JP6159594B2 - Observation value processing apparatus, observation value processing method, and observation value processing program - Google Patents

Description

  The present invention relates to an observation value processing apparatus, an observation value processing method, and an observation value processing program.

  By performing spatial interpolation on observation values obtained from a plurality of sensors arranged at different positions, an observation value at an arbitrary position where no sensor is present is estimated. As a result, an estimated value (estimated value of the observed value) at that position can be obtained without installing a sensor at the position where the observed value is desired to be acquired.

  The technology described in “Non-patent Document 1“ Visualization of distribution and transition of public land prices for Tokyo 23 wards and all-use areas by spatio-temporal kriging ”uses kriging to interpolate land price information. ing. In addition to the land price information, the technique described in Non-Patent Document 1 uses information on the nearest station, distance to the nearest station, floor area ratio, land area, front road width, presence of sewage, and interpolation. The improvement of accuracy is aimed at.

  In the technique of “a solar radiation amount estimating apparatus and a solar radiation amount estimating method” according to Patent Document 1, Kriging is used to interpolate the solar radiation amount. In the technique according to Patent Document 1, a function having a cloud velocity coefficient and an attenuation coefficient is used for interpolation in order to consider the influence of the cloud.

  In the technique of “Observation Data Estimation Method and Observation Data Estimation Program” according to Patent Document 2, in order to reduce the interpolation error due to the observation value, interpolation is performed for each component by principal component analysis on the observation value, and the interpolation value To integrate.

  In the technique of “observation value reliability evaluation apparatus, observation value reliability evaluation method and observation value reliability evaluation program” according to Patent Document 3, the reliability of observation values is evaluated. The reliability in the technique according to Patent Document 3 is a reliability based on the variation of the representative values of the observed values and the method, and the reliability using the distribution of the observed values is used.

Here, in the technique according to Non-Patent Document 1, interpolation accuracy is improved by using not only the target variable but also auxiliary variables. The target variable is an observation value to be interpolated, and the auxiliary variable is a variable other than the target variable.
In the technique according to Patent Document 1 and the technique related to Patent Document 2, the observation value itself is analyzed to improve the estimation accuracy of the interpolation value.
In the technique according to Patent Document 3, the reliability using the distribution of observed values is used.

JP2013-045572A JP 2008-058109 A JP 2013-036887 A

Ryo Inoue, Hidenori Shimizu, Yutaro Yoshida, Yuzuru Li, “Visualization of distribution and transition of public land prices in Tokyo 23 wards and all-use areas by spatio-temporal kriging”, GIS-Theory and Application, 17 (1 ), 2009, p. 13-24

However, in the observation value interpolation method according to the related art as described above, it has been desired to further improve the interpolation accuracy.
For example, an observation value obtained from a sensor includes an error, and the error occurrence probability and the magnitude of the error differ depending on the installation state (installation condition) of the sensor, the characteristics of each sensor, and the like. In the observation value interpolation method according to the technology, such an error is not taken into consideration, which is a factor that degrades the interpolation accuracy.

  The present invention has been made in view of such circumstances, and it is an object of the present invention to provide an observation value processing apparatus, an observation value processing method, and an observation value processing program that can improve the accuracy of observation value interpolation. To do.

  (1) In order to solve the above-described problem, the observation value processing apparatus according to one aspect of the present invention detects an observation target having a correlation with a position by a plurality of sensors provided at a plurality of different positions. A collection unit that collects the observed values, and a reliability setting unit that sets a reliability that is uncorrelated with the distribution of the position of the sensor for each observation value collected by the collection unit or for each sensor And an observation value for performing an interpolation process on the position of the observation value collected by the collection unit based on the observation value collected by the collection unit, the position of the sensor, and the reliability set by the reliability setting unit. A processing unit.

  (2) One aspect of the present invention is the observation value processing apparatus according to (1) described above, wherein the reliability setting unit is based on the observation value collected by the collection unit and a predetermined reference value. A reliability determination unit for determining the degree;

  (3) One aspect of the present invention is the observation value processing apparatus according to any one of (1) or (2) described above, wherein the observation value processing unit is an observation collected by the collection unit. A kriging unit that obtains a weighting value of kriging based on a value and a position of the sensor; an observation value collected by the collecting unit; a position of the sensor; a weighting value of kriging obtained by the kriging unit; and the reliability An interpolation processing unit that performs an interpolation process on the position of the observation values collected by the collection unit based on the reliability set by the setting unit.

  (4) In order to solve the above-described problem, an observation value processing method according to one aspect of the present invention is an observation target in which a collection unit has a correlation with a position by a plurality of sensors provided at a plurality of different positions. The observation value that is the result of detecting the detection value is collected, and the reliability setting unit has a reliability that is uncorrelated with the distribution of the position of the sensor for each observation value collected by the collection unit or for each sensor. The observation value processing unit sets the observation value collected by the collection unit based on the observation value collected by the collection unit, the position of the sensor, and the reliability set by the reliability setting unit. Interpolate the position.

  (5) One aspect of the present invention is the observation value processing method according to (4), wherein the reliability setting unit is collected by the collection unit by a reliability determination unit included in the reliability setting unit. The reliability is determined based on the observed value and a predetermined reference value.

  (6) One aspect of the present invention is the observation value processing method according to any one of (4) or (5) described above, wherein the observation value processing unit includes the kriging included in the observation value processing unit. A weighting value of kriging is obtained based on the observation value collected by the collection unit and the position of the sensor, and the observation value collected by the collection unit by the interpolation processing unit included in the observation value processing unit, Based on the position of the sensor, the weighting value of kriging acquired by the kriging unit, and the reliability set by the reliability setting unit, an interpolation process related to the position is performed on the observation values collected by the collection unit.

  (7) In order to solve the above-described problem, an observation value processing program according to one aspect of the present invention is an observation target in which a collection unit has a correlation with a position by a plurality of sensors provided at a plurality of different positions. The first step of collecting the observation values that are the results of detecting, and the reliability setting unit for each observation value collected by the collection unit or for each sensor, the distribution of the position of the sensor The second step of setting the uncorrelated reliability, and the observation value processing unit, based on the observation value collected by the collection unit, the position of the sensor and the reliability set by the reliability setting unit, It is a program for making a computer perform the 3rd step which performs the interpolation process regarding a position about the observation value collected by the said collection part.

  (8) According to one aspect of the present invention, in the observation value processing program according to (7), in the second step, the reliability setting unit includes a reliability determination unit included in the reliability setting unit. The reliability is determined based on the observation value collected by the collection unit and a predetermined reference value.

  (9) According to one aspect of the present invention, in the observation value processing program according to any one of (7) or (8) described above, in the third step, the observation value processing unit A kriging unit included in the observation value processing unit acquires a weighting value of kriging based on the observation value collected by the collection unit and the position of the sensor, and the interpolation processing unit included in the observation value processing unit includes the collection unit. The position of the observation value collected by the collection unit based on the observation value collected by the sensor, the position of the sensor, the weighting value of kriging acquired by the kriging unit, and the reliability set by the reliability setting unit Interpolation processing is performed.

  According to the present invention, the accuracy of observation value interpolation can be improved.

It is a block diagram which shows the schematic structure of the observation value processing apparatus which concerns on one Embodiment of this invention. It is a figure which shows schematic structure of the observation value sensor database which concerns on one Embodiment of this invention. It is a figure which shows schematic structure of the reliability database which concerns on one Embodiment of this invention. It is a figure which shows the example of the interpolation value at the time of performing the spatial interpolation (kriging) in consideration of the reliability based on one Embodiment of this invention.

Hereinafter, embodiments of the present invention will be described with reference to the drawings.
In the present embodiment, a plurality of sensors that detect the same physical quantity are arranged at different positions, and an observation value detected by one sensor and an observation detected by another sensor are detected with respect to the observation values detected by the plurality of sensors. An observation value that satisfies the condition that there is a correlation according to a difference in the arrangement (for example, distance) between these sensors is set as a processing target. More specifically, in this embodiment, an observation value that is a target in normal kriging is a processing target.
In the present embodiment, for example, when a plurality of sensors are arranged in a concentrated manner, or when the observation target is irrelevant to the sensor arrangement position (no correlation), the processing is not performed. Suppose there is.
In this embodiment, a spatial interpolation method based on kriging is used, but a spatial interpolation method based on another method that replaces kriging may be used.

[Description of Observation Value Processing Device According to this Embodiment]
FIG. 1 is a block diagram showing a schematic configuration of an observation value processing apparatus 1 according to an embodiment of the present invention.
The observation value processing apparatus 1 according to the present embodiment includes a collection storage unit 11 and an interpolation unit 12.
The collection storage unit 11 includes a collection unit 21 and an observation value sensor database 22.
The interpolation unit 12 includes a reliability determination unit 31, a reliability database 32, a kriging unit 33, and an interpolation processing unit 34.

In the present embodiment, a plurality (n) of sensors are arranged at different positions. Among these n sensors, there may be a sensor that detects (observes) the same physical quantity and a sensor that detects (observes) different physical quantities. In the present embodiment, physical quantities (observed values) detected by these n sensors are referred to as an observed value A1, an observed value A2, ..., an observed value An, respectively.
Here, various physical quantities may be used as the observation target. For example, temperature, humidity, amount of sunlight, amount of rainfall, wind direction, air volume, and the like can be used.

  The collection unit 21 is connected to each sensor or is connected to an external device that stores observation values detected by each sensor. And the collection part 21 inputs and collects the value (observation value A1-An) obtained by each sensor. The collection unit 21 stores (stores) the collected observation values A <b> 1 to An in the observation value sensor database 22.

The observation value sensor database 22 is a database that stores observation values and information related to sensors, and stores information (observation value information 1001) of observation values A1 to An input from the collection unit 21. Here, in this embodiment, the observation value from each sensor changes with time, and since the observation value from each sensor cannot always be acquired at the same time, the function of saving the observation value (observation value sensor database) 22) is provided in the observation value processing apparatus 1.
The observation value sensor database 22 stores information on each sensor (sensor information 1002) and information on the type of observation value input from each sensor (observation value type information 1003). Here, the sensor information 1002 and the observation value type information 1003 can be acquired based on information (sensor detailed information) input from an external device to the observation value processing device 1 as one configuration example. For example, the observation value processing apparatus 1 may be set in advance, or may be set in the observation value processing apparatus 1 by a user operation at an arbitrary timing.
The observation value sensor database 22 outputs the stored observation value information 1001, sensor information 1002, and observation value type information 1003 to the kriging unit 33.
In addition, the observation value sensor database 22 outputs the stored observation value information 1001 to the reliability determination unit 31.

FIG. 2 is a diagram showing a schematic configuration of the observation value sensor database 22 according to the embodiment of the present invention.
The observation value sensor database 22 according to the present embodiment includes a table (observation value information table) 101 for storing observation value information 1001, a table (sensor information table) 102 for storing sensor information 1002, and observation value type information 1003. A table (observation value type information table) 103 is stored.

The observation value information table 101 stores information in association with four items of observation value ID, observation date and time, observation value, and sensor ID.
The information of the observation value ID item is identification information (ID) given for each observation value. In the present embodiment, serial number information is assigned to each observation value as the observation value ID by the collection unit 21 or the observation value sensor database 22.
The information on the observation date / time item is information on the date / time of observation, and in this embodiment, information on the date / time when the observation value is detected by the sensor. In the present embodiment, information in the format of “A.D.-Month-Day T Time” is used as the observation date and time. As a specific example, “2013-07-07T12: 00” is 17:00 on July 7, 2013. Is represented by the information. In the present embodiment, observation date and time information is included in the observation values A <b> 1 to An by each sensor and extracted by the collection unit 21 or the observation value sensor database 22. As another configuration example, date information when the observation value is acquired by the collection unit 21 or the observation value sensor database 22 may be used as the observation date.

The information of the observation value item is information of the observation value itself, and is represented by numerical information such as “19.6” as a specific example. In the present embodiment, the information on the observation value itself is included in the observation values A1 to An by each sensor and extracted by the collection unit 21 or the observation value sensor database 22.
The information of the sensor ID item is identification information (ID) for each sensor that detects the observed value, and is information of a number such as “1” or “2” in the present embodiment. In the present embodiment, sensor ID information is included in the observation values A <b> 1 to An by each sensor and extracted by the collection unit 21 or the observation value sensor database 22. Based on the sensor ID, it is possible to identify the sensor in which the observation value is detected.

  Here, the observation date / time is assumed to be different with respect to the observation value acquired from each sensor, that is, the observation value from each sensor may not be acquired at the same time. For this reason, in the actual interpolation process, for example, the observed values from a plurality of sensors are processed together for a predetermined period (for example, a certain period). In this case, the observation date and time are required to specify the period.

The sensor information table 102 stores information in association with four items of sensor ID, latitude, longitude, and observation target ID.
The information of the sensor ID item is the same information as the sensor ID stored in the observation value information table 101.
The information on the latitude item is information indicating the latitude of the place (position) where the sensor is installed, and is expressed by numerical information such as “35.6” as a specific example.
The information on the item of longitude is information representing the longitude of the place (position) where the sensor is installed, and is represented by numerical information such as “139.7” as a specific example.
The information of the observation target ID item is identification information (ID) for each observation target (in this embodiment, an object for each type of observation value), and in this embodiment, a number such as “1” or “2”. Information.

Here, for example, these pieces of information are included in the detailed sensor information input to the observation value sensor database 22 of the observation value processing apparatus 1 from the outside, and are extracted by the observation value sensor database 22 or other information. It is set by the method.
As information indicating the location (position) where the sensor is installed, in this embodiment, information on latitude and longitude is used, but other information may be used.

The observation value type information table 103 stores information in association with two items of the observation target ID and the observation value type.
The information of the observation target ID item is the same information as the observation target ID stored in the sensor information table 102.
The information of the observation type item is information indicating the type of the observation value. As a specific example, information indicating the type of the observation value such as “temperature” and “humidity” (in the example of FIG. 2, Words).

Here, for example, these pieces of information are included in the detailed sensor information input to the observation value sensor database 22 of the observation value processing apparatus 1 from the outside, and are extracted by the observation value sensor database 22 or other information. It is set by the method.
In the present embodiment, by combining the content of the sensor information table 102 and the content of the observation value type information table 103, the observation position (latitude and longitude in which the sensor exists) and the observation value are compared with the observation value. Types can be associated.
As another configuration example, when there is only one type of observation value, storage and processing of information indicating the type of observation value (for example, information in the observation value type information table 103) may be omitted. Good.

The reliability determination unit 31 acquires the observation value information 1001 from the observation value sensor database 22, acquires the observation value from the acquired observation value information 1001, and obtains the reliability that is the probability of each acquired observation value. The obtained reliability information is stored (stored) in the reliability database 32.
Here, in the present embodiment, information other than the observation value distribution (data distribution) is used for the reliability. The reason for this is that Kriging performs interpolation by using the observed value distribution (in this embodiment, spatial interpolation). If the value related to the observed value distribution is used as the reliability, the observed value distribution is doubled. This is because it may be applied. Furthermore, when using a value that has a correlation with the observed value as the reliability, it can be handled by co-kriging (multivariate kriging) that uses an auxiliary variable (a variable that has a correlation with the observed value) in addition to the observed value and the position. Since this is possible, in this embodiment, a value that has no correlation with the distribution of observed values (in this embodiment, the positional distribution) is used as the reliability.

For this reason, in this embodiment, the reliability determination part 31 determines reliability based on the information (external reference information) used as the reference | standard input from an external apparatus.
Here, as the external reference information, for example, a true value corresponding to the observed value or an error from the true value is used (used). If a true value does not exist, it is assumed that instead, a value that is not a true value but is as accurate as possible (which is better) is used. As a specific example, when the observation target is air temperature, the temperature information of AMeDAS (common name of the regional weather observation system that is an unmanned observation facility of the Japan Meteorological Agency) with strict installation conditions can be used as external reference information. The reliability of the observed value can be obtained by comparing the observed value (temperature) of the sensor with the temperature of AMeDAS around the sensor. Usually, the greater the error between the observed value and the true value (or its substitute value), the lower the reliability, and various relationships may be used as the correspondence between the error and the reliability. .
As another configuration example, the reliability may be set in the observation value processing apparatus 1 by the user (by hand) for each observation value or for each sensor. In this configuration, the reliability determination unit 31 may not be provided.

The reliability database 32 is a database that stores reliability information, and stores reliability information (reliability information 1011) input from the reliability determination unit 31.
In addition, the reliability database 32 outputs the stored reliability information 1011 to the interpolation processing unit 34.

FIG. 3 is a diagram showing a schematic configuration of the reliability database 32 according to the embodiment of the present invention.
The reliability database 32 according to the present embodiment includes a table (reliability information table for each observation value) 201 that stores reliability information 1011 for each observation value.
The reliability information table 201 for each observation value stores information associated with two items of observation value ID and reliability.
The information of the observation value ID item is the same information as the observation value ID stored in the observation value information table 101 in the observation value sensor database 22.
The information of the item of reliability is reliability information, and is represented by numerical information such as “15.5” as a specific example.

Thus, in this embodiment, the reliability is set for each observation value ID. The reason why the reliability can be set and stored for each observation value ID, not for each sensor ID as in this embodiment, is that the reliability changes depending on the observation value ID even for the same sensor. This is to make it possible.
On the other hand, as another configuration example, a configuration in which the reliability is set for each sensor instead of for each observation value may be used. In this configuration, since it is not necessary to calculate the reliability for each observation value, if the reliability information set for each sensor is stored in a database (for example, the reliability database 32) and reused, for example, the reliability It is more efficient than setting and saving the degree every time.
In the present embodiment, since normalization is performed in the interpolation processing unit 34, there is no particular limitation on the level of reliability before normalization.

  The kriging unit 33 acquires the observation value information 1001, sensor information 1002, and observation value type information 1003 from the observation value sensor database 22, and outputs each observation value and its weight value (weight value) in kriging to the interpolation processing unit 34. Do (send). In the present embodiment, the kriging unit 33 specifies the type of observation value for the observation value sensor database 22 and handles the type of observation specified from the observation value sensor database 22 in order to handle only the observation value as the target variable. Get the value. Further, the kriging unit 33 acquires information on the position of the sensor (in this embodiment, the latitude and longitude in which the sensor exists) necessary for kriging from the observation value sensor database 22.

Here, in the present embodiment, in order to perform the interpolation considering the reliability, it is necessary to obtain the weight used in normal kriging, but it is not necessary to obtain the actually interpolated value.
In interpolation at an arbitrary position by kriging, a weighted average calculation shown in Expression (1) is performed.
In Expression (1), N represents the number (total number) of observation values to be subjected to kriging (in this embodiment, the types of observation values are the same), i represents an integer of 1 to N, s _i (in this embodiment, the latitude and longitude in the position of the sensor) is the i th observation position represent, Z (s _i) represents the observed value at observation positions s _i, lambda _i is weighted with respect to the observation position s _i Represents a value, and Σ represents a total sum of a predetermined range (in the example of Expression (1), i = 1 to i = N). In addition, s _o represents a predicted position (in this embodiment, a position where no sensor exists and is to be interpolated), and Z _krig (s _o ) represents a predicted value (interpolated value) at the predicted position s _o . .
In the present embodiment, among the observation values A1 to An from the n sensors shown in FIG. 1, N (N is the same as the observation target (type of observation value) detected by the sensor). It shows a case where the number is n).

In the present embodiment, the kriging unit 33 acquires observation value information 1001, sensor information 1002, and observation value type information 1003 from the observation value sensor database 22, and each observation value Z (s _i ) and each weighted value λ _{i in} kriging. Is output to the interpolation processing unit 34.
Here, each weight value λ _i is obtained by an operation according to the kriging by the kriging unit 33 (for example, an operation method using a variogram).

The interpolation processing unit 34 acquires each observed value Z (s _i ) and each corresponding weight value λ _i for each i in kriging from the kriging unit 33, acquires reliability information 1011 from the reliability determination unit 31, Based on these pieces of information, an interpolation value (predicted value) is obtained by interpolating the observed value for an arbitrary position (predicted position (in this embodiment, a position where no sensor exists and performs interpolation)). To get.
In addition, the interpolation processing unit 34 outputs the acquired interpolation value (predicted value). This interpolation value is used, for example, inside or outside the observation value processing apparatus 1. As a specific example, the observed value processing apparatus 1 or another apparatus uses a process for storing (storing) an interpolation value in a database or the like, a process for presenting (for example, displaying) the interpolation value to a user, or the like. Thus, it is possible to perform processing for executing other predetermined calculations.

Here, in the present embodiment, the interpolation processing unit 34 performs an interpolation process in consideration of the reliability for each observation value. The interpolation processing unit 34 performs interpolation using a method based on kriging, and specifically uses the reliability at the time of weighted average in kriging using equation (2), equation (3), and equation (4). Weighing is done.
In the formula (2), _{c i} represents the reliability of the observed value _{Z (s i), Z (} s o) represents a predicted value at the predicted position _{s o} (interpolated values).

Equation (3) is a constraint on the reliability c _i . In this constraint, the value calculated by Expression (2) is corrected by setting the average value of c _i λ _i (the total value in the example of Expression (3)) to 1. If all the reliability values are the same, the same value as that of normal kriging is obtained as the interpolation value. As described above, the influence on each observed value varies depending on the relative value when a certain reliability is compared with other reliability, not the value of the reliability alone.

Further, the interpolation processing unit 34 obtains the reliability c _i obtained by performing the normalization shown in the equation (4) in order to satisfy the constraint of the equation (3).
In Expression (4), j is used as a variable similar to i. In Expression (4), c ′ _i (or c ′ _j ) represents the reliability before normalization, and is determined by the reliability determination unit 31 and stored in the reliability database 32 in this embodiment. Represents the reliability value itself. That is, in the present embodiment, the interpolation processing unit 34 normalizes the reliability c ′ _i before normalization acquired from the reliability database 32, and uses the reliability c _i after normalization to predict the predicted value Z (s _o ) Is calculated.
Such normalization enables interpolation using reliability in weighted average in kriging. Thus, ultimately, the interpolation processing unit 34 obtains the predicted value in the predicted position _{s o} (interpolated value) Z _{(s o).}
As another configuration example, the reliability determination unit 31 or the reliability database 32 performs normalization of reliability, and the normalized reliability c _i acquired from the reliability database 32 by the interpolation processing unit 34 is used as it is. It may be configured.

FIG. 4 is a diagram illustrating an example of an interpolation value when spatial interpolation (kriging) is performed in consideration of reliability according to an embodiment of the present invention.
The table shown in FIG. 4, the observation position _{s i,} the horizontal position x, and vertical position y, the observed value Z _{(s i),} the coordinate position of the predicted position _{s o} (x, y) = Kriging weight λ _i in (2, 2), reliability c ′ _i before normalization, reliability c _i after normalization, and (x, y) = (2) after correction (normalization) , 2) are shown in association with examples of the weight values c _i λ _i . This is an example of an interpolation value at a specific point when the interpolation method considering the reliability according to the present embodiment is used.
In the example of FIG. 4, it is assumed that observed values Z (s ₁ ) to Z (s ₅ ) are acquired at five points (s _{1 to} s ₅ ) on the xy plane.
Note that the coordinate position (x, y) on the xy plane in the example of FIG. 4 is uniquely specified from the latitude and longitude in this embodiment, and substantially uses the latitude and longitude. It is the same.

In the example of FIG. 4, the coordinate position of the predicted position s _o (x, y) = is shown an example of weights lambda _i of Kriging in point (2,2), by kriging using equation (1) The obtained interpolation value Z _krig (s _{(2, 2)} ) at the coordinate position (x, y) = (2, ₂ ) was about 15.06. On the other hand, as in the present embodiment, the reliability c ′ _i is set for each observation value Z (s _i ), and the reliability c ′ _i (the one used is the normalized reliability c _i ). The interpolated value Z (s _{(2, 2)} ) at the coordinate position (x, y) = (2, ₂ ) was about 14.73.
The reason for this result is that the influence of the observation values Z (s ₁ ) and Z (s ₂ ) at the observation positions s ₁ and s ₂ with low reliability is relatively small in this embodiment. is there. In the present embodiment, for example, by appropriately setting the reliability using a value based on an error, it is possible to perform spatial interpolation with less error in consideration of the influence of the error.

As described above, the observation value processing apparatus 1 according to this embodiment includes the collection storage unit 11 that collects and stores information on observation value information and sensor installation status, and the observation values collected and stored. Interpolator 12 that outputs an interpolated value at a position other than the observed position using kriging from the above information, and interpolator 12 performs kriging using the reliability of each observed value.
Further, in the observation value processing apparatus 1 according to the present embodiment, the collection and storage unit 11 collects and stores information on the observation value and the observation position for the observation value that depends on (correlates with) the observation position.
In the observation value processing apparatus 1 according to the present embodiment, the kriging unit 33 of the interpolation unit 12 is fixed from the observation position based on information collected and stored (for example, observation value information and observation position information). For an arbitrary position within the range, the observed value and its position information are used to perform processing relating to kriging, thereby determining a weight value when calculating an interpolation value by kriging.

Further, in the observation value processing apparatus 1 according to the present embodiment, the interpolation unit 12 does not have a correlation (correlation in the spatial distribution) with the observation value itself, for example, as the reliability value indicating the certainty of the observation value. It is possible to use an independent value, a value based on an error of an observed value, or a value that can be arbitrarily set. As an example, the reliability determination unit 31 of the interpolation unit 12 determines the reliability based on the information collected and stored by the collection storage unit 11 and the external reference information.
Further, in the observation value processing apparatus 1 according to the present embodiment, the interpolation processing unit 34 uses the weighting value in kriging obtained by the kriging unit 33 and the reliability determined by the reliability determination unit 31, except for the observation position. The interpolation value at the position of is determined. The interpolation processing unit 34 acquires an interpolation value in consideration of the observed value, its position information, and the reliability.

  As described above, in the observation value processing apparatus 1 according to the present embodiment, when spatial interpolation is performed using kriging on observation values obtained from a plurality of sensors arranged at different positions, reliability ( For example, since the kriging is corrected by the reliability for each observation value or the reliability for each sensor, the accuracy of the interpolation can be improved as compared with kriging that does not consider the reliability. In the observed value processing apparatus 1 according to the present embodiment, for example, even if the observed values are distributed in exactly the same manner, a more accurate interpolation value is obtained based on the reliability by setting different reliability. It becomes possible.

[Specific Example of Effects Obtained by Observation Value Processing Device According to this Embodiment]
In the present embodiment, for example, a state is assumed in which the reliability that is the certainty can be acquired for each observation value. The observed value obtained from the sensor includes an error, and the error generation probability and the error size vary depending on the installation state (installation condition) of the sensor, the characteristics of each sensor, and the like. In the case of the temperature (observed value) obtained from the sensor installed in the alley, the observed value is affected by the height of the sensor installation location, the ground material, the air volume, and the like. In the case of AMeDAS installed by the Japan Meteorological Agency, there are fine rules regarding installation so that correct observations can be obtained, but in reality there are also sensors installed under looser conditions. As described above, the error of the observation value varies depending on the installation status of the sensor. For this reason, improvement in interpolation accuracy can be expected by affecting the reliability of the observation value itself during the interpolation process as in this embodiment. In the present embodiment, for a plurality of observed values, different reliability is set for the observed values themselves, and the observed values are interpolated in consideration of the reliability.

On the other hand, with the technique according to Patent Document 2 and the technique according to Patent Document 3, interpolation is performed by analyzing or transforming the observation value itself, but there is a limit to improving the accuracy only by that. For this reason, when another index can be used as in the present embodiment, an improvement in interpolation accuracy can be expected by using the index.
In addition, as an attempt to interpolate a target variable using a plurality of variables, co-kriging, which is a conventional technique, uses an auxiliary variable that has a correlation with the target variable (see, for example, Non-Patent Document 1). For example, when trying to interpolate the temperature, it is conceivable to use the date and time or the amount of solar radiation as the auxiliary variable. However, it can be said that the date and the amount of solar radiation are correlated with the temperature. In the technique according to Patent Document 3, an observation value distribution is used as the reliability. On the other hand, in the present embodiment, the variation in the observed value and the magnitude of the error in the observed value itself are used as the reliability. In general, since such variations and errors are not correlated with the target variable, it is considered that the effect of handling the reliability as in the present embodiment is low in co-kriging.
As described above, the conventional techniques of kriging and co-kriging cannot handle the reliability that is not correlated with the observed value, so that the interpolation accuracy is limited in that respect.

[Configuration example according to the above embodiment]
As one configuration example, a collection that collects observation values as a result of detecting an observation target (in the above embodiment, temperature, humidity, etc.) having a correlation with a position by a plurality of sensors provided at a plurality of different positions. And the distribution of the position of the sensor for each observation value (or for each sensor) collected by the collection unit (or the function of the collection unit 21 in the example of FIG. 1). Reliability setting unit (in the example of FIG. 1, the function of the reliability determination unit 31), the observed value collected by the collection unit, the position of the sensor, and the reliability set by the reliability setting unit On the basis of the observation value processing unit (in the above embodiment, N observation values), the observation value processing unit (in the example of FIG. 1, the kriging unit 33 and the interpolation processing) performs an interpolation process on the position. Function of the unit 34), (In the above embodiments, the observed value processing apparatus 1) observations processing apparatus including a.

  As one configuration example, in the observation value processing device, the reliability setting unit calculates the reliability based on the observation value collected by the collection unit and a predetermined reference value (external reference information in the example of FIG. 1). A determination unit for determining reliability (in the example of FIG. 1, the function of the reliability determination unit 31) is included.

As an example of the configuration, in the observation value processing apparatus, the observation value processing unit is configured to use a weighting value of kriging based on the observation value collected by the collecting unit and the position of the sensor (in the above embodiments, the weighting value λ _i ) (In the example of FIG. 1, the function of the kriging unit 33), the observed value collected by the collecting unit, the position of the sensor, the weighting value of the kriging obtained by the kriging unit, and the reliability An interpolation processing unit (in the example of FIG. 1, the function of the interpolation processing unit 34) that performs an interpolation process on the position of the observation values collected by the collection unit based on the reliability set by the degree setting unit .

  As one configuration example, the collection unit collects observation values that are a result of detecting an observation target having a correlation with the position by a plurality of sensors provided at a plurality of different positions, and the reliability setting unit collects the collection For each observation value collected by the unit or for each sensor, a reliability that is uncorrelated with the distribution of the position of the sensor is set, and the observation value processing unit collects the observation value collected by the collection unit, An observed value processing method (in the above embodiment, an observed value) that performs interpolation processing related to the position of the observed value collected by the collecting unit based on the position of the sensor and the reliability set by the reliability setting unit. A method performed in the processing apparatus 1).

  As an example of the configuration, in the observed value processing method, the reliability setting unit is configured such that the reliability determination unit included in the reliability setting unit has the reliability based on the observation value collected by the collection unit and a predetermined reference value. Determine the degree.

  As an example of the configuration, in the observation value processing method, the observation value processing unit is a weighting value of kriging based on the observation value collected by the collection unit and the position of the sensor by the kriging unit included in the observation value processing unit. Is set by the interpolation processing unit included in the observation value processing unit by the observation value collected by the collection unit, the position of the sensor, the weighting value of kriging acquired by the kriging unit, and the reliability setting unit. Based on the reliability, the position-related interpolation processing is performed on the observation values collected by the collection unit.

  As one configuration example, a first step in which the collection unit collects observation values that are a result of detecting an observation target having a correlation with a position by a plurality of sensors provided at a plurality of different positions, and reliability setting A second step of setting a reliability that is uncorrelated with the distribution of the position of the sensor for each observation value collected by the collection unit or for each sensor, and an observation value processing unit, A third step of performing an interpolation process on the position of the observation value collected by the collection unit based on the observation value collected by the collection unit, the position of the sensor, and the reliability set by the reliability setting unit. Is an observation value processing program for causing a computer to execute (in the above embodiment, a program for processing performed in the observation value processing device 1).

  As an example of the configuration, in the observation value processing program, in the second step, the reliability setting unit is configured such that the reliability determination unit included in the reliability setting unit includes an observation value collected by the collection unit and a predetermined value. The reliability is determined based on a reference value.

  As an example of the configuration, in the observation value processing program, in the third step, the observation value processing unit is configured such that the observation value collected by the collection unit and the position of the sensor by the kriging unit included in the observation value processing unit. The weighting value of kriging is acquired based on the observation value collected by the collecting unit, the position of the sensor, the weighting value of kriging acquired by the kriging unit, and the interpolation processing unit included in the observation value processing unit. Based on the reliability set by the reliability setting unit, the position-related interpolation processing is performed on the observation values collected by the collection unit.

[Specific application examples]
As an example, the configuration according to the above embodiment can be applied to a system in which specific devices are installed at a plurality of different positions and a sensor is provided in each of the plurality of specific devices.
Here, as a specific device, for example, a wireless base station device such as a mobile phone system can be used. In this case, for example, the function of the observation value processing apparatus 1 is provided at a higher level with respect to the plurality of base station apparatuses, and information on the observation values acquired by the sensors provided in each of the plurality of base station apparatuses is displayed. The data is acquired (received) and collected by wired or wireless communication with the base station apparatus provided with each sensor.

[Summary of the above embodiments]
As mentioned above, although embodiment of this invention was explained in full detail with reference to drawings, the concrete structure is not restricted to this embodiment, The design change etc. of the range which does not deviate from the summary of this invention are included.

  Further, as one configuration example, a program for realizing the functions of the apparatus according to the above-described embodiment (for example, the observation value processing apparatus 1) is recorded on a computer-readable recording medium and recorded on the recording medium. The processing can be performed by loading the executed program into a computer system and executing it.

The “computer system” referred to here may include an operating system (OS) and hardware such as peripheral devices.
The “computer-readable recording medium” means a flexible disk, a magneto-optical disk, a ROM (Read Only Memory), a writable nonvolatile memory such as a flash memory, a portable medium such as a DVD (Digital Versatile Disk), A storage device such as a hard disk built in a computer system.

Further, the “computer-readable recording medium” refers to a volatile memory (for example, DRAM (DRAM)) in a computer system that becomes a server or a client when a program is transmitted via a network such as the Internet or a communication line such as a telephone line. Dynamic Random Access Memory)) that holds a program for a certain period of time is also included.
The program may be transmitted from a computer system storing the program in a storage device or the like to another computer system via a transmission medium or by a transmission wave in the transmission medium. Here, the “transmission medium” for transmitting a program refers to a medium having a function of transmitting information, such as a network (communication network) such as the Internet or a communication line (communication line) such as a telephone line.
Further, the above program may be for realizing a part of the functions described above. Furthermore, what can implement | achieve the function mentioned above in combination with the program already recorded on the computer system, and what is called a difference file (difference program) may be sufficient.

DESCRIPTION OF SYMBOLS 1 ... Observation value processing apparatus, 11 ... Collection storage part, 12 ... Interpolation part, 21 ... Collection part, 22 ... Observation value sensor database, 31 ... Reliability determination part, 32 ... Reliability database, 33 ... Kriging part, 34 ... Interpolation processing unit 101 ... Observation value information table 102 ... Sensor information table 103 ... Observation value type information table 201 ... Reliability information table 1001 ... Observation value information 1002 ... Sensor information 1003 ... Observation value type information 1011 ... Reliability information

Claims (9)

A collection unit that collects observation values that are a result of detecting an observation target having a correlation with a position by a plurality of sensors provided at a plurality of different positions;
For each observation value collected by the collection unit or for each sensor, a reliability setting unit that sets a degree of reliability that is uncorrelated with the distribution of the position of the sensor;
An observation value processing unit that performs an interpolation process on the position of the observation values collected by the collection unit based on the observation values collected by the collection unit, the position of the sensor, and the reliability set by the reliability setting unit When,
An observation value processing apparatus comprising: The reliability setting unit includes a reliability determination unit that determines the reliability based on the observation value collected by the collection unit and a predetermined reference value.
The observation value processing apparatus according to claim 1. The observation value processing unit
A kriging unit that obtains a weighting value of kriging based on the observation value collected by the collection unit and the position of the sensor;
Observations collected by the collection unit based on the observation values collected by the collection unit, the position of the sensor, the weighting value of kriging acquired by the kriging unit, and the reliability set by the reliability setting unit An interpolation processing unit that performs an interpolation process on a position for a value,
The observation value processing apparatus according to claim 1 or 2. The collection unit collects observation values that are the results of detecting an observation target having a correlation with a position by a plurality of sensors provided at a plurality of different positions,
A reliability setting unit sets a reliability that is uncorrelated with the distribution of the position of the sensor for each observation value collected by the collection unit or for each sensor,
Interpolation processing related to the position of the observation value collected by the collection unit based on the observation value collected by the collection unit, the position of the sensor, and the reliability set by the reliability setting unit. I do,
Observation value processing method. The reliability setting unit determines the reliability based on an observation value and a predetermined reference value collected by the collection unit by a reliability determination unit included in the reliability setting unit.
The observation value processing method according to claim 4. The observation value processing unit obtains a weighting value of kriging based on the observation value collected by the collection unit and the position of the sensor by the kriging unit included in the observation value processing unit, and the observation value processing unit includes Based on the observation value collected by the collection unit, the sensor position, the weighting value of kriging acquired by the kriging unit, and the reliability set by the reliability setting unit by the interpolation processing unit, the collection unit Interpolate the position for the observations collected by
The observation value processing method according to any one of claims 4 and 5. A first step of collecting an observation value, which is a result of detecting an observation target having a correlation with a position by a plurality of sensors provided at a plurality of different positions;
A second step in which a reliability setting unit sets a reliability that is uncorrelated with the distribution of the position of the sensor for each observation value collected by the collection unit or for each sensor;
Interpolation processing related to the position of the observation value collected by the collection unit based on the observation value collected by the collection unit, the position of the sensor, and the reliability set by the reliability setting unit. A third step of performing
Observation processing program to make the computer execute. In the second step, the reliability setting unit determines the reliability based on an observation value collected by the collection unit and a predetermined reference value by a reliability determination unit included in the reliability setting unit. ,
The observation value processing program according to claim 7. In the third step, the observation value processing unit obtains a weighting value of kriging based on the observation value collected by the collection unit and the position of the sensor by the kriging unit included in the observation value processing unit, The observed value collected by the collecting unit, the sensor position, the weighting value of kriging acquired by the kriging unit, and the reliability set by the reliability setting unit by the interpolation processing unit included in the observed value processing unit On the basis of the position of the observation value collected by the collection unit,
The observation value processing program according to any one of claims 7 and 8.

Images