Prediction model making device and method, exhaust gas concentration control system and method

Abstract

A prediction model creation device and method, as well as an exhaust gas concentration control system and method, are capable of rapidly creating an exhaust gas concentration prediction model that improves the accuracy of predicting the concentration of gas components contained in exhaust gas discharged from a furnace. The prediction model creation device, used to create an exhaust gas concentration prediction model for predicting the concentration of gas components contained in exhaust gas discharged from a furnace, comprises: an image acquisition unit that acquires an image of the combustion area of ​​the furnace; an extraction unit that converts the image into color image data and extracts feature quantities from the color image data; an exhaust gas concentration acquisition unit that acquires the concentration of gas components contained in exhaust gas downstream of the combustion area after a predetermined set time has passed since the image was captured; and a model creation unit that creates the exhaust gas concentration prediction model by machine learning learning data that establishes a correspondence between the feature quantities extracted by the extraction unit and the concentrations of the gas components obtained by the exhaust gas concentration acquisition unit.

Description

Technical Field

[0001] The present invention relates to a prediction model making device and a prediction model making method for making an exhaust gas concentration prediction model for predicting the concentration of gas components contained in the exhaust gas discharged from a furnace, and an exhaust gas concentration control system and an exhaust gas concentration control method for controlling the concentration of gas components contained in the exhaust gas discharged from a furnace using the exhaust gas concentration prediction model. Background Art

[0002] For example, Patent Document 1 discloses the following technology: using an estimation model (exhaust gas concentration prediction model) produced by machine learning of learning data, the CO concentration (furnace process value) is estimated based on an image (feature quantity) taken of the inside of the furnace, and the learning data is formed by establishing a correspondence between image information obtained from the image taken of the inside of the furnace and a state quantity (e.g., CO concentration) after a specified time has passed from the time the image is taken.

[0003] Prior art literature

[0004] Patent Literature

[0005] Patent Document 1: Japanese Patent Application Publication No. 2019-074240

[0006] However, although the feature amount is preferably information that improves the prediction accuracy of the exhaust gas concentration prediction model, there is no disclosure regarding a specific method for extracting the feature amount from image data of the combustion region of the equipment. Summary of the Invention

[0007] The present invention has been made in view of the above-mentioned problems, and an object of the present invention is to provide a prediction model creation device and a prediction model creation method capable of improving the prediction accuracy of an exhaust gas concentration prediction model.

[0008] In order to achieve the above-mentioned purpose, the prediction model making device of the present invention makes an exhaust gas concentration prediction model for predicting the concentration of gas components contained in the exhaust gas discharged from the furnace, wherein the prediction model making device comprises: an image acquisition unit, which acquires an image of the combustion area of ​​the furnace; an extraction unit, which converts the image into color image data and extracts feature quantities from the color image data; an exhaust gas concentration acquisition unit, which acquires the concentration of the gas components contained in the exhaust gas after a predetermined set time has passed from the time when the image is taken at a location downstream of the combustion area; and a model making unit, which makes the exhaust gas concentration prediction model by mechanically learning learning data in which a correspondence is established between the feature quantities extracted by the extraction unit and the concentrations of the gas components obtained by the exhaust gas concentration acquisition unit.

[0009] In order to achieve the above-mentioned purpose, the prediction model production method of the present invention produces an exhaust gas concentration prediction model for predicting the concentration of gas components contained in the exhaust gas discharged from a furnace, wherein the prediction model production method includes the following steps: obtaining an image of the combustion area of ​​the furnace; converting the image into color image data, and extracting feature quantities from the color image data; obtaining the concentration of the gas components contained in the exhaust gas after a predetermined set time has passed from the time when the image was captured at a location downstream of the combustion area; and producing the exhaust gas concentration prediction model by mechanically learning learning data obtained by establishing a correspondence between the feature quantities and the concentration of the gas components contained in the exhaust gas after a set time has passed from the time when the image was captured.

[0010] Effects of the Invention

[0011] According to the prediction model preparation device and prediction model preparation method of the present invention, the prediction accuracy of the exhaust gas concentration prediction model can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 This is a diagram schematically showing the structure of a furnace in which the prediction model creation device according to the first embodiment is installed.

[0013] Figure 2 This is a schematic functional block diagram of the prediction model creation device according to the first embodiment.

[0014] Figure 3 is a diagram showing color image data.

[0015] Figure 4 This is a diagram schematically showing the structure of a furnace in which the prediction model creation device according to the second embodiment is installed.

[0016] Figure 5 This is a schematic functional block diagram of the prediction model creation device according to the second embodiment.

[0017] Figure 6 This is a diagram schematically showing the configuration of an exhaust gas concentration control system according to a third embodiment.

[0018] Figure 7 It is a diagram for explaining the operation and effects of the exhaust gas concentration control system according to the third embodiment.

[0019] Figure 8 This is a diagram schematically showing the configuration of an exhaust gas concentration control system according to a modified example of the third embodiment.

[0020] Figure 9 This is a flowchart illustrating a prediction model creation method according to one embodiment.

[0021] Figure 10 1 is a flow chart showing an exhaust gas concentration control method according to one embodiment.

[0022] Description of reference numerals:

[0023] 1...Prediction model making device;

[0024] 2...image acquisition unit;

[0025] 4...Extraction Department;

[0026] 6...Exhaust gas concentration acquisition unit;

[0027] 8...Model Making Department;

[0028] 10...Process data acquisition department;

[0029] 50...Exhaust gas concentration control system;

[0030] 52...shooting device;

[0031] 54...Adjustment device;

[0032] 100...furnace;

[0033] 116...filming device;

[0034] 130...burning area;

[0035] 140...Gas Analyzer;

[0036] 142...pressure gauge;

[0037] 144...Temperature sensor;

[0038] 146...Heat recovery boiler;

[0039] 160...color image data;

[0040] A...color area;

[0041] Eg...exhaust gas;

[0042] S2...image acquisition step;

[0043] S4...feature extraction step;

[0044] S6...concentration acquisition step;

[0045] S8...Model making steps;

[0046] S52...shooting steps;

[0047] S54...Adjustment steps. DETAILED DESCRIPTION

[0048] The following describes a prediction model creation device, an exhaust gas concentration control system, a prediction model creation method, and an exhaust gas concentration control method according to an embodiment of the present invention, based on the accompanying drawings. This embodiment illustrates one aspect of the present invention and does not limit the present invention. The present invention can be modified arbitrarily within the scope of the technical concept of the present invention.

[0049] <First embodiment>

[0050] (Furnace structure)

[0051] Figure 1 1 is a diagram schematically showing the structure of a furnace 100 in which the prediction model creation device 1 according to the first embodiment is installed. Figure 1 In the illustrated embodiment, furnace 100 is a grate-type waste incinerator that uses municipal waste, industrial waste, or biomass as solid fuel Fs. It should be noted that furnace 100 is not limited to a grate-type waste incinerator. Hereinafter, the direction in which solid fuel Fs moves within furnace 100 is referred to as movement direction W1.

[0052] like Figure 1 As shown, the furnace 100 includes a hopper 102 , a feeder 104 , a combustion chamber 106 , an air supply 108 , and a flue 110 .

[0053] Solid fuel Fs fed from hopper 102 accumulates in passage 103 extending toward combustion chamber 106. Feeder 104 is disposed in passage 103 and is configured to reciprocate along the extending direction (movement direction W1) of passage 103. Feeder 104 pushes solid fuel Fs accumulated in passage 103 toward combustion chamber 106.

[0054] The combustion chamber 106 includes a grate 112 (grate) corresponding to the bottom of the combustion chamber 106 and onto which the solid fuel Fs pushed into the combustion chamber 106 falls. The grate 112 is configured to move the solid fuel Fs on the grate 112 in a direction away from the receiving port 114 of the combustion chamber 106 (from the upstream side to the downstream side of the movement direction W1). The combustion chamber 106 includes a drying zone 128, a combustion zone 130, and a post-combustion zone 132, which are arranged in sequence from the upstream side to the downstream side of the movement direction W1. In the drying zone 128, the solid fuel Fs is dried using the heat within the combustion chamber 106. In the combustion zone 130, the flame Fr is increased to cause the solid fuel Fs to combust. In the post-combustion zone 132, the unburned burnt materials in the combustion zone 130 are completely combusted. The solid fuel Fs dried, burned, and post-combusted in the combustion chamber 106 becomes ash Fa, which is discharged outside the furnace 100.

[0055] The burning area 130 is photographed by a photographing device 116 such as a visible light camera. Figure 1 In the illustrated embodiment, the combustion area 130 is imaged by a camera 116 installed at the furnace end 118 of the combustion chamber 106. It should be noted that the camera 116 is not limited to a visible light camera, as long as it can capture the combustion area 130. For example, the camera 116 may also be an infrared camera. Furthermore, the location of the camera 116 is not limited to the furnace end 118 of the combustion chamber 106.

[0056] The air supply device 108 is configured to supply primary air for combustion of the solid fuel Fs and secondary air for further combustion of the exhaust gas Eg generated by the combustion of the solid fuel Fs to the combustion chamber 106. Figure 1 In the illustrated embodiment, the air supply device 108 includes an air supply pipe 120 and a blower 122 provided in the air supply pipe 120. A portion of the air flowing in the air supply pipe 120 is supplied as primary air from the grate 112 to the lower portion of the combustion chamber 106 via the first flow regulating valve 124, and the remaining portion is supplied as secondary air from the side wall of the combustion chamber 106 to the upper portion of the combustion chamber 106 via the second flow regulating valve 126. That is, the air supply device 108 functions as a primary air supply device that supplies primary air to the lower portion of the combustion chamber 106, and functions as a secondary air supply device that supplies secondary air to the upper portion of the combustion chamber 106. It should be noted that, in Figure 1 In the illustrated embodiment, primary air is supplied to each of the drying zone 128 , the combustion zone 130 , and the post-combustion zone 132 of the combustion chamber 106 .

[0057] The flue 110 communicates with the upper portion of the combustion chamber 106 and allows exhaust gas Eg generated by the combustion of the solid fuel Fs to flow through. A gas analyzer 140 is provided at the outlet 111 of the flue 110. The gas analyzer 140 measures the concentrations of gas components such as oxygen and carbon monoxide contained in the exhaust gas Eg discharged from the furnace 100.

[0058] It should be noted that in Figure 1 In the illustrated embodiment, the furnace 100 is configured as part of an incineration system 150. Figure 1As shown, the incineration system 150 includes a furnace 100, a cooling tower 152, a dust collector 154, and a chimney 156. The cooling tower 152 is connected to the outlet portion 111 of the flue 110 of the furnace 100. The cooling tower 152 reduces the temperature of the exhaust gas Eg discharged from the furnace 100 (the flue 110 of the furnace 100). The dust collector 154 is connected to the cooling tower 152. The dust collector 154 captures fly ash contained in the exhaust gas Eg discharged from the cooling tower 152. The chimney 156 is connected to the dust collector 154. The chimney 156 discharges the exhaust gas Eg discharged from the dust collector 154 to the outside of the incineration system 150, such as into the atmosphere.

[0059] (Structure of Prediction Model Creation Device)

[0060] The prediction model creation device 1 installed in the furnace 100 acquires data (images and gas component concentrations) from the imaging device 116 and the gas analyzer 140, respectively. Based on this acquired data, it creates an exhaust gas concentration prediction model that predicts the concentrations of gas components contained in the exhaust gas Eg discharged from the furnace 100. It should be noted that the prediction model creation device 1 may also be configured to be electrically connected to the imaging device 116 and the gas analyzer 140, respectively, to acquire data in real time. Alternatively, the prediction model creation device 1 may be configured to store data previously acquired from the imaging device 116 and the gas analyzer 140 in a database, and then acquire the data from the database.

[0061] The prediction model creation device 1 is a computer such as an electronic control device, and includes a processor such as a CPU and a GPU (not shown), a memory such as a ROM and a RAM, and an I / O interface. The prediction model creation device 1 realizes the various functional units of the prediction model creation device 1 by causing the processor to perform operations (calculations, etc.) according to the instructions of the program loaded into the memory. Figure 2 , each functional unit of the prediction model creation device 1 according to one embodiment will be described.

[0062] Figure 2 This is a schematic functional block diagram of the prediction model creation device 1 according to the first embodiment. Figure 2 As shown, the prediction model creation device 1 includes an image acquisition unit 2 , an extraction unit 4 , an exhaust gas concentration acquisition unit 6 , and a model creation unit 8 .

[0063] The image acquisition unit 2 acquires an image of the combustion area 130 of the furnace 100 from the imaging device 116 .

[0064] The extraction unit 4 converts the image acquired by the image acquisition unit 2 into color image data 160. Specifically, the extraction unit 4 performs clustering processing to convert the image acquired by the image acquisition unit 2 into color image data 160 divided into a plurality of color regions A based on color information. The extraction unit 4 then extracts features from the color image data 160.

[0065] An example of "dividing an image into multiple color areas A based on color information through clustering processing" will be described. The color information refers to the RGB color components, and the multiple color areas A are each set so that the RGB color components do not overlap with each other through clustering processing. The extraction unit 4 decomposes the image acquired by the image acquisition unit 2 into the RGB color components for each pixel and identifies the color area A containing that pixel. It should be noted that color information is not limited to the RGB color components and may also include brightness or chroma.

[0066] The clustering algorithm is not particularly limited, and a known clustering algorithm can be used. For example, an algorithm that can specify the number of clusters, such as k-means, or an algorithm that automatically determines the number of clusters, such as flowsom, can be used for clustering.

[0067] Figure 3 is a diagram showing color image data 160. Figure 3 In the illustrated embodiment, color image data 160 is divided into seven color regions A through clustering. These regions, in descending order of brightness, include a first color region A1 (A), a second color region A2 (A), a third color region A3 (A), a fourth color region A4 (A), a fifth color region A5 (A), a sixth color region A6 (A), and a seventh color region A7 (A). The first through seventh color regions A1 through A7 are converted to black and white (grayscale) values, with the values ​​becoming darker as one moves from the first color region A1 to the seventh color region A7.

[0068] Next, an example of "extracting a feature from color image data" will be described. The extraction unit 4 calculates the total number of pixels divided into the first color area A1 and extracts this total number of pixels as a feature. The extraction unit 4 then extracts the total number of pixels in the first color area A1 at predetermined intervals (e.g., every second). The extraction unit 4 also calculates the total number of pixels at predetermined intervals for each of the second through seventh color areas A2 through A7, and extracts each total number of pixels as a feature.

[0069] It should be noted that, in the first embodiment, the feature quantity includes the total number of pixels of all color areas (first color area A1 to seventh color area A7) in the plurality of color areas A, but the present invention is not limited to this method. The feature quantity only needs to include the total number of pixels of at least one color area A in the plurality of color areas A. In some embodiments, the feature quantity includes the total number of pixels of the color area A that has a high correlation with the concentration of the gas component in the plurality of color areas A. Here, the color area A with a high correlation can be a color area A with a correlation higher than a predetermined correlation, or a color area A selected from the plurality of color areas A in order of high correlation.

[0070] The exhaust gas concentration acquisition unit 6 acquires the concentration of gas components contained in the exhaust gas Eg downstream of the combustion zone 130 after a predetermined set time t has elapsed since the image was captured. In other words, the exhaust gas concentration acquisition unit 6 acquires the concentration of gas components measured by the gas analyzer 140 after the set time t has elapsed since the image from which the total number of pixels was extracted was captured. The set time t is calculated based on, for example, the length of the flue 110 and the distance from the combustion zone 130 to the gas analyzer 140.

[0071] Model creation unit 8 creates an exhaust gas concentration prediction model by performing machine learning on learning data that establishes a correspondence between the total number of pixels (features) extracted by extraction unit 4 and the concentrations of gas components acquired by exhaust gas concentration acquisition unit 6. Specifically, model creation unit 8 performs regression analysis on multiple learning data sets whose images were captured at different times to create a regression equation for calculating the concentration of the gas component based on the total number of pixels.

[0072] (Functions / Effects of Prediction Modeling Device)

[0073] According to the findings of the present inventors, the concentration of the gas components contained in the exhaust gas Eg discharged from the furnace 100 has a high correlation with the total number of pixels extracted from the color image data 160 divided into multiple color areas A according to color information, in the information obtained from the image of the combustion area 130 of the furnace 100.

[0074] According to the first embodiment, the exhaust gas concentration prediction model created by the model creation unit 8 is created through machine learning of learning data, which is data that establishes a correspondence between the total number of pixels extracted from the color image data 160 and the concentration of the gas component measured by the gas analyzer 140 at a set time t from the time the image from which the total number of pixels was extracted was captured. Therefore, the prediction accuracy of the exhaust gas concentration prediction model can be improved. In addition, since clustering processing is applied when converting the image to the color image data 160, there is no need to prepare teacher data. In addition, clustering processing does not require manual adjustment work such as pre-processing and threshold setting. Therefore, it is possible to facilitate the automation of feature extraction.

[0075] The gas analyzer 140 is sometimes located downstream of the source of the exhaust gas Eg. Therefore, the concentration of the gas component measured by the gas analyzer 140 includes a time lag between the source of the exhaust gas Eg and the point of measurement. To address this issue, according to the first embodiment, the exhaust gas concentration acquisition unit 6 acquires the concentration of the gas component measured by the gas analyzer 140 at the time a set time t has elapsed since the image was captured. Therefore, by creating learning data that takes this time lag into account, the prediction accuracy of the exhaust gas concentration prediction model can be improved.

[0076] The total number of pixels is one type of information that can be quickly extracted. According to the first embodiment, since the extraction unit 4 extracts the total number of pixels as a feature value, the model creation unit 8 can quickly create an exhaust gas concentration prediction model.

[0077] <Second embodiment>

[0078] A prediction model creation device 1 according to a second embodiment of the present invention will be described. The second embodiment differs from the first embodiment in that a process data acquisition unit 10 is further provided. The remaining configuration is the same as that described in the first embodiment. In the second embodiment, components identical to those of the first embodiment are denoted by the same reference numerals, and detailed description thereof will be omitted.

[0079] (Furnace structure)

[0080] First, the structure of the furnace 100 in which the prediction model creation device 1 according to the second embodiment is installed will be described. Figure 4 This figure schematically shows the structure of a furnace 100 equipped with the prediction model creation device 1 according to the second embodiment. The furnace 100 according to the second embodiment further includes a pressure gauge 142, a temperature sensor 144, and a heat recovery boiler 146 for measuring process data of the furnace 100.

[0081] The pressure gauge 142 is provided at the upper portion of the combustion chamber 106 to measure the pressure in the combustion chamber 106 (inside the furnace 100). The temperature sensor 144 is provided at the flue 110 to measure the temperature of the exhaust gas Eg flowing in the flue 110 as the temperature in the furnace 100. The heat recovery boiler 146 is provided at the flue 110 to generate steam using the thermal energy of the exhaust gas Eg flowing in the flue 110. In this embodiment, the heat recovery boiler 146 is configured to be able to measure the amount of steam generated by the heat recovery boiler 146. It should be noted that in Figure 4 In the illustrated embodiment, the temperature sensor 144 , the heat recovery boiler 146 , and the gas analyzer 140 are provided in the flue 110 in this order from the upstream side toward the downstream side in the flow direction of the exhaust gas Eg. However, the present invention is not limited to this embodiment.

[0082] (Structure of Prediction Model Creation Device)

[0083] Figure 5 This is a schematic functional block diagram of the prediction model creation device 1 according to the second embodiment. Figure 4 and Figure 5 As shown in the example, the prediction model creation device 1 obtains data (images, temperature / pressure in the furnace 100, steam volume, and concentration of gas components) from the imaging device 116, the gas analyzer 140, the pressure gauge 142, the temperature sensor 144, and the heat recovery boiler 146, and based on the obtained data, creates an exhaust gas concentration prediction model that predicts the concentration of gas components contained in the exhaust gas Eg discharged from the furnace 100.

[0084] like Figure 5 As shown in FIG. 1 , the prediction model creation device 1 further includes a process data acquisition unit 10 for acquiring process data. Figure 5 In the illustrated embodiment, the process data acquisition unit 10 acquires the pressure within the combustion chamber 106 (the pressure within the furnace 100) measured by the pressure gauge 142, the temperature of the exhaust gas Eg flowing through the flue 110 (the temperature within the furnace 100) measured by the temperature sensor 144, and the amount of steam generated by the heat recovery boiler 146. The process data acquisition unit 10 acquires process data measured at the same time as the image was captured. It should be noted that in the second embodiment, the process data includes the pressure within the furnace 100, the temperature within the furnace 100, and the amount of steam, but the present invention is not limited to this embodiment. The process data only needs to include at least one of the pressure within the furnace 100, the temperature within the furnace 100, and the amount of steam.

[0085] The model generation unit 8 creates an exhaust gas concentration prediction model by machine learning based on learning data that establishes a correlation between the total number of pixels (features) extracted by the extraction unit 4, the concentrations of gas components obtained by the exhaust gas concentration acquisition unit 6, and the process data obtained by the process data acquisition unit 10. Specifically, the model generation unit 8 performs linear regression analysis on multiple learning data sets with different image capture times to create a regression equation for calculating the concentration of the gas component based on the process data and the total number of pixels. It should be noted that the basis function in the linear regression can also be a polynomial, a Gaussian function, or other function to minimize the error in the predicted value based on the regression analysis.

[0086] (Functions / Effects of Prediction Modeling Device)

[0087] The concentration of gas components contained in the exhaust gas Eg exhausted from the furnace 100 is also highly correlated with the process data. According to the second embodiment, since the exhaust gas concentration prediction model is created taking the process data into consideration, the prediction accuracy of the exhaust gas concentration prediction model can be further improved.

[0088] In some embodiments, the exhaust gas concentration acquisition unit 6 also acquires the concentration of the gas component contained in the exhaust gas Eg at the time the image was captured, i.e., the pre-correction concentration. The model creation unit 8 then creates an exhaust gas concentration prediction model by performing machine learning on learning data that correlates the total number of pixels (feature quantity) extracted by the extraction unit 4, the concentration of the gas component and the pre-correction concentration obtained by the exhaust gas concentration acquisition unit 6, and the process data obtained by the process data acquisition unit 10.

[0089] According to the findings of the present inventors, the error (RMSE) of the exhaust gas concentration prediction model can be reduced by performing machine learning on learning data that also includes the pre-correction concentration. This structure further improves the prediction accuracy of the exhaust gas concentration prediction model by also considering the pre-correction concentration when creating the exhaust gas concentration prediction model.

[0090] <Third embodiment>

[0091] The exhaust gas concentration control system 50 of the third embodiment of the present invention is described. The exhaust gas concentration control system 50 uses the exhaust gas concentration prediction model produced by the above-mentioned prediction model production device 1 to control the concentration of gas components contained in the exhaust gas Eg discharged from the furnace 100. In the third embodiment, the exhaust gas concentration control system 50 uses the exhaust gas concentration prediction model produced by the prediction model production device 1 of the first embodiment to control the concentration of oxygen. In another embodiment, the exhaust gas concentration control system 50 uses the exhaust gas concentration prediction model produced by the prediction model production device 1 of the second embodiment to control the concentration of oxygen. It should be noted that the concentration of the gas component controlled by the exhaust gas concentration control system 50 is not limited to the concentration of oxygen, for example, it can also be the concentration of carbon monoxide or the concentration of nitrogen oxides (NOx).

[0092] (Structure of exhaust gas concentration control system)

[0093] Figure 6 FIG. 5 is a diagram schematically showing the configuration of an exhaust gas concentration control system 50 according to a third embodiment. Figure 6 As shown, the exhaust gas concentration control system 50 includes an imaging device 52 and an adjustment device 54 .

[0094] The imaging device 52 captures the combustion area 130 of the furnace 100. In the second embodiment, the imaging device 52 is the same as the imaging device 116 described above and is installed at the furnace tail 118 of the combustion chamber 106. It should be noted that in another embodiment, the imaging device 52 may be a device different from the imaging device 116 as long as it is configured to capture the combustion area 130 of the furnace 100.

[0095] The adjustment device 54 is electrically connected to the imaging device 52 and acquires the image captured by the imaging device 52 in real time. The adjustment device 54 extracts the total number of pixels (feature quantity) from the captured image using the above-described method and inputs the extracted total number of pixels into the exhaust gas concentration prediction model created by the prediction model creation device 1. The adjustment device 54 adjusts the amount of air supplied to the furnace 100 based on the oxygen concentration (hereinafter referred to as the predicted value X1) output by the exhaust gas concentration prediction model based on the total number of pixels.

[0096] It should be noted that in Figure 6 In the illustrated embodiment, the adjustment device 54 is provided separately from the prediction model creation device 1 and obtains the exhaust gas concentration prediction model from the prediction model creation device 1. In another embodiment, the adjustment device 54 is configured to function as the prediction model creation device 1 and input the total number of pixels extracted from the captured image into the exhaust gas concentration prediction model created by the adjustment device 54 itself.

[0097] An example of the operation of the adjustment device 54 of the third embodiment will be described. Figure 6 In the illustrated embodiment, the adjustment device 54 is electrically connected to the blower 122, the first flow control valve 124, and the second flow control valve 126. By sending electrical signals (hereinafter referred to as "instructions") to the blower 122, the first flow control valve 124, and the second flow control valve 126, the adjustment device 54 can adjust the fan speed of the blower 122, the opening of the first flow control valve 124, and the opening of the second flow control valve 126. Furthermore, when the predicted value X1 is less than a reference value of the adjustment device 54, the adjustment device 54 instructs the blower 122 to increase the fan speed, the first flow control valve 124 to increase the opening, and the second flow control valve 126 to increase the opening. In other words, when the predicted value X1 is less than the reference value of the adjustment device 54, the adjustment device 54 increases the amount of air supplied to the furnace 100. The reference value of the adjustment device 54 is a pre-set value, and is a value at which the combustion state of the furnace 100 becomes unstable when the concentration of oxygen contained in the exhaust gas Eg discharged from the furnace 100 becomes lower than this value. The reference value is set to, for example, 2 wet% or less. If the combustion state is stable, the oxygen concentration at the outlet of the furnace 100 (the outlet portion 111 of the flue 110) fluctuates within a range of approximately 1 wet%. Therefore, if the combustion state is unstable, the reference value will rarely become less than 2 wet%. Therefore, by setting the reference value to less than 2 wet%, the adjustment device 54 can increase the amount of air supplied to the furnace 100 at the timing of the transition period to the unstable state (that is, at an appropriate time), thereby restoring the stable state.

[0098] In some embodiments, the amount of air supplied to the furnace 100 is determined based on the difference between the predicted value X1 and a preset target value. The target value is set to be greater than the reference value of the adjustment device 54. This target value can be set arbitrarily. For example, the oxygen concentration contained in the exhaust gas Eg is set to 3 wet % so that the air ratio is 1.2.

[0099] It should be noted that in the third embodiment, the adjustment device 54 adjusts the fan speed of the blower 122, the opening of the first flow control valve 124, and the opening of the second flow control valve 126, respectively. However, the present invention is not limited to this embodiment. The adjustment device 54 may also adjust at least one of the fan speed of the blower 122, the opening of the first flow control valve 124, and the opening of the second flow control valve 126.

[0100] When the predicted value X1 returns to the reference value by increasing the amount of air supplied to the furnace 100, the adjustment device 54 instructs the blower 122 to reduce its fan speed, instructs the first flow control valve 124 to reduce its opening, and instructs the second flow control valve 126 to reduce its opening. Specifically, when the predicted value X1 returns to the reference value, the adjustment device 54 reduces the amount of air supplied to the furnace 100. In some embodiments, when the predicted value X1 returns to the reference value, the adjustment device 54 adjusts the fan speed of the blower 122, the opening of the first flow control valve 124, and the opening of the second flow control valve 126 to their values ​​before adjustment (increase) by the adjustment device 54.

[0101] (Function / Effect of Exhaust Gas Concentration Control System)

[0102] Figure 7 : is a diagram for explaining the operation and effect of the exhaust gas concentration control system 50 according to the third embodiment. Figure 7 , a graph is shown with oxygen concentration on the vertical axis and time on the horizontal axis. The solid line represents the oxygen concentration output from the exhaust gas concentration prediction model (predicted value X1). The dashed line represents the oxygen concentration obtained in real time from the gas analyzer 140 (hereinafter referred to as the measured value X2). The dotted line represents the oxygen concentration obtained by shifting the dashed line by an amount corresponding to the set time t (hereinafter referred to as the corrected value X3).

[0103] As described above, the oxygen concentration measured by the gas analyzer 140 includes a time lag from the generation point of the exhaust gas Eg to the measurement point. Therefore, it may be inappropriate to adjust the amount of air supplied to the furnace 100 based on the actual measurement value X2. Figure 7 As shown, the actual measurement value X2 is delayed by an amount corresponding to the set time t, and it is appropriate to adjust the amount of air supplied to the furnace 100 based on the correction value X3.

[0104] According to the third embodiment, the oxygen concentration contained in the exhaust gas Eg exhausted from the furnace 100 is controlled based on the oxygen concentration (predicted value X1) predicted by the exhaust gas concentration prediction model created with time lag in mind. Therefore, compared to controlling the oxygen concentration based on the actual measured value X2, the oxygen concentration contained in the exhaust gas Eg exhausted from the furnace 100 can be controlled at an earlier stage. Furthermore, according to the third embodiment, the control of the oxygen concentration contained in the exhaust gas Eg exhausted from the furnace 100 can be automated.

[0105] When the oxygen concentration in the exhaust gas Eg is low, the carbon monoxide concentration in the exhaust gas Eg increases, often leading to unstable combustion conditions in the furnace 100. According to the third embodiment, when the predicted value X1 (oxygen concentration) is less than a reference value, the amount of air supplied to the furnace 100 is increased. This stabilizes the combustion conditions in the furnace 100 and reduces the carbon monoxide concentration in the exhaust gas Eg discharged from the furnace 100.

[0106] On the other hand, increasing the amount of air supplied to the furnace 100 may promote the generation of nitrogen oxides when the solid fuel Fs is burned in the combustion chamber 106. According to the third embodiment, when the predicted value X1 returns to the reference value by increasing the amount of air supplied to the furnace 100, the amount of air supplied to the furnace 100 is reduced. Therefore, the concentration of carbon monoxide contained in the exhaust gas Eg discharged from the furnace 100 can be reduced while also reducing the concentration of nitrogen oxides contained in the exhaust gas Eg.

[0107] <Modification of the Third Embodiment>

[0108] In the third embodiment, the adjusting device 54 adjusts the amount of air supplied to the furnace 100 , but the adjusting device 54 may adjust the amount of solid fuel Fs supplied to the furnace 100 instead of or in addition to this. Figure 8 FIG. 1 is a diagram schematically showing the configuration of an exhaust gas concentration control system 50 according to a modified example of the third embodiment. Figure 8 As shown, the adjustment device 54 is electrically connected to the feeder 104 and can adjust the reciprocating speed of the feeder 104 by sending an electrical signal to the feeder 104 .

[0109] An example of the operation of the adjustment device 54 according to a modified example of the third embodiment will be described. When the predicted value X1 is less than the reference value of the adjustment device 54, the adjustment device 54 instructs the feeder 104 to decrease its reciprocating speed. Specifically, when the predicted value X1 is less than the reference value of the adjustment device 54, the adjustment device 54 reduces the amount of solid fuel Fs supplied to the furnace 100.

[0110] According to such a configuration, when the predicted value X1 is smaller than the reference value, the amount of solid fuel Fs supplied to the furnace 100 is reduced, thereby reducing the concentration of carbon monoxide contained in the exhaust gas Eg exhausted from the furnace 100 .

[0111] In some embodiments, the adjustment device 54 is configured to adjust the speed of the grate 112, and when the predicted value X1 is less than a reference value of the adjustment device 54, the speed of the grate 112 is reduced. With this configuration, when the predicted value X1 is less than the reference value, the stirring of the solid fuel Fs supplied to the combustion chamber 106 is suppressed, thereby reducing the concentration of carbon monoxide contained in the exhaust gas Eg discharged from the furnace 100.

[0112] <Prediction model creation method>

[0113] The prediction model preparation method of the present invention is a method of preparing an exhaust gas concentration prediction model for predicting the concentration of gas components contained in the exhaust gas Eg exhausted from the furnace 100 . Figure 9 FIG. 1 is a flowchart illustrating a method for creating a prediction model according to an embodiment of the present invention. Figure 9 As shown, the prediction model creation method includes an image acquisition step S2 , a feature extraction step S4 , a concentration acquisition step S6 , and a model creation step S8 .

[0114] In the image acquisition step S2, an image of the combustion area 130 of the furnace 100 is acquired. In the feature extraction step S4, the image acquired in the image acquisition step S2 is converted into color image data 160 divided into a plurality of color areas A according to color information by clustering processing, and feature quantities are extracted from the color image data 160. In the concentration acquisition step S6, the concentration of the gas components contained in the exhaust gas Eg after a predetermined set time t has passed since the image was captured is acquired at a location downstream of the combustion area 130. In the model making step S8, a waste gas concentration prediction model is made by performing machine learning on the learning data obtained by establishing a correspondence between the feature extracted in the feature extraction step S4 and the concentration of the gas component obtained in the concentration acquisition step S6. It should be noted that the concentration acquisition step S6 only needs to be executed before the model making step S8 and is not limited to the step S8. Figure 9 the way exemplified.

[0115] According to this prediction model creation method, the exhaust gas concentration prediction model is created through machine learning of learning data. This learning data is obtained by using clustering processing to convert an image of the combustion area 130 of the furnace 100 into color image data 160 divided into multiple color areas A by color. The data is then mapped to the concentration of gas components contained in the exhaust gas Eg after a set time has passed since the image was captured. This improves the prediction accuracy of the exhaust gas concentration prediction model. Furthermore, by applying clustering processing when converting the image to color image data 160, it is no longer necessary to prepare teacher data, allowing for rapid creation of the exhaust gas concentration prediction model.

[0116] <Exhaust gas concentration control method>

[0117] The exhaust gas concentration control method of the present invention is a method of controlling the concentration of gas components contained in the exhaust gas Eg exhausted from the furnace 100 using the exhaust gas concentration prediction model created by the above-mentioned prediction model creation method. Figure 10 : is a flow chart showing an exhaust gas concentration control method according to one embodiment. Figure 10 As shown, the exhaust gas concentration control method includes a photographing step S52 and an adjusting step S54.

[0118] In the imaging step S52, an image of the combustion area 130 of the furnace 100 is captured. In the adjustment step S54, at least one of the amount of air supplied to the furnace 100 and the amount of solid fuel Fs supplied to the furnace is adjusted based on the concentration of the gas component output by inputting the image captured in the imaging step S52 into the exhaust gas concentration prediction model.

[0119] According to such an exhaust gas concentration control method, the concentration of the gas component is controlled based on the (predicted) concentration of the gas component output from the exhaust gas concentration prediction model, so the concentration of the gas component can be appropriately controlled compared to the case where the concentration of the gas component is controlled based on the concentration of the gas component obtained in real time.

[0120] The contents described in each of the above-mentioned embodiments can be understood, for example, as follows.

[0121] [1] The prediction model preparation device (1) of the present invention prepares an exhaust gas concentration prediction model for predicting the concentration of gas components contained in exhaust gas (Eg) discharged from a furnace (100), wherein:

[0122] The prediction model making device (1) comprises:

[0123] An image acquisition unit (2) that acquires an image of the combustion area (130) of the furnace;

[0124] an extraction unit (4) that converts the image into color image data (160) and extracts a feature value from the color image data;

[0125] an exhaust gas concentration acquisition unit (6) that acquires, at a location downstream of the combustion region, the concentration of the gas component contained in the exhaust gas after a predetermined set time has elapsed from the time when the image is captured; and

[0126] A model making unit (8) makes the exhaust gas concentration prediction model by performing machine learning on learning data in which the feature quantity extracted by the extraction unit and the concentration of the gas component acquired by the exhaust gas concentration acquisition unit establish a correspondence relationship.

[0127] According to the findings of the present inventors, the concentration of gas components contained in exhaust gas is highly correlated with the feature quantity (information) extracted from color image data, which is obtained from an image of the combustion area of ​​a furnace. According to the structure described in [1] above, the exhaust gas concentration prediction model is created by machine learning of learning data, which is obtained by converting an image of the combustion area of ​​a furnace into color image data and establishing a correspondence between the feature quantity extracted from the color image data and the concentration of gas components contained in the exhaust gas after a set time has passed since the image was captured. Therefore, the prediction accuracy of the exhaust gas concentration prediction model can be improved.

[0128] [2] In some embodiments, based on the structure described in [1] above,

[0129] The color image data includes data (A) obtained by dividing the image into a plurality of color regions according to color information through clustering processing.

[0130] According to the structure described in [2] above, since clustering processing is applied when converting an image into color image data, it is not necessary to prepare teacher data. In addition, clustering processing does not require manual adjustment work such as pre-processing and threshold setting. Therefore, it is possible to easily automate the extraction of feature quantities.

[0131] [3] In some embodiments, based on the structure described in [1] or [2] above,

[0132] The feature amount includes the area of ​​at least one of the plurality of color regions.

[0133] The area of ​​a color region is one type of information that can be quickly acquired. According to the configuration described in [3] above, since the feature quantity includes the area of ​​the color region, it is possible to quickly prepare learning data.

[0134] [4] In some embodiments, based on the structure described in any one of [1] to [3] above,

[0135] The prediction model preparation device further includes a process data acquisition unit (10) that acquires process data including at least one of the temperature in the furnace, the pressure in the furnace, and the amount of steam generated by combustion of fuel in the furnace.

[0136] The model creation unit creates the exhaust gas concentration prediction model by performing machine learning on learning data in which the feature value extracted by the extraction unit, the concentration of the gas component acquired by the exhaust gas concentration acquisition unit, and the process data acquired by the process data acquisition unit are associated with each other.

[0137] The concentration of gas components contained in the exhaust gas discharged from the furnace is also highly correlated with the process data. According to the structure described in [4] above, since the exhaust gas concentration prediction model is also created taking the process data into consideration, the prediction accuracy of the exhaust gas concentration prediction model can be further improved.

[0138] [5] In some embodiments, based on the structure described in [4] above,

[0139] The exhaust gas concentration acquisition unit further acquires the concentration of the gas component contained in the exhaust gas at the time when the image is captured, that is, the pre-correction concentration.

[0140] The model making unit makes the exhaust gas concentration prediction model by performing machine learning on learning data formed by establishing a correspondence between the feature value extracted by the extraction unit, the concentration of the gas component and the pre-correction concentration obtained by the exhaust gas concentration acquisition unit, and the process data obtained by the process data acquisition unit.

[0141] According to the findings of the present inventors, the error (RMSE) of the exhaust gas concentration prediction model can be reduced by performing machine learning on learning data that also includes the pre-correction concentration. According to the structure described in [5] above, since the exhaust gas concentration prediction model is created by also taking the pre-correction concentration into account, the prediction accuracy of the exhaust gas concentration prediction model can be further improved.

[0142] [6] The exhaust gas concentration control system (50) of the present invention uses the exhaust gas concentration prediction model produced by the prediction model production device described in any one of [1] to [5] above to control the concentration of gas components contained in the exhaust gas discharged from the furnace, wherein:

[0143] The exhaust gas concentration control system (50) comprises:

[0144] a photographing device (52) for photographing the combustion area of ​​the furnace; and

[0145] An adjustment device (54) is configured to adjust at least one of the amount of air supplied to the furnace and the amount of fuel supplied to the furnace based on the concentration of the gas component output by inputting the feature quantity extracted from the captured image captured by the imaging device into the exhaust gas concentration prediction model.

[0146] Adjusting the amount of air or fuel supplied to the furnace based on the concentration of the gas component obtained in real time is sometimes inappropriate. According to the configuration described in [6] above, the concentration of the gas component is controlled based on the (predicted) concentration of the gas component output from the exhaust gas concentration prediction model. Therefore, the concentration of the gas component can be appropriately controlled compared to a case where the concentration of the gas component is controlled based on the concentration of the gas component obtained in real time. In addition, according to the configuration described in [6] above, the control of the concentration of the gas component can be automated.

[0147] [7] In some embodiments, based on the structure described in [6] above,

[0148] The gas component includes oxygen,

[0149] When the oxygen concentration output from the exhaust gas concentration prediction model is lower than a reference value of the adjustment device, the adjustment device increases the amount of air supplied to the furnace.

[0150] When the oxygen concentration in the exhaust gas is low, the carbon monoxide concentration in the exhaust gas increases, and the furnace combustion state often becomes unstable. According to the configuration described in [7] above, when the oxygen concentration output from the exhaust gas concentration prediction model is lower than the reference value, the amount of air supplied to the furnace is increased. As a result, the carbon monoxide concentration in the exhaust gas discharged from the furnace can be reduced.

[0151] [8] In some embodiments, based on the structure described in [7] above,

[0152] The amount of air supplied to the furnace is determined based on the difference between the oxygen concentration output from the exhaust gas concentration prediction model and a preset target value.

[0153] According to the configuration described in [8] above, the amount of air supplied to the furnace is determined based on the difference between the oxygen concentration output from the exhaust gas concentration prediction model and the target value, thereby making it possible to bring the oxygen concentration contained in the exhaust gas discharged from the furnace close to the target value.

[0154] [9] In some embodiments, based on the structure described in [7] or [8] above,

[0155] When the oxygen concentration output from the exhaust gas concentration prediction model returns to a reference value of the adjustment device, the adjustment device reduces the amount of air supplied to the furnace.

[0156] Increasing the amount of air supplied to the furnace may promote the generation of nitrogen oxides when burning fuel in the furnace. According to the configuration described in [9] above, when the oxygen concentration output from the exhaust gas concentration prediction model returns to a reference value by increasing the amount of air supplied to the furnace, the amount of air supplied to the furnace is reduced. Therefore, it is possible to reduce the concentration of carbon monoxide contained in the exhaust gas discharged from the furnace while reducing the concentration of nitrogen oxides contained in the exhaust gas.

[0157]

[10] In some embodiments, based on the structure described in any one of [6] to [9] above,

[0158] The gas component includes oxygen,

[0159] The adjusting device reduces the amount of fuel supplied to the furnace when the oxygen concentration output from the exhaust gas concentration prediction model is lower than a reference value of the adjusting device.

[0160] According to the configuration described in

[10] above, when the oxygen concentration output from the exhaust gas concentration prediction model is lower than the reference value, the amount of fuel supplied to the furnace is reduced. This reduces the concentration of carbon monoxide contained in the exhaust gas discharged from the furnace.

[0161]

[11] The prediction model preparation method of the present invention prepares an exhaust gas concentration prediction model for predicting the concentration of gas components contained in the exhaust gas discharged from the furnace, wherein:

[0162] The prediction model preparation method comprises:

[0163] Step (S2), in which an image of the combustion area of ​​the furnace is obtained;

[0164] Step (S4), in which the image is converted into color image data divided into a plurality of color regions according to color information by clustering processing, and a feature quantity is extracted from the color image data;

[0165] Step (S6), in which the concentration of the gas component contained in the exhaust gas after the predetermined set time has elapsed from the time when the image is captured is obtained at a location downstream of the combustion area; and

[0166] Step (S8), in which the exhaust gas concentration prediction model is produced by mechanically learning the learning data obtained by establishing a correspondence between the feature quantity and the concentration of the gas component contained in the exhaust gas after a set time has passed since the image was captured.

[0167] According to the method described in

[11] , the exhaust gas concentration prediction model is created by machine learning of learning data obtained by converting an image of the combustion area of ​​the furnace into color image data and establishing a correspondence between the feature quantities extracted from the color image data and the concentrations of gas components contained in the exhaust gas after a set time has passed since the image was captured. Therefore, the prediction accuracy of the exhaust gas concentration prediction model can be improved.

[0168]

[12] The exhaust gas concentration control method of the present invention uses the exhaust gas concentration prediction model produced by the prediction model production method described in

[11] to control the concentration of gas components contained in the exhaust gas discharged from the furnace, wherein:

[0169] The exhaust gas concentration control method includes:

[0170] a photographing step (S52), in which the combustion area of ​​the furnace is photographed; and

[0171] An adjustment step (S54), in which at least one of the amount of air supplied to the furnace and the amount of fuel supplied to the furnace is adjusted based on the concentration of the gas component output by inputting the feature quantity extracted from the captured image captured in the capturing step into the exhaust gas concentration prediction model.

[0172] According to the method described in

[12] above, since the concentration of the gas component is controlled based on the (predicted) concentration of the gas component output from the exhaust gas concentration prediction model, the concentration of the gas component can be appropriately controlled compared to the case where the concentration of the gas component is controlled based on the concentration of the gas component obtained in real time.

Claims

1. A prediction model generating device for generating an exhaust gas concentration prediction model for predicting the concentration of gas components contained in exhaust gas discharged from a grate-type waste incinerator, wherein: The grate-type waste incinerator is a structure provided with a flue, through which the exhaust gas generated by the combustion of solid fuel flows. A gas analyzer is provided at the outlet of the flue, and measures the concentration of the gas components contained in the exhaust gas discharged from the grate-type waste incinerator. The prediction model making device comprises: an image acquisition unit configured to acquire an image of a combustion area of ​​the grate-type waste incinerator; an extraction unit that converts the image into color image data by clustering the image into a plurality of color regions based on RGB values, and extracts a total number of pixels of at least one of the plurality of color regions that has a correlation with the concentration of the gas component that is greater than or equal to a predetermined value as a feature value; an exhaust gas concentration acquiring unit that acquires, at an outlet of the flue located downstream of a point where the exhaust gas is generated, using the gas analyzer, a concentration of the gas component contained in the exhaust gas after a predetermined time period calculated based on the length of the flue and the distance from the combustion area to the gas analyzer has elapsed since the image was captured; and A model creating unit creates the exhaust gas concentration prediction model by performing machine learning on learning data in which the feature amount extracted by the extraction unit and the concentration of the gas component acquired by the exhaust gas concentration acquisition unit are associated with each other.

2. The prediction model creation device according to claim 1, wherein: The prediction model preparation device further includes a process data acquisition unit that acquires process data including at least one of a temperature in the grate-type waste incinerator, a pressure in the grate-type waste incinerator, and an amount of steam generated by combustion of fuel in the grate-type waste incinerator. The model creation unit creates the exhaust gas concentration prediction model by performing machine learning on learning data in which the feature value extracted by the extraction unit, the concentration of the gas component acquired by the exhaust gas concentration acquisition unit, and the process data acquired by the process data acquisition unit are associated with each other.

3. The prediction model creation device according to claim 2, wherein: The exhaust gas concentration acquisition unit further acquires the concentration of the gas component contained in the exhaust gas at the time when the image is captured, that is, the pre-correction concentration. The model making unit makes the exhaust gas concentration prediction model by performing machine learning on learning data formed by establishing a correspondence between the feature value extracted by the extraction unit, the concentration of the gas component and the pre-correction concentration obtained by the exhaust gas concentration acquisition unit, and the process data obtained by the process data acquisition unit.

4. A waste gas concentration control system, which uses the waste gas concentration prediction model produced by the prediction model production device according to any one of claims 1 to 3 to control the concentration of gas components contained in the waste gas discharged from the grate-type waste incinerator, wherein: The exhaust gas concentration control system comprises: a photographing device for photographing the combustion area of ​​the grate-type waste incinerator; and An adjustment device is configured to adjust at least one of the amount of air supplied to the grate-type waste incinerator and the amount of fuel supplied to the grate-type waste incinerator based on the concentration of the gas component output by inputting the feature quantity extracted from the captured image captured by the capturing device into the exhaust gas concentration prediction model.

5. The exhaust gas concentration control system according to claim 4, wherein: The gas component includes oxygen, When the oxygen concentration output from the exhaust gas concentration prediction model is lower than a reference value of the adjustment device, the adjustment device increases the amount of air supplied to the grate-type waste incinerator.

6. The exhaust gas concentration control system according to claim 5, wherein: The amount of air supplied to the grate-type waste incinerator is determined based on the difference between the oxygen concentration output from the exhaust gas concentration prediction model and a preset target value.

7. The exhaust gas concentration control system according to claim 5 or 6, wherein: When the oxygen concentration output from the exhaust gas concentration prediction model returns to a reference value of the adjustment device, the adjustment device reduces the amount of air supplied to the grate-type waste incinerator.

8. The exhaust gas concentration control system according to any one of claims 4 to 6, wherein: The gas component includes oxygen, When the oxygen concentration output from the exhaust gas concentration prediction model is lower than a reference value of the adjustment device, the adjustment device reduces the amount of fuel supplied to the grate-type waste incinerator.

9. A method for producing a prediction model for producing an exhaust gas concentration prediction model for predicting the concentration of gas components contained in exhaust gas discharged from a grate-type waste incinerator, wherein: The grate-type waste incinerator is a structure provided with a flue, through which the exhaust gas generated by the combustion of solid fuel flows. A gas analyzer is provided at the outlet of the flue, and measures the concentration of the gas components contained in the exhaust gas discharged from the grate-type waste incinerator. The prediction model preparation method comprises the following steps: Obtaining an image of a combustion area of ​​the grate-type waste incinerator; converting the image into color image data by dividing the image into a plurality of color regions based on RGB values ​​through clustering processing, and extracting a total number of pixels of at least one of the plurality of color regions having a correlation with the concentration of the gas component that is greater than or equal to a predetermined value as a feature value; acquiring, by the gas analyzer, at an outlet of the flue located downstream of a point where the flue gas is generated, a concentration of the gas component contained in the flue gas after a predetermined time period calculated based on the length of the flue and the distance from the combustion area to the gas analyzer has elapsed since the image was captured; and The exhaust gas concentration prediction model is created by performing machine learning on learning data in which the feature value is associated with the concentration of the gas component contained in the exhaust gas after the set time has elapsed from the time when the image is captured.

10. A method for controlling exhaust gas concentration, comprising controlling the concentration of gas components contained in the exhaust gas discharged from the grate-type waste incinerator using the exhaust gas concentration prediction model produced by the prediction model production method according to claim 9, wherein: The exhaust gas concentration control method includes: a photographing step, in which the combustion area of ​​the grate-type waste incinerator is photographed; and An adjustment step, in which at least one of the amount of air supplied to the grate-type waste incinerator and the amount of fuel supplied to the grate-type waste incinerator is adjusted based on the concentration of the gas component output by inputting the feature quantity extracted from the captured image captured in the capturing step into the exhaust gas concentration prediction model.

Images