unit 47 actual value reception determination unit 48 reception unit

Claims (9)

A sensor that measures a measurement target and outputs an actual measurement value as a measurement result;
A prediction unit that calculates a predicted value of the measurement result using any one of a plurality of prediction models for predicting the measurement result;
A transmission unit that transmits the actual measurement value to a server device when a difference between the actual measurement value and the predicted value is equal to or greater than a predetermined threshold;
A change unit that changes the prediction model according to the transmission frequency of the actual measurement data;
A sensor control device comprising: The said change part changes the said prediction model in steps to the said prediction model based on a more complicated approximate expression, or to the said prediction model based on predetermined order, so that the said transmission frequency is high. Item 2. The sensor control device according to Item 1. The changing unit changes the prediction model stepwise to the prediction model based on a simpler approximate expression or in the reverse order of the order when the transmission frequency is equal to or less than a predetermined value for a predetermined time. The sensor control device according to claim 2. The prediction unit
Using the first prediction model and the second prediction model corresponding to the relationship before and after the change when changing in a stepwise manner among the plurality of prediction models, calculating a prediction value for each of the same events,
The change unit indicates the number of times that a difference between the actual measurement value and the prediction value is equal to or greater than a predetermined threshold value in a predetermined time when the prediction is performed using the first prediction model currently applied. A second frequency indicating the number of times that the difference between the actual measurement value and the predicted value is equal to or greater than a predetermined threshold in a predetermined time, when the prediction is performed using the second prediction model, rather than the first frequency. The sensor control device according to any one of claims 1 to 3, wherein when the direction is smaller, the first prediction model is changed to the second prediction model. When the sensor control device is operating using a battery, or when the sensor control device is operating using an environmental power generation device and is not generating power, the changing unit is configured according to the transmission frequency of the actual measurement data. The sensor control device according to any one of claims 1 to 4, wherein the prediction model is changed. The change unit calculates a power balance in the sensor control device when the sensor control device is operated using a battery, or is operated using an environmental power generation device. 5. The sensor control device according to claim 1, wherein when the power consumption is larger, the prediction model is changed according to a transmission frequency of the actual measurement data. The sensor control device further includes:
A measurement time adjustment unit for adjusting the measurement time of the sensor according to the predicted value;
The sensor control apparatus according to any one of claims 1 to 6, further comprising: On the computer,
From the sensor, measure the measurement object and obtain the actual measurement value as the measurement result.
Calculating a predicted value of the measurement result using a prediction model of any of a plurality of prediction models for predicting the measurement result;
If the difference between the measured value and the predicted value is greater than or equal to a predetermined threshold, the measured value is transmitted to the server device,
A sensor control program for executing a process of changing the prediction model according to the transmission frequency of the actual measurement data. Computer
From the sensor, measure the measurement object and obtain the actual measurement value as the measurement result.
Calculating a predicted value of the measurement result using a prediction model of any of a plurality of prediction models for predicting the measurement result;
If the difference between the measured value and the predicted value is greater than or equal to a predetermined threshold, the measured value is transmitted to the server device,
The sensor control method, wherein the prediction model is changed according to a transmission frequency of the actual measurement data.

Images