Method for inspecting the state of flocculation of suspended matter, system for inspecting the state of flocculation of suspended matter, and water treatment system

Abstract

To provide an inspection method of flocculation states of suspended matters which can be achieved at lower cost while considering secular changes of flocculation states of flocs.SOLUTION: An inspection method of flocculation states of suspended matters includes the steps of: introducing water containing suspended matters and flocculants to a cell 20 having light transmissivity; continuously taking images of a first image of a first observation part 21 provided in an intermediate part of the cell at predetermined time intervals from the front side of the cell while applying illumination light 12 from a back side of the cell; calculating brightness dispersion in the first image with respect to each first image; and determining whether or not the flocculants are enough based on the secular changes of brightness dispersion of the first image.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a method and system for inspecting the flocculation state of suspended solids in turbid water or muddy water, and more particularly to a method and system capable of determining whether the amount of flocculant added to flocculate suspended solids is sufficient or insufficient.The present invention also relates to a water treatment system for treating water containing suspended solids. [Background technology]

[0002] Industrial wastewater and wastewater from various business facilities undergo the necessary water treatment, such as the removal of harmful substances, before being discharged into sewers, rivers, etc. If this wastewater is turbid and contains suspended solids (SS), the SS must be removed. Similarly, turbid water generated by construction and civil engineering work must have the SS removed before being discharged into rivers, etc.

[0003] In turbid water treatment, a common method is to use a flocculant to separate small particles into larger flocs, causing them to settle, resulting in solid-liquid separation, and then discharging the clarified water. In this process, it is preferable to limit the amount of flocculant added to the minimum necessary amount to reduce treatment costs. However, with industrial and facility wastewater, the water quality and SS concentration are prone to change depending on factors such as the changeover of work performed on-site and the operating status of the equipment, and with turbid water from construction work, the weather and progress of the work make it difficult to know in advance the appropriate amount of flocculant to be added.

[0004] A conventional method for estimating the appropriate amount of coagulant to be added is the jar test, in which raw water samples are collected in multiple beakers and the amount of coagulant added is changed to observe the state of coagulation. However, because the jar test takes 15 to 20 minutes to produce results, it is difficult to quickly adjust the amount of coagulant added in response to fluctuations in water quality and SS concentration while water treatment is in progress.

[0005] As a method for determining the amount of flocculant to be added, Patent Document 1 describes adding a flocculant to wastewater to form flocs, photographing the flocs with a digital camera or the like, and inputting the obtained image data into a flocculant determination model constructed using a machine learning algorithm to determine whether the flocculant addition rate is excessive or insufficient.

[0006] Furthermore, Patent Document 2 describes a water treatment system that identifies the contours of each floc shown in a series of time-series images of flocs formed by adding a flocculant to a liquid containing suspended solids and stirring the liquid, generates characteristic information about the flocs based on the identified contours, and controls the addition of the flocculant based on the generated characteristic information. [Prior art documents] [Patent documents]

[0007] [Patent Document 1] Japanese Patent Application Publication No. 2020-065964 [Patent Document 2] Patent Publication No. 2021-103146 Summary of the Invention [Problem to be solved by the invention]

[0008] The aggregation state of flocs continues to change over time due to complex transport phenomena such as aggregation due to particle collisions and mutual interference between particles. In contrast, a single image data shows the aggregated flocs at the moment the image is captured. Therefore, there is a high possibility that the determination model described in Patent Document 1, which is constructed using image data at a certain moment as learning data, will not be able to achieve sufficient accuracy in determination.

[0009] In Patent Document 2, a series of photographed images of flocs taken in time series is used. However, it is not an easy task to identify the contours of flocs from the series of photographed images or to create training data for machine learning to determine the relationship between images of flocs and their contours. In addition, a high-resolution photographing device and an information processing device with high processing capabilities are required, which increases the cost of configuring the system.

[0010] The present invention has been made in consideration of the above, and aims to provide a method for inspecting the flocculation state of suspended matter that can be implemented at a lower cost while taking into account changes over time in the flocculation state of flocs. [Means for solving the problem]

[0011] The method for inspecting the state of suspended matter coagulation of the present invention comprises the steps of: introducing water containing suspended matter and a coagulant into a translucent cell; successively photographing first images of a first observation section provided in the middle of the cell from the front of the cell at predetermined time intervals while illuminating the cell from behind; calculating the variance of brightness in each of the first images; and judging whether the coagulant is sufficient or insufficient based on the change over time in the variance of brightness of the first images.

[0012] Here, the middle part of the cell refers to the region that does not include the upper and lower ends of the cell. The luminance variance includes all indices of luminance variation, such as statistical variance, standard deviation, and coefficient of variation.

[0013] With this configuration, it is possible to determine whether the amount of flocculant is insufficient or not at a lower cost while taking into consideration the change over time in the flocculation state of the flocs.

[0014] Preferably, the flocculant is an inorganic flocculant and an organic flocculant, and the determining step is a step of determining whether or not each of the inorganic flocculant and the organic flocculant is sufficient. According to the method for inspecting the flocculation state of suspended solids of the present invention, even when both an inorganic flocculant and an organic flocculant are used, the sufficiency of each flocculant can be determined.

[0015] Preferably, the method for inspecting the state of suspended solids flocculation further includes a step of performing machine learning using the change over time in the variance of brightness of the first image and the amount of flocculant sufficiency as training data to generate in advance a judgment model in which the change over time in the variance of brightness of the first image is used as an input and the amount of flocculant sufficiency is used as an output, and the judgment step is a step of judging whether the amount of flocculant sufficiency is high or low using the judgment model.By using the judgment model created by machine learning, whether the amount of flocculant sufficiency is high or low can be accurately judged regardless of the experience or skill of an operator.

[0016] More preferably, the method for inspecting the state of suspended matter aggregation further comprises a step of performing machine learning of the judgment model using the change over time in variance of brightness of the first image that served as the basis for the judgment step as training data, thereby further improving the accuracy of the judgment model.

[0017] Alternatively, preferably, the method for inspecting the state of suspended solids coagulation when using the inorganic and organic coagulants further comprises the steps of: taking continuous second images of a second observation section located above the first observation section of the cell from the front of the cell at predetermined time intervals while illuminating the cell from behind; calculating the average brightness of each of the second images; taking continuous third images of a third observation section located above the first observation section of the cell and above or below the second observation section, with the back surface of the cell shielded by a dark-colored light-shielding member, from the front of the cell at predetermined time intervals while illuminating the cell from behind; and calculating the average brightness of each of the third images.The determination step is a step of determining whether the coagulant is sufficient or insufficient based on the change over time in the variance of the brightness of the first images, the change over time in the average brightness of the second images, and the change over time in the average brightness of the third images. By utilizing information obtained from the second and third observation units in addition to the first observation unit, it is possible to further improve the accuracy of determining whether the amount of flocculant is sufficient or insufficient.

[0018] Preferably, the method for inspecting the state of suspended solids flocculation when using the second observation unit and the third observation unit further includes a step of performing machine learning using the change over time in the variance of brightness of the first image, the change over time in the average brightness of the second image, the change over time in the average brightness of the third image, and the amount of flocculant as training data to generate in advance a judgment model whose input is the change over time in the variance of brightness of the first image, the change over time in the average brightness of the second image, and the change over time in the average brightness of the third image, and whose output is the amount of flocculant.The judgment step is a step of judging the amount of flocculant using the judgment model.

[0019] The suspended matter coagulation state inspection system of the present invention comprises a light-transmitting cell into which water containing suspended matter and a coagulant is introduced, a light that illuminates the cell from behind, a first observation unit provided in the middle of the cell, an imaging unit that continuously takes first images of the first observation unit from the front of the cell at predetermined time intervals, and a calculation unit that calculates the variance of brightness in each of the first images and determines whether the coagulant is sufficient or insufficient based on the change over time in the variance of brightness of the first images.

[0020] Preferably, the above-mentioned suspended matter coagulation state inspection system further has a judgment model generation means for generating a judgment model in which the input is the change in brightness variance over time of the first image and the output is the sufficiency or lack of the coagulant by performing machine learning using the change in brightness variance over time of the first image and the sufficiency or lack of the coagulant as training data, and the calculation unit judges the sufficiency or lack of the coagulant using the judgment model.

[0021] The water treatment system of the present invention includes any one of the above-described systems for inspecting the state of suspended solids flocculation, a flocculation unit that adds the flocculant to raw water and stirs it, and a solid-liquid separation tank that stores the raw water to which the flocculant has been added and stirred and allows the flocs to settle, and the system for inspecting the state of suspended solids flocculation takes an aliquot of the raw water to which the flocculant has been added and stirred from a pipe that transfers the raw water to which the flocculant has been added and stirred from the flocculation unit to the solid-liquid separation tank, and determines whether the flocculant is sufficient or insufficient. [Effects of the Invention]

[0022] According to the method for inspecting the state of flocculation of suspended solids of the present invention, by continuously capturing first images, which are images of the first observation section, it is possible to determine whether the flocculant is sufficient or insufficient, taking into account changes in the state of flocculation over time. Furthermore, according to the method for inspecting the state of flocculation of suspended solids of the present invention, the variance in brightness of the first images is calculated and used as the basis for the determination, which makes it possible to keep costs low in terms of the cost of the device and the working time. [Brief explanation of the drawings]

[0023] [Figure 1] FIG. 1A is a diagram showing the configuration of an embodiment of a system for inspecting the state of suspended matter aggregation, and FIG. 1B is a front view of a cell. [Figure 2] FIG. 1 is a flow chart of an embodiment of a method for inspecting the state of suspended matter aggregation; FIG. 2 is a flow chart of an overall process; and FIG. [Figure 3] 1 is a diagram showing a configuration of a water treatment system according to an embodiment. [Figure 4] 10A shows the transition of the average brightness of the third image, B shows the transition of the average brightness of the second image, and C shows the transition of the brightness variance of the first image when the concentration of suspended matter is high and the color is light. [Figure 5] 10A shows the transition of the average luminance of the third image, B shows the transition of the average luminance of the second image, and C shows the transition of the luminance variance of the first image when the concentration of suspended matter is high and the color is dark. [Figure 6] 10A shows the transition of the average brightness of the third image, B shows the transition of the average brightness of the second image, and C shows the transition of the brightness variance of the first image when the concentration of suspended matter is low and the color is light. [Figure 7] 10A shows the transition of the average brightness of the third image, B shows the transition of the average brightness of the second image, and C shows the transition of the brightness variance of the first image when the concentration of suspended matter is low and the color is dark. DETAILED DESCRIPTION OF THE INVENTION

[0024] Before describing the embodiments, we will briefly explain the function of flocculants. The surfaces of suspended solids (SS) particles are generally negatively charged, and small SS particles repel each other, forming a stably dispersed colloidal solution. Therefore, natural settling of small SS particles takes a long time, so flocculants are used to flocculate them. Inorganic flocculants, such as polyaluminum chloride (PAC) and aluminum sulfate, neutralize the surface charge of SS particles with cations, weakening their mutual repulsion and causing flocculation. Cationic organic flocculants are also sometimes used for the same purpose. Organic flocculants are linear water-soluble polymers that adsorb to SS particles and cross-link them, flocculating the SS and forming larger flocs. While inorganic or cationic organic flocculants are sometimes used alone to treat turbid water associated with construction work, a commonly used method involves sequentially adding and stirring an inorganic flocculant as a primary flocculant and an organic flocculant as a secondary flocculant. In this case, an anionic organic flocculant with the opposite ionicity to the inorganic flocculant is used.

[0025] 1A, a system 10 for inspecting the state of suspended solids coagulation in this embodiment includes a cell 20 into which raw water that has been stirred with the addition of a coagulant is introduced, a light 12 that illuminates the cell 20 from behind, a camera 13 that photographs the cell 20 from the front, and a calculation unit 14 that performs various calculations. The raw water is water containing SS that is the target of water treatment.

[0026] The cell 20 is a vertically elongated rectangular parallelepiped container. The cell is translucent. Preferably, the front surface of the cell 20 facing the camera 13 is transparent, and the back surface facing the light source 12 is preferably finished in a matte finish or the like to diffuse light. Such a cell can be produced, for example, by attaching a light-diffusing diffusion film 25 to the back surface of a transparent cell on its entire periphery. The inner wall surface of the cell may be coated with a hydrophilic or water-repellent material to prevent the adhesion of sludge and the like. Referring also to FIG. 1B, the cell 20 has a first observation area 21, a second observation area 22, and a third observation area 23.

[0027] The light 12 illuminates the cell 20 from behind. As the light 12, for example, a vertically elongated linear or planar light, similar to the cell 20, can be used.

[0028] Camera 13 is an imaging unit that captures images of cell 20 from the front. Camera 13 continuously captures a predetermined number of images at predetermined time intervals: a first image of a first observation unit, a second image of a second observation unit, and a third image of a third observation unit. More specifically, in the configuration shown in FIG. 1A, one camera 13 captures the entire front of cell 20 to obtain one overall image. This overall image includes all of the first to third images, and a calculation unit 14 (described later) crops the first, second, and third observation units from the overall image to obtain the first, second, and third images, respectively. Note that the number of cameras 13 is not limited to one; for example, an imaging unit may be configured with multiple cameras, and each observation unit may be captured by a separate camera.

[0029] Referring to FIG. 1B, the first observation unit 21 is provided in the middle of the cell 20 in the height direction, excluding the upper and lower ends. The first observation unit is preferably provided in a region extending from the center to the bottom of the cell in the height direction. In water introduced into the cell, SS aggregates to form flocs, which grow and settle through the first observation unit. Qualitatively, the larger the floc, the greater the brightness variance in the image in the first image, which is a transmission image through the first observation unit. Furthermore, as the flocs descend from above into the first observation unit, the brightness variance increases, and as they pass through the first observation unit and descend downward, the brightness variance decreases again. Information about the settling of flocs can be obtained primarily from the first image, which is a transmission image through the first observation unit.

[0030] The second observation unit 22 is provided above the first observation unit 21 of the cell 20. Qualitatively, the more SS there is, the more scattered light there is, and the lower the brightness of the second image, which is a transmission image of the second observation unit. Information about the supernatant water can be obtained primarily from the second image, which is a transmission image of the second observation unit.

[0031] The third observation unit 23 is located above the first observation unit 21 of the cell 20. The second observation unit 22 and the third observation unit 23 can be positioned vertically. The back surface of the cell (the surface facing the illumination unit 12) of the third observation unit is shielded from light by a dark, preferably black, light-shielding member 24. The light-shielding member can be, for example, a dark sheet or sticker attached to the back surface of the cell. The third image of the third observation unit is not a transmitted image. Direct light from the illumination unit 12 is blocked by the light-shielding member 24 and does not enter the third observation unit. However, light that enters the areas above and below the light-shielding member is scattered in those areas and some of it reaches the third observation unit. Qualitatively, the higher the SS, the higher the brightness of the third image, because light incident from the upper and lower areas is further scattered within the third observation unit, increasing the amount of light directed toward the camera 13. Similar to the second image, the third image primarily provides information about the supernatant water.

[0032] Both the second and third images provide information mainly about the supernatant water. The reason for providing both the second observation unit 22 and the third observation unit 23 is to accommodate turbidity of different colors. For light-colored turbidity, the state of the supernatant water can be more clearly read from the third observation unit, which has a black background. For dark-colored turbidity, the state of the supernatant water can be more clearly read from the second observation unit, which is directly illuminated by the light source 12 and can acquire a transmission image (second image).

[0033] The second observation unit 22 and the third observation unit 23 can be omitted, but by using the information obtained from the second observation unit and the third observation unit in addition to the first observation unit 21, the amount of flocculant can be more accurately determined.

[0034] The calculation unit 14 is electrically connected to the camera 13 by wire or wirelessly. The calculation unit 14 controls the camera 13 to take pictures, performs various processes on the images acquired by the camera 13, and calculates statistics such as the variance of the brightness of the first image and the average brightness of the second and third images. Hereinafter, these statistics will be collectively referred to as "statistics of the first to third images." The calculation unit also constitutes, together with a machine learning program, a determination model generation means for generating a model for determining whether the amount of flocculant is sufficient or insufficient. A computer such as a PC can be used as the calculation unit.

[0035] The calculation unit 14 may be installed at the location where the SS flocculation state inspection is performed, or may be installed in a remote location and connected to the camera 13 via a network such as the Internet. Furthermore, the calculation unit 14 may be physically configured as a single device, or may be configured by combining multiple devices. Preferably, the calculation unit 14 has a first calculation unit that controls the camera, etc., installed at the location where the SS flocculation state inspection is performed, and a second calculation unit that determines whether the flocculant is sufficient or insufficient, etc., installed in a remote location and connected to the first calculation unit via the Internet, etc. This allows a common judgment model to be used even when the SS flocculation state inspection is performed at multiple locations. Details of the processing by the calculation unit will be described later.

[0036] Next, the method for inspecting the SS aggregation state according to this embodiment will be described with reference to the flow chart of FIG.

[0037] Figure 2A shows the process of testing raw water with an unknown SS content and determining whether the amount of coagulant to be added is sufficient or insufficient, with a judgment model already generated by machine learning. The method for generating the judgment model will be described later.

[0038] Referring to FIG. 2A, (S1) raw water to which a flocculant has been added is introduced into cell 20.

[0039] The type of flocculant added to the raw water is not particularly limited. The flocculant may be an inorganic flocculant, an organic flocculant, or both an inorganic and an organic flocculant. In this case, two or more inorganic flocculants may be used as the inorganic flocculant, and two or more organic flocculants may be used as the organic flocculant. When both an inorganic flocculant and an organic flocculant are used as the flocculant, a mixture of the inorganic and organic flocculants may be added to the raw water, or the inorganic and organic flocculants may be added sequentially and stirred. In addition, the properties of the flocculant added to the raw water are not particularly limited, and the flocculant may be in the form of a powder, liquid, or slurry.

[0040] (S2) While illuminating the cell 20 from behind with a light source 12, the entire front surface of the cell is photographed with a camera 13. A predetermined number of images are continuously captured at a predetermined interval. The time interval between images is, for example, 0.1 to 5 seconds, preferably 1 to 3 seconds. If the interval is too long, it becomes difficult to understand the characteristics of the time-dependent changes in the statistical quantities. On the other hand, a short interval between images is not particularly problematic, but shortening the interval beyond this will result in a small effect compared to the increased amount of data. Furthermore, the number of images captured is preferably set to a number that can be captured within 10 seconds to 10 minutes, more preferably 1 to 5 minutes, depending on the time interval between images. This is because if the time from the start to the end of image capture is too short, the time-dependent changes in the statistical quantities of the first to third images may be overlooked. On the other hand, if the time from the start to the end of image capture is too long, it becomes difficult to perform feedback control by correcting the amount of flocculant added in real time in the water treatment facility.

[0041] (S3) When the camera 13 has taken a predetermined number of images, the calculation unit 14 processes the obtained images.

[0042] 2B, (S31) the entire image of the front surface of cell 20 captured by camera 13 is taken into calculation unit 14. Next, the following steps S32 to S34 are performed for each entire image.

[0043] (S32) Since the entire image includes the first image, the second image, and the third image, the calculation unit 14 extracts the first to third images by cutting out the portions of the first observation unit 21, the second observation unit 22, and the third observation unit 23 from this entire image.

[0044] (S33) The calculation unit 14 converts the first to third images into a grayscale image. The grayscale image has, for example, 256 gradations. Note that step S33 may be performed before step S32, and the first to third images may be extracted after converting the entire image into a grayscale image. However, if the first to third images are extracted first, the extraction process for the first to third images may be facilitated by, for example, drawing an outline or trim mark surrounding each observation portion in a conspicuous color such as red or yellow on the front of the cell 20 to indicate the area of each observation portion.

[0045] (S34) The calculation unit 14 calculates the statistics of the first to third images. Specifically, the variance of the brightness of each pixel in the first image is calculated, and the average brightness of each pixel in the second and third images is calculated. This completes the processing for each image.

[0046] (S35) The calculation unit 14 creates arrays of statistics calculated from all the acquired images. Specifically, three arrays are created: an array in which the variances of the brightness of the first images are arranged in the order in which they were taken, an array in which the average brightness of the second images is arranged in the order in which they were taken, and an array in which the average brightness of the third images is arranged in the order in which they were taken.

[0047] In the above description of steps S2 to S3, after camera 13 has completed capturing a predetermined number of images, calculation unit 14 captures all of the full images together (S31), and steps S32 to S35 shown in Fig. 2B are performed. However, the order of the steps may be changed if possible. For example, each time camera 13 captures a full image, it may be captured by calculation unit 14, and after the predetermined number of full images have been captured by the calculation unit, steps S32 to S35 shown in Fig. 2B may be performed. Alternatively, for example, each time camera 13 captures a full image, it may be captured by calculation unit 14, and steps S32 to S34 may be performed on that full image, and step S35 may be performed when the capturing is complete.

[0048] Returning to FIG. 2A, (S4) prepares input values for the judgment model. The input values include the array of statistics created in step S35. The input values preferably also include other information. Examples of other information include the shade and color of turbidity, pH, turbidity, oxidation-reduction potential (ORP), electrical conductivity (EC), alkalinity, zeta potential, and water temperature regarding the properties of raw water, and pH, turbidity, ORP, EC, alkalinity, and zeta potential regarding the properties of treated water.

[0049] (S5) The input value is input into the trained decision model.

[0050] (S6) The determination model takes into consideration the feature amounts derived from the arrangement of the statistics of the first to third images, and outputs the determination result of whether the amount of flocculant is sufficient or insufficient.

[0051] Preferably, the arrangement of the statistics of the first to third images and other information that have been input to the determination model and that have served as the basis for the determination are used as new learning data for further learning of the determination model.

[0052] When the calculation unit 14 is composed of a first calculation unit installed at the location where the SS aggregation state test is performed and a second calculation unit installed in a remote location, the processing by the calculation unit described above is shared and performed by the first calculation unit and the second calculation unit. The first calculation unit controls the image capture by the camera 13 (S2) and captures the entire image from the camera (S31). The image processing (S32 to S34) and creation of an array of statistics (S35) may be performed by either the first calculation unit or the second calculation unit. When these processes (S32 to S35) are performed by the first calculation unit, there is no need to send image data from the first calculation unit to the second calculation unit, which reduces the communication load and makes problems such as communication delays less likely to occur. On the other hand, when these processes (S32 to S35) are performed by the second calculation unit, high processing power is not required of the first calculation unit, thereby reducing equipment costs. In preparing input values for the judgment model (S4), information on the properties of the raw water and treated water is prepared by a first calculation unit installed at the testing site, and the arrangement of statistics is performed by the first calculation unit or the second calculation unit that performed image processing. The input values are input to the judgment model stored in the second calculation unit (S5), and the second calculation unit determines whether the flocculant is sufficient or insufficient and notifies the first calculation unit (S6).

[0053] Next, a method for generating a decision model in advance using machine learning will be described.

[0054] The difference between the training data used to generate a judgment model and the input values for the trained judgment model is that the former contains the correct answer for the flocculant deficiency. The training data creation method is the same as the process shown in Figure 2, except that instead of water with an unknown flocculant deficiency, water with a known flocculant deficiency is used. Examples of water with a known flocculant deficiency include synthetic turbid water prepared by mixing fine powder with tap water, or actual industrial wastewater, whose flocculant deficiency has been determined using a jar test or similar. Water with a known flocculant deficiency is introduced into cell 20 (S1). The following steps, similar to those shown in Figure 2, are then used to create the training data: the statistical array of the first to third images, other information, and the flocculant deficiency.

[0055] The judgment model generating means is composed of a calculation unit 14 and a machine learning program. The above-mentioned training data is input to the judgment model generating means, and machine learning is performed to generate a judgment model.

[0056] Next, a water treatment system equipped with the above-described inspection system 10 will be described.

[0057] Referring to FIG. 3, a water treatment system 30 of this embodiment includes a raw water tank 31 for storing raw water to be treated, a flocculation unit 32 for forming flocs by adding a flocculant to the raw water and stirring it, a solid-liquid separation tank 38 for storing the water to which the flocculant has been added and stirred and for allowing the flocs to settle, and an SS flocculation state inspection system 10.

[0058] The flocculation unit 32 has an inorganic flocculant addition unit 33 that adds an inorganic flocculant to raw water, a line mixer 34 that agitates the water to which the inorganic flocculant has been added, an organic flocculant addition unit 35 that is provided downstream of the line mixer and adds an organic flocculant, and a reaction tower 36 that agitates the water to which the inorganic and organic flocculants have been added to promote the flocculation reaction. Note that the equipment for agitating the water to which the flocculant has been added is not limited to this, and for example, a storage tank may be provided instead of the line mixer or reaction tower, and the water in the tank may be agitated.

[0059] Raw water is pumped from raw water tank 31 by pump P and transported through pipe 39. An appropriate amount of inorganic flocculant is added to the raw water in inorganic flocculant addition section 33 installed in pipe 39, and the water is agitated in line mixer 34 installed downstream. The water leaving line mixer 34 is further transported through pipe 40, where an appropriate amount of organic flocculant is added in organic flocculant addition section 35 installed in pipe 40. The water is agitated in reactor 36 installed downstream to promote the flocculation reaction. The water leaving reactor 36 is transported through pipe 41 to solid-liquid separation tank 38. In solid-liquid separation tank 38, flocs continue to grow and settle. The supernatant water from the solid-liquid separation tank is monitored for turbidity using a sensor (not shown). If the water meets the required purity, it is subjected to pH adjustment and other treatments before being discharged into a river or other water source. The water containing the settled flocs is discharged from the bottom of the solid-liquid separation tank and further treated as thickened sludge.

[0060] The water to which the inorganic and organic coagulants have been added and stirred branches off from pipe 41, which transfers the water from the reaction tower 36 to the solid-liquid separation tank 38, and is sent to the inspection system 10 through pipe 42. Once the water in the inspection system cell 20 and pipe 42 has been exchanged, valve 43 on pipe 42 is closed, the transfer of water to the inspection system is temporarily stopped, and the water in the cell is allowed to stand. Further growth and sedimentation of flocs progresses within the cell, and the above-described inspection method determines whether the amount of inorganic and / or organic coagulant added is insufficient. If it is determined that the amount of inorganic and / or organic coagulant added is insufficient, the amount of inorganic or organic coagulant added that is determined to be insufficient is adjusted.

[0061] In addition, the properties of the raw water and the properties of the treated water downstream of the reaction tower 36 are measured by various sensors not shown and transmitted to the calculation unit 14, and as described above, are used as learning data for the judgment model together with the arrangement of statistics of the first to third images.

[0062] After one test is completed, the inside of the cell 20 is washed with tap water or the like as needed, and inorganic and organic coagulants are added. The water is stirred and sent back to the testing system 10 through the pipe 42, and the test is repeated. At this time, the tested water is returned from the cell 20 to the raw water tank 31. [Example]

[0063] In order to explain the meaning of the statistics of the first to third images, an example of training data measured using the inspection system 10 will be shown.

[0064] Experiments were conducted by varying the type and concentration of SS, the amount of inorganic coagulant added, and the amount of organic coagulant added. Tap water was placed in a beaker, and a predetermined amount of a specified powder was added. Synthetic turbid water was prepared by stirring with a benchtop mixer. A predetermined amount of inorganic coagulant (PAC) was added to the synthetic turbid water and rapidly stirred for 1 minute. A predetermined amount of organic coagulant (anionic polymer compound) was added and rapidly stirred for 30 seconds, followed by slow stirring for 10 seconds. The pump was started, and treated water containing both inorganic and organic coagulants was introduced through the inlet at the bottom of the cell. After confirming that the treated water level had reached the outlet at the top side of the cell, the pump was stopped after 2 seconds. Photography began immediately after the pump was stopped, and 100 full-frame images were taken at approximately 2-second intervals. The cell size was 60 × 60 × 330 mm, and the full-frame resolution was 2592 × 1944 pixels.

[0065] Using the method described above, the first image of the first observation area, the second image of the second observation area, and the third image of the third observation area were extracted from each full-image. Each image was converted to a 256-level grayscale, and the variance of the brightness of the first image, the average brightness of the second image, and the average brightness of the third image were calculated. After performing the above processing on 100 full-images, each statistical value was stored in an array in the order in which they were taken. The array of statistics for the first to third images, the amount of inorganic coagulant added, and the amount of organic coagulant added were input into a machine learning program as training data. The machine learning program was created using the programming language PYTHON and its standard library (Tensorflow).

[0066] Table 1 shows the type and concentration of SS, the amount of inorganic coagulant added, and the amount of organic coagulant added. The abbreviation "X" means that the amount of inorganic coagulant added is sufficient, "x" means that the amount of inorganic coagulant added is insufficient, "Y" means that the amount of organic coagulant added is sufficient, and "y" means that the amount of organic coagulant added is insufficient. Below, each experiment will be referred to as "XY," "Xy," "xY," or "xy" depending on the amount of inorganic coagulant added and the amount of organic coagulant added.

[0067] [Table 1]

[0068] Figures 4 to 7 show the changes in statistics over time. In each figure, A is the average brightness of the third image against a black background, B is the average brightness of the second image, and C is the brightness variance of the first image. Please note that the range of values on the vertical axis differs in each figure. In addition, in the measurement results for tap water with a turbidity of 0 and no powder mixed in, the average brightness of the third image was approximately 70, the average brightness of the second image was approximately 250, and the brightness variance of the first image was 2 to 3.

[0069] Figure 4 shows the results when the SS concentration is high and the turbidity is whitish and light in color. Referring to Figure 4A, in XY, the brightness decreases over time, approaching 70, the value when the turbidity is 0. In Xy, the flocs probably do not grow to a sufficient size, causing slow settling. In xY, the brightness gradually decreases, but it can be seen that some SS remains in the supernatant. In xy, there is no settling of SS, and there is almost no change in brightness.

[0070] Referring to Figure 4B, there was almost no change in brightness in xy. The others approached the 250 value at 0 turbidity over time.

[0071] Referring to Figure 4C, in xy, the brightness was almost uniform, and the variance remained close to 0, with little change. In XY, the brightness variance rapidly decreased from an initial high value and approached 0. In xY, the initial variance was lower than in XY due to insufficient coagulation, and then rapidly approached 0. In Xy, the growth of flocs is thought to be the rate-limiting factor, and the brightness variance increased after the start of measurement, peaked, and then decreased to approach 0.

[0072] The results shown in Figures 5 to 7 generally show similar trends to those in Figure 4. However, referring to Figure 5A, when the SS concentration is high and the turbidity is a dark gray color, the average brightness is sometimes below 70, which is the value when the turbidity is 0, which is thought to be because light absorption by SS is dominant. Therefore, it was found that when the SS concentration is high and the color is dark, the information from the third observation section is not very useful.

[0073] In addition, overall, when the color is light, the information from the third observation area, which has a black background, more clearly shows the trends due to differences in the amount of coagulant added, while when the color is dark, the information from the second observation area, where light from the lighting 12 is directly incident from the back, more clearly shows the trends due to differences in the amount of coagulant added.

[0074] It was confirmed that the change in brightness variance over time of the first image reflects the insufficiency of the inorganic and organic coagulants added, and that the change in brightness variance over time of the first image can be used to determine whether the inorganic and organic coagulants have been added. In addition, the change in the average brightness of the second and third images over time also reflects the insufficiency of the inorganic and organic coagulants added, and it was thought that the accuracy of the determination could be improved by taking these into consideration together.

[0075] The present invention is not limited to the above-described embodiment, and various modifications are possible within the scope of the technical concept thereof. [Explanation of symbols]

[0076] 10. Inspection system for suspended solids coagulation status 12. Lighting 13 Camera (imaging unit) 14 Arithmetic section 20 cells 21 First Observation Section 22 Second Observation Section 23 Third Observation Section 24 Light blocking material 25 Diffusion Film 30 Water Treatment Systems 31 Raw Water Tank 32 Floc forming section 33 Inorganic flocculant addition section 34 Line Mixer 35 Organic flocculant addition section 36 Reaction tower 38 Solid-liquid separation tank 39~42 Piping 43 Valve

Claims

1. introducing water containing suspended solids and a flocculant into a light-transmitting cell; a step of continuously capturing first images of a first observation unit provided in an intermediate portion of the cell from the front of the cell at predetermined time intervals while illuminating the cell from behind; calculating, for each of the first images, a variance of brightness in the first image; a determination step of determining whether the aggregating agent is sufficient or insufficient based on a change over time in the variance of brightness of the first image; A method for inspecting the flocculation state of suspended matter, comprising:

2. The flocculant is an inorganic flocculant and an organic flocculant, The determination step is a step of determining whether the inorganic flocculant and the organic flocculant are sufficient or insufficient, The method for inspecting the state of suspended solids aggregation according to claim 1.

3. The method further includes a step of performing machine learning using the time-dependent change in variance of brightness of the first image and the sufficiency or lack of the flocculant as training data to generate in advance a determination model in which the time-dependent change in variance of brightness of the first image is used as an input and the sufficiency or lack of the flocculant is used as an output, The determination step is a step of determining whether the flocculant is sufficient or insufficient using the determination model.

3. The method for inspecting the state of suspended solids aggregation according to claim 1 or 2.

4. The method further includes a step of performing machine learning of the determination model using a time-dependent change in variance of luminance of the first image that is the basis of the determination step as learning data. The method for inspecting the state of suspended matter aggregation according to claim 3.

5. a step of continuously capturing second images of a second observation section provided above the first observation section of the cell from in front of the cell at predetermined time intervals while illuminating the cell from behind; determining an average luminance of each of the second images; a step of continuously capturing third images of a third observation section, which is located above the first observation section of the cell and above or below the second observation section, and whose back surface is shielded from light by a dark-colored light-shielding member, from the front of the cell at predetermined time intervals while illuminating the cell from behind; and calculating an average brightness of each of the third images, the determining step is a step of determining whether the aggregating agent is insufficient or not based on a time-dependent change in variance of brightness of the first image, a time-dependent change in average brightness of the second image, and a time-dependent change in average brightness of the third image.

3. The method for inspecting the state of suspended solids aggregation according to claim 1 or 2.

6. The method further comprises a step of performing machine learning using the change over time of the variance of brightness of the first image, the change over time of the average of brightness of the second image, the change over time of the average of brightness of the third image, and the amount of the flocculant as training data, thereby generating in advance a determination model in which the input is the change over time of the variance of brightness of the first image, the change over time of the average of brightness of the second image, and the change over time of the average of brightness of the third image, and the output is the amount of the flocculant; The determination step is a step of determining whether the flocculant is sufficient or insufficient using the determination model. The method for inspecting the state of suspended matter aggregation according to claim 5.

7. a light-transmitting cell into which water containing suspended matter and a flocculant is introduced; a light for illuminating the cell from behind; a first observation section provided in a middle portion of the cell; an imaging unit that continuously captures first images of the first observation unit from in front of the cell at predetermined time intervals; a calculation unit that calculates a variance of brightness in each of the first images and determines whether the aggregating agent is sufficient or insufficient based on a change over time in the variance of brightness in the first images; A system for inspecting the flocculation state of suspended matter.

8. Further, a determination model generating means is provided for generating a determination model in which an input is the change in variance of brightness of the first image over time and an amount of the flocculant is insufficient by performing machine learning using the change in variance of brightness of the first image over time and an amount of the flocculant is insufficient as training data, The calculation unit determines whether the flocculant is sufficient or insufficient using the determination model. The system for inspecting the state of suspended matter aggregation according to claim 7.

9. A system for inspecting the state of suspended matter aggregation according to claim 7 or 8; a flocculation unit that adds the coagulant to raw water and stirs it; a solid-liquid separation tank for storing the raw water to which the coagulant has been added and stirred, and for allowing flocs to settle, the system for inspecting the state of suspended solids flocculation collects a portion of the raw water, to which the flocculant has been added and stirred, from a pipe that transfers the raw water from the flocculation unit to the solid-liquid separation tank, and determines whether the flocculant is sufficient or insufficient. Water treatment system.

Images