JP2021196330A - Diffusion source position estimation device and diffusion source position estimation method - Google Patents

Abstract

To provide a diffusion source position estimation device capable of accurately estimating a position of a diffusion source of a diffusion substance.SOLUTION: A diffusion source position estimation device 100 for estimating a position of a diffusion source of a diffusion substance diffused in the air in a predetermined area includes a data acquisition part 33 for acquiring concentration information of the diffusion substance from a plurality of observation devices arranged in the predetermined area, and a processing part 10 for processing the acquired data. The processing part 10 constructs a virtual space stimulating a predetermined area; arranges virtual observation points corresponding to the observation devices and a plurality of virtual diffusion sources diffusing the virtual diffusion substance into the virtual space, in the virtual space; determines diffusion source estimation indices indicating the magnitude of an influence degree given to the virtual observation points by the virtual diffusion substance, for each of the virtual diffusion sources, by a statistical method; repeats processing of rearranging an arrangement position of the virtual diffusion source until the diffusion source estimation index is maximized, and calculating the diffusion source estimation indices; and estimates the position where the diffusion source estimation index is maximized as the diffusion resource.SELECTED DRAWING: Figure 1

Description

The present invention relates to a diffusion source position estimation device and a diffusion source position estimation method.

When a substance (diffusing substance) that is harmful to the human body is released into the atmosphere due to an accident such as a crack in a pipe laid in a factory or chemical plant, the pipe is cracked (diffusion source) anywhere in the site. ) Must occur promptly and necessary measures must be taken to prevent the damage from spreading.

In Patent Document 1, a plurality of virtual diffusion sources are set, and the measured concentration (concentration information) at the actual observation point (measuring device) and the influence function value from the virtual diffusion source on the virtual observation point corresponding to the actual observation point are set. It is disclosed to estimate the virtual diffusion source with the smallest residual norm as the location of the diffusion source.

Japanese Patent No. 5433559

Patent Document 1 attempts to estimate the position of the diffusion source by using the so-called minimum residual method. From the formulation of this method, even if the influence function from the diffusion source position and the distribution tendency of the measured concentration (concentration information) at the actual observation point (measurement device) are different in the entire predetermined area, the distribution tendency is different within the predetermined area. The residual tends to be small when only the relatively large parts roughly match. Therefore, in this case, there is a problem that the position of the diffusion source is easily estimated and may not be appropriate as the position of the diffusion source.

The present invention has been made to solve the above-mentioned problems, and an object of the present invention is to provide a diffusion source position estimation device and a diffusion source position estimation method capable of accurately estimating the position of a diffusion source of a diffuser. And.

In order to achieve the above object, the diffusion source position estimation device of the present invention is a diffusion source position estimation device that estimates the position of the diffusion source of the diffusion substance diffused in the atmosphere in a predetermined area, and is arranged in the predetermined area. Information processing is provided with an information acquisition unit (for example, data acquisition unit 33) that acquires concentration information of diffused substances from a plurality of observation devices, and an information processing unit (for example, processing unit 10) that processes the acquired information. The department constructs a virtual space that simulates a predetermined area, and in the virtual space, arranges a virtual observation point corresponding to the observation device and a plurality of virtual diffusion sources that diffuse the virtual diffusion substance in the virtual space. Using a statistical method, a diffusion source estimation index indicating the high degree of reproducibility of the current state of the virtual diffusion material is obtained for each virtual diffusion source, and the placement position of the virtual diffusion source is rearranged until the diffusion source estimation index is maximized. The process of calculating the diffusion source estimation index is repeated, and the position where the diffusion source estimation index is maximized is estimated as the diffusion source. Other aspects of the invention will be described in embodiments described below.

According to the present invention, the position of the diffusion source of the diffusing substance can be estimated accurately.

It is a figure which shows the structure of the diffusion source position estimation apparatus which concerns on embodiment. It is a figure which shows the example of the virtual space corresponding to the actual predetermined area in the 1st estimation mode. It is a figure which shows the example of setting of the diffusion source candidate in a virtual space. It is a flowchart which shows the process (the 1) of the 1st estimation mode of the diffusion source position estimation apparatus. It is a flowchart which shows the processing (the 2) of the 1st estimation mode of the diffusion source position estimation apparatus. It is a figure which shows the example of setting of the diffusion source candidate in the 2nd estimation mode. It is a figure which shows the example of resetting the diffusion source candidate in the 2nd estimation mode. It is a figure which shows the example estimated as the diffusion source position in the 2nd estimation mode. It is a flowchart which shows the processing of the 2nd estimation mode of the diffusion source position estimation apparatus. It is a figure which shows the example of the reach of the diffusing material in a complex airflow field. It is a figure which shows the position of the virtual diffusion source in a predetermined area. It is a layout configuration diagram showing the actual observation point, the summit of Mt. Tsukuba, and the position of the actual diffusion source in the diffusion experiment. It is a layout drawing which shows the setting (the 1) of the virtual emission point of the comparative example. It is a layout drawing which shows the setting (the 2) of the virtual emission point of a comparative example. It is a layout drawing which shows the setting (the 3) of the virtual emission point of the comparative example. It is a layout drawing which shows the setting (the 4) of the virtual emission point of the comparative example. It is a layout drawing which shows the setting (the 5) of the virtual emission point of the comparative example. It is a layout drawing which shows the position of the virtual diffusion source set up to the estimation of the diffusion source position which concerns on embodiment. It is explanatory drawing which enlarged the area of the square thick line part of FIG. 18, and showed the transition of a diffusion source candidate.

First, the features of the present invention will be described.
Factories, chemical plants, etc. that are the targets of estimation in actual operation of the diffusion source location generally have an airflow field environment in which branches, mergers, vortices, etc. exist due to the influence of obstacles, topography, etc. Under such an airflow field environment, the distribution tendency of the diffuser over the entire predetermined area (the shape of the reach of the diffuser and the shade of the diffuser within the reach) strongly depends on the position of the diffuser. Therefore, the distribution tendency of the diffusing substance in the entire predetermined area is important information for estimating the position of the diffusing source. An example of the distribution tendency of the diffuser is shown in FIG.

FIG. 10 is a diagram showing an example of the reach of a diffuser in a complex airflow field. Under the airflow field where branching, merging, and vortices exist, the distribution tendency of the diffusing material is greatly different due to the slight change in the position of the diffusing source. In the case of FIG. 10A, the diffusing material reaches one side of the obstacle in the leeward of the diffusion source, and in the case of FIG. 10B, the diffusing material reaches both sides of the obstacle in the leeward of the diffusion source. is doing. The position of the diffusion source is slightly upwind in the case of FIG. 10 (b) than in the case of FIG. 10 (a), and the position of the diffusion source is an obstacle in the case of FIG. 10 (b). It is located on the extension of the axis of.

In the present invention, the range of reach of the diffusing material is obtained from the comparison between the measured concentration (concentration information) at the actual observation point (measuring device) and the influence function value from the virtual diffusion source at the virtual observation point corresponding to each actual observation point. The shape and the degree of matching of the relative shades of the diffuser within the reach are quantified and considered. Therefore, in the present invention, the position of the diffusion source of the diffusing substance can be estimated accurately.

Embodiments for carrying out the present invention will be described in detail with reference to the drawings as appropriate.
FIG. 1 is a diagram showing a configuration of a diffusion source position estimation device 100 according to an embodiment. The diffusion source position estimation device 100 includes a processing unit (information processing unit) 10, a storage unit 20, an input unit 31, an output unit 32, and a data acquisition unit 33. The processing unit 10 is an observation information acquisition unit 11 that acquires observation information via a data acquisition unit (information acquisition unit) 33, a virtual diffusion source setting unit 12 that sets a virtual diffusion source in a predetermined area, and a virtual grid setting unit 13. , The influence function calculation unit 14, the diffusion source estimation index calculation unit 15, and the diffusion source position estimation unit 16.

The storage unit 20 contains information 21 (concentration information and position information) at the observation point, position information 22 of the virtual diffusion source, an influence function 23 on the virtual observation point, and a first diffusion source estimation index 24 for each virtual diffusion source ( The index P ^m ), the second diffusion source estimation index 25 (index V ^m ) for each virtual diffusion source, the diffusion source estimation index 26 based on the index P ^m and the index V ^{m (index Z m} ), and the like are stored.

The observation information acquisition unit 11 acquires the measurement data of each observation device through the wired or wireless network via the data acquisition unit 33, and stores the acquired measurement data in the information 21 at the observation point of the storage unit 20.

The virtual diffusion source setting unit 12 sets a plurality of virtual diffusion sources at arbitrary positions in a predetermined area of the virtual space, and stores the set positions in the position information 22 of the virtual diffusion source of the storage unit 20.

In the second estimation mode described later, the virtual grid setting unit 13 constructs a rectangular grid in which grid points are arranged centered on the "diffusion source candidate" specified in the first estimation mode described later. A "virtual diffusion source" is set for each grid point, and this is stored in the position information 22 of the virtual diffusion source of the storage unit 20.

The influence function calculation unit 14 calculates the influence function value and stores it in the influence function value 23 that gives the virtual observation point of the storage unit 20. The influence function value is calculated from the position information of the virtual diffusion source, the position information of the observation point in the virtual space (position information of the observation device), and the elapsed time since the diffuser started virtual diffusion from the virtual diffusion source. Will be done. The detailed calculation method of the influence function value will be described later.

The diffusion source estimation index calculation unit 15 uses the influence function value 23 on each virtual observation point and the concentration information at each virtual observation point to obtain a virtual diffusion substance for each virtual diffusion source by a statistical method. ^{An index P m} indicating the degree of agreement of the reach range ^{and an index V m} indicating the degree of agreement of the relative shades of the virtual diffuser are calculated, and these are used as the first diffusion source estimation index 24 and the second of the storage unit 20. It is stored in the diffusion source estimation index 25 of. Then, the diffusion source estimation index calculation unit 15 subtracts the second diffusion source estimation index 25 from the first diffusion source estimation index 24 after normalizing the values in the virtual diffusion source to be evaluated in the corresponding calculation step. Then, the index Z ^m is calculated, and this is stored in the diffusion source estimation index 26 of the storage unit 20.

The diffusion source position estimation unit 16 identifies a virtual diffusion source that maximizes the diffusion source estimation index 26.

In the above configuration, the processing unit 10 constructs a virtual space simulating a predetermined area, and in the virtual space, a plurality of virtual diffusions that diffuse a virtual observation point corresponding to an observation device and a virtual diffusion substance in the virtual space. Place the source and. ^{After placement, a diffusion source estimation index (index Z m} ) indicating the magnitude of the influence of the virtual diffusion substance on the virtual observation point is obtained for each virtual diffusion source by a statistical method, and the diffusion source estimation index is maximized. Until then, the process of rearranging the placement position of the virtual diffusion source and calculating the diffusion source estimation index is repeated, and the position where the diffusion source estimation index is maximized is estimated as the diffusion source.

That is, in the present embodiment, an existing observation device and a virtual diffusion source that virtually diffuses a diffusing substance are arranged in the virtual space. After placement, the diffusion source candidate is obtained based on the virtual diffusion source and the concentration information at the observation point based on the statistical method, the virtual diffusion source is rearranged, the diffusion source candidate is obtained again, and the value of the diffusion source candidate is large. It is assumed that the direction of becoming is the direction in which the diffusion source exists. Then, this process is repeated until the value of the diffusion source candidate is not updated, and the position of the diffusion source candidate when the value is not updated is estimated as the diffusion source (diffusion source position). This makes it possible to accurately estimate the position of the diffusion source of the diffusing substance.

Hereinafter, a specific example will be described.
The diffusion source position estimation device 100 has a first estimation mode and a second estimation mode.
<First estimation mode>
The first estimation mode is a mode for narrowing down where the diffusion source position is located in a predetermined area.

FIG. 2 is a diagram showing an example of a virtual space corresponding to an actual predetermined area in the first estimation mode. This will be described with reference to FIG. 1 as appropriate. It is assumed that a diffusing substance diffused into the atmosphere is released from a certain point in a predetermined area of real space. However, at first, it is not known from where in the predetermined area the diffusing substance is released, that is, the position of the diffusing source.

(S1) An observation device for measuring actual observation information (concentration information of diffusing substances in the atmosphere) is installed at an arbitrary position in a predetermined area. The number of observation devices installed shall be at least two.
In the example of FIG. 2, five observation devices (observation devices 1 to 5) are installed in a predetermined area. The observation information acquisition unit 11 acquires the measurement data of each observation device through a wired or wireless network.

(S2) A virtual space corresponding to an actual predetermined area is constructed. At that time, as shown in FIG. 2, a specific position in a predetermined area is set as an origin (0,0), and the device is installed in an actual predetermined area. The position information (X-axis and Y-axis position information from the origin) of each observation device is set in the virtual space. That is, the processing unit 10 constructs a virtual space simulating a predetermined area. The virtual space has coordinate axis information indicating that one point where the horizontal axis and the vertical axis of the surrounding frame intersect is the origin, the horizontal axis direction is the X axis, and the vertical axis is the Y axis direction.

(S3) The virtual diffusion source setting unit 12 sets a plurality of virtual diffusion sources at arbitrary positions in a predetermined area of the virtual space. In the example shown in FIG. 2, four virtual diffusion sources (virtual diffusion sources 1 to 4) are set in a predetermined area. That is, the processing unit 10 arranges a virtual observation point corresponding to the observation device and a plurality of virtual diffusion sources for diffusing the virtual diffuser corresponding to the diffuser at an arbitrary position in the virtual space.

The influence function calculation unit 14 calculates the influence function value in each virtual diffusion source. As described above, the influence function value is calculated from the position information of the virtual diffusion source and the position information of the virtual observation point, and the elapsed time from the start of virtual diffusion of the diffuser from the virtual diffusion source. In FIG. 2, for example, the influence function values on the observation devices 1 to 5 are calculated for the virtual diffusion source 1.

That is, the processing unit 10 obtains the concentration information at the virtual observation point with respect to the virtual diffusion substance diffused from the virtual diffusion source by using the atmospheric diffusion calculation model. In order to obtain the influence function value, diffusion source information (position and emission amount) is required in addition to topographical information and meteorological information as the calculation conditions. However, since the diffusion source information is not known, tentative values are used for the diffusion source information (position and emission amount). As the atmospheric dispersion calculation model, a plume puff model, an in-cell particle method, a Lagrangian particle model, and the like are known.

At the stage where the concentration (influence function) of the virtual diffuser changes with time in the virtual space after the virtual diffuser starts diffusing, the elapsed time from the start of diffusion of the virtual diffuser is not taken into consideration. Using a stationary model, it is sufficient time after the start of diffusion until the concentration (influence function) of the virtual diffuser does not change over time in the situation where the weather conditions and the amount of the virtual diffuser released do not change in the virtual space. Use a stationary model that does not require elapsed time at the stage where it can be considered that has elapsed.

(S4) Using the concentration information of each observation device measured in (S1) and the influence function value obtained in (S3) above, the diffusion source estimation index calculation unit 15 in each virtual diffusion source uses the diffusion source estimation index calculation unit 15. Calculate the diffusion source estimation index. The diffusion source estimation index is obtained by obtaining the following two quantities (index P ^m and index V ^m ) and using a statistical calculation method called principal component analysis to obtain these values. The index P ^m and the index V ^m are shown below.

The number of observation points where the concentration information of the observation device ≠ 0 and the influence function value ≠ 0 corresponding to the position of the concentration information, and the concentration information of the observation device = 0 and the influence function corresponding to the position of the concentration information Value obtained by dividing the sum of the number of observation points where the value = 0 by the total number of observation points: index P ^{m
Further, P m *} normalized so that the average 0 and the variance 1 in the entire set ^{of the indexes P m} in the virtual diffusion source set in the current step (first estimation mode) is obtained. Here, the index P ^m and the index P ^{m *} are indexes indicating the degree of agreement between the reach of the diffusing substance and the virtual diffusing substance.

The value obtained by dividing the influence function value in which the concentration information of the observation device ≠ 0 and the value of the influence function corresponding to the position of the concentration information ≠ 0 by the corresponding concentration information is converted by the common logarithm as necessary. Set dispersion: index V ^{m
Further, V m *} normalized so that the average 0 and the variance 1 in the entire set ^{of the index V m} in the virtual diffusion source set in the current step (first estimation mode) is obtained. Here, the index V ^m and the index V ^{m *} are indexes indicating the degree of matching of the relative shades of the diffusing substance and the virtual diffusing substance.

Then, the diffusion source estimation index calculation unit 15 obtains the ^{index Z m by the following equation.}
Z ^m = P ^{m *} -V ^{m *}
In the example shown in FIG. 2, the diffusion source estimation index is calculated for each of the virtual diffusion sources 1 to 4. Among them, the one having the maximum value of the diffusion source estimation index is specified, and it is set as the "diffusion source candidate" which is the diffusion source position candidate in the next step.

FIG. 3 is a diagram showing an example of setting a diffusion source candidate in a virtual space. In the example of FIG. 3, it is assumed that the diffusion source estimation index of the virtual diffusion source 1 shows the maximum value.

(S5) The diffusion source position estimation unit 16 excludes other virtual diffusion sources (virtual diffusion sources 2 to 4 shown in FIG. 3) from the diffusion source position candidates.
The processing of the first estimation mode described above will be further described with reference to FIG.

FIG. 4 is a flowchart showing processing (No. 1) of the first estimation mode of the diffusion source position estimation device. The processing unit 10 obtains observation information (S11), constructs a virtual space simulating a predetermined area (S12), and sets a virtual diffusion source (S13). Then, the processing unit 10 calculates an influence function value for each virtual diffusion source (S14), ^{and calculates an index Z m} which is a diffusion source estimation index for each virtual diffusion source (S15).

The processing unit 10 sets the largest of the calculated diffusion source estimation indexes as the “diffusion source candidate” (S16), and excludes virtual diffusion sources other than the “diffusion source candidate” from the “diffusion source position” candidates. (S17), and the processing of the first estimation mode is terminated.

FIG. 5 is a flowchart showing the processing (No. 2) of the first estimation mode of the diffusion source position estimation device. FIG. 5 is a modified example of the processing of the first estimation mode. As shown in FIG. 5, after obtaining the observation information (S11), the processing unit 10 may set the position of the observation point where the observation concentration is maximum as the “diffusion source candidate” (S16A). As a result, the processing time can be shortened.

<Second estimation mode>
The second estimation mode is a mode for further estimating the diffusion source position in a predetermined area based on the "diffusion source candidate" in the first estimation mode.

FIG. 6 is a diagram showing an example of setting a diffusion source candidate in the second estimation mode.
(S1A) The virtual grid setting unit 13 constructs a rectangular grid in which grid points are arranged centered on the "diffusion source candidate" specified in the first estimation mode, and "virtual" is assigned to each of the grid points. Set the "diffusion source". Next, the influence function calculation unit 14 calculates the influence function value in each virtual diffusion source. Then, the diffusion source estimation index calculation unit 15 calculates the diffusion source estimation index in each of the virtual diffusion sources.

(S2A) The diffusion source position estimation unit 16 compares the diffusion source estimation index and the diffusion source estimation index of the diffusion source candidate in all the calculated second virtual sources, and the virtual diffusion in which the diffusion source estimation index is maximized. Identify the location of the source. In the example shown in FIG. 6, it is assumed that the diffusion source estimation index in the virtual diffusion source 3 on the upper right is the maximum (maximum virtual diffusion source).

Here, FIG. 6 is an example in which the diffusion source candidate and the maximum virtual diffusion source do not match. In this case, the maximum virtual diffusion source is selected as the diffusion source candidate until the diffusion source candidate and the maximum virtual diffusion source match. By replacing with, the virtual lattice and the virtual diffusion source are reset so that the diffusion source candidate becomes the center, and the step of obtaining the diffusion source estimation index with the reset diffusion source candidate and the virtual diffusion source as the evaluation target is repeated, and the diffusion source is repeated. The position where the candidate and the maximum virtual diffusion source match is estimated as the "diffusion source position". In FIG. 6, due to the above relationship, the diffusion source estimation index in the virtual diffusion source 3 on the upper right is maximized, so the virtual grid is reset so that the virtual diffusion source 3 is at the center (see FIG. 7).

FIG. 7 is a diagram showing an example of resetting the diffusion source candidate in the second estimation mode.
(S3A) The virtual diffusion source setting unit 12 replaces the “virtual diffusion source (= maximum virtual diffusion source)” with the maximum diffusion source estimation index in (S2A) with a “diffusion source candidate”. Other virtual diffusion sources and diffusion source candidates before resetting are excluded from the diffusion source location candidates. In the subsequent calculation of the diffusion source estimation index, the virtual diffusion source excluded from the diffusion source position candidates and the diffusion source candidate before replacement are excluded from the calculation target of the diffusion source estimation index.

The flow of (S1A) to (S3A) is repeated. Until the following loop escape condition is satisfied, the "maximum virtual diffusion source" is replaced with the "diffusion source candidate", and this flow is repeated.
(Loop escape condition): "Diffusion source candidate" and "Maximum virtual diffusion source" match.
In FIG. 7, it is assumed that the diffusion source estimation index in the virtual diffusion source 3 on the upper right is the maximum (maximum virtual diffusion source) as in FIG.

FIG. 8 is a diagram showing an example of estimating the diffusion source position in the second estimation mode.
(S4A) In FIG. 8, when the diffusion source estimation index in the replaced diffusion source candidate becomes maximum, that is, when the loop escape condition is satisfied, the diffusion source position estimation unit 16 determines that this position is the “diffusion substance”. It is presumed to be the "diffusion source position", and the processing of the second estimation mode is terminated.
The processing of the second estimation mode described above will be further described with reference to FIG.

FIG. 9 is a flowchart showing the processing of the second estimation mode of the diffusion source position estimation device. The processing unit 10 obtains the observation information (S21), constructs a rectangular grid in which grid points are arranged around the diffusion source candidates (S22), and rearranges the virtual diffusion source at each grid point of the rectangular grid. (S23).

The processing unit 10 calculates the influence function value for each virtual diffusion source (S24), calculates the diffusion source candidate and the diffusion source estimation index for all the virtual diffusion sources (S25), and calculates among the virtual diffusion sources. The one with the largest diffusion source estimation index is estimated as the "maximum virtual diffusion source" (S26).

The processing unit 10 determines whether or not the "diffusion source candidate" and the "maximum virtual diffusion source" match (S27), and if they do not match (S27, No), the "maximum virtual diffusion source" is selected as the "diffusion source candidate". (S29), and the process returns to S22. On the other hand, when the "diffusion source candidate" and the "maximum virtual diffusion source" match (S27, Yes), the position of the "diffusion source candidate" is estimated to be the "diffusion source position" (S28), and the second estimation mode is set. finish.

The diffusion source position estimation device 100 has been described above, but a calculation example will be shown below.
(Calculation example)
FIG. 11 is a diagram showing the position of the virtual diffusion source in the predetermined area. In FIG. 11, four virtual diffusion sources are arranged, and observation points 1 to 15 are arranged around them. Using the concentration information of each observation point measured at each observation point and the influence function value, the diffusion source estimation index calculation unit 15 calculates the diffusion source estimation index at each virtual diffusion source. Table 1 shows a list of arrangement coordinates of each observation point (gas concentration sensor) and observation values. Table 2 shows a list of position coordinates of each observation point.

Then, the value of the influence function on each observation point in each virtual diffusion source is obtained, and the list of the obtained influence function values is shown in Table 3.

Next, for each virtual diffusion source number m (here, 1 to 4), three quantities (indexes P ^m , V ^m , Z ^m ) are obtained, and these values are used as a statistical calculation method called principal component analysis. The procedure for obtaining the diffusion source estimation index by using it is shown.

As described above, the index P ^m is the number of observation points where the concentration information of the observation device is ≠ 0 and the influence function value corresponding to the position of the concentration information is ≠ 0, and the concentration information of the observation device is 0 and the index P m is described above. It is a value obtained by dividing the sum of the number of observation points at which the value of the influence function corresponding to the position of the concentration information = 0 by the total number of observation points. Here, the index P ^m is an index indicating the degree of agreement between the reach of the diffusing substance and the virtual diffusing substance.

As described above, the index V ^m is the concentration information corresponding to the concentration information ≠ 0 of the observation device and the influence function value ≠ 0 corresponding to the position of the concentration information in a certain virtual diffusion source. It is the concentration of the set of values obtained by converting the divided values by the common logarithm. Here, the index V ^m is an index indicating the degree of matching of the relative shades of the diffusing substance and the virtual diffusing substance.

(Calculation of indicators P ^m and V ^m)
Here, an example of obtaining ^{P 1} and V ¹ for the virtual diffusion source 1 (virtual diffusion source number m = 1) is shown. Table 4 shows the observed values extracted from Table 1 and the values of the influence function of the virtual diffusion source 1 extracted from Table 3.

P ¹ can be obtained by simply comparing the values in Table 4, counting up, and dividing by the total number of observation points. There are four observation points, observation points 1, 5, 10, and 13, which satisfy the observation value = 0 and the value of the influence function = 0, and the number of observation points satisfying the observation value ≠ 0 and the value of the influence function ≠ 0. Is 8 points of observation points 3, 4, 7, 8, 9, 11, 12, and 14.
Therefore, P ¹ = (4 + 8) / 15 = 12/15.

For V ¹ , first, the observed value ≠ 0 and the value of the influence function ≠ 0 are the observation points 3, 4, 7, 8, 9, 11, as seen when ^{P 1 was derived.} There are 8 points of 12 and 14. Values / observations of influence function at these observation points examined a set of common logarithm converted value, the variance (indicative of statistics) becomes V ^1. The set is as follows.
(0.380, -0.332, 0.020, -0.568, -0.418, -0.386, 1.959, 0.247)

ここで、ｎは要素数、μは集合の平均値である。 The variance V of the set is represented by equation (1).

Here, n is the number of elements and μ is the average value of the set.

Therefore, V ¹ = 0.585.
It should be noted that n = 8 and μ = 0.115.
Thus, the virtual diffusion source 1 is ^P 1, ^{V 1} was calculated.

Hereinafter, when the other virtual diffusion sources are obtained in the same manner, they are a set ^{of P m} from the virtual diffusion sources 1 to 4 which are the virtual diffusion sources to be evaluated in the current step ^{(P 1} , P ² , P ³ , P). ⁴ ) is
(12/15, 13/15, 13/15, 14/15)
Is.
The set ^{of V m (V 1} , V ² , V ³ , V ⁴ ) from the virtual diffusion sources 1 to 4, which are the virtual diffusion sources to be evaluated in the current step, is
(0.585, 0.780, 1.489, 0.692)
Is.

Next, the set is normalized so that it becomes a set with a variance of 1 and a mean of 0, and (P ^{1 *} , P ^{2 *} , P ^{3 *} , P ^{4 *} ), (V ^{1 *} , V ^{2 *} , V ^{3 *} , V ^{4 *} ) is obtained from the equations (2) and (3).

From the above, if you apply specific numerical values,
The set ^{of indexes P m * (P 1 *} , P ^{2 *} , P ^{3 *} , P ^{4 *} ) from the virtual diffusion sources 1 to 4, which are the virtual diffusion sources to be evaluated in the current step, is
(-1.414, 0.0000, 0.0000, 1.414)
Is.
The set ^{of indexes V m * (V 1 *} , V ^{2 *} , V ^{3 *} , V ^{4 *} ) from the virtual diffusion sources 1 to 4, which are the virtual diffusion sources to be evaluated in the current step, is
(-0.850, -0.300, 1.699, -0.549)
Is.

式（４）において、指標Ｐ^ｍ＊がより大きいほど、拡散物質および仮想拡散物質の到達範囲の一致度が高いことを示し、指標Ｖ^ｍ＊がより小さいほど、拡散物質と仮想拡散物質の相対的な濃淡の一致度が高いことを示す。従って、指標Ｐ^ｍ＊が大きく、かつ、指標Ｖ^ｍ＊が小さいほど、拡散源推定指標である指標Ｚ^ｍは大きくなり、指標Ｚ^ｍが大きいほど、対応する仮想拡散源が拡散源位置に近いことを示す。 (Calculation of index Z ^m)
The index Z ^m is obtained from the equation (4).

In the formula (4), the ^{larger the index P m *, the} higher the degree of agreement between the reach of the diffusing substance and the virtual diffusing substance, and the smaller the ^{index V m *, the relative the diffusing substance and the virtual diffusing substance.} It shows that the degree of matching of shades is high. Therefore, the ^{larger the index P m * and the smaller the index V m *} , the larger the ^{index Z m,} which is the diffusion source estimation index, and the larger the index Z ^m , the closer the corresponding virtual diffusion source is to the diffusion source position. Show that.

Applying a specific numerical value to equation (4),
The set ^{of indexes Z m (Z 1} , Z ² , Z ³ , Z ⁴ ) from the virtual diffusion sources 1 to 4, which are the virtual diffusion sources to be evaluated in the current step, is
(-0.565, 0.300, -1.699, 1.964)
Is.
The one having the largest diffusion source estimation index in the set of indexes Z ^m is defined as the central grid (diffusion source candidate) of the grid setting in the next step (for example, the second estimation mode). Since what the maximum among the diffusion source estimation indicators of the is a Z ^4, virtual diffusion source 4 is defined as a diffusion source candidate.

(Verification using diffusion experiment of sulfur hexafluoride)
Next, a diffusion experiment of sulfur hexafluoride in the area around Mt. Tsukuba conducted by the Japan Atomic Energy Research Institute on November 6, 1985 from 15:00 to 16:30 (airflow at the eastern foot of Mt. Tsukuba) shown in Non-Patent Document 1. A calculation example showing the diffusion source position by the diffusion source position estimation device 100 according to the embodiment using the observed value in (branch generation) will be described together with a comparative example. The estimation method of the comparative example is a method of estimating the position of the diffusion source by using the minimum residual method shown in Patent Document 1. In both the method of this embodiment and the method of the comparative example, "Kakusansuket" (registered trademark) detailed numerical calculation function (Hitachi Power Solutions Co., Ltd.) was used to calculate the influence function value.

[Non-Patent Document 1] Michio Tsunoda, Takashi Hayashi, Takashi Adachi, "Atmospheric Dispersion Experiment Data Vol.3 TSUKUBA84, TSUKUBA85" Japan Atomic Energy Research Institute, (1988)

FIG. 12 is an arrangement configuration diagram showing actual observation points (〇 and ●), the summit of Mt. Tsukuba (▲), and the actual position of the diffusion source (◇) in the diffusion experiment. The predetermined area of the diffusion experiment is an area of 15,000 m in each of the east-west direction and the north-south direction with the summit of Mt. Tsukuba as the center point. For convenience, the east-west direction is set as the X coordinate (the positive direction is the east), and the north-south direction is set as the Y coordinate (the positive direction is the north). In this diffusion experiment, there are 41 actual observation points in the predetermined area, and among the 41 actual observation points, the actual observation point with the maximum actual concentration is indicated by ●, and the other actual observation points are indicated by 〇. ..

FIG. 13 is a layout diagram showing the setting (No. 1) of the virtual emission point of the comparative example. FIG. 14 is a layout diagram showing the setting (No. 2) of the virtual emission point of the comparative example. FIG. 15 is a layout diagram showing the setting (No. 3) of the virtual emission point of the comparative example. FIG. 16 is a layout diagram showing the setting (No. 4) of the virtual diffusion source of the comparative example. FIG. 17 is a layout diagram showing the setting (No. 5) of the virtual emission point of the comparative example. With reference to FIGS. 13 to 17, the narrowing down of virtual emission points will be described. The virtual emission points are indicated by x in FIGS. 13 to 17. Further, the "virtual emission point" of Patent Document 1 corresponds to the "virtual diffusion source".

As shown in Fig. 13, the actual position of the diffusion source is unknown, so to cover the actual observation points, the first stage (No. 1) is virtual at intervals of 4500 m from north to south and east to west around the summit of Mt. Tsukuba. A grid was set, and 9 virtual emission points were set on the grid points. At each virtual emission point, the effect function value on the actual observation point is obtained. The north, south, east and west are the first stage, centering on the virtual emission point where the residual norm between the measured concentration (concentration information) and the influence function value from the virtual diffusion source on the virtual observation point corresponding to the actual observation point is minimized. The virtual emission point is set so that it is half the size (2250 m) (see Fig. 14, Part 2), and this enforcement is repeated until the setting of the 5th stage virtual grid, and the setting is made on the 5th stage virtual grid. Of the virtual emission points (see FIG. 17, No. 5), the virtual emission point with the smallest residual norm was estimated as the emission point.

In the example of the comparative example, the residual was minimized at the virtual emission point near the actual diffusion source in the trial of the second stage (No. 2), but the residual was after the implementation of the third stage (No. 3). The position of the virtual emission point that becomes the minimum moves away from the actual diffusion source, and finally the residual becomes the minimum at the virtual emission point near the actual observation point where the measured concentration becomes the maximum, and this position is used as the emission point. Estimated.

As a result, it was estimated that the source location was near the summit of the mountain range on the north side of the eastern foot basin of Mt. Tsukuba. The estimated position was 2416 m away from the actual source (diffusion source position).

The reason for this is that even if the distribution tendency of the influence function value from the diffusion source position and the measured concentration (concentration information) at the actual observation point (measuring device) in the entire predetermined area is different, it is relatively large within the predetermined area. It is considered that this is due to the problem caused by the formulation of the prior art that the residual tends to be small when only the above-mentioned parts roughly match.

Next, it is a calculation example which shows the diffusion source position by the diffusion source position estimation apparatus 100 which concerns on embodiment.
FIG. 18 is a layout diagram showing the positions of the virtual diffusion sources set up to the estimation of the diffusion source position according to the embodiment. FIG. 19 is an explanatory diagram showing the transition of diffusion source candidates while enlarging the region of the thick square line portion of FIG. 18 (the region where the X and Y coordinates are 6000 m to 12000 m). Virtual diffusion sources are indicated by x in FIGS. 18 and 19.

In this embodiment, according to FIG. 5, the position of the actual observation point where the measured concentration is maximum is set as the diffusion source candidate, and the calculation is performed using a rectangular grid having 400 m in each of the east-west and north-south directions according to FIG. The vicinity of the position of the diffusion source was estimated as the diffusion source position.

As a result, it was possible to estimate the source location in the basin at the eastern foot of Mt. Tsukuba. A position 217 m away from the actual source could be estimated as the source position. As described above, according to the diffusion source position estimation device 100 of the present embodiment, the position of the diffusion source of the diffusing substance can be estimated accurately as compared with the estimation method of the comparative example.

According to FIG. 5, the diffusion source candidate first defined in the present application is in the vicinity of the position estimated as the emission point in the example of the comparative example, and as a result, from the second stage (No. 2) of the comparative example. The transition of the virtual diffusion source in which the residual norm is minimized in the fifth stage (No. 5) and the transition of the diffusion source candidate in each calculation step in FIG. 9 of the present application take the opposite path, and in the present application, they are actually It is considered that the vicinity of the position of the diffusion source is estimated as the position of the diffusion source, and the problem in the comparative example is solved in this embodiment.

The diffusion source position estimation device 100 described above has the following features.
A diffusion source position estimation device 100 that estimates the position of a diffusion source of a diffuser that has diffused into the atmosphere in a predetermined area, and data acquisition that acquires concentration information of the diffuser from a plurality of observation devices arranged in the predetermined area. A unit 33 and a processing unit 10 are provided.

The processing unit 10 constructs a virtual space simulating a predetermined area (S12), and in the virtual space, a virtual observation point corresponding to an observation device and a plurality of virtual diffusion sources for diffusing a virtual diffusion substance in the virtual space. ^{(S13), a diffusion source estimation index (index Z m} ) indicating the high degree of reproducibility of the current state of the virtual diffusion substance is obtained for each virtual diffusion source (S15), and the diffusion source estimation index is used. Until it is maximized, the process of rearranging the placement position of the virtual diffusion source and calculating the diffusion source estimation index is repeated (S22 to S27), and the position where the diffusion source estimation index is maximized is estimated as the diffusion source (S22 to S27). S28). This makes it possible to accurately estimate the position of the diffusion source of the diffusing substance.

In the first estimation mode, the processing unit 10 sets one of the virtual diffusion sources as a diffusion source candidate by a predetermined method (S16), and in the second estimation mode, a rectangle centered on the diffusion source candidate. A grid is constructed (S22), virtual diffusion sources are rearranged at the grid points of each rectangular grid (S23), the diffusion source estimation index in the diffusion source candidates is recalculated (S25), and the rearranged virtual diffusion is performed. The diffusion source estimation index is calculated for each source, the largest one is estimated as the maximum virtual diffusion source (S26), and the maximum virtual diffusion source is replaced with the diffusion source candidate until the maximum virtual diffusion source and the diffusion source candidate match. By repeating the process of rearranging the virtual diffusion source (S27), the position where the diffusion source estimation index is maximized is estimated (S28).

The data acquisition unit 33 acquires concentration information, and the processing unit 10 uses at least the concentration information and position information (S11) at the virtual observation point and the position information of the virtual diffusion source, and uses an atmospheric diffusion calculation model. The influence function value that each virtual diffusion source gives to each virtual observation point is calculated (S14), and the virtual diffusion is performed by a statistical method using each influence function value and the concentration information at each virtual observation point. ^{For each source, the first diffusion source estimation index (index Pm} ) indicating the degree of agreement of the reach of the virtual diffuser and the second diffusion source estimation index (index) indicating the degree of agreement of the relative shades of the virtual diffuser. V ^m ) and is calculated, and the second diffusion source estimation index is subtracted from the first diffusion source estimation index ^{to calculate the diffusion source estimation index (index Z m} ) (S15).

In the first estimation mode, the diffusion source position estimation device 100 sets a virtual observation point showing the largest concentration information as the first diffusion source candidate, or calculates a diffusion source estimation index for each virtual diffusion source. The virtual diffusion source showing the largest value among them is set as a diffusion source candidate.

The position information of the virtual observation point and the virtual diffusion source in the virtual space has the origin at one point where the horizontal axis and the vertical axis of the surrounding frame in the virtual space intersect, the horizontal axis direction is the X axis direction, and the vertical axis is the Y axis direction. It is the coordinate axis information shown as (see FIG. 2).

According to the diffusion source position estimation device 100 of the present embodiment, the shape of the reachable range of the diffuser of the measured concentration (concentration information) at the actual observation point (measurement device) and the influence function from the virtual diffusion source, and within the reachable range. By quantifying and considering the degree of matching of the relative shades of the diffusing material, the position of the diffusing source of the diffusing material can be estimated accurately.

10 Processing unit (information processing unit)
11 Observation information acquisition unit 12 Virtual diffusion source setting unit 13 Virtual grid setting unit 14 Impact function calculation unit 15 Diffusion source estimation index calculation unit 16 Diffusion source position estimation unit 20 Storage unit 21 Information at the observation point (concentration information and position information)
22 Position information of virtual diffusion source 23 Effect function on virtual observation point 24 First diffusion source estimation index (index P ^m )
25 Second diffusion source estimation index (index V ^m )
26 Diffusion source estimation index (index Z ^m )
31 Input unit 32 Output unit 33 Data acquisition unit (information acquisition unit)
100 Diffusion source position estimation device (estimator)

Claims (6)

A diffusion source position estimation device that estimates the position of the diffusion source of a diffuser that has diffused into the atmosphere in a predetermined area.
It is provided with an information acquisition unit that acquires concentration information of the diffuser from a plurality of observation devices arranged in the predetermined area, and an information processing unit that processes the acquired information.
The information processing unit
Build a virtual space that simulates the predetermined area,
In the virtual space, a virtual observation point corresponding to the observation device and a plurality of virtual diffusion sources for diffusing a virtual diffusing substance are arranged in the virtual space.
By a statistical method, a diffusion source estimation index indicating the high reproducibility of the current state of the virtual diffusion substance was obtained for each virtual diffusion source.
Until the diffusion source estimation index is maximized, the process of rearranging the placement position of the virtual diffusion source and calculating the diffusion source estimation index is repeated, and the position where the diffusion source estimation index is maximized is the diffusion. A diffusion source position estimator characterized by estimating as a source. The information processing unit
In the first estimation mode, one of the virtual diffusion sources is set as a diffusion source candidate by a predetermined method.
In the second estimation mode, a rectangular lattice centered on the diffusion source candidate is constructed, the virtual diffusion source is rearranged at each grid point of the rectangular lattice, and the diffusion source estimation index in the diffusion source candidate is set. While recalculating, the diffusion source estimation index is calculated for each of the rearranged virtual diffusion sources, and the largest one is estimated as the maximum virtual diffusion source until the maximum virtual diffusion source and the diffusion source candidate match. The first aspect of the present invention is to estimate the position where the diffusion source estimation index is maximized by replacing the maximum virtual diffusion source with the diffusion source candidate and repeating the process of rearranging the virtual diffusion source. The diffusion source position estimator described. The information processing unit
Using at least the concentration information and position information at the virtual observation point and the position information of the virtual diffusion source, the influence function of each virtual diffusion source on each virtual observation point is calculated by the atmospheric dispersion calculation model. Calculate and
A first diffusion indicating the degree of agreement of the reach of the virtual diffusion substance for each virtual diffusion source by a statistical method using each of the influence functions and the concentration information at each virtual observation point. A source estimation index and a second diffusion source estimation index indicating the degree of agreement between the relative shades of the virtual diffuser are calculated, and the second diffusion source estimation index is subtracted from the first diffusion source estimation index. The diffusion source position estimation device according to claim 2, wherein the diffusion source estimation index is calculated. In the first estimation mode, the diffusion source position estimation device sets the virtual observation point showing the largest concentration information as the first diffusion source candidate, or the diffusion source for each virtual diffusion source. The diffusion source position estimation device according to claim 2, wherein an estimation index is calculated and the virtual diffusion source showing the largest value among them is set as a diffusion source candidate. The position information of the virtual observation point and the virtual diffusion source in the virtual space has the origin at one point where the horizontal axis and the vertical axis of the peripheral frame in the virtual space intersect, the horizontal axis direction is the X axis, and the vertical axis is the vertical axis. The diffusion source position estimation device according to claim 1, wherein the coordinate axis information is shown as the Y-axis direction. It is a diffusion source position estimation method that estimates the position of the diffusion source of a diffuser that has diffused into the atmosphere in a predetermined area.
The estimation device is
The concentration information of the diffuser is acquired from a plurality of observation devices arranged in the predetermined area, and the concentration information is obtained.
Build a virtual space that simulates the predetermined area,
In the virtual space, a virtual observation point corresponding to the observation device and a plurality of virtual diffusion sources for diffusing a virtual diffusing substance are arranged in the virtual space.
By a statistical method, a diffusion source estimation index indicating the high reproducibility of the current state of the virtual diffusion source is obtained for each virtual diffusion source.
Until the diffusion source estimation index is maximized, the process of rearranging the placement position of the virtual diffusion source and calculating the diffusion source estimation index is repeated, and the position where the diffusion source estimation index is maximized is the diffusion. A diffusion source location estimation method characterized by estimating as a source.

Images