Evaluation support system, evaluation support method, and evaluation support program

Abstract

This invention provides an evaluation support system, an evaluation support method, and an evaluation support program for efficiently evaluating the drying status of an object. [Solution] The support server 20 is connected to the estimation model storage unit 23 and includes a control unit 21 that evaluates the drying status of the mud. The estimation model storage unit 23 stores an estimation model that predicts the drying status of the mud, which is generated by machine learning using captured images including cracks that have occurred in the drying status of the mud and a drying index as training information. The control unit 21 acquires captured images of the mud, inputs the captured images into the estimation model to calculate a drying index, and predicts the drying status of the mud according to the drying index.

Description

Technical Field

[0001] The present disclosure relates to an evaluation support system, an evaluation support method, and an evaluation support program that support evaluation of the drying status of sludge.

Background Art

[0002] River water, which serves as a raw material for tap water, is taken into a water purification plant through an intake facility connecting the river and the water purification plant. Since river water contains sludge (mud and sand), part of these is removed in a grit chamber arranged at the inlet of the water purification facility. Final sludge extraction is performed in a solar drying bed arranged at the outlet of the water purification facility (see, for example, Patent Document 1). In the solar drying bed described in this document, a water-permeable layer is formed on the bottom surface of a storage tank, and water-containing sludge charged onto the water-permeable layer is dried by sunlight. This solar drying bed is provided with a plurality of strip-shaped water-permeable sheets on the water-permeable layer before the water-containing sludge is charged.

[0003] Sludge laid on a solar drying bed is dried and dewatered over a period of several months by utilizing solar heat and natural wind. The dewatered sludge is reused as a raw material for cement, soil for playgrounds, gardens, and the like.

[0004] In solar drying, since drying is performed using natural phenomena, portions of the sludge where drying has progressed and portions where drying has not progressed may coexist. In order to achieve a uniform drying status, the difference between the portion where drying has progressed and the portion where drying has not progressed is eliminated during drying. For this purpose, sludge is agitated to introduce air into the sludge and ensure air permeability. Heavy machinery or the like is regularly used for this sludge agitation. The timing of agitation is determined based on the result of checking the drying status by visual inspection by an administrator.

Prior Art Literature

Patent Literature

[0005]

Patent Literature 1

Summary of the Invention

[0006] If the drying of the mud is judged subjectively by the manager based on its surface condition, it may be difficult to accurately determine the timing of stirring. [Means for solving the problem]

[0007] The evaluation support system that solves the above problems includes a control unit connected to an estimation model storage unit, which evaluates the drying status of an object. The estimation model storage unit stores an estimation model that is generated by machine learning using captured images of the object including cracks that have occurred in the object's drying status and a drying index as training information, and which predicts the drying status of the object. The control unit acquires captured images of the object, inputs the captured images into the estimation model to calculate a drying index, and predicts the drying status of the object according to the drying index. [Effects of the Invention]

[0008] According to this disclosure, the drying status of an object can be efficiently evaluated. [Brief explanation of the drawing]

[0009] [Figure 1] This is an explanatory diagram of the structure of the evaluation support system in the embodiment. [Figure 2] This is an explanatory diagram of the hardware configuration in the embodiment. [Figure 3] This is an explanatory diagram of the captured image in the embodiment. [Figure 4] This is an explanatory diagram of the drying of mud in the embodiment. [Figure 5] This is an explanatory diagram illustrating the division of a captured image in the embodiment. [Figure 6] This is an explanatory diagram of a partial image in an embodiment. [Figure 7] This is an explanatory diagram for evaluating captured images in the embodiment. [Figure 8] This is an explanatory diagram of the processing procedure for the evaluation support method in the embodiment. [Figure 9] This is an explanatory diagram of the processing procedure for the evaluation support method in the embodiment. [Figure 10] These are explanatory diagrams illustrating the evaluation of captured images in the embodiment, where (a) is an explanatory diagram of the images after 12 days, (b) after 33 days, (c) after 44 days, (d) after 64 days, (e) after 80 days, (f) after 100 days, (g) after 140 days, and (h) after 182 days. [Figure 11] This is an explanatory diagram illustrating the change in drying conditions over time in the embodiment. [Modes for carrying out the invention]

[0010] Below, an embodiment of the evaluation support system, evaluation support method, and evaluation support program will be described using Figures 1 to 11. As shown in Figure 1, the evaluation support method of this embodiment evaluates the drying status of mud (sludge or natural mud in this embodiment), which is soil containing water, on a sun-drying bed. In this embodiment, we assume the case where the drying status of the entire mud (including the intermediate layer) is evaluated from captured images. For this purpose, the evaluation support system CS1 uses a network-connected camera 10, support server 20, and user device 30.

[0011] (Example hardware configuration) Figure 2 shows an example of the hardware configuration of the information processing device H10, which functions as a shooting device 10, a support server 20, a user device 30, etc.

[0012] The information processing device H10 includes a communication device H11, an input device H12, a display device H13, a storage device H14, and a processor H15. Note that this hardware configuration is an example, and other hardware may be included.

[0013] Communication device H11 is an interface that performs data transmission and reception by establishing a communication path with other devices, such as a network interface or a wireless interface.

[0014] The input device H12 is a device that accepts input from a user or the like (e.g., a mouse, a keyboard, etc.). The display device H13 is a device that displays various types of information (e.g., a display, a touch panel, etc.).

[0015] The storage device H14 is a storage unit that stores data and various programs for executing various functions of the imaging device 10, the support server 20, and the user device 30. Examples of the storage device H14 include ROM, RAM, and a hard disk.

[0016] The processor H15 controls each process by using programs and data stored in the storage device H14. Examples of the processor H15 include a CPU, an MPU, and the like. The processor H15 executes various processes corresponding to various types of processing by loading a program stored in ROM or the like into RAM.

[0017] The processor H15 is not limited to performing software processing for all processes executed by itself. For example, the processor H15 may include a dedicated hardware circuit (e.g., an application-specific integrated circuit: ASIC) that performs hardware processing for at least part of the processes executed by itself. The processor includes a CPU and memories such as RAM and ROM, and the memory stores program codes or instructions configured to cause the CPU to execute processes. The memory, that is, a computer-readable medium, includes any available medium that can be accessed by a general-purpose or special-purpose computer.

[0018] (Drying of mud) As shown in the image 500 illustrated in FIG. 3, cracks (cracks M2) form in mud M1 (object) when it dries. These cracks M2 include primary cracks that form first, and secondary cracks that connect the primary cracks.

[0019] As shown in FIG. 4, the moisture in the mud M1 is discharged from the bottom to the drainage layer D1. Furthermore, the water flows along the cracked cracks M2. Drying progresses as saturated gas evaporates from the top surface of the mud and the surfaces of the cracks. Thus, the cracks M2 that form in the mud are related to the progress of drying.

[0020] (Configuration of the evaluation support system) The imaging device 10 is a device (camera) for photographing the mud to be evaluated. This camera is positioned to photograph the mud within a predetermined field of view from above the object to be evaluated.

[0021] The support server 20 is a computer device that assists in evaluating the drying status of the mud. This support server 20 comprises a control unit 21, a teacher information storage unit 22, an estimation model storage unit 23, and an evaluation information storage unit 24.

[0022] The control unit 21 executes the acquisition, learning, and prediction processes by running the evaluation program stored in the memory unit. As a result, the control unit 21 functions as an acquisition unit 211, a learning unit 212, and a prediction unit 213.

[0023] The acquisition unit 211 acquires the captured image from the imaging device 10. The learning unit 212 generates a predictive model for predicting the drying status using machine learning with training data prepared based on the captured images. The prediction unit 213 evaluates the drying status of the mud by using a prediction model generated by machine learning.

[0024] The training information storage unit 22 stores training information used for machine learning. This training information is recorded during the learning process. The training information includes information about the dryness score, which evaluates the dryness of partial images of captured mud images.

[0025] A partial image is an image of a predetermined size obtained by dividing a captured image into a predetermined number of pixels. As shown in Figure 5, the captured image 510 is divided into partial images of predetermined sizes.

[0026] The dryness score is a numerical value (dryness index) that evaluates the degree of cracking according to the dryness conditions. Here, the degree of cracking is determined by the shape of the crack (e.g., width). This allows us to evaluate the dryness of the area contained in a partial image. As shown in Figure 6, a drying score is assigned to each partial image 520 according to the shape (width) of the crack. For example, for each partial image 520, a drying score is assigned to the degree of cracking on an 8-point scale from "0 (no crack) to 1 (maximum crack)".

[0027] The estimation model storage unit 23 stores an estimation model for evaluating the drying status. This estimation model is recorded when a learning process has been performed. The estimation model takes captured images as input and outputs a score indicating the dryness level. As shown in Figure 7, by using the estimation model, a heatmap is output that evaluates the captured image 530 with a score.

[0028] The evaluation information storage unit 24 stores evaluation management information for images captured using the imaging device 10. This evaluation management information is recorded when an image is registered. The evaluation management information includes information about the shooting date, the captured image, and the drying status.

[0029] The shooting date information refers to the date and time the image was taken. The captured image is an image acquired from the imaging device 10. The drying status information is information related to the results of evaluating the drying status of the mud. In this embodiment, the statistical value of the score of the partial image included in the captured image (overall drying score) is used. Furthermore, the drying status information records the moisture content of this mud (surface layer and middle layer) measured with a moisture meter.

[0030] User device 30 is a computer terminal used by the user who manages the mud.

[0031] (Learning process) Next, we will explain the learning process using Figure 8. First, the control unit 21 of the support server 20 performs the image acquisition process (step S11). Specifically, the acquisition unit 211 of the control unit 21 acquires the captured image from the imaging device 10. Next, the control unit 21 of the support server 20 performs image segmentation processing (step S12). Specifically, the learning unit 212 of the control unit 21 generates partial images by dividing the captured image acquired from the imaging device 10 into multiple regions of a predetermined size.

[0032] Next, the control unit 21 of the support server 20 performs a drying status evaluation process (step S13). Specifically, the learning unit 212 of the control unit 21 outputs a partial image to the user device 30. The learning unit 212 then obtains a drying score corresponding to the partial image from the user device 30.

[0033] Next, the control unit 21 of the support server 20 performs the drying status registration process (step S14). Specifically, the learning unit 212 of the control unit 21 records teacher information, which associates the drying score with the partial image, in the teacher information storage unit 22.

[0034] Next, the control unit 21 of the support server 20 performs machine learning processing (step S15). Specifically, when the number of training information records in the training information storage unit 22 exceeds a predetermined amount, the learning unit 212 of the control unit 21 generates an estimation model for predicting the drying status using machine learning based on the training information. For example, deep learning-based image recognition or semantic segmentation can be used for machine learning. The learning unit 212 records the generated estimation model in the estimation model storage unit 23.

[0035] (Predictive processing) Next, we will explain the prediction process using Figure 9. First, the control unit 21 of the support server 20 executes the image acquisition process (step S21). Specifically, the acquisition unit 211 of the control unit 21 waits for the timing of the photo shoot (once a day). Then, when the timing of the photo shoot arrives, the acquisition unit 211 acquires the image from the photo shoot device 10.

[0036] Next, the control unit 21 of the support server 20 performs filtering (step S22). Specifically, the prediction unit 213 of the control unit 21 identifies water surface reflections and weeds in the captured image acquired from the imaging device 10 through image recognition (for example, object recognition). The prediction unit 213 then performs filtering to exclude areas containing the identified water surface reflections and weeds.

[0037] Next, the control unit 21 of the support server 20 performs image segmentation processing in the same manner as in step S12 (step S23). Next, the control unit 21 of the support server 20 performs evaluation processing (step S24). Specifically, the prediction unit 213 of the control unit 21 calculates the dryness score for each partial image by inputting the partial images into the estimation model.

[0038] Next, the prediction unit 213 calculates a statistical value (overall dry score) of the dryness score for each sub-image. As the statistical value, the average value obtained by dividing the sum of the scores of each sub-image by the number of sub-images is used. Note that the statistical value is not limited to this, and the mean, median, etc. may also be used.

[0039] Figure 10(a) shows image 531 taken after 12 days, with an overall drying score of "0.06". Figure 10(b) shows image 532 taken after 33 days, with an overall drying score of "0.16". Figure 10(c) shows image 533 taken after 44 days, with an overall drying score of "0.28". Figure 10(d) shows image 534 taken after 64 days, with an overall drying score of "0.46". Figure 10(e) shows image 535 taken after 80 days, with an overall drying score of "0.50". Figure 10(f) shows image 536 taken after 100 days, with an overall drying score of "0.51". Figure 10(g) shows image 537 taken after 140 days, with an overall drying score of "0.48". Figure 10(h) shows image 538 taken after 182 days, with an overall drying score of "0.46". The prediction unit 213 then records evaluation management information, including the shooting date, the captured image, and information regarding the drying status, in the evaluation information storage unit 24.

[0040] Next, the control unit 21 of the support server 20 performs a determination process to determine whether stirring is necessary (step S25). Specifically, the prediction unit 213 of the control unit 21 compares the overall drying score with a reference value. If the overall drying score exceeds the reference value, it determines that stirring is necessary.

[0041] If it is determined that stirring is necessary (if the answer is "YES" in step S25), the control unit 21 of the support server 20 performs a warning process (step S26). Specifically, the prediction unit 213 of the control unit 21 outputs a guidance message to the user device 30 indicating that stirring is necessary.

[0042] Figure 11 shows graphs evaluating the time-dependent changes in each indicator (drying score, mud height, and moisture content) of mud according to the number of days elapsed. Mud height is measured using mud detection sensors positioned vertically or by using images that include a measuring tape placed next to the mud. Moisture content can be measured using a moisture meter. Mud height decreases from the start of drying. However, the moisture content of the intermediate layer (inside) does not decrease. The moisture content of the surface layer decreases sharply immediately after the drying score saturates. In other words, the change in the drying score reflects the difference in moisture content between the surface and intermediate layers of the mud. Therefore, by stirring the mud after the drying score saturates, the surface and intermediate layers mix, which can promote overall drying.

[0043] (Effect of the embodiment) In this embodiment, the captured image of the mud includes the degree of cracking corresponding to the cracks that have formed in the mud, so the dryness of the mud can be evaluated.

[0044] (Effects of the embodiment) (1) In this embodiment, the control unit 21 of the support server 20 performs image segmentation processing (step S12) and drying status evaluation processing (step S13). This makes it possible to evaluate the degree of cracking based on cracks caused by drying using partial images.

[0045] (2) In this embodiment, the control unit 21 of the support server 20 performs machine learning processing (step S15). This makes it possible to predict the drying status from the captured image. (3) In this embodiment, the control unit 21 of the support server 20 performs image acquisition processing (step S21) and filtering processing (step S22). This makes it possible to acquire the image to be evaluated.

[0046] (4) In this embodiment, the control unit 21 of the support server 20 executes the evaluation process (step S24). This makes it possible to determine the drying status of the object to be evaluated. (5) In this embodiment, if stirring is determined to be necessary (if the answer is "YES" in step S25), the control unit 21 of the support server 20 performs a warning process (step S26). This encourages stirring to promote drying. This makes it possible to determine the appropriate stirring timing by heavy machinery, etc., and thus shorten the sun drying period.

[0047] This embodiment can be implemented with the following modifications. This embodiment and the following modifications can be combined with each other to the extent that they do not contradict each other technically. • In the above embodiment, mud was evaluated as the target material, but it is not limited to mud as long as it cracks when it dries. In the above embodiment, the control unit 21 of the support server 20 performs a drying status registration process (step S14). Specifically, the learning unit 212 of the control unit 21 records a drying score in association with a partial image. Here, the index used to evaluate the image can be any index that indicates the drying status, and is not limited to the drying score. For example, an index indicating the stirring time may be used.

[0048] In the above embodiment, the control unit 21 of the support server 20 performs a drying status registration process (step S14). The training data is not limited to images. For example, in addition to this, since mud shrinks when it dries, the height of the mud may also be included in the training information.

[0049] In the above embodiment, the control unit 21 of the support server 20 performs image segmentation processing (steps S12, S23). Here, training information may be generated using the entire image without segmenting it.

[0050] In the above embodiment, the control unit 21 of the support server 20 performs filtering (step S22). The target of filtering is not limited to water surface reflections or weeds. The target of this filtering can be anything that hinders the determination of cracks, for example, partial shadows may be filtered. Edge enhancement processing may also be used in combination.

[0051] In the above embodiment, the control unit 21 of the support server 20 performs a determination process to determine whether stirring is necessary (step S25). Here, the timing of stirring may be determined by considering future weather information. In this case, the control unit 21 obtains weather information that affects drying, such as future temperature, humidity, sunshine, and rainfall, from a weather information site published on the internet. The control unit 21 then generates a model to predict the drying score using the weather information. For example, a predictive model is generated that calculates the future drying score from future weather information using training information that associates past weather information with the drying score using machine learning. [Explanation of Symbols]

[0052] CS1...Evaluation support system, M1...Mud, M2...Crack, 10...Photography device, 20...Support server, 21...Control unit, 211...Acquisition unit, 212...Learning unit, 213...Prediction unit, 22...Teacher information storage unit, 23...Estimation model storage unit, 24...Evaluation information storage unit, 30...User device.

Claims

1. An evaluation support system comprising a control unit connected to an estimated model memory unit and used to evaluate the drying status of an object, The estimation model storage unit stores an estimation model that is generated by machine learning using captured images including cracks that have occurred in the drying state of the object and a drying index as training information, and which predicts the drying state of the object. The control unit, Capture an image of the object, An evaluation support system characterized by inputting the captured image into the estimation model to calculate a drying index and predicting the drying status of the object according to the drying index.

2. The evaluation support system according to claim 1, characterized in that the control unit predicts the drying status of sludge placed on a sun-drying bed as the target object.

3. The evaluation support system according to claim 1, characterized in that the control unit acquires the drying index at different timings and predicts the drying status according to the change in the drying index over time.

4. The evaluation support system according to claim 3, characterized in that the control unit guides the stirring of the object according to the drying status.

5. The evaluation support system according to claim 1, characterized in that the control unit uses images obtained by dividing the captured image into predetermined sizes.

6. A method for supporting the evaluation of drying conditions using an evaluation support system that includes a control unit connected to an estimation model memory unit and which evaluates the drying conditions of an object, The estimation model storage unit stores an estimation model that is generated by machine learning using captured images including cracks that have occurred in the drying state of the object and a drying index as training information, and which predicts the drying state of the object. The control unit, Capture an image of the object, An evaluation support method characterized by inputting the captured image into the estimation model to calculate a drying index and predicting the drying status of the object according to the drying index.

7. A program that assists in evaluating the drying status using an evaluation support system that is connected to an estimation model memory unit and includes a control unit for evaluating the drying status of an object, The estimation model storage unit stores an estimation model that is generated by machine learning using captured images including cracks that have occurred in the drying state of the object and a drying index as training information, and which predicts the drying state of the object. The control unit, Capture an image of the object, An evaluation support program characterized by inputting the captured image into the estimation model to calculate a drying index and using it as a means to predict the drying status of the object according to the drying index.

Images