JP7617740B2 - Information processing device, information processing method, and program - Google Patents

Description

The present disclosure relates to an information processing device, an information processing method, and a program.

Patent document 1 describes an image processing method for estimating the filling volume of a container based on satellite images.

Special table 2019-514123 publication

However, the image processing method described in Patent Document 1 has the problem that although it can estimate the container's fill volume at the time the image was taken based on each image, it cannot estimate whether the contents of the container are in use, i.e., whether the equipment is in operation.

The present disclosure has been made to solve the above problem, and aims to provide an information processing device, an information processing method, and a program that can estimate the operating status of equipment.

In order to solve the above problem, the information processing device according to the present disclosure includes an acquisition unit that acquires detection information detected in the air, including status information corresponding to the status of a specified emission within a specified area space on the ground, and an operation status estimation unit that estimates the operation status of equipment associated with the area space based on the detection information.

The information processing method disclosed herein includes a step of acquiring detection information detected in the air, the detection information including status information corresponding to the status of a specified emission within a specified area space on the ground, and a step of estimating the operating status of equipment associated with the area space based on the detection information.

The program disclosed herein causes a computer to execute the steps of acquiring detection information detected in the air, including status information corresponding to the status of a specified emission within a specified area on the ground, and estimating the operating status of equipment associated with the area based on the detection information.

The information processing device, information processing method, and program disclosed herein make it possible to estimate the operating status of equipment.

1 is a block diagram showing an example of a basic configuration of an information processing device according to an embodiment of the present disclosure. FIG. 2 is a schematic diagram showing a configuration example of facility information according to the first embodiment of the present disclosure. FIG. 2 is a schematic diagram illustrating an example of an area space according to the first embodiment of the present disclosure. FIG. 2 is a schematic diagram illustrating an example of an area space according to the first embodiment of the present disclosure. FIG. 2 is a schematic diagram illustrating an example of an area space according to the first embodiment of the present disclosure. FIG. 1 is a schematic diagram showing an example of the configuration of a trained model according to a first embodiment of the present disclosure. 5 is a flowchart illustrating an operation example of the information processing device according to the first embodiment of the present disclosure. 5 is a flowchart illustrating an operation example of the information processing device according to the first embodiment of the present disclosure. 5 is a flowchart illustrating an operation example of the information processing device according to the first embodiment of the present disclosure. 5 is a flowchart illustrating an operation example of the information processing device according to the first embodiment of the present disclosure. 5 is a flowchart illustrating an operation example of the information processing device according to the first embodiment of the present disclosure. 5 is a flowchart illustrating an operation example of the information processing device according to the first embodiment of the present disclosure. 5 is a flowchart illustrating an operation example of the information processing device according to the first embodiment of the present disclosure. FIG. 13 is a schematic diagram showing an example configuration of a trained model according to a second embodiment of the present disclosure. FIG. 13 is a schematic diagram showing an example configuration of a trained model according to a second embodiment of the present disclosure. 10 is a flowchart illustrating an example of an operation of the information processing device according to the second embodiment of the present disclosure. 10 is a flowchart illustrating an operation example of an information processing device according to a second embodiment of the present disclosure. 10 is a flowchart illustrating an operation example of an information processing device according to a second embodiment of the present disclosure. FIG. 11 is a schematic diagram showing an example of an area space according to a second embodiment of the present disclosure. FIG. 11 is a schematic diagram showing an example of an area space according to a second embodiment of the present disclosure. 13 is a flowchart illustrating an operation example of the information processing device according to the third embodiment of the present disclosure. 13 is a flowchart illustrating an operation example of the information processing device according to the third embodiment of the present disclosure. FIG. 13 is a schematic diagram illustrating an example of an area space according to a third embodiment of the present disclosure. FIG. 13 is a schematic diagram illustrating an example of an area space according to a third embodiment of the present disclosure. FIG. 13 is a schematic diagram illustrating an example of an area space according to a third embodiment of the present disclosure. 13 is a flowchart illustrating an example of an operation of the information processing device according to the fourth embodiment of the present disclosure. 13 is a flowchart illustrating an example of an operation of the information processing device according to the fourth embodiment of the present disclosure. FIG. 13 is a schematic diagram showing an example of an area space according to a fourth embodiment of the present disclosure. 13 is a flowchart illustrating an operation example of the information processing device according to the fifth embodiment of the present disclosure. FIG. 1 is a system diagram illustrating an example of a business flow according to an embodiment of the present disclosure. FIG. 1 is a schematic block diagram illustrating a configuration of a computer according to at least one embodiment.

<Basic configuration example of each embodiment>
Hereinafter, an information processing device, an information processing method, and a program according to an embodiment of the present disclosure will be described with reference to the drawings. Note that in each drawing, the same or corresponding components are designated by the same reference numerals and descriptions thereof will be omitted as appropriate.

FIG. 1 is a block diagram showing an example of the basic configuration of an information processing device according to an embodiment of the present disclosure. The basic configuration of the information processing device 1 shown in FIG. 1 is the basic configuration of the information processing device of the first to fifth embodiments described below, and the information processing device according to the first to fifth embodiments can be configured by modifying (including omissions and additions) part of the configuration of the information processing device 1 shown in FIG. 1. In the description of each embodiment, the modified parts from the information processing device 1 shown in FIG. 1 will be mainly described.

The information processing device 1 shown in FIG. 1 can be configured using a computer and its peripheral devices such as a mouse, a touch panel, a display device, etc., and has the following units as a functional configuration consisting of a combination of hardware such as a computer and software such as a program executed by the computer. That is, the information processing device 1 has an acquisition unit 11, an operation state estimation unit 12, a production volume estimation unit 13, a location information acquisition unit 14, a state estimation unit 15, a trained model 16, detection information 17, equipment information 18, and a model learning unit 19. The trained model 16 includes at least one of a general trained model 161 and an individual trained model 162.

The acquisition unit 11 acquires detection information detected in the air, including status information corresponding to the status of a specific emission 41 in a specific area space 51 on the ground, for example from the detection information providing system 2, and stores it as detection information 17 in a storage device or the like provided in a computer constituting the information processing device 1. The detection information is visible light image data, thermal infrared image data, etc. corresponding to the area space 51 acquired from the air by an artificial satellite 21 (or a drone, an aircraft, a high-rise building), etc.

The state information contained in the detection information is, for example, information that is presumed to be included when the emission 41 is present in the visible light image data obtained by capturing the areal space 51, and can be defined by the luminance values of each RGB (color) channel, the luminance value of a predetermined gray level, and the shape and size of the range having that luminance value. Alternatively, the state information contained in the detection information is, for example, information that is presumed to be included when the emission 41 is present in the thermal infrared image data obtained by capturing the areal space 51, and can be defined by the luminance value of a predetermined gray level, and the shape and size of the range having that luminance value. Alternatively, the state information contained in the detection information may be, for example, information (such as information corresponding to the ground or a building) that is presumed to be included when the emission 41 is not present in the visible light image data or thermal infrared image data obtained by capturing the areal space 51.

The emissions 41 are substances emitted from equipment 71 such as production equipment in a plant, and are, for example, white smoke, gas, heat, mist, etc. In this case, the chimney 61 that emits the emissions 41 is the analysis target that serves as the basis for determining the state of the emissions 41. The analysis target can be specified by the structure that emits the emissions or the emission location. However, the analysis target may be multiple objects such as multiple chimneys, cooling towers, drains, etc., or the entire plant or part of a plant equipped with multiple pieces of equipment. The area space 51 is defined to surround the analysis target (in this case, the chimney 61).

The operation state estimation unit 12 estimates the operation state of the equipment 71 associated with the area space 51 based on the detection information 17. Based on the state information corresponding to the state of the discharged material 41 contained in the detection information 17, the operation state estimation unit 12 estimates that the equipment 71 is operating if the discharge unit 41 is discharging, and estimates that the equipment 71 is not operating (non-operating) if the discharge unit 41 is not discharging, for example.

The operating state estimation unit 12 can estimate the operating state using a trained model 16 that inputs detection information and outputs a classification result that classifies the operating state into at least operating or non-operating. The trained model 16 is, for example, a model that has a neural network as an element, and the weighting coefficients between the neurons of each layer of the neural network are optimized by machine learning so that a desired solution is output for a large amount of input data. The trained model 16 is, for example, composed of a combination of a program that performs calculations from input to output and weighting coefficients (parameters) used in the calculations. The general-purpose trained model 161 is, for example, a model that has a neural network as an element, and is a model that has been trained using multiple detection information obtained in multiple areal spaces 51 as teacher data, and can be commonly applied to multiple areal spaces 51. The individual trained model 162 is, for example, a model that has a neural network as an element, and is a model that has been retrained for each areal space 51 using the weighting coefficients between the neurons of each layer of a neural network (general trained model 161) that has been trained using multiple detection information corresponding to multiple different areal spaces 51 as teacher data as initial values. An individual learning model 162 is prepared for each areal space 51 and applied to each areal space 51. By setting the weighting coefficients obtained by the general-purpose trained model 161 as initial values, even if the amount of detection information 17 obtained for each areal space 51 is small, the model can be trained efficiently by using the detection information 17 obtained in different areal spaces 51 as training data.

The trained model 16 (general trained model 161 and individual trained model 162) may further input meteorological information corresponding to the area space 51 and output a classification result. The trained model 16 (general trained model 161 and individual trained model 162) may further input a previous classification result and output the classification result. For example, for equipment that is operated continuously for a certain period of time, the operating state of the estimation target is affected by the operating state a predetermined time ago (e.g., one day), so the estimation accuracy can be improved by recursively inputting the previous estimation result.

When the equipment 71 is a production equipment, the production volume estimation unit 13 estimates the production volume of the equipment 71 (production equipment) based on the operation state of the equipment 71 estimated by the operation state estimation unit 12. The production volume estimation unit 13 estimates the production volume to be a predetermined volume when the equipment is in operation, and estimates the production volume to be a predetermined volume (or zero) that is smaller than when the equipment is in operation, based on, for example, equipment information (production volume when in operation, or information for estimating the production volume when in operation) included in the equipment information 18 prepared in advance.

The position information acquisition unit 14 acquires position information of one or more mobile terminals 31 within a predetermined region corresponding to the area space 51, for example, from the position information providing system 3. The state estimation unit 15 estimates a non-operating state when the operation state of the equipment 71 is non-operating based on the position information acquired by the position information acquisition unit 14. The non-operating state is, for example, a state that indicates whether the equipment is normally non-operating or abnormally non-operating (for example, due to construction or the like). For example, the state estimation unit 15 can estimate normal non-operating state when there is no significant change in the position information by comparing with the position information obtained during operation, and estimate abnormal non-operating state when there is a significant change in the position information.

When the emissions 41 are white smoke and the detection information 17 is color or monochrome image data, the production volume estimation unit 13 can estimate the production volume by correcting it according to the season. The ease of imaging white smoke is affected by temperature and humidity, and is also affected by the season, so the correction can reduce the estimation error.

In addition, when the emissions 41 are heat and the detection information 17 is thermal infrared image data, the operation state estimation unit 12 may estimate the operation state of the equipment 71 based on a predetermined temperature difference based on the detection information 17. For example, the operation state can be estimated from the difference between the inlet temperature and the outlet temperature of the cooling water.

In addition, when the facility 71 is a production facility and a power generation facility, if the operation state estimation unit 12 cannot estimate the operation state based on the detection information 17, the production volume estimation unit 13 may estimate the production volume of the power generation facility based on the configuration of the system to which the power generation facility belongs.

In addition, the model creation unit 19 uses the detection information 17, etc. to learn or re-learn the trained model 16.

As described above, according to the information processing device 1 of this embodiment, the operating status of the equipment 17 can be estimated based on detection information (detection information 17) detected in the air, which includes status information corresponding to the status of a specified emission material 41 within a specified area space 51 on the ground.

First Embodiment
Next, an information processing device, an information processing method, and a program according to the first embodiment of the present disclosure will be described with reference to Figs. 2 to 13. Fig. 2 is a schematic diagram showing a configuration example of facility information according to the first embodiment of the present disclosure. Figs. 3 to 5 are schematic diagrams showing an example of an area space according to the first embodiment of the present disclosure. Fig. 6 is a schematic diagram showing a configuration example of a trained model according to the first embodiment of the present disclosure. Figs. 7 to 13 are flowcharts showing an operation example of the information processing device according to the first embodiment of the present disclosure.

In the information processing device 1 according to the first embodiment, the location information acquisition unit 14, the state estimation unit 15, and the individual trained model 162 can be omitted from the configuration shown in FIG. 1. In the first embodiment and each of the following embodiments, the target for estimating the operating state is a production facility (power generation facility) that produces electricity. In the first embodiment, the detection information is color image data.

2 shows an example of the configuration of the equipment information 18 shown in FIG. 1 in the first embodiment. The equipment information 18 shown in FIG. 2 includes an attribute information database 181 and a geographic information database 182. The attribute information database 181 stores the "object No.", which is the identification number of the object to be analyzed (such as the chimney 61), the attribute information (output, fuel type, plant name) indicating the attributes of the equipment 71, and the geographic information (the latitude and longitude P1, P2, P3, and P4 of the four vertices when the areal space 51 is a quadrilateral polygon), in association with the "unit name" which is the name of the equipment 71. The geographic information database 182 stores the "object No." and the geographic information (the latitude and longitude P1, P2, P3, and P4 of the four vertices when the areal space 51 is a quadrilateral polygon), in association with each other.

The facility information 18 can be set using maps (satellite images, etc.) and coordinate information obtained from a base map service 301 that provides map information, public information on the Internet, etc. 302, information obtained from an external information service (external service) 303, and information estimated from the appearance and dimensions (appearance and dimensions of an object) 304 of an object (facility, etc.) obtained from image data.

(Processing flow of the information processing device of the first embodiment)
In the information processing device 1 of the first embodiment, the production volume of the equipment associated with the analysis target is calculated according to the following steps (1) to (9).

(1) For the image (detection information 17) obtained by photographing, an analysis object (such as a plant, a chimney, a cooling tower, etc.) suitable for estimating production volume based on the state of emissions (white smoke, gas, mist, etc.) is identified, and geographic information is added to the area space centered on the identified analysis object. For example, for a satellite image (such as an image of the plant area space 501 shown in Figure 3), the position of the power plant's chimney 601 can be marked as a polygon ("rectangle") at four points P1, P2, P3, and P4 as shown in Figure 4, and geographic information can be added to the area space 511 centered on the analysis object. Note that in Figures 3 and 4, the chimney 601 is a chimney for releasing exhaust gas 401 shared by power generation units 701 and 702. Alternatively, for the image of the plant area space 501 shown in FIG. 3, for example, by marking the analysis object while specifying the position of the cooling tower 602 with four points P1, P2, P3, and P4 as shown in FIG. 5, geographic information can be added to the area space 512 centered on the analysis object.

(2) The attribute information of the analysis target is associated with the area space to which the geographic information has been assigned, and stored as, for example, facility information 18 shown in FIG. 2. For example, the attribute information of the object is associated with the marked chimney for the power plant based on publicly available information on the Internet, external services, and appearance and dimensions, and stored in the attribute information database 181.

(3) Images corresponding to the areal space based on the geographic information are acquired from any image provider. For example, images for the target period (e.g., daily images for several months or several years) are downloaded for the analysis target, and stored as detection information 17.

(4) The acquired images are classified into images for learning and images for inference. For example, the downloaded images are classified into images for learning and images for inference, for example, by random sampling.

(5) Status information (operation information) required to estimate the production volume of the analysis target is assigned to the learning images, images in which emissions are being emitted are labeled as "operating," images in which emissions are not being emitted are labeled as "stopped," and images in which it is difficult to distinguish are labeled as "unknown," and a learning model is used to learn, generating a classifier (trained model 16) that automatically infers the status information of any image. For example, learning images are labeled as Active (operating), images in which white smoke is not being emitted are labeled as Inactive (not operating), and images in which it is unclear whether white smoke is being emitted are labeled as Unknown, and learning is performed using the learning model (convolutional neural network) to generate a trained model.

(6) The image for inference is input to the classifier (trained model 16), and status information (operation information) of the analysis target is obtained as the classification result. By inputting the image for inference into the model, the operating status of the plant (Active (operating)/Inactive (not operating)/Unknown (unknown)) is estimated.

(7) Based on the classification results, combine the attribute information of the analysis target to obtain the production volume of the analysis target. The estimated operating state of the plant and attribute information such as output from (2) are used to estimate the power generation volume. For example, if minute white smoke cannot be detected and the plant is inactive, the power generation volume will be too low, and if the plant is active but all of its associated units are not operating, the power generation volume will be too high. Therefore, instead of using "Active"/"Inactive" = 1/0, a correction is made using, for example, 0.7/0.3.

(8) Find parameters for accurately classifying actual production volume based on changes in classification accuracy and classification results due to seasonal fluctuations, and generate a production volume estimation formula for any set. For example, corrections are made for changes in classification accuracy due to seasonal fluctuations as follows. That is, since the classification accuracy of white smoke is highest in December, January, and February, and the detection accuracy of white smoke is lowest in July and August (the degree of moisture condensation varies due to the effects of temperature and humidity), if the production volumes in December, January, and February are set as the standard (1x), the production volume in July and August is set at 1.5x, and the production volume from September to November and April to June is set at 1.3x.

(9) Calculate the production volume of the analysis target in any set. For example, aggregate the estimated power generation volume by area.

(Example of trained model configuration)
Next, an example of the configuration of the trained model 16 (general trained model 161) will be described with reference to FIG. 6. The input to the trained model 16 shown in FIG. 6 is an array of m×n×3 elements for each RGB channel representing the areal space 511, etc. The data structure of one color image is made up of pixel values of an m-row, n-column matrix according to the image size and resolution for each of the three RGB channels, which becomes the input to the convolutional neural network. In addition, the output side is the probability of being classified into each of the three classification numbers ("operating", "not operating", and "unknown") as the output of the softmax function at the final stage of the model, and the sum of the three is 1 (in the example shown in FIG. 6, the probability of each classification is "operating" = 0.78, "not operating" = 0.12, and "unknown" = 0.10).

(Example of operation of information processing device)
Next, an example of the operation of the information processing device 1 during the image acquisition process will be described with reference to FIG. 7. In the image acquisition process, the information processing device 1 first selects an image (step S111), then specifies an analysis target for the image (step S112), then assigns geographic information to the analysis target (step S113), then assigns attribute information to the analysis target (step S114), and stores the geographic/attribute information regarding the analysis target (step S115). The information processing device 1 repeatedly executes the processes of steps S111 to S115 until the selection is completed (until "Y" is obtained in step S116). On the other hand, when the selection is completed (when "Y" is obtained in step S116), the information processing device 1 acquires an image based on the geographic information (step S117) and stores the acquired image (step S118).

Next, referring to FIG. 8, an example of the operation of the information processing device 1 during the learning phase will be described. In the learning phase, the information processing device 1 first classifies images into learning images and inference images (step S121), then selects learning images (step S122), then displays the selected learning images on a predetermined display device (step S123), and then labels the learning images (step S124). The information processing device 1 repeatedly executes the processes of steps S122 to S124 until the selection is completed (until "Y" is obtained in step S125). On the other hand, when the selection is completed (when "Y" is obtained in step S125), the information processing device 1 performs learning using the labeled learning images (step S126) and stores them as a learned model (step S127).

Next, an example of the operation of the information processing device 1 in the estimation phase will be described with reference to Figures 9 to 13. In the estimation phase, the information processing device 1 acquires an image for inference (step S131), inputs the image for inference to the trained model (step S132), and outputs (records) information about the operating state (step S133). The information processing device 1 repeatedly executes the processes of steps S131 to S133 until acquisition is completed (until "Y" is obtained in step S134). When acquisition is completed (when "Y" is obtained in step S134), the information processing device 1 calculates the amount of power generation based on the information about the operating state and the attribute information (step S135), corrects the amount of power generation based on the classification accuracy (step S136), counts the amount of power generation for each area (step S137), and outputs the counted result (step S138).

10 and 11 show a specific example of the process of step S135. In this example, the process of step S135 is composed of the processes of steps S1351 to S1355 shown in FIG. 10 and FIG. 11. In step S1351, the plant operation state is arranged in chronological order for an arbitrary object (hereinafter, an example of "object No. 1"). In step S1352, operation = 1, non-operation = 0, n/a (missing) and unknown = "most recent 0/1" are assigned. In step S1353, using a constant k (0<k<1), k is assigned to 1 and 1-k is assigned to 0. In step S1354, the total output value of the units associated with the object (1200 MW in the case of object No. 1) is multiplied. Then, in step S1355, the above process is similarly performed for objects No. 1 to N to obtain the estimated power generation amount that can be grasped from the object.

Figure 12 also shows a specific example of the processing of step S136 as step S1361. In step S1361, since the degree of condensation of moisture contained in exhaust gas varies depending on the outside air temperature and humidity, in order to correct the power generation amount based on the classification accuracy of the classifier (trained model), a parameter α that changes with the season (month) is multiplied, and the estimated power generation amount (MWh) is calculated over a further 24 hours (however, the example in Figure 12 is the value before multiplying by 24 hours).

Figure 13 also shows a specific example of the processing of step S137 as step S1371. In step S1371, the estimated power generation amount calculated based on other unit information (area, country, fuel type, company name, etc.) associated with each object is tallied (Figure 13 shows an example of tallying the estimated power generation amount by area (prefecture)).

Second Embodiment
Next, an information processing device, an information processing method, and a program according to the second embodiment of the present disclosure will be described with reference to FIGS. 14 to 20. FIGS. 14 and 15 are schematic diagrams showing a configuration example of a trained model according to the second embodiment of the present disclosure. FIG. 14 shows the trained model 16 shown in FIG. 1 as trained model 16b, the general-purpose trained model 161 shown in FIG. 1 as trained model 161b, and the individual trained model 162 shown in FIG. 1 as trained model 162b. FIG. 15 shows the trained model 16 shown in FIG. 1 as trained model 16c, the general-purpose trained model 161 shown in FIG. 1 as trained model 161c, and the individual trained model 162 shown in FIG. 1 as trained model 162c. FIGS. 16 to 18 are flowcharts showing an example of the operation of an information processing device according to the second embodiment of the present disclosure. FIGS. 19 and 20 are schematic diagrams showing an example of an areal space according to the second embodiment of the present disclosure.

In addition, in the information processing device 1 according to the second embodiment, the location information acquisition unit 14 and the state estimation unit 15 can be omitted from the configuration shown in FIG. 1.

The second embodiment improves the accuracy of estimating production volume compared to the first embodiment by performing learning and estimation taking into account meteorological information (temperature, humidity, etc.) in the vicinity of the analysis target (plant, chimney, etc.) (Figures 14 and 15).

In addition, the weight matrix set in the convolutional neural network set in the classifier (trained models 161b and 161c) created in the same manner as in the first embodiment is used as the initial value of the weights of the classifier (trained models 162b and 162c), and when inferring images of each plant, three classifications of "operating", "not operating", and "unknown" are assigned to the images acquired of the object in the plant, label information is prepared, and additional re-learning is performed to improve the classification accuracy.

When learning is performed as in the first embodiment, a plant (object) is not specified, but learning is performed across multiple types of objects. For example, if 200 chimneys throughout Japan are targeted and 100 images are acquired from each chimney, the total number of images will be 20,000. In the first embodiment, this data is used as the basis for giving initial values to the parameters Wij using an arbitrary method, and a single general-purpose trained model is generated using the data and labels.

On the other hand, in the second embodiment, in order to improve accuracy, the weights Wij of the trained models (trained models 161b and 161c) trained in the same manner as in the first embodiment are set as initial values, and the classifiers (trained models 162b and 162c) are trained again using the data of each object and its label. In this way, the number of classifiers is generated as many as the number of objects. For example, if there are 200 chimneys, 200 classifiers will be created, and each created classifier has parameters corresponding to each object, so while versatility is reduced, classification accuracy is improved. The classifier generated in the first embodiment is called a general-purpose classifier, and the classifier newly generated in the second embodiment is called an individual classifier. Note that, depending on the attribute information of the plant, the previous judgment result affects the current judgment result, so the previous judgment result can be recursively incorporated into the classification algorithm (trained models 161c and 162c). Note that multiple memories M are newly provided in the trained models 161c and 162c shown in FIG. 15.

(Processing flow of the information processing device of the second embodiment)
In the information processing device 1 of the second embodiment, the production volume of the equipment associated with the analysis target is calculated according to the following steps (1) to (9).

(1) to (3), (4), (7), and (9) are the same as in the first embodiment. However, in the process of (3) in the second embodiment, weather information for the date and time when the image obtained from the geographic information in (3) in the first embodiment was captured is obtained (FIGS. 19 and 20). For the weather information, for example, known weather information for the location closest to the location of the analysis target (such as the nearest city) is used.

(5) Learn using weather information and generate a classifier.

(6) Weather information and the image for inference are input to the classifier, and status information of the subject to be analyzed is obtained as the classification result.

(8) Weather information is used when generating the classifier, eliminating the need for seasonal corrections as in the first embodiment.

(Example of operation of information processing device)
Next, an example of the operation of the information processing device 1 during the image acquisition process will be described with reference to FIG. 16. In the image acquisition process, the information processing device 1 first selects an image (step S141), then specifies an analysis target for the image (step S142), then assigns geographic information to the analysis target (step S143), then assigns attribute information to the analysis target (step S144), and then stores the geographic/attribute information regarding the analysis target (step S145). The information processing device 1 repeatedly executes the processes of steps S141 to S145 until the selection is completed (until "Y" is obtained in step S146). On the other hand, when the selection is completed (when "Y" is obtained in step S146), the information processing device 1 acquires an image based on the geographic information (step S147), acquires weather information for the date and time when the image was taken based on the geographic information (step S148), and stores the acquired weather information and image data (step S149).

Next, referring to FIG. 17, an example of the operation of the information processing device 1 during the learning phase will be described. In the learning phase, the information processing device 1 first classifies weather information and image data into learning and inference images (step S151), then selects learning images (step S152), then displays the selected learning images on a predetermined display device (step S153), and then labels the learning images (step S154). The information processing device 1 repeatedly executes the processes of steps S152 to S154 until the selection is completed (until "Y" is obtained in step S155). On the other hand, when the selection is completed (when "Y" is obtained in step S155), the information processing device 1 performs learning using the labeled learning images (step S156), stores them as a learned model (step S157), and re-learns for each analysis target (step S158).

Next, referring to FIG. 18, an example of the operation of the information processing device 1 during the estimation phase will be described. In the estimation phase, the information processing device 1 acquires an image for inference (step S161), inputs the image for inference to the trained model (step S162), and outputs (records) information about the operating state (step S163). The information processing device 1 repeatedly executes the processes of steps S161 to S163 until acquisition is completed (until "Y" is obtained in step S164). When acquisition is completed (when "Y" is obtained in step S164), the information processing device 1 calculates the amount of power generation based on the information about the operating state and the attribute information (step S165), totals the amount of power generation for each area (step S166), and outputs the totaled result (step S167).

Third Embodiment
Next, an information processing device, an information processing method, and a program according to a third embodiment of the present disclosure will be described with reference to Fig. 21 to Fig. 25. Fig. 21 and Fig. 22 are flowcharts showing an example of the operation of the information processing device according to the third embodiment of the present disclosure. Fig. 23 to Fig. 25 are schematic diagrams showing an example of an area space according to the third embodiment of the present disclosure.

The third embodiment differs from the first and second embodiments in that it enables additional estimation of status information (non-operating state) regarding whether the analysis target (plant, chimney, etc.) is shut down, for example, due to large-scale construction work or for other reasons. As a result, there are cases where it is not clear whether exhaust gas is being generated just by the system determining whether or not exhaust gas is being generated. Since both shutdowns due to large-scale construction work and shutdowns for other reasons correspond to cases where exhaust gas is not being generated, being able to recognize such status information proves that exhaust gas is not being generated, improving the value of the information.

(Processing flow of the information processing device of the third embodiment)
In the information processing device 1 of the third embodiment, when the operating status of the equipment associated with the analysis target is non-operating in the following steps (1) to (7), the non-operating status (cause of non-operation) is estimated based on the location information of the mobile terminal 31.

(1) For the image obtained by photographing, geographic information is assigned to the areal space within the range in which the population is counted (areal space 501, which is a specified region corresponding to areal space 511 shown in FIG. 23), without being limited to the range in which the analysis target is specified.

(2) Obtain location information for each of multiple devices based on geographic information.

(3) Divide the areal space of the image into multiple meshes 5011 with a vertical and horizontal distance of k1 (Figures 24 and 25). If each device (mobile terminal 31) is present in any mesh 5011, enter the number of devices in that mesh, and if not, enter 0. If the areal space 501 to be analyzed is not rectangular, expand the length vertically and horizontally, fill it with 0s, and create an array with l1 vertical and m1 horizontal.

(4) This array is classified into a learning array and an inference array. If at least a portion of the image (area space 511) acquired in the first embodiment is included in the area space 501 specified in the third embodiment, matching is performed. However, it is kept independent as a classifier.

(5) For the learning sequences, the sequence during a period of major construction such as a regular inspection is labeled as "large-scale construction in progress" as status information for the analysis target, and the sequence during other periods is labeled as "normal."

(6) As in the second embodiment, a general classifier and an individual classifier are created using the labeled data, and the inference array is used to classify the plant status. (The relationship between the general classifier and the individual classifier is the same as that in the first embodiment, where the general classifier is used, and the parameters Wji known in the first embodiment are used as initial values, and the individual classifier is generated as a result of re-learning using the images and labels of each object.)

(7) In the third embodiment, the plant status information inferred by the first and second embodiments as "shutdown" is further classified as "large-scale construction in progress" or "normal operation."

(Example of operation of information processing device)
Next, an example of the operation of the information processing device 1 during the location information acquisition process and the learning phase process will be described with reference to Fig. 21. During the location information acquisition and learning phase process, the information processing device 1 first acquires device location information based on the geographic information of the analysis target (step S171), divides the image into meshes (step S172), determines whether or not a device exists in the mesh (step S173), and if a device exists (if "Y" in step S173), sets the number of pieces of location information for the mesh. The information processing device 1 repeatedly executes the processes of steps S173 and S174 until the process is completed for all meshes of all images (until both steps S175 and S176 are "Y").

Next, the information processing device 1 classifies the sequences into learning sequences and inference sequences (step S177), selects a learning sequence (step S178), displays the selected learning sequence (step S179), and labels the learning sequence (step S180). Until the selection is completed (until "Y" is displayed in step S181), the information processing device 1 repeatedly executes the processes of steps S178 to S180.

Next, the information processing device 1 performs learning using the labeled learning sequence (step S182), stores it as a learned model (step S183), and generates an individual classifier (step S184).

Next, referring to FIG. 22, an example of the operation of the information processing device 1 during the processing of the estimation phase will be described. During the processing of the estimation phase, the information processing device 1 first acquires an inference array (step S191), then inputs the inference array to the trained model (step S192), and estimates the non-operating state (step S193). Until acquisition is completed (until "Y" is obtained in step S194), the information processing device 1 repeatedly executes the processing of steps S191 to S193. Next, the information processing device 1 determines whether or not the plant has an estimated operating state (step S195), and if the plant has an estimated operating state (if "Y" is obtained in step S195), the information processing device 1 integrates the results (step S196) and outputs the tabulated results (step S197). On the other hand, if the plant has not an estimated operating state (if "N" is obtained in step S195), the information processing device 1 outputs the tabulated results (step S197).

Fourth Embodiment
Next, an information processing device, an information processing method, and a program according to a fourth embodiment of the present disclosure will be described with reference to Fig. 26 to Fig. 28. Fig. 26 and Fig. 27 are flowcharts showing an example of the operation of the information processing device according to the fourth embodiment of the present disclosure. Fig. 28 is a schematic diagram showing an example of an area space according to the fourth embodiment of the present disclosure.

The fourth embodiment deals with cases where the production volume is difficult to estimate depending on the analysis object (plant, chimney, etc.) in the first to third embodiments. For example, on cloudy days, it may be difficult to estimate the production volume because it is difficult to determine emissions (white smoke, gas, fog, etc.) from images obtained by photographing. Therefore, in the fourth embodiment, by using images photographed by an infrared probe, it becomes possible to handle a larger number of analysis objects for which production volume is estimated.

(Processing flow of the information processing device of the fourth embodiment)
In the information processing device 1 of the fourth embodiment, the production volume of the equipment associated with the analysis target is calculated in the following steps (1) to (5).

(1) For any given analysis target, draw semicircles of radius r centered on the inlet and outlet ports for the cooling water intake area (area space 515 in FIG. 28) and the warm wastewater discharge area (area space 514 in FIG. 28), specify an area space for each, and assign geographic information. Furthermore, for any given analysis target, specify area spaces 516 and 517 for the exhaust gas outlet duct (chimney 601) and the boiler building, respectively, and assign geographic information.

(2) The attribute information of the analysis target is associated with the area space to which the geographic information has been assigned, and stored as, for example, facility information 18 shown in FIG. 2. For example, the attribute information of the object is associated with the marked chimney for the power plant based on publicly available information on the Internet, external services, and appearance and dimensions, and stored in the attribute information database 181.

(3) Depending on the performance of the remote sensing device (thermal infrared image sensor), luminance information of the infrared sensor is obtained for the image acquired based on the above area space.
(3-1) Obtain brightness information for the pixels of the inlet and outlet ports for the cooling water intake area and the warm wastewater discharge area, and obtain the difference between the average brightness value on the inlet side (area space 515) and each pixel on the outlet side (area space 514).
(3-2) For the exhaust gas outlet duct and the boiler building, luminance information of the respective infrared sensors is obtained for the images acquired based on the above-mentioned area spaces 516 and 517.

(4) Each difference value is multiplied by a constant k2 proportional to the distance from the exit port, and the values are classified into a learning array and an inference array.

Reason for setting constant k2: The surface temperature of seawater, rivers, and lakes decreases the farther away from the outlet port, so data from a more distant location should be taken into account in determining the operating status even if it is a small change from the inlet temperature. A difference of 1°C at a distant location is more reliable in determining operation than a difference of 1°C at a nearby location.
In addition, for the exhaust gas outlet duct and the boiler building, the acquired luminance information is classified into a learning array and an inference array without multiplying it by the constant k2.
When an image acquired in another embodiment is included in a plant specified in the fourth embodiment, the image is associated with the plant, but the image is made independent as a classifier.

(5) Using the luminance information classified into the learning array, status information is added and labeling is performed as in the first embodiment, and learning is performed using a learning model to generate a classifier. The relationship between the general-purpose classifier and the individual classifier is the same as that in the first embodiment, where the general-purpose classifier is the one in which the parameters Wji known in the first embodiment are set as initial values, and the individual classifier is generated as a result of re-learning using the images and labels of each object.

(Example of operation of information processing device)
Next, an example of the operation of the information processing device 1 during data acquisition and learning phase processing will be described with reference to Fig. 26. The information processing device 1 first acquires luminance information from an image (thermal infrared image) (step S201), and if it is a predetermined area (if "Y" in step S203), acquires the difference in luminance information between areas (step S204), and corrects the luminance information (step S205). If it is not a predetermined area (if "N" in step S203), or if the luminance information has been corrected (step S205), the information processing device 1 determines whether or not processing of all images has been completed (step S206), and repeatedly executes the processing from step S203 to step S205 until it is completed (until "Y" in step S206).

Next, the information processing device 1 classifies the sequences into learning sequences and inference sequences (step S207), selects a learning sequence (step S208), displays the selected learning sequence (step S209), and labels the learning sequence (step S210). Until the selection is completed (until "Y" is displayed in step S211), the information processing device 1 repeatedly executes the processes from step S208 to step S210.

Next, the information processing device 1 performs learning using the labeled learning sequence (step S212), stores it as a learned model (step S213), and generates an individual classifier (step S214).

Next, referring to FIG. 27, an example of the operation of the information processing device 1 during the estimation phase will be described. In the estimation phase, the information processing device 1 acquires data for inference (step S221), inputs the data for inference to the trained model (step S222), and outputs (records) information about the operating state (step S223). The information processing device 1 repeatedly executes the processes of steps S221 to S223 until acquisition is completed (until "Y" is obtained in step S224). When acquisition is completed (when "Y" is obtained in step S224), the information processing device 1 calculates the amount of power generation based on the information about the operating state and the attribute information (step S225), tallying up the amount of power generation for each area (step S226), and outputs the tallying up result (step S227).

Fifth Embodiment
Next, an information processing device, an information processing method, and a program according to a fifth embodiment of the present disclosure will be described with reference to Fig. 29. Fig. 29 is a flowchart showing an operation example of the information processing device according to the fifth embodiment of the present disclosure.

In the fifth embodiment, it is possible to estimate the production volume when the production volume and status information cannot be estimated in the first to fourth embodiments.

(Processing flow of the information processing device of the fifth embodiment)
In the information processing device 1 of the fifth embodiment, the production volume of the equipment associated with the analysis target is calculated according to the following steps (1) to (4).

(1) Identify the system to which the analysis object (plant, chimney, etc.) whose production volume you want to estimate is subordinate, and make a list of the components of that system (including the analysis object). If the analysis object whose production volume you want to estimate is a power plant, the area under the jurisdiction of the central load dispatching center of a company or other entity responsible for controlling frequency and voltage is considered the "subordinate system."

The components that need to be listed are the power generation facilities, large-scale demand facilities, substations, and power transmission networks related to the demand and supply of electricity, and their attribute information needs to include information such as the type of power generation (thermal, wind, hydroelectric, etc.), output, heat rate, load change rate, time periods when power generation is possible, power generation constraints, plant shutdown information, contracted power volume and voltage, subordinate substations, and transmission line distance.

(2) Prioritize the listed components for operation using attribute information, etc. For example, if there are 10 wind power generation facilities, 30 thermal power generation facilities, and 10 hydroelectric power generation facilities as power generation facilities, a score is assigned in advance to prioritize each facility. The higher the number, the higher the priority, so all renewable energy sources are assigned a score of 100, thermal power generation supercritical units are assigned 60, sub-critical units are assigned 45, and so on, and the score setting is further varied according to the efficiency (heat rate) of the unit, etc.

(3) Based on the total demand in the system, the analysis targets are considered to be operating in descending order of score. For example, when estimating the amount of power generated on January 10, 2019 at a plant in the area under the jurisdiction of Hokkaido Electric Power Network Co., Ltd., first, the power demand for that time period is estimated based on weather conditions and past performance. Next, for example, if there is a total power demand of 4,000 MWh (publicly available information), power generation equipment is selected according to score (for example, in descending order of score) to reach the required amount, and the equipment generating power in the system is identified, taking into account the time periods when power generation is possible and power generation constraints.

(4) Based on the identified equipment, use it as an estimate of the production volume of the subject of analysis.

(Example of operation of information processing device)
Next, an example of the operation of the information processing device 1 will be described with reference to FIG.

In the operation example shown in FIG. 29, first, the information processing device 1 determines whether or not estimation is possible using images and position information (step S231). If it is possible (if "Y" in step S231), the information processing device 1 executes the learning phase processing and the like in the same manner as in embodiment 1 and the like (step S232). On the other hand, if it is not possible (if "N" in step S231), the information processing device 1 identifies the system to which the analysis target is dependent (step S233), identifies the components of the dependent system (step S234), calculates the scores of the components (step S235), calculates the total demand (step S236), estimates the operating state (step S237), and estimates the production volume (step S238).

<Business flow>
Next, a business flow (business model) according to each of the above embodiments will be described with reference to Fig. 30. Fig. 30 is a system diagram showing an example of a business flow according to an embodiment of the present disclosure.

The information processing device 1 according to each of the above embodiments is operated by a business management company 809 shown in FIG. 30. The information processing device 1 provides the estimated production volume (e.g., power generation volume) as data 808 indicating a power generation volume trend to a mobile terminal 807 managed by a customer 805. When the power generation volume trend satisfies, for example, conditions set by the customer 805, the mobile terminal 807 issues a transaction signal 806 indicating a favorable opportunity for a transaction such as an investment. In this case, the business management company 809 acquires detection information 803 such as satellite image data from the data supply company 802, and also acquires arbitrary data 804 from the available data 801. In this case, the fee for the exchange of information flows from the customer 805 to the business management company 809 via the sales channel 810, and also flows from the business management company 809 to the data supply company 802.

(Action and Effect)
According to the information processing device, the information processing method, and the program configured as described above, the operating state of the facility can be estimated.

Other Embodiments
Although the embodiments of the present disclosure have been described in detail above with reference to the drawings, the specific configuration is not limited to this embodiment, and design changes and the like that do not depart from the gist of the present disclosure are also included.

Computer Configuration
FIG. 31 is a schematic block diagram illustrating a configuration of a computer according to at least one embodiment.
The computer 90 comprises a processor 91 , a main memory 92 , a storage 93 , and an interface 94 .
The information processing device 1 described above is implemented in a computer 90. The operations of each of the above-mentioned processing units are stored in the form of a program in a storage 93. The processor 91 reads the program from the storage 93, loads it in the main memory 92, and executes the above-mentioned processing in accordance with the program. The processor 91 also secures storage areas in the main memory 92 corresponding to each of the above-mentioned storage units in accordance with the program.

The program may be for realizing some of the functions to be performed by the computer 90. For example, the program may be for realizing the functions by combining with other programs already stored in the storage, or by combining with other programs implemented in other devices. In other embodiments, the computer may include a custom LSI (Large Scale Integrated Circuit) such as a PLD (Programmable Logic Device) in addition to or instead of the above configuration. Examples of PLDs include PAL (Programmable Array Logic), GAL (Generic Array Logic), CPLD (Complex Programmable Logic Device), FPGA (Field Programmable Gate Array), etc. In this case, some or all of the functions realized by the processor may be realized by the integrated circuit.

Examples of storage 93 include HDD (Hard Disk Drive), SSD (Solid State Drive), magnetic disk, magneto-optical disk, CD-ROM (Compact Disc Read Only Memory), DVD-ROM (Digital Versatile Disc Read Only Memory), semiconductor memory, etc. Storage 93 may be an internal medium directly connected to the bus of computer 90, or an external medium connected to computer 90 via interface 94 or a communication line. In addition, when this program is distributed to computer 90 via a communication line, computer 90 that receives the program may expand the program in main memory 92 and execute the above-mentioned processing. In at least one embodiment, storage 93 is a non-transitory tangible storage medium.

<Additional Notes>
The information processing device 1 described in each embodiment can be understood, for example, as follows.

(1) The information processing device 1 according to the first aspect includes an acquisition unit 11 that acquires detection information detected in the air, including status information corresponding to the status of a specified emission material 41 within a specified area space 51 on the ground, and an operation status estimation unit 12 that estimates the operation status of equipment 71 associated with the area space 51 based on the detection information. According to this aspect and each of the following aspects, the operation status of the equipment can be estimated.

(2) The information processing device 1 according to the second aspect is the information processing device 1 according to (1), in which the equipment 71 is production equipment, and further includes a production volume estimation unit 13 that estimates the production volume of the production equipment based on the operating state.

(3) The information processing device 1 according to the third aspect is the information processing device 1 according to (1) or (2), in which the operation state estimation unit 12 estimates the operation state using a trained model 16 that inputs the detection information and outputs a classification result that classifies the operation state into at least operation or non-operation.

(4) The information processing device 1 according to the fourth aspect is the information processing device 1 according to (3), in which the trained model 16 includes a model (individually trained model 162) that has been retrained for each areal space using, as initial values, the weighting coefficients Wij between neurons in each layer of a neural network trained using multiple pieces of detection information corresponding to multiple different areal spaces as training data.

(5) The information processing device 1 according to the fifth aspect is the information processing device 1 according to (3) or (4), in which the trained model 16 further inputs meteorological information corresponding to the areal space and outputs the classification result.

(6) The information processing device 1 according to the sixth aspect is the information processing device 1 according to any one of (3) to (5), in which the trained model 16 further inputs the previous classification result and outputs the classification result.

(7) The information processing device 1 according to the seventh aspect is the information processing device 1 according to any one of (1) to (6), and further includes a location information acquisition unit 14 that acquires location information of one or more mobile terminals 31 within a predetermined area (area space 501) corresponding to the area space 511, and a state estimation unit 15 that estimates a non-operating state when the operating state is non-operating based on the location information.

(8) The information processing device 1 according to the eighth aspect is the information processing device 1 according to any one of (1) to (7), in which the emission 41 is white smoke, and the detection information is color or monochrome image data.

(9) The information processing device 1 according to the ninth aspect is the information processing device 1 according to (2), in which the emission 41 is white smoke, the detection information is color or monochrome image data, and the production volume estimation unit 13 estimates the production volume by correcting it according to the season.

(10) The information processing device 1 according to the tenth aspect is the information processing device 1 according to any one of (1) to (7), in which the emission 41 is heat and the detection information is thermal infrared image data.

(11) The information processing device 1 according to the eleventh aspect is the information processing device 1 according to (2), in which the operation state estimation unit 13 estimates the operation state of the equipment based on a predetermined temperature difference based on the detection information.

Reference Signs List 1 Information processing device 11 Acquisition unit 12 Operation state estimation unit 13 Production amount estimation unit 14 Position information acquisition unit 15 State estimation unit 16 Trained model 31 Mobile terminal 41 Waste 51 Area space 71 Equipment 31 Mobile terminal 162 Individual trained model

Claims (15)

an acquisition unit that acquires detection information detected in the air, the detection information including status information corresponding to a status of a predetermined emission within a predetermined area space on the ground;
an operation state estimation unit that estimates an operation state of a facility associated with the area space based on the detection information;
Equipped with
the operation state estimation unit estimates the operation state using a trained model that receives the detection information and outputs a classification result that classifies the operation state into at least operation or non-operation;
The trained model includes a model that is retrained for each of the areal spaces using weighting coefficients between neurons in each layer of a neural network trained using a plurality of pieces of detection information corresponding to a plurality of different areal spaces as training data as initial values.
Information processing device . an acquisition unit that acquires detection information detected in the air, the detection information including status information corresponding to a status of a predetermined emission within a predetermined area space on the ground;
an operation state estimation unit that estimates an operation state of a facility associated with the area space based on the detection information;
Equipped with
the operation state estimation unit estimates the operation state using a trained model that receives the detection information and outputs a classification result that classifies the operation state into at least operation or non-operation;
The trained model further inputs meteorological information corresponding to the areal space and outputs the classification result.
Information processing device . The trained model further inputs meteorological information corresponding to the areal space and outputs the classification result.
The information processing device according to claim 1 . an acquisition unit that acquires detection information detected in the air, the detection information including status information corresponding to a status of a predetermined emission within a predetermined area space on the ground;
an operation state estimation unit that estimates an operation state of a facility associated with the area space based on the detection information;
Equipped with
the operation state estimation unit estimates the operation state using a trained model that receives the detection information and outputs a classification result that classifies the operation state into at least operation or non-operation;
The trained model is an information processing device that further inputs the previous classification result and outputs the classification result. The information processing device according to claim 1 , wherein the trained model further inputs a previous classification result and outputs the classification result. an acquisition unit that acquires detection information detected in the air, the detection information including status information corresponding to a status of a predetermined emission within a predetermined area space on the ground;
an operation state estimation unit that estimates an operation state of a facility associated with the area space based on the detection information;
a location information acquisition unit that acquires location information of one or more mobile devices within a predetermined area corresponding to the area space;
a state estimation unit that estimates a non-operating state when the operating state is non-operating based on the position information;
An information processing device comprising : a location information acquisition unit that acquires location information of one or more mobile devices within a predetermined area corresponding to the area space;
a state estimation unit that estimates a non-operating state when the operating state is non-operating based on the position information;
The information processing device according to claim 1 , further comprising: the emissions are white smoke,
The information processing apparatus according to claim 1 , wherein the detection information is color or monochrome image data. an acquisition unit that acquires detection information detected in the air, the detection information including status information corresponding to a status of a predetermined emission within a predetermined area space on the ground;
an operation state estimation unit that estimates an operation state of a facility associated with the area space based on the detection information;
a production volume estimation unit that estimates a production volume of the production facility based on the operation state;
Equipped with
the emissions are white smoke,
the detection information is color or monochrome image data,
The production amount estimation unit estimates the production amount by correcting it according to the season.
Information processing device . the exhaust is heat;
The information processing apparatus according to claim 1 , wherein the detection information is thermal infrared image data. The information processing device according to claim 10 , wherein the operation state estimation unit estimates the operation state of the facility based on a predetermined temperature difference based on the detection information. The facility is a production facility,
The information processing device according to claim 1 , further comprising a production volume estimation unit that estimates a production volume of the production facility based on the operation state. an acquisition unit that acquires detection information detected in the air, the detection information including status information corresponding to a status of a predetermined emission within a predetermined area space on the ground;
an operation state estimation unit that estimates an operation state of a facility associated with the area space based on the detection information;
a production volume estimation unit that estimates a production volume of the production facility based on the operation state;
Equipped with
When the production facility is a power generation facility, if the operation state estimation unit cannot estimate the operation state based on the detection information, the production amount estimation unit estimates the production amount of the power generation facility based on a configuration of a system to which the power generation facility belongs.
Information processing device . acquiring airborne sensed detection information including condition information corresponding to a given emission condition within a given area space on the ground;
estimating an operating state of a facility associated with the area space based on the detection information;
Including,
The step of estimating the operating state includes estimating the operating state using a trained model that inputs the detection information and outputs a classification result that classifies the operating state into at least operating or non-operating;
The trained model includes a model that is retrained for each of the areal spaces using weighting coefficients between neurons in each layer of a neural network trained using a plurality of pieces of detection information corresponding to a plurality of different areal spaces as training data as initial values.
Information processing methods. acquiring airborne sensed detection information including condition information corresponding to a given emission condition within a given area space on the ground;
estimating an operating state of a facility associated with the area space based on the detection information;
A program for causing a computer to execute the following:
The step of estimating the operating state includes estimating the operating state using a trained model that inputs the detection information and outputs a classification result that classifies the operating state into at least operating or non-operating;
The trained model includes a model that is retrained for each of the areal spaces using weighting coefficients between neurons in each layer of a neural network trained using a plurality of pieces of detection information corresponding to a plurality of different areal spaces as training data as initial values.
program .

Images