Program, information processing method, information processor and combustion image analysis system

Abstract

To provide a program, etc. for outputting an analysis result on a combustion state.SOLUTION: A program acquires a combustion image of biomass fuel taken by an imaging section 26 provided in a biomass combustion furnace 1 and causes a computer to execute processing for outputting an analysis result on a combustion state on the basis of the acquired combustion image. Preferably, the program displays a burnout point with respect to each section on the basis of the combustion image. Preferably, when the combustion image of the biomass fuel is input, the program outputs information on a combustion area by inputting the acquired combustion image to a learning model learned to output the information on the combustion area included in the combustion image, and specifies the burning point with respect to each section on the basis of the information on the output combustion area.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present disclosure relates to a program, an information processing method, an information processing device, and a combustion image analysis system. [Background technology]

[0002] Patent Document 1 discloses a biomass fuel facility that stabilizes combustion operation by accurately measuring the moisture content of biomass fuel before it is combusted in a combustion furnace. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2022-69028 Summary of the Invention [Problem to be solved by the invention]

[0004] However, the invention of Patent Document 1 has a problem in that it is not possible to output analysis results regarding the combustion state based on the combustion image of biomass fuel.

[0005] In one aspect, an object of the present invention is to provide a program or the like that outputs analysis results relating to the combustion state. [Means for solving the problem]

[0006] A program according to one aspect acquires a combustion image of biomass fuel captured by an imaging unit provided in a biomass combustion furnace, and outputs an analysis result regarding the combustion state based on the acquired combustion image. [Effects of the Invention]

[0007] In one aspect, it is possible to output analytical results regarding the combustion state. [Brief explanation of the drawings]

[0008] [Figure 1] FIG. 1 is an explanatory diagram showing an overview of a combustion image analysis system. [Figure 2] FIG. 1 is an explanatory diagram illustrating a biomass combustion furnace. [Figure 3] FIG. 2 is a block diagram showing the hardware configuration of a server. [Figure 4] FIG. 10 is an explanatory diagram showing an example of a combustion image. [Figure 5] FIG. 10 is an explanatory diagram showing a method for identifying a burned area by binarization. [Figure 6] FIG. 1 is a schematic diagram of a combustion area identification model. [Figure 7] FIG. 10 is an explanatory diagram showing a method for identifying a burnout point. [Figure 8] FIG. 10 is an explanatory diagram showing time-series changes in the burnout point. [Figure 9] FIG. 10 is an explanatory diagram showing an example of an additional image displayed in association with a graph showing time-series changes in the burn-out point; [Figure 10] FIG. 10 is an explanatory diagram showing time series changes in the overall average brightness value and the burned area average brightness value. [Figure 11] FIG. 1 is a schematic diagram of a deposition area identification model. [Figure 12] FIG. 10 is an explanatory diagram showing a method for identifying the center of gravity position. [Figure 13] FIG. 10 is an explanatory diagram showing time-series changes in the center of gravity position in the X-axis direction for the deposition area. [Figure 14] FIG. 10 is an explanatory diagram showing time-series changes in the center of gravity position in the Y-axis direction for the deposition area. [Figure 15] FIG. 10 is an explanatory diagram showing the proportion of the deposition area in the combustion image in chronological order. [Figure 16] FIG. 10 is an explanatory diagram showing the ratio of the deposition area to the combustion area in time series. [Figure 17] 10 is a flowchart showing a procedure for generating a combustion area identification model. [Figure 18] 10 is a flowchart showing a procedure for generating a deposition area identification model. [Figure 19] 1 is a flowchart showing the processing of the combustion image analysis system. [Figure 20] 10 is a flowchart showing a process for acquiring information about a combustion area. [Figure 21] 10 is a flowchart illustrating a process for obtaining information about a deposition area. DETAILED DESCRIPTION OF THE INVENTION

[0009] (Embodiment 1) Figure 1 is an explanatory diagram showing an overview of a combustion image analysis system. The combustion image analysis system includes a biomass combustion furnace 1 and an information processing device 2. When a combustion image is acquired by an imaging unit provided in the biomass combustion furnace 1, the combustion image analysis system outputs an analysis result of the combustion state.

[0010] The information processing device 2 is an information processing device operated by an operator of the biomass combustion furnace 1, and is, for example, a server computer (hereinafter referred to as a server) or a personal computer (hereinafter referred to as a computer), etc. In this embodiment, the information processing device 2 will be described as the server 2.

[0011] FIG. 2 is an explanatory diagram illustrating a biomass combustion furnace. The biomass combustion furnace 1 includes a fuel discharger 11, a traveling stoker 12, an air supplier 13, an outlet 14, and an imaging unit 26. The traveling stoker 12 includes a belt unit 121 that connects multiple fire grates in a ring shape, and a drive wheel 122 and a driven wheel 123 that are arranged inside the belt unit 121. The belt unit 121 is arranged at the bottom of the biomass combustion furnace 1. The drive wheel 122 and the driven wheel 123 are cylindrical. The central axes of the drive wheel 122 and the driven wheel 123 are arranged horizontally. The central axes of the drive wheel 122 and the driven wheel 123 are arranged parallel to each other. The drive wheel 122 rotates counterclockwise, causing the belt unit 121 to rotate counterclockwise. As the belt unit 121 rotates, the driven wheel 123 also rotates counterclockwise. The belt portion 121 is stretched almost horizontally above the drive wheel 122 and the driven wheel 123. The fuel discharger 11 is arranged so that the fuel discharger 11 is located above the drive wheel 122. The air supplier 13 is arranged below the driven wheel 123. The outlet 14 is arranged below the drive wheel 122. The imaging unit 26 is arranged in a position where it can capture an image of the upper surface of the belt portion 121.

[0012] The fuel discharger 11 discharges biomass fuel onto the driven wheel 123 side of the upper surface of the belt portion 121. The belt portion 121 transports the discharged biomass fuel from right to left in FIG. 1. The biomass fuel is burned during the transport process by the belt portion 121. The air supplier 13 blows air for combustion of the biomass fuel from the bottom to the top of the belt portion 121. The combustion ash of the biomass fuel is discharged from the belt portion 121 to the discharge port 14. The imaging unit 26 captures an image of the process of the biomass fuel burning on the belt portion 121. In this embodiment, the image captured by the imaging unit 26 will be described as a combustion image.

[0013] 3 is a block diagram showing the hardware configuration of the server 2. The server 2 includes a control unit 21, a storage unit 22, a communication unit 23, an input unit 24, a display unit 25, an imaging unit 26, a mass storage unit 27, and a reading unit 28. The above-mentioned units are connected to each other via a bus. The control unit 21 is configured using one or more processors such as a CPU (Central Processing Unit), an MPU (Micro-Processing Unit), or a GPU (Graphics Processing Unit). The control unit 21 executes various information processing operations performed by the server 2 by appropriately executing a control program 22P (program product) stored in the storage unit 22.

[0014] The storage unit 22 includes RAM (Random Access Memory) or ROM (Read Only Memory), etc. The storage unit 22 pre-stores the control program 22P executed by the control unit 21 and various data required for executing the control program 22P. The storage unit 22 temporarily stores data generated when the control unit 21 executes the control program 22P. The communication unit 23 transmits information to and from the biomass combustion furnace 1. The input unit 24 is an input interface such as a mouse, a stylus pen, or a touch panel, and accepts operational input from the user. The display unit 25 is, for example, a liquid crystal display or an organic EL (Electroluminescence) display, etc. The display unit 25 may be a touch panel integrated with the input unit 24.

[0015] The imaging unit 26 is an imaging device such as a CCD (Charge Coupled Device) camera, a CMOS (Complementary Metal Oxide Semiconductor) camera, a hyperspectral camera, etc. The imaging unit 26 captures combustion images while the biomass combustion furnace 1 is in operation.

[0016] The mass storage unit 27 includes a RAM, a ROM, or the like. The mass storage unit 27 includes a combustion area identification model 271 and an accumulation area identification model 272. The combustion area identification model 271 and the accumulation area identification model 272 are trained models generated by machine learning. Details of the combustion area identification model 271 and the accumulation area identification model 272 will be described later.

[0017] The reading unit 28 reads information stored in a portable storage medium 2a, such as a CD (Compact Disc)-ROM, a DVD (Digital Versatile Disc)-ROM, a USB (Universal Serial Bus) memory, or an SD (Secure Digital) card. The control program 22P and various data stored in the storage unit 22 may be stored in the storage unit 22 by the control unit 21 reading them from the portable storage medium 2a via the reading unit 28. Alternatively, the control program 22P and various data stored in the storage unit 22 may be stored in the storage unit 22 by the control unit 21 downloading them from an external device (not shown) via the communication unit 23.

[0018] In this embodiment, the memory unit 22 and the mass storage unit 27 may be configured as an integrated memory device. The mass storage unit 27 may be configured with a plurality of memory devices. The mass storage unit 27 may be an external memory device connected to the server 2. The combustion area identification model 271 and the deposition area identification model 272 may be stored in an external server connected to the server 2.

[0019] In this embodiment, the functions of the server 2 may be implemented separately, in which case the functions of the server 2 are implemented by a plurality of servers, computers, etc.

[0020] FIG. 4 is an explanatory diagram showing an example of a combustion image. The upper side of the combustion image is the driven wheel 123 side of the upper surface of the belt part 121. The lower side of the combustion image is the drive wheel 122 side of the upper surface of the belt part 121. The vertical direction toward the bottom of the combustion image is the traveling direction of the belt part 121. The belt part 121 transports biomass fuel from the top to the bottom of the combustion image. The horizontal direction of the combustion image is the width direction of the belt part 121. The combustion image illustrates the area where biomass combustion is occurring (hereinafter referred to as the fuel area) and other areas. The origin (black dot) is the upper left part of the combustion image and is identified when the combustion image is acquired.

[0021] The combustion state of biomass fuel varies depending on the type, shape, and moisture content of the biomass fuel. If unburned fuel occurs due to changes in the combustion state, problems such as a decrease in the utilization efficiency of the biomass fuel and an increase in combustion ash can occur. Based on their many years of experience, workers use acquired combustion images to change combustion conditions such as the traveling stoker's movement speed, biomass fuel supply rate, and air volume. However, changing combustion conditions is a very nerve-wracking task even for experienced workers, so effective use of combustion images was desired.

[0022] Therefore, the combustion image analysis system of this embodiment outputs analysis results regarding the combustion state from the combustion image. The analysis results regarding the combustion state include, for example, the position where the biomass fuel in the biomass combustion furnace burns out (hereinafter referred to as the burnout point), the brightness value regarding the combustion image, and the accumulation area where the biomass fuel accumulates without burning (hereinafter referred to as the accumulation area). The burnout point and brightness value are identified based on the combustion area included in the combustion image.

[0023] First, a method for identifying a burned area will be described. The burned area is identified by binarization or a learning model. An example using binarization will be described. FIG. 5 is an explanatory diagram showing a method for identifying a burned area by binarization. The burned area (hatched portion) is illustrated in the burned image of FIG. 5. The position of each pixel in the burned image of FIG. 5 is represented by coordinates (X, Y) in an image coordinate system in which the X axis extends from the origin (0, 0) in the upper left corner of the image to the right (the width direction of the belt part 121) and the Y axis extends downward from the origin (the direction of travel of the belt part 121). The X axis and Y axis are represented by 0 to 100, for example, when the origin is used as the reference point.

[0024] The control unit 21 acquires a combustion image captured by the imaging unit 26. The control unit 21 identifies the time when the combustion image was acquired. The control unit 21 uses binarization to identify areas in the combustion image where the brightness value is equal to or greater than a certain level as combustion areas (hatched areas). The control unit 21 acquires a segmentation image in which the combustion area is identified. The control unit 21 acquires position information of the combustion area based on the origin. The position information of the combustion area is identified by coordinate values on the X and Y axes. The control unit 21 acquires the number of pixels in the combustion area based on the position information of the combustion area. The control unit 21 calculates the average brightness value of each pixel in the entire combustion image (bold frame) (hereinafter referred to as the overall average brightness value). Specifically, the control unit 21 calculates the overall average brightness value by dividing the sum of the brightness values of each pixel included in the combustion image by the number of pixels included in the combustion image. The control unit 21 calculates the average brightness value of each pixel in the combustion area (hereinafter referred to as the combustion area average brightness value). Specifically, the control unit 21 calculates the average brightness value of the combustion area by dividing the sum of the brightness values of each pixel included in the combustion area by the number of pixels included in the combustion area. The control unit 21 stores the combustion image, the time when the combustion image was acquired, the segmentation image, the position information of the combustion area, the number of pixels in the combustion area, the overall average brightness value, and the combustion area average brightness value in the memory unit 22.

[0025] An example using a learning model will be described. FIG. 6 is a schematic diagram of a combustion area identification model. The combustion area identification model 271 is a model trained to output information about the combustion area included in a combustion image when a combustion image of biomass fuel is input. The information about the combustion area is, for example, a segmentation image that identifies the combustion area (hatched area) or position information corresponding to the combustion area. In this embodiment, an example will be described in which a segmentation image that identifies the combustion area is used as the information about the combustion area, but this is not limited to this. The combustion area identification model 271 is, for example, a model that performs semantic segmentation. The combustion area identification model 271 uses a U-net (U-Shaped Network) in which a convolution layer, a pooling layer, an upsampling layer, and a softmax layer are symmetrically arranged. The combustion area identification model 271 may also use SegNet, R-CNN, or the like.

[0026] The procedure for generating the combustion area identification model 271 will be described. As a preparatory step for generating the combustion area identification model 271, annotation is performed on the acquired combustion image. The operator uses a mouse, a stylus pen, or the like provided as the input unit 24 to create an annotation image that identifies the portion corresponding to the combustion area from the combustion image displayed on the display unit 25. A set of the many combustion images used for annotation and the annotation image is stored in the mass storage unit 27 as training data for generating the combustion area identification model 271.

[0027] The control unit 21 acquires a segmentation image that identifies a combustion area from the combustion image by inputting the combustion image included in the training data into the combustion area identification model 271. The control unit 21 adjusts the parameters of the combustion area identification model 271 so that the difference between the acquired segmentation image and the annotation image included in the training data is minimized.

[0028] The generation process of the combustion area identification model 271 may be performed by an external computer (not shown) or the like. In this case, the generated combustion area identification model 271 is deployed to the server 2 from the external computer or the like via a network.

[0029] The control unit 21 inputs the combustion image into the combustion area identification model 271, thereby acquiring a segmentation image in which the combustion area is identified from the combustion image. In the segmentation image of FIG. 6, a U-shaped combustion area is identified. The identified combustion area is hatched. The subsequent processing is the same as in the example using binarization, so a description thereof will be omitted.

[0030] A method for identifying the burnout point from the combustion area will be described. FIG. 7 is an explanatory diagram showing a method for identifying the burnout point. The fuel image in FIG. 7 shows a detection area (bold frame), a combustion area (hatched area), multiple division lines (dotted lines) that divide the combustion image, and the burnout point (bold line) for each division. As in FIG. 5, the position of each pixel in the combustion image is indicated by coordinates in an image coordinate system in which the X axis extends rightward from the origin and the Y axis extends downward from the origin. The X axis and Y axis are indicated, for example, by 0 to 100, with the origin as the reference point. The detection area is, for example, a rectangular area that is set in advance to identify the burnout point. Multiple division lines divide the detection area at equal intervals in the X axis direction.

[0031] The control unit 21 identifies a detection area from the segmentation image using the origin as a reference. The detection area is an area for identifying the burnout point for each section from a combustion image captured of biomass fuel burning on the belt unit 121. The detection area is pre-stored in the memory unit 22 as coordinates in an image coordinate system of the X and Y axes. The detection area in FIG. 7 is a rectangular area bounded by X1 to X2 in the X-axis direction and Y1 to Y2 in the Y-axis direction. The detection area can be changed according to the embodiment. The control unit 21 divides the detection area at equal intervals using multiple division lines. The detection area in FIG. 7 is divided into sections A, B, C, D, and E in the X-axis direction. In FIG. 7, the detection area is divided into five sections, but this is not limited to this. The number of sections can be changed, for example, from 3 to 9. The control unit 21 identifies the range of each section in the X-axis direction. In FIG. 7, the range of section B in the X-axis direction is b1 to b2. The control unit 21 reads out the position information of the combustion area from the memory unit 22. The control unit 21 compares the position information of the combustion area with the range of each section in the X-axis direction to identify the lowest point of the combustion area included in each section as the burnout point. The burnout point of each section is indicated by a value between 0 and 100. In Figure 7, the burnout point of section B is 87. The control unit 21 stores the burnout point of each section in the memory unit 22 in association with the combustion image, the time the combustion image was acquired, the segmentation image, the position information of the combustion area, the number of pixels in the combustion area, the overall average brightness value, and the average brightness value of the combustion area. The control unit 21 performs similar processing on multiple acquired combustion images. The control unit 21 displays a graph on the display unit 25 showing the time series change of the burnout point based on the burnout points of each section acquired from the multiple combustion images.

[0032] In Fig. 7, the control unit 21 identifies the burnout point from the segmentation image acquired from the combustion area identification model 271, but the burnout point can also be identified in a similar manner from the binarized combustion image of Fig. 5. In Fig. 5, the proportion of the combustion area occupying the entire image is larger than in Fig. 7. In this case, the control unit 21 identifies the burnout point using the portion of the combustion area identified in Fig. 5 that is included within the range of the detection region.

[0033] FIG. 8 is an explanatory diagram showing the time-series changes in the burnout point. The vertical axis of the graph indicates the position of the burnout point, and the horizontal axis indicates the time when the combustion image was acquired. The graph shown in FIG. 8 includes a first graph (solid or dotted line) showing the time-series changes in the burnout point of each section and a second graph (thick line) showing the average value of multiple first graphs. More specifically, the first graph showing the time-series changes in the burnout points of sections A, B, C, D, and E is shown with different types of solid and dotted lines. The second graph shows the average value of the first graph showing the time-series changes in the burnout points of sections A, B, C, D, and E. From FIG. 8, workers can confirm how the burnout point of each section changes over time. Based on the time-series changes in the burnout point, workers can control combustion conditions such as the traveling stoker's movement speed, biomass fuel supply rate, and air volume.

[0034] FIG. 9 is an explanatory diagram showing an example of an additional image displayed in association with a graph showing time-series changes in the burn-out point. The additional image d1 is displayed in association with the graph shown in FIG. 8 when a predetermined position (black dot) on the graph shown in FIG. 8 is operated. The additional image d1 includes a segmentation image linked to the predetermined position on the graph shown in FIG. 8 and information on the burn-out point of each segment. The predetermined position on the graph is, for example, a position on the graph shown in FIG. 8. Methods for operating a position on the graph include, for example, touching and tapping with a stylus pen, and moving and clicking a cursor with a mouse. In the following explanation, a click operation on a position on the graph with a mouse (hereinafter referred to as a click operation) will be used as an example.

[0035] The control unit 21 determines whether a click operation has been received at a position on the graph. If the control unit 21 determines that a click operation has been received at a position on the graph, the control unit 21 identifies the time and type of graph (multiple first graphs or second graph) corresponding to the position on the graph. In the example of FIG. 9, the control unit 21 identifies the time (15:56) and the first graph (A) corresponding to the position on the graph where the click operation was received. The control unit 21 reads out from the storage unit 22 a segmentation image corresponding to the time (15:56) and the burnout point of section A corresponding to the time (15:56) and the first graph (A). The control unit 21 generates an additional image d1 by adding information about the burnout point of section A to the read segmentation image. The control unit 21 displays the additional image d1 generated by the above-described process in association with the graph shown in FIG. 8.

[0036] When the control unit 21 receives a click operation on a position on the first graph (A), it may display the burnout point of section A in a specific display format (for example, a thick line and a solid line, etc.) and display the other burnout points from section B to section E in other display formats (for example, a dotted line, etc.) in an additional image d1 associated with the graph of Fig. 8. Furthermore, the control unit 21 may display the burnout point of section A in black and the burnout points from section B to section E in other colors (for example, yellow, etc.) in an additional image d1 associated with the graph of Fig. 8.

[0037] When the control unit 21 receives a click operation on a position on the second graph in Fig. 9, it generates an additional image d1 by adding a straight line indicating the average value of the burn-out point of each section to the read segmentation image. Specifically, in the additional image d1, a straight line indicating the average value of the burn-out point of each section is displayed along the X-axis direction in the range from section A to section E of the segmentation image. The control unit 21 displays the additional screen d1 generated by the above-described processing in association with the graph shown in Fig. 8.

[0038] When a click operation is received on a position on the second graph, the control unit 21 may display an additional image d1 that displays the other burn-out points from section A to section D in addition to a straight line indicating the average value of the burn-out points of each section, in association with the graph of Fig. 8. When a click operation is received on a position on the first graph or a position on the second graph, the control unit 21 may display an additional image d1 to which burn-out points, etc. have been added in advance, in association with the graph of Fig. 8.

[0039] A method for utilizing brightness values related to combustion images will be described. The control unit 21 reads out the overall average brightness value and the combustion area average brightness value from the storage unit 22. The control unit 21 performs the same processing on the multiple combustion images it has acquired. Based on the overall average brightness value and the combustion area average brightness value acquired from the multiple combustion images, the control unit 21 displays a graph on the display unit 25 showing the time series changes in the overall average brightness value and the combustion area average brightness value.

[0040] FIG. 10 is an explanatory diagram showing the time series changes in the overall average brightness value and the combustion area average brightness value. The vertical axis of the graph indicates the brightness value, and the horizontal axis indicates the time when the combustion image was acquired. The graph shown in FIG. 10 includes a graph (dotted line) showing the time series changes in the overall average brightness value and a graph (solid line) showing the time series changes in the combustion area average brightness value. Based on the time series changes in the overall average brightness value and the combustion area average brightness value, workers can control combustion conditions such as the traveling stoker's movement speed, biomass fuel supply rate, and air volume.

[0041] The procedure for identifying an accumulation area will be described. FIG. 11 is a schematic diagram of an accumulation area identification model. The accumulation area identification model 272 is a model trained to output a segmentation image in which an accumulation area is identified when an image of biomass fuel combustion is input. The accumulation area identification model 272 is, for example, a model that performs semantic segmentation. As shown in FIG. 11, the accumulation area identification model 272 uses a U-net in which a convolution layer, a pooling layer, an upsampling layer, and a softmax layer are symmetrically arranged. The accumulation area identification model 272 may also use SegNet, R-CNN, or the like.

[0042] The procedure for generating the accumulation area identification model 272 will be described. As a preparatory step for generating the accumulation area identification model 272, annotation is performed on captured combustion images. An operator uses a mouse, a stylus pen, or the like provided as the input unit 24 to create an annotation image that identifies the accumulation area from the combustion image displayed on the display unit 25. A set of the many combustion images used for annotation and the annotation image is stored in the mass storage unit 27 as training data for generating the accumulation area identification model 272.

[0043] The control unit 21 acquires a segmentation image in which the accumulation area (hatched portion) is identified from the combustion image by inputting the combustion image included in the training data into the accumulation area identification model 272. The control unit 21 adjusts the parameters of the accumulation area identification model 272 so that the difference between the acquired segmentation image and the annotation image included in the training data is minimized.

[0044] The process of generating the deposition area identification model 272 may be performed by an external computer or the like (not shown). In that case, the generated deposition area identification model 272 is deployed to the server 2 from the external computer or the like via a network.

[0045] In this embodiment, the control unit 21 identifies the combustion area and the deposition area using two learning models (combustion area identification model 271 and deposition area identification model 272), but is not limited to this. The control unit 21 may acquire the combustion area and the deposition area using one learning model. In this case, the control unit 21 generates the learning model using training data consisting of a set of many combustion images and annotation images that identify the combustion area and the deposition area from the combustion images.

[0046] The control unit 21 reads out the combustion image from the storage unit 22. The control unit 21 inputs the combustion image into the accumulation area identification model 272, thereby acquiring a segmentation image in which the accumulation area is identified from the combustion image. In the segmentation image of FIG. 11, a bowl-shaped accumulation area is identified. The identified accumulation area is hatched.

[0047] The method for acquiring the segmentation image in which the deposition area is identified is not limited to the above-mentioned method, and the control unit 21 may acquire the segmentation image by binarization.

[0048] Fig. 12 is an explanatory diagram showing a method for identifying the center of gravity position. The segmentation image in Fig. 12 shows the accumulation area (hatched portion) and the center of gravity position of the accumulation area. As in Fig. 5, the position of each pixel in the combustion image is shown by coordinates in an image coordinate system in which the X axis extends rightward from the origin and the Y axis extends downward from the origin. The X axis and Y axis are expressed as 0 to 100, for example, when the origin is used as the reference point.

[0049] The control unit 21 acquires position information of the accumulation area based on the origin. The position information of the accumulation area is displayed using coordinate values on the X and Y axes. The control unit 21 acquires the number of pixels of the accumulation area based on the position information of the accumulation area. The control unit 21 identifies the position of the center of gravity of the accumulation area from the position information of the accumulation area. The position of the center of gravity of the accumulation area is displayed using coordinate values on the X and Y axes. In Figure 12, the position of the center of gravity of the accumulation area is (50, 40). The control unit 21 performs similar processing on the multiple combustion images acquired. Based on information related to the accumulation area, the control unit 21 displays on the display unit 25 a graph showing the time series change in the position of the center of gravity of the accumulation area, a graph showing the proportion of the accumulation area in the combustion image over time, and a graph showing the ratio of the accumulation area and the combustion area over time.

[0050] 13 is an explanatory diagram showing time-series changes in the center of gravity position in the X-axis direction for the accumulation area. The vertical axis of the graph shows time-series changes in the center of gravity position (X-axis). The horizontal axis of the graph shows elapsed time.

[0051] FIG. 14 is an explanatory diagram showing the time-series change in the center of gravity position in the Y-axis direction for the accumulation area. The vertical axis of the graph shows the time-series change in the center of gravity position (Y-axis). The horizontal axis of the graph shows the elapsed time. Workers can confirm the time-series change in the center of gravity position of the accumulation area based on FIGS. 13 and 14.

[0052] FIG. 15 is an explanatory diagram showing the proportion of the deposition area in the combustion image in chronological order. The vertical axis of the graph shows the percentage of the accumulated area in the combustion image (number of pixels in the accumulated area / number of pixels in the entire combustion image), and the horizontal axis shows the elapsed time. Workers can check the percentage of the accumulated area in the combustion image over time.

[0053] FIG. 16 is an explanatory diagram showing the ratio of the accumulation area to the combustion area over time. The vertical axis of the graph shows the ratio of the accumulation area to the combustion area (number of pixels in the accumulation area / number of pixels in the combustion area + number of pixels in the accumulation area), and the horizontal axis shows the elapsed time. Workers can check the ratio of the accumulation area to the combustion area over time. Based on the time-series data on the accumulation area described above, workers can control combustion conditions such as the traveling stoker's movement speed, biomass fuel supply rate, and air volume.

[0054] 17 is a flowchart showing the procedure for generating a combustion area identification model. The control unit 21 acquires training data consisting of a set of a large number of combustion images and annotation images in which combustion areas are identified from the combustion images from the mass storage unit 27 (step S101). The control unit 21 uses the acquired training data to generate a combustion area identification model 271 that takes combustion images as input and outputs segmentation images in which combustion areas are identified (step S102). The control unit 21 stores the generated combustion area identification model 271 in the mass storage unit 27 (step S103).

[0055] 18 is a flowchart showing the procedure for generating an accumulation area identification model. The control unit 21 acquires training data consisting of a set of a large number of combustion images and annotation images that identify accumulation areas from the combustion images from the mass storage unit 27 (step S201). The control unit 21 uses the acquired training data to generate an accumulation area identification model 272 that takes combustion images as input and outputs segmentation images that identify accumulation areas (step S202). The control unit 21 stores the generated accumulation area identification model 272 in the mass storage unit 27 (step S203).

[0056] 19 is a flowchart showing the processing of the combustion image analysis system. The control unit 21 acquires a combustion image captured by the imaging unit 26 (step S301). The control unit 21 identifies the time when the combustion image was acquired (step S302). The control unit 21 starts a subroutine for acquiring information about the combustion area (step S303). The control unit 21 starts a subroutine for acquiring information about the accumulation area (step S304). After performing the processes from step S301 to step S304 for multiple combustion images taken at different times, the control unit 21 proceeds to step S305. The control unit 21 reads out the burn-out point of each section acquired from the multiple combustion images from the storage unit 22 (step S305). The control unit 21 displays a graph showing the time-series change in the burn-out point on the display unit 25 (step S306). The control unit 21 determines whether a click operation has been received on a position on the graph displayed in step S306 (step S307). If the control unit 21 determines that a click operation has not been received at a position on the graph displayed in step S306 (step S307: NO), the control unit 21 returns to step S306. If the control unit 21 determines that a click operation has been received at a position on the graph displayed in step S306 (step S307: YES), the control unit 21 identifies the time and graph type corresponding to the position on the graph (step S308). The control unit 21 reads out a segmentation image corresponding to the time and the burn-out points of each section corresponding to the time and graph type from the storage unit 22 (step S309). The control unit 21 generates an additional image by adding information about the burn-out points of each section to the read segmentation image (step S310). The control unit 21 displays the generated additional image in association with the graph displayed in step S306 (step S311).

[0057] The control unit 21 reads out the overall average brightness value and the burned area average brightness value obtained from the multiple combustion images from the storage unit 22 (step S312). The control unit 21 displays a graph showing the time series changes in the overall average brightness value and the burned area average brightness value on the display unit 25 (step S313). The control unit 21 reads out the center of gravity position, the number of pixels in the burned area, and the number of pixels in the accumulated area obtained from the multiple combustion images from the storage unit 22 (step S314). The control unit 21 displays on the display unit 25 a graph showing the time series changes in the center of gravity position of the accumulated area, a graph showing the proportion of the accumulated area in the combustion images over time, and a graph showing the proportion of the accumulated area and the burned area over time (step S315).

[0058] The subroutine for S303 will be described in detail. Fig. 20 is a flowchart showing the process of acquiring information about the combustion area. The control unit 21 acquires a segmentation image in which the combustion area is identified from the combustion image based on the binarization or combustion area identification model 271 (step S401). The control unit 21 acquires position information of the combustion area based on the origin (step S402). The control unit 21 acquires the number of pixels of the combustion area based on the position information of the combustion area (step S403). The control unit 21 calculates an overall average brightness value from the segmentation image (step S404). The control unit 21 calculates the average brightness value of the burned area from the segmentation image (step S405). The control unit 21 identifies a detection area from the segmentation image using the origin as a reference (step S406). The control unit 21 divides the detection area into equal intervals using multiple dividing lines (step S407). The control unit 21 identifies the range of each division in the X-axis direction (step S408). The control unit 21 compares the position information of the burned area identified in step S402 with the range of each division identified in step S408, and identifies the lowest point of the burned area included in each division as the burn-out point (step S409). The control unit 21 stores the segmentation image, the position information of the burned area, the number of pixels of the burned area, the overall average brightness value, the burn-out area average brightness value, and the burn-out point of each division in the memory unit 22, linking them to the burned image and the time the burned image was acquired (step S410). The control unit 21 returns the process to step S304.

[0059] The subroutine related to S304 will be described in detail. FIG. 21 is a flowchart showing the process of acquiring information about the accumulation area. The control unit 21 acquires a segmentation image in which the accumulation area is identified from the combustion image using the accumulation area identification model 272 (step S501). The control unit 21 acquires position information of the accumulation area based on the origin (step S502). The control unit 21 acquires the number of pixels of the accumulation area based on the position information of the accumulation area (step S503). The control unit 21 identifies the position of the center of gravity of the accumulation area from the position information of the accumulation area (step S504). The control unit 21 stores the segmentation image, the position information of the accumulation area, the number of pixels, and the position of the center of gravity in the storage unit 22, linking them to the combustion image and the time when the combustion image was acquired (step S505). The control unit 21 returns the process to step S305.

[0060] According to this embodiment, the combustion image analysis system can identify the burnout point, the brightness value of the combustion image, and the accumulation area from the combustion image as indicators of the combustion state.

[0061] According to this embodiment, the combustion image analysis system generates a graph showing a time series change in the burnout point, A graph showing the time series changes in the overall average brightness value and the burned area average brightness value, as well as a graph showing the time series changes in the accumulation area, can be displayed.

[0062] According to this embodiment, workers using the combustion image analysis system can utilize graphs showing the time series changes in the burnout point, graphs showing the time series changes in the overall average brightness value and the average brightness value of the combustion area, and graphs showing the time series changes in the accumulation area to manage combustion conditions.

[0063] The features described in each of the above embodiments can be combined with each other. Furthermore, the independent claims and dependent claims described in the claims can be combined with each other in any and all combinations, regardless of the reference format. Furthermore, the claims use a format in which a claim references two or more other claims (multiple claim format), but this is not limited to this. Multiple claims (multi-multi claims) that reference at least one other multiple claim may also be used.

[0064] The embodiments disclosed herein are to be considered in all respects as illustrative and not restrictive. The scope of the present invention is defined by the claims, not by the above meaning, and is intended to include all modifications within the meaning and scope of the claims. [Explanation of symbols]

[0065] 1. Biomass combustion furnace 11 Fuel release machine 12 Traveling Stalker 121 Belt section 122 Drive wheels 123 Driven Wheel 13 Air supply machine 14 Outlet 2. Information processing device (server) 21 Control section 22 Memory section 22P control program 23 Communications Department 24 Input section 25 Display section 26 Imaging unit 27 Mass storage 271 Combustion Area Identification Model 272 Deposition Area Identification Model 28 Reading unit 2a Portable storage media

Claims

1. An image of the combustion of biomass fuel is acquired by an imaging unit provided in the biomass combustion furnace; Based on the acquired combustion image, an analysis result regarding the combustion state is output. A program that causes a computer to perform a process.

2. Based on the combustion image, the burnout point is displayed for each section. The program according to claim 1.

3. inputting the acquired combustion image into a learning model that has been trained to output information about a combustion area included in a combustion image of a biomass fuel when the combustion image is input, thereby outputting information about the combustion area; The burnout point is identified for each section based on the output information about the combustion area. The program according to claim 2.

4. Display a graph showing the time series change of the burnout point The program according to claim 2 or 3.

5. The graph includes a first graph showing a time series change in the burnout point for each section, and a second graph showing an average value of a plurality of the first graphs. The program according to claim 4.

6. determining whether an operation on a predetermined position on the graph has been accepted; When it is determined that an operation on the predetermined position has been received, the combustion image and the burnout point corresponding to the predetermined position are displayed. The program according to claim 4.

7. Based on the combustion image, a graph showing a time series change in the average brightness value of the combustion image and a graph showing a time series change in the average brightness value of the combustion area included in the combustion image are displayed. The program according to claim 1 or 2.

8. When a combustion image of biomass fuel is input, a learning model that has been trained to output an accumulation area where the biomass fuel is accumulated without being burned outputs the accumulation area by inputting the acquired combustion image. The program according to claim 1 or 2.

9. Display a graph showing the time series changes in the deposition area The program according to claim 8.

10. An image of the combustion of biomass fuel is acquired by an imaging unit provided in the biomass combustion furnace; Based on the acquired combustion image, an analysis result regarding the combustion state is output. An information processing method in which processing is performed by a computer.

11. An information processing device including a control unit, The control unit An image of the combustion of biomass fuel is acquired by an imaging unit provided in the biomass combustion furnace; Based on the acquired combustion image, an analysis result regarding the combustion state is output. Information processing device.

12. an imaging unit that captures images of combustion of biomass fuel in a biomass combustion furnace; an information processing device that outputs analysis results regarding the combustion state based on the combustion image acquired from the imaging unit; A combustion image analysis system comprising:

Images