JP7430008B2 - Analysis equipment, analysis method and analysis program - Google Patents

Description

The present invention relates to an analysis device, an analysis method, and an analysis program.

Conventionally, buildings built on the ground have been analyzed based on images taken by artificial satellites. For example, Patent Document 1 listed below describes a method of extracting a parameter vector from an image taken by a satellite, determining the height and width of an object, and generating an idealized image of the object. . In Patent Document 1, the filling volume of the object is determined by comparing the image of the object with an idealized image. Note that the object in Patent Document 1 is mainly a cylindrical container having a floating roof structure.

JP 2019-514123 Publication

For example, when analyzing the operating status of an object such as a factory, images of the object may be taken over a long period of time using a satellite and the changes over time may be analyzed. However, it is sometimes difficult to determine the operating status of an object from its appearance, making it difficult to accurately analyze the operating status.

Therefore, the present invention provides an analysis device, an analysis method, and an analysis program that can analyze the operating status of an object with higher accuracy.

An analysis device according to one aspect of the present invention includes an acquisition unit that acquires satellite data obtained by observing a target object in a wavelength band below thermal infrared by a satellite and meteorological data of a region where the target object exists; A generation unit that generates predicted data that is predicted to be observed in the thermal infrared wavelength band of satellite data based on data observed in wavelength bands other than infrared and meteorological data; It includes a detection unit that detects a change in the object based on a comparison between data observed in an infrared wavelength band and predicted data.

According to this aspect, data observed in the thermal infrared wavelength band among satellite data changes depending on the surface temperature and exhaust temperature of the object, even if there is no change in the appearance of the object. . Therefore, by comparing the predicted data with satellite data observed in the thermal infrared wavelength band, it is possible to analyze with higher precision the operational status of objects that cannot be determined from their appearance.

In the above aspect, the generation unit may generate the prediction data using a trained model generated by supervised learning using satellite data and weather data observed in the past as learning data.

According to this aspect, the predicted data is generated to reproduce the data observed in the thermal infrared wavelength band under normal conditions with no change in the object, and the predicted data is generated to reproduce the data observed in the thermal infrared wavelength band. Changes in the object can be detected by the deviation between the two.

The above aspect may further include a learning unit that generates a trained model by supervised learning using learning data.

According to this aspect, it is possible to generate a trained model that reproduces data observed in the thermal infrared wavelength band during normal times when there is no change in the object.

In the above aspect, the learning unit may generate a learned model for each object.

According to this aspect, it is possible to generate a trained model according to the weather characteristics of the area where the target object exists.

In the above aspect, when the target object is a factory, the detection unit detects the factory when the value obtained by subtracting the value of data observed in the thermal infrared wavelength band among the satellite data from the value of the predicted data is greater than the threshold value. It may be detected that at least a portion of the system is out of operation.

According to this aspect, it is possible to detect that the amount of heat released from the factory is smaller than in normal times, and to detect that at least a part of the factory has stopped operating.

The above aspect further includes a calculation unit that divides the area into a plurality of sections and calculates feature amounts of satellite data regarding the plurality of sections, and the detection section detects changes related to objects in the plurality of sections based on the feature amounts. may be detected.

According to this aspect, a local change in the object can be detected, and a location where the change has occurred can be specified.

In the above aspect, the detection unit may detect when a change regarding the object occurs based on the difference between the average of the feature amounts before the reference time and the average of the feature amounts after the reference time.

According to this aspect, it is possible to detect when a change regarding the object occurs.

An analysis method according to another aspect of the present invention includes acquiring satellite data obtained by observing a target object in a wavelength band below thermal infrared by a satellite and meteorological data of a region where the target object exists; Based on data observed in wavelength bands other than infrared and meteorological data, prediction data that is predicted to be observed in thermal infrared wavelength band among satellite data is generated, and thermal infrared detecting a change in the object based on a comparison between data observed in an external wavelength band and predicted data.

According to this aspect, data observed in the thermal infrared wavelength band among satellite data changes depending on the surface temperature and exhaust temperature of the object, even if there is no change in the appearance of the object. . Therefore, by comparing the predicted data with satellite data observed in the thermal infrared wavelength band, it is possible to analyze with higher precision the operational status of objects that cannot be determined from their appearance.

An analysis program according to another aspect of the present invention uses a calculation unit included in an analysis device to collect satellite data obtained by observing an object in a wavelength band below thermal infrared by a satellite and meteorological data of a region where the object exists. The acquisition unit generates predicted data that is predicted to be observed in the thermal infrared wavelength band of the satellite data based on the data observed in the wavelength band other than the thermal infrared among the satellite data and weather data. and a detection unit that detects changes in the target object based on a comparison between satellite data observed in the thermal infrared wavelength band and predicted data.

According to this aspect, data observed in the thermal infrared wavelength band among satellite data changes depending on the surface temperature and exhaust temperature of the object, even if there is no change in the appearance of the object. . Therefore, by comparing the predicted data with satellite data observed in the thermal infrared wavelength band, it is possible to analyze with higher precision the operational status of objects that cannot be determined from their appearance.

According to the present invention, it is possible to provide an analysis device, an analysis method, and an analysis program that can analyze the operating status of an object with higher accuracy.

1 is a diagram showing a network configuration of an analysis system according to an embodiment of the present invention. FIG. 2 is a diagram showing functional blocks of the analysis device according to the present embodiment. FIG. 1 is a diagram showing the physical configuration of an analysis device according to the present embodiment. FIG. 2 is a diagram showing predicted data generated by the analysis device according to the present embodiment and data observed in a thermal infrared wavelength band. FIG. 3 is a diagram showing a difference between predicted data generated by the analysis device according to the present embodiment and data observed in a thermal infrared wavelength band. FIG. 3 is a diagram showing the average feature amount of satellite data before the reference time and the average feature amount of the satellite data after the reference time calculated by the analysis device according to the embodiment. It is a flowchart of the 1st process performed by the analysis device concerning this embodiment. It is a flowchart of the 2nd process performed by the analysis device concerning this embodiment.

Embodiments of the present invention will be described with reference to the accompanying drawings. In addition, in each figure, those with the same reference numerals have the same or similar configurations.

FIG. 1 is a diagram showing a network configuration of an analysis system 100 according to an embodiment of the present invention. The analysis system 100 includes a satellite base station 50 that receives satellite data from a satellite S flying in a satellite orbit in space, a database DB that records satellite data, and an analysis device 10 that analyzes the satellite data.

The satellite S is equipped with an optical sensor and transmits satellite data including images of the ground taken by the optical sensor to the satellite base station 50. Satellite data includes at least images taken of the ground in a wavelength band below thermal infrared. Here, the thermal infrared wavelength band may be 8 to 15 μm, but may also include adjacent wavelength bands. The satellite data may include, for example, images taken in a plurality of wavelength bands included from 400 nm to 15 μm.

The satellite base station 50 receives satellite data from the satellite S and records it in the database DB. The database DB is connected to a communication network N such as the Internet. The analysis device 10 acquires satellite data stored in the database DB via the communication network N.

FIG. 2 is a diagram showing functional blocks of the analysis device 10 according to this embodiment. The analysis device 10 includes an acquisition section 11 , a generation section 12 , a detection section 13 , a learning section 14 , and a calculation section 15 .

The acquisition unit 11 acquires satellite data obtained by observing an object using a satellite S in a wavelength band below thermal infrared, and meteorological data of a region where the object exists. Here, the target object may be a factory, a cylindrical or rectangular container, a crude oil tank, a building such as a building, a commercial facility such as an amusement park or a zoo, or a moving object such as an airplane traveling on the same route. , adjacent buildings, etc. The meteorological data may be obtained via the communication network N, and includes, for example, cumulative precipitation, solar radiation, upper cloud amount, middle cloud amount, lower cloud amount, surface pressure, sea level pressure, relative humidity, total cloud amount, temperature, and east/west. Includes wind and north-south winds.

The generation unit 12 generates predicted data that is predicted to be observed in the thermal infrared wavelength band among the satellite data, based on data observed in the wavelength band other than the thermal infrared among the satellite data and meteorological data. do. Data observed in wavelength bands other than thermal infrared is data that measures the appearance of the object, and data observed in the thermal infrared wavelength band is data that measures the object's surface temperature and exhaust gas from the object. This is data obtained by measuring temperature. The generation unit 12 generates prediction data that predicts the surface temperature of the object and the temperature of exhaust gas from the object, based on data obtained by measuring the appearance of the object and weather data.

The generation unit 12 generates prediction data using a trained model 12a generated by supervised learning using satellite data and meteorological data observed in the past as learning data. The trained model 12a may be configured by, for example, a gradient boosting decision tree or a neural network, and uses as input data observed in wavelength bands other than thermal infrared and meteorological data among the learning data. It is generated through supervised learning to reproduce data observed in the thermal infrared wavelength band. In this way, predicted data is generated to reproduce the data observed in the thermal infrared wavelength band under normal conditions when there are no changes in the object, and the predicted data is compared with the data observed in the thermal infrared wavelength band. Changes in the object can be detected by the deviation of the .

The detection unit 13 detects changes regarding the object based on a comparison between data observed in the thermal infrared wavelength band of the satellite data and predicted data. Of the satellite data, data observed in the thermal infrared wavelength band changes depending on the surface temperature and exhaust temperature of the object, even if there is no change in the appearance of the object. Therefore, by comparing the predicted data with satellite data observed in the thermal infrared wavelength band, it is possible to analyze with higher precision the operational status of objects that cannot be determined from their appearance.

The learning unit 14 generates a trained model 12a by supervised learning using learning data. For example, when the trained model 12a is configured by a gradient boosting decision tree, the learning unit 14 inputs data observed in a wavelength band other than thermal infrared and meteorological data among the learning data, and inputs predicted data and learning data. A decision tree may be constructed using gradient boosting so that the value of the loss function that measures the difference between the data observed in the thermal infrared wavelength band among the data for use in the thermal infrared wavelength band becomes small. Further, when the trained model 12a is configured by a neural network, the learning unit 14 inputs data observed in a wavelength band other than the thermal infrared among the learning data and weather data, and inputs the thermal infrared among the learning data. The learned model 12a is generated by optimizing the parameters of the neural network by error backpropagation method or the like so as to output data observed in an outside wavelength band. In this way, it is possible to generate a trained model 12a that reproduces data observed in the thermal infrared wavelength band during normal times when there is no change in the object. Here, supervised learning is not limited to the algorithms described above, and other machine learning algorithms may be used.

The learning unit 14 may generate a learned model 12a for each object. In that case, learning data is prepared for each object. Since weather conditions may vary depending on the region where the target object is located, by generating a trained model 12a for each target object, it is possible to generate a trained model 12a that corresponds to the weather characteristics of the region where the target object is located. .

The calculation unit 15 divides the area where the target object is present into a plurality of sections, and calculates feature amounts of satellite data regarding the plurality of sections. Then, the detection unit 13 detects changes regarding the object in the plurality of sections based on the feature amount. Thereby, local changes in the object can be detected, and the location where the change has occurred can be specified.

FIG. 3 is a diagram showing the physical configuration of the analysis device 10 according to this embodiment. The analysis device 10 includes a CPU (Central Processing Unit) 10a corresponding to a calculation section, a RAM (Random Access Memory) 10b corresponding to a storage section, a ROM (Read Only Memory) 10c corresponding to a storage section, and a communication section 10d. , an input section 10e, and a display section 10f. These components are connected to each other via a bus so that they can transmit and receive data. In this example, a case will be described in which the analysis device 10 is composed of one computer, but the analysis device 10 may be realized by combining a plurality of computers. Further, the configuration shown in FIG. 3 is an example, and the analysis device 10 may have configurations other than these, or may not have some of these configurations.

The CPU 10a is a control unit that performs control related to the execution of an analysis program stored in the RAM 10b or ROM 10c, and performs calculations and processing of data. The CPU 10a is a calculation unit that executes a program (analysis program) for detecting changes in a target object based on satellite data and weather data. The CPU 10a receives various data from the input section 10e and the communication section 10d, and displays the calculation results of the data on the display section 10f or stores them in the RAM 10b.

The RAM 10b is a storage unit in which data can be rewritten, and may be formed of, for example, a semiconductor storage element. The RAM 10b may store data such as analysis programs executed by the CPU 10a, satellite data, and weather data. Note that these are just examples, and the RAM 10b may store data other than these, or some of them may not be stored.

The ROM 10c is a storage section from which data can be read, and may be composed of, for example, a semiconductor storage element. The ROM 10c may store, for example, an analysis program or data that is not rewritten.

The communication unit 10d is an interface that connects the analysis device 10 to other devices. The communication unit 10d may be connected to a communication network N such as the Internet.

The input unit 10e receives data input from the user, and may include, for example, a keyboard and a touch panel.

The display unit 10f visually displays the calculation results by the CPU 10a, and may be configured by, for example, an LCD (Liquid Crystal Display). The display unit 10f may display satellite data.

The analysis program may be provided by being stored in a computer-readable storage medium such as the RAM 10b or ROM 10c, or may be provided via a communication network connected by the communication unit 10d. In the analysis device 10, the various operations described using FIG. 2 are realized by the CPU 10a executing the analysis program. Note that these physical configurations are merely examples, and do not necessarily have to be independent configurations. For example, the analysis device 10 may include an LSI (Large-Scale Integration) in which a CPU 10a, a RAM 10b, and a ROM 10c are integrated.

FIG. 4 is a diagram showing predicted data generated by the analysis device 10 according to the present embodiment and data observed in the thermal infrared wavelength band. In the figure, the vertical axis shows the value of one pixel of data observed in the TIR (Thermal Infra-Red) wavelength band, and the horizontal axis shows the date. The satellite data used in this example is satellite data when a factory is photographed as an object, and the values shown on the vertical axis indicate temperature changes in the factory among images taken in the thermal infrared wavelength band. It is the value of the pixel to be reflected, and more specifically, it is the value of the pixel corresponding to the area where the factory exists. The factory area includes exhaust vents, specific heat generating equipment, etc.

In the figure, data D observed in the thermal infrared wavelength band among the satellite data is shown as a solid line, and predicted data P expected to be observed in the thermal infrared wavelength band among the satellite data is shown in a broken line. ing. Although both data almost match, there is a relatively large deviation at timing T1 in mid-2017.

FIG. 5 is a diagram showing the difference Diff between the predicted data P generated by the analysis device 10 according to the present embodiment and the data D observed in the thermal infrared wavelength band. In the figure, the vertical axis shows the score (diff_score(TIR)) representing the difference Diff, and the horizontal axis shows the date. The score representing the difference Diff is the value obtained by subtracting the actually measured data D from the predicted data P.

The difference Diff obtained by subtracting the actually measured data D from the predicted data P is relatively large at timing T1, and it can be seen that a change regarding the object has occurred at timing T1. When the target object is a factory, the detection unit 13 detects that when the value of the difference Diff, which is obtained by subtracting the value of data D observed in the thermal infrared wavelength band of the satellite data from the value of predicted data P, is larger than the threshold value. , a shutdown of at least a portion of the factory may be detected. In this way, it is possible to detect that the amount of heat released from the factory is smaller than in normal times, and to detect that at least part of the factory has stopped operating. Furthermore, in this embodiment, pixel value data corresponding to an area where a factory exists is described as an example, but in an area where a building or house exists, data corresponding to a device that exhausts heat such as an air conditioner outdoor unit, etc. is described as an example. Observations similar to those in this embodiment can be made using pixel values, and usage conditions can be detected.

FIG. 6 is a diagram showing the average feature amount of satellite data before the reference time and the average feature amount of the satellite data after the reference time calculated by the analysis device 10 according to the present embodiment. The analysis device 10 acquires an image IMG obtained by observing a certain area in a predetermined wavelength band (for example, a visible light wavelength band). Here, a plurality of images IMG are photographed in chronological order. The analysis device 10 divides the image IMG into a plurality of sections to obtain a plurality of partial images IMG_1 to IMG_N. Here, each of the plurality of partial images IMG_1 to IMG_N includes a plurality of partial images arranged in chronological order. Then, the analysis device 10 calculates the feature amount of the satellite data regarding the plurality of sections by calculating the feature amount from the plurality of partial images IMG_1 to IMG_N. The analysis device 10 calculates a feature amount for each of the plurality of partial images IMG_1 to IMG_N arranged in chronological order, and calculates a time-series change in the feature amount. The analysis device 10 may calculate feature amounts from the plurality of partial images IMG_1 to IMG_N using a variational autoencoder or a generative adversarial network.

The detection unit 13 may detect when a change regarding the object occurs based on the difference diff between the average of the feature values f before the reference time and the average of the feature values f after the reference time. FIG. 6 shows a case where the average of the feature quantities f before the reference time is 0.2, and the average of the feature quantities f after the reference time is -1.1. The detection unit 13 may detect the timing at which the absolute value of the difference between the average of the feature quantities f before the quasi-time and the average of the feature quantities f after the reference time becomes the maximum as the timing at which a change regarding the object occurs. . In this way, it is possible to detect when a change regarding the object occurs.

FIG. 7 is a flowchart of the first process executed by the analysis device 10 according to the present embodiment. The first process is a process of detecting changes in the object using satellite data observed in a wavelength band below thermal infrared.

First, the analysis device 10 acquires satellite data and meteorological data observed by a satellite in a wavelength band below thermal infrared regarding a region where a target object exists (S10). The analysis device 10 then generates predicted data that is predicted to be observed in the thermal infrared wavelength band among the satellite data, based on the data observed in the wavelength band other than the thermal infrared among the satellite data and the meteorological data. is generated (S11).

Then, the analysis device 10 detects changes regarding the object based on a comparison between data observed in the thermal infrared wavelength band of the satellite data and predicted data (S12).

FIG. 8 is a flowchart of the second process executed by the analysis device 10 according to this embodiment. The second process is a process of detecting local changes in the object using satellite data observed in a wavelength band below near-infrared.

First, the analysis device 10 acquires satellite data observed by a satellite in a wavelength band other than thermal infrared regarding the area where the target object exists (S20). After that, the analysis device 10 divides the area into a plurality of sections and calculates the feature amount of the satellite data regarding the plurality of sections (S21).

The analysis device 10 then detects changes regarding the object in the plurality of sections based on the feature amount (S22).

As a modification of this embodiment, when monitoring an active volcano as a target object, the area where the target object exists is divided into a plurality of sections, and satellite data regarding the plurality of sections is monitored using wavelengths other than thermal infrared. Based on the data observed in the satellite data and meteorological data, we generate predicted data that is predicted to be observed in the thermal infrared wavelength band of the satellite data, and combine the generated predicted data and the thermal infrared wavelength band of the satellite data. By comparing the data with the data observed in the wavelength range of A warning notice is sent to residents near the object. Furthermore, the monitoring target can be applied to solar thermal power generation such as Stirling engines, flare stacks of crude oil mining facilities, etc.

The embodiments described above are intended to facilitate understanding of the present invention, and are not intended to be interpreted as limiting the present invention. Each element included in the embodiment, as well as its arrangement, material, conditions, shape, size, etc., are not limited to those illustrated, and can be changed as appropriate. Further, it is possible to partially replace or combine the structures shown in different embodiments.

Claims (9)

an acquisition unit that acquires satellite data obtained by observing an object in a wavelength band below thermal infrared by a satellite and meteorological data of a region where the object exists;
Generating predicted data that is predicted to be observed in a thermal infrared wavelength band among the satellite data, based on data observed in a wavelength band other than thermal infrared among the satellite data and the meteorological data. Department and
a detection unit that detects a change regarding the object based on a comparison between data observed in a thermal infrared wavelength band of the satellite data and the predicted data;
An analysis device equipped with. The generation unit generates the prediction data using a trained model generated by supervised learning using the satellite data and the weather data observed in the past as learning data.
The analysis device according to claim 1. further comprising a learning unit that generates the learned model by supervised learning using the learning data;
The analysis device according to claim 2. The learning unit generates the learned model for each object,
The analysis device according to claim 3. When the target object is a factory, the detection unit detects the detection unit when the value obtained by subtracting the value of the data observed in the thermal infrared wavelength band among the satellite data from the value of the predicted data is greater than a threshold value. detecting an outage of at least a portion of the factory;
An analysis device according to any one of claims 1 to 4. further comprising a calculation unit that divides the area into a plurality of sections and calculates a feature amount of the satellite data regarding the plurality of sections,
The detection unit detects a change regarding the object in the plurality of sections based on the feature amount.
An analysis device according to any one of claims 1 to 5. The detection unit detects when a change regarding the object occurs based on the difference between the average of the feature amounts before the reference time and the average of the feature values after the reference time.
The analysis device according to claim 6. An analysis method executed by an analysis device,
Obtaining satellite data of an object observed by a satellite in a wavelength band below thermal infrared and meteorological data of the area where the object exists;
Generating predicted data that is predicted to be observed in a thermal infrared wavelength band among the satellite data, based on data observed in a wavelength band other than thermal infrared among the satellite data and the meteorological data. and,
detecting a change regarding the object based on a comparison between data observed in a thermal infrared wavelength band of the satellite data and the predicted data;
analysis methods including The calculation unit included in the analysis device is
an acquisition unit that acquires satellite data obtained by observing an object in a wavelength band below thermal infrared by a satellite and meteorological data of a region where the object exists;
Generating predicted data that is predicted to be observed in a thermal infrared wavelength band among the satellite data, based on data observed in a wavelength band other than thermal infrared among the satellite data and the meteorological data. a detection unit that detects changes regarding the object based on a comparison between data observed in a thermal infrared wavelength band of the satellite data and the predicted data;
An analysis program that functions as

Images