Water purification treatment monitoring system, water purification treatment monitoring device, information processing device, program, and water purification treatment monitoring method

Abstract

To provide a water purification treatment monitoring system, water purification treatment monitoring device, information processing device, program and water purification treatment monitoring method which accurately determine a flocculation failure without installing a special sensor.SOLUTION: A water purification treatment monitoring system comprises: storage means which stores a plurality of prediction models according to water quality information; water quality information acquisition means which acquires the water quality information of raw water; image acquisition means which acquires an image of treatment water in a flocculation sedimentation process; selection means which selects one prediction model corresponding to the water quality information acquired by the water quality information acquisition means from the plurality of prediction models; and determination means which determines a flocculation failure of floc in the treatment water on the basis of the selected one prediction model and the image acquired by the image acquisition means.SELECTED DRAWING: Figure 2

Description

[Technical Field]

[0001] The present invention relates to a water purification treatment monitoring system, an information processing device, a water purification treatment monitoring device, a program, and a water purification treatment monitoring method. [Background technology]

[0002] BACKGROUND ART Conventionally, in various water treatment fields such as drinking water treatment, sewage treatment, industrial water treatment, and industrial wastewater treatment, coagulation sedimentation methods have been used as a method for removing suspended solids from water to be treated.

[0003] In the coagulation sedimentation method, suspended matter in the water to be treated is coagulated using a coagulant, and the resulting flocs (aggregates) are precipitated to remove the suspended matter from the water to be treated. Therefore, in water treatment technology that uses the coagulation sedimentation method, it is essential to ensure that the formation of flocs using the coagulant proceeds smoothly.

[0004] Therefore, for example, monitoring techniques are known in which operators monitor floc formation by patrolling, or in which the state of floc formation is monitored. For example, Patent Document 1 describes a monitoring technique in which the zeta potential of flocs is measured using images and the injection of a pH adjuster or a coagulant is controlled. Patent Document 2 describes a technique in which a sensor is installed to measure the particle size of flocs and detect abnormalities at an early stage. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] Japanese Patent Application Laid-Open No. 2017-136545 [Patent Document 2] Japanese Patent Application Laid-Open No. 2015-192960 Summary of the Invention [Problem to be solved by the invention]

[0006] Here, it is difficult for an operator who is not sufficiently skilled to perform the monitoring properly.

[0007] Furthermore, in order to apply the techniques of Patent Documents 1 and 2, it is necessary to install a special sensor.

[0008] In view of the above circumstances, an object of the present invention is to provide a water purification treatment monitoring system, a water purification treatment monitoring device, an information processing device, a program, and a water purification treatment monitoring method that can accurately determine coagulation failure without installing a special sensor. [Means for solving the problem]

[0009] A water purification treatment monitoring system according to an embodiment of the present invention includes: a storage means for storing a plurality of prediction models according to water quality information; a water quality information acquisition means for acquiring water quality information of raw water; an image acquisition means for acquiring an image of treated water in a coagulation and sedimentation process; a selection means for selecting one prediction model from the plurality of prediction models that corresponds to the water quality information acquired by the water quality information acquisition means; a determining means for determining the coagulation failure of flocs in the treated water based on the one selected prediction model and the image acquired by the image acquiring means; Equipped with.

[0010] A water purification treatment monitoring system according to an embodiment of the present invention includes: a storage means for storing a plurality of prediction models according to water quality information, which are machine-learned using images of treated water in the coagulation-sedimentation process and the coagulation state of flocs as training data; a water quality information acquisition means for acquiring water quality information of raw water; an image acquisition means for acquiring an image of treated water in a coagulation and sedimentation process; a selection means for selecting one prediction model from the plurality of prediction models that corresponds to the water quality information acquired by the water quality information acquisition means; an output means for outputting a floc aggregation state predicted from the image acquired by the image acquisition means using the one selected prediction model; a determining means for determining whether flocs in the treated water are coagulated properly based on the coagulation state of the flocs; Equipped with.

[0011] A water purification treatment monitoring device according to one embodiment of the present invention includes: a storage means for storing a plurality of prediction models according to water quality information; a water quality information acquisition means for acquiring water quality information of raw water; an image acquisition means for acquiring an image of treated water in a coagulation and sedimentation process; a selection means for selecting one prediction model from the plurality of prediction models that corresponds to the water quality information acquired by the water quality information acquisition means; a determining means for determining the coagulation failure of flocs in the treated water based on the one selected prediction model and the image acquired by the image acquiring means; Equipped with.

[0012] An information processing device according to an embodiment of the present invention includes: a storage means for storing a plurality of prediction models according to water quality information; a water quality information acquisition means for acquiring water quality information of raw water; an image acquisition means for acquiring an image of treated water in a coagulation and sedimentation process; a selection means for selecting one prediction model from the plurality of prediction models that corresponds to the water quality information acquired by the water quality information acquisition means; a determining means for determining the coagulation failure of flocs in the treated water based on the one selected prediction model and the image acquired by the image acquiring means; To configure a water purification treatment monitoring system comprising: The apparatus includes at least one of the storage means, the water quality information acquisition means, the image acquisition means, the selection means, and the determination means.

[0013] A program according to an embodiment of the present invention includes: The computer that monitors the water purification process a storage means for storing a plurality of prediction models according to water quality information; a water quality information acquisition means for acquiring water quality information of raw water; an image acquisition means for acquiring an image of treated water in a coagulation and sedimentation process; a selection means for selecting one prediction model from the plurality of prediction models that corresponds to the water quality information acquired by the water quality information acquisition means; a determining means for determining the coagulation failure of flocs in the treated water based on the one selected prediction model and the image acquired by the image acquiring means; and make it work.

[0014] A program according to an embodiment of the present invention includes: The system is configured by a plurality of information processing devices connected to each other so as to be able to communicate with each other, a storage means for storing a plurality of prediction models according to water quality information; a water quality information acquisition means for acquiring water quality information of raw water; an image acquisition means for acquiring an image of treated water in a coagulation and sedimentation process; a selection means for selecting one prediction model from the plurality of prediction models that corresponds to the water quality information acquired by the water quality information acquisition means; a determining means for determining the coagulation failure of flocs in the treated water based on the one selected prediction model and the image acquired by the image acquiring means; In the water purification treatment monitoring system, one of the plurality of information processing devices is The means functions as at least one of the storage means, the water quality information acquisition means, the image acquisition means, the selection means, and the determination means.

[0015] A water purification treatment monitoring method according to one embodiment of the present invention includes: storing a plurality of prediction models according to water quality information; acquiring water quality information of raw water; acquiring an image of treated water in a coagulation and sedimentation process; selecting one prediction model from the plurality of prediction models that corresponds to the acquired water quality information; determining the degree of flocculation failure of the flocs in the treated water based on the selected one prediction model and the acquired image; Includes. [Effects of the Invention]

[0016] According to the water purification treatment monitoring system, water purification treatment monitoring device, information processing device, program, and water purification treatment monitoring method according to one embodiment of the present invention, coagulation failure can be accurately determined without installing a special sensor. [Brief explanation of the drawings]

[0017] [Figure 1] FIG. 1 is a diagram showing a schematic configuration of a typical water treatment system. [Figure 2] 1 is a block diagram showing a schematic configuration of a water purification treatment monitoring system according to an embodiment of the present invention. [Figure 3] 10 is a flowchart showing a determination operation of the information processing device. [Figure 4] 10 is a flowchart showing a procedure for creating training data for training a prediction model. [Figure 5] FIG. 10 is a diagram showing the correspondence between information indicating turbidity and water temperature, which are water quality information, and a prediction model. DETAILED DESCRIPTION OF THE INVENTION

[0018] Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

[0019] Water treatment systems are used, for example, in water purification plants to treat raw water taken from rivers, etc. However, the raw water to be treated in water treatment systems is not limited to river water, etc.

[0020] Here, a schematic configuration of an example of a water treatment system is shown in Figure 1. The water treatment system 100 shown in Figure 1 includes a receiving well 10, a chemical mixing basin 20, a flocculation basin 30 whose interior is divided into three flocculation sections 31, 32, and 33, and a settling basin 40.

[0021] In the water treatment system 100, raw water supplied from a receiving well 10 is mixed with a coagulant in a chemical mixing basin 20 using an agitator 21. Furthermore, coagulant-mixed water obtained by mixing the raw water with the coagulant in the chemical mixing basin 20 is supplied to a flocculation basin 30, and flows sequentially through a first flocculation section 31, a second flocculation section 32, and a third flocculation section 33 in the flocculation basin 30.

[0022] In the flocculation basin 30, flocs grow in the coagulant-mixed water under agitation by agitators 34, 35, and 36 provided in each of the flocculation sections 31, 32, and 33. The time that the treated water remains in the flocculation basin 30 is, for example, about 20 to 40 minutes, but is not limited to this example. The flocs in the coagulant-mixed water that flow out of the flocculation basin 30 are then settled in a settling basin 40.

[0023] The flocs that have settled in the settling tank 40 are scraped up by a scraper 41 provided at the bottom of the settling tank 40, and then discharged outside the settling tank 40 for treatment.

[0024] The treated water obtained by settling the flocs grown in the flocculation basin 30 in the settling basin 40 is supplied to the filtration tank 50. In the filtration tank 50, small flocs that could not be removed by the settling process are removed by filtering through a layer of sand or gravel or a filter.

[0025] In this water treatment system 100, the intensity of the agitation in the chemical mixing basin 20 and the flocculation basin 30 is not particularly limited and can be determined using known methods. Specifically, the intensity of the agitation can be determined in advance according to the quality of the raw water, and controlled to be in accordance with the measured quality of the raw water.

[0026] The water treatment system 100 also includes a water purification treatment monitoring device 60 that determines whether flocculation has failed in the treatment water using a flocculant and notifies the outside.

[0027] Furthermore, the water treatment system 100 includes a water quality information acquisition device 70 that acquires water quality information of the raw water in the receiving well 10 and an image acquisition device 80 that acquires images of the treated water, as devices that provide the water purification treatment monitoring device 60 with information necessary for determining the poor coagulation of flocs.

[0028] A water purification treatment monitoring system according to one embodiment of the present invention will be described below. The water purification treatment monitoring system shown in Fig. 2 includes a water purification treatment monitoring device 60, a water quality information acquisition device 70, and an image acquisition device 80.

[0029] (Outline of the water quality information acquisition device) An overview of the water quality information acquisition device 70 will be described. The water quality information acquisition device 70 is a device that acquires water quality information about raw water. In this embodiment, the water quality information acquisition device 70 is a turbidity meter that measures the turbidity of the raw water in the receiving well 10. Note that the "turbidity" referred to here may be, for example, the turbidity defined in the Japan Water Works Association's "Drinking Water Testing Method," the turbidity defined in JIS K0101 "Industrial Water Testing Method," or "NTU" (Nephelometric Turbidity Unit). The water quality information acquisition device 70 can also acquire water temperature, organic matter concentration, etc.

[0030] (Image acquisition device overview) An overview of the image acquisition device 80 will be described. The image acquisition device 80 is a device that acquires images of treated water in a coagulation and sedimentation process. The image acquisition device 80 is generally inexpensive compared to various common sensors that detect water quality. In this specification, "image" includes not only still images but also videos. When the acquired image is a video, it is desirable to acquire snapshot images at regular intervals or to acquire images by dividing the video into small pieces in order to reduce the load on the database.

[0031] In this embodiment, the image acquisition device 80 is a camera that is installed in the final stage of the flocculation basin 30 and acquires images of the final stage. In the final stage, flocs become larger and the rotation speed of the flocculator is slower than in stages before the final stage. Therefore, by installing the image acquisition device 80 in the final stage, the image acquisition device 80 can capture images of flocs more clearly, improving the image acquisition accuracy. Furthermore, in some water purification plants, a camera is already installed in the final stage, and by using an existing camera, the introduction cost can be reduced.

[0032] The image acquisition device 80 may be installed in the final stage of the flocculation basin 30 or in a stage before the final stage. The image acquisition device 80 may also be installed in the chemical mixing basin 20 or the settling basin 40. Installing the image acquisition device 80 in the chemical mixing basin 20 allows for earlier determination of the floc coagulation state. Installing the image acquisition device 80 at the entrance to the settling basin 40 allows for clearer image acquisition. The installation location of the image acquisition device 80 is determined by balancing the need for early determination of the floc coagulation state with the clarity of the images acquired. The better the performance of the image acquisition device 80, the earlier it can be installed, allowing for earlier determination of the floc coagulation state. Generally, the rotation speed of the flash mixer in the chemical mixing basin 20 is faster than the rotation speed of the flocculator installed in the flocculation basin 30. Therefore, when installing the image acquisition device 80 in the chemical mixing basin 20, it is preferable to use an image acquisition device with higher accuracy. The image acquisition device 80 may also be installed in the pipeline (flow path) from the chemical mixing basin 20 to the flocculation basin 30, and if the chemical mixing basin 20 is a two-stage type, it is preferable to install the image acquisition device 80 in the chemical mixing basin in the latter stage.

[0033] (Outline of water purification monitoring device) Next, an overview of the water purification process monitoring device 60 will be described. The water purification process monitoring device 60 is a device that determines the failure of floc coagulation in treated water. The water purification process monitoring device 60 is installed at any location, such as a water treatment plant. In this embodiment, the water purification process monitoring device 60 is used in combination with a water quality information acquisition device 70 and an image acquisition device 80 external to the water purification process monitoring device 60.

[0034] (Hardware configuration of water purification treatment monitoring device) Next, a detailed description will be given of the hardware configuration of the water purification treatment monitoring device 60. The water purification treatment monitoring device 60 includes a communication unit 61, a display unit 62, an operation unit 63, a storage unit 64, and a control unit 66.

[0035] The communication unit 61 is one or more interfaces that communicate with external devices wirelessly or via wires. In this embodiment, the communication unit 61 is capable of communicating with the water quality information acquisition device 70 and the image acquisition device 80. Communication between the communication unit 61 and the water quality information acquisition device 70 and the image acquisition device 80 is performed via any device such as a PLC, but a configuration in which communication is performed without the use of such a device is also possible.

[0036] The display unit 62 is a display device, such as a liquid crystal display or an OEL (organic electroluminescence) display, but is not limited to these and may be any display device.

[0037] The operation unit 63 is an input interface that accepts user operations. The input interface may be, for example, a pointing device such as a mouse, physical keys, or a touch panel that is integral with the display unit 62, but is not limited to these and may be any input interface.

[0038] The storage unit 64 is a storage device including one or more memories. The memory may be, for example, a semiconductor memory, a magnetic memory, an optical memory, or the like, but is not limited to these, and may be any memory. The storage unit 64 functions as, for example, a primary storage device or a secondary storage device. The storage unit 64 is, for example, built into the water purification treatment monitoring device 60, but may also be configured to be externally connected to the water purification treatment monitoring device 60 via any interface.

[0039] The memory unit 64 stores multiple prediction models corresponding to water quality information. In this embodiment, the memory unit 64 stores a first prediction model corresponding to raw water turbidity of less than 2 degrees, a second prediction model corresponding to raw water turbidity of 2 degrees or more but less than 10 degrees, and a third prediction model corresponding to raw water turbidity of 10 degrees or more. This can improve the prediction accuracy of each prediction model compared to, for example, storing only one prediction model. When an image of treated water is input, each prediction model classifies the coagulation state of the treated water into one of "normal," "poor coagulation due to insufficient coagulant injection rate," "poor coagulation due to inappropriate pH," and "poor coagulation due to inappropriate alkalinity," and outputs the result.

[0040] In this embodiment, the number of prediction models is three, but the number may be two, four, or more. Increasing the number of prediction models by further subdividing turbidity, for example, can improve the accuracy of determining poor agglutination. However, since it may be difficult to increase the amount of training data for each prediction model without duplication, it is preferable to set the number of prediction models so as to improve the accuracy of determination while taking into account the amount of training data that can be obtained.

[0041] In this embodiment, a prediction model that has been trained using training data is stored in the storage unit 64, but a prediction model before training may also be stored in the storage unit 64. In this case, the prediction model is trained by a training means 661, which will be described later.

[0042] The control unit 66 is one or more processors. The processor may be, for example, a general-purpose processor or a dedicated processor specialized for a particular process, but is not limited to these and may be any processor. The control unit 66 controls the overall operation of the water purification treatment monitoring device 60.

[0043] (Software configuration of water purification treatment monitoring device) Next, we will explain the software configuration of the water purification treatment monitoring device 60. One or more programs used to control the operation of the water purification treatment monitoring device 60 are stored in the memory unit 64. When the one or more programs are loaded by the control unit 66, they cause the control unit 66 to function as a learning means 661, a selection means 662, a determination means 663, a notification means 664, a storage means 665, a water quality information acquisition means 666, and an image acquisition means 667.

[0044] The learning means 661 is a means for training the prediction model on training data. The learning means 661 is a means for training the prediction model to learn the relationship between the input image and the output state of floc coagulation in the treated water by machine learning. The output may also include the cause of poor floc coagulation. Causes of poor coagulation include insufficient coagulant injection rate, improper pH, and improper alkalinity. The prediction model is, for example, a supervised learning model. Any method can be applied to machine learning.

[0045] The selection means 662 is a means for selecting one prediction model corresponding to the water quality information acquired by the water quality information acquisition device 70 from among the multiple prediction models stored in the memory unit 64. For example, the selection means 662 selects a first prediction model when the raw water turbidity acquired by the water quality information acquisition device 70 is less than 2 degrees, selects a second prediction model when the raw water turbidity is 2 degrees or more and less than 10 degrees, and selects a third prediction model when the raw water turbidity is 10 degrees or more.

[0046] The determining means 663 is a means for determining the poor flocculation of flocs in the treated water based on the one selected prediction model and the image acquired by the image acquisition device 80. The determining means 663 can also estimate and output the cause of the poor flocculation.

[0047] The notification means 664 is a means for notifying an external person (e.g., a monitor) of the determination result of the determination means 663 regarding the floc coagulation failure via the display unit 62. The notification means 664 can warn of the coagulation failure and its cause, and may be incorporated into the display unit 62. The notification means 664 may notify the determination result to a control device of the water treatment system 100 (e.g., a control device for controlling the amount of coagulant injection).

[0048] The storage means 665 is a means for storing data or information in the memory unit 64. In this embodiment, the storage means 665 stores a plurality of prediction models according to water quality information.

[0049] The water quality information acquisition means 666 is a means for acquiring raw water quality information from the water quality information acquisition device 70 .

[0050] The image acquisition means 667 is a means for acquiring an image of the treated water in the coagulation and sedimentation process from the image acquisition device 80.

[0051] (Procedure for creating training data) The procedure for creating the training data will be described with reference to FIG.

[0052] Step S100: The water quality information acquisition device 70 measures the water quality information of the raw water. In this embodiment, the water quality information acquisition device 70 is a turbidity meter, and acquires information indicating the turbidity in the receiving well 10 as the water quality information of the raw water.

[0053] Step S101: The image capture device 80 captures images of the treated water during the coagulation and sedimentation process. In this embodiment, the image capture device 80 is a camera, and captures images of the treated water at the final stage of the flocculation basin 30 or a stage before that, or at the entrance to the sedimentation basin 40 or somewhere along the way. The image capture device 80 can be installed inside or outside the treated water. The images include video.

[0054] In one example, one image contains approximately 10 flocs. However, the number of flocs contained in one image can be determined arbitrarily. Furthermore, capturing an image against a screen background unifies the background color and limits the capture range to the depth direction from the camera to the screen position. In other words, objects located farther than the screen are not captured. When the image capture device 80 captures a distant object, it becomes difficult to identify the particle size of the flocs from the captured image. On the other hand, limiting the capture range makes it easier to calculate the particle size of the flocs, for example, from the number of pixels of the flocs in the captured image.

[0055] Step S102: The operator or developer determines the flocculation state of the treated water whose image was acquired in step S101. The flocculation state serves as the correct label for the prediction model. In this embodiment, the flocculation state can be one of four states: "normal," "poor flocculation due to insufficient flocculant injection rate," "poor flocculation due to inappropriate pH," and "poor flocculation due to inappropriate alkalinity." One of the conditions for determining a state as normal may be that the average particle size and number of flocs (the number of flocs contained in a unit volume of treated water) are within a target range. Another condition for determining a state as normal may be that the shape of the flocs is normal (e.g., spherical). These criteria can be understood from images acquired during training and evaluation. Therefore, the prediction of the flocculation state made by the prediction model is considered to be based on these criteria that can be understood from images.

[0056] The conditions for determining that the floc coagulation state of the treated water is normal may include that the settling velocity of the flocs is appropriate. As described above, the image may be a still image or a moving image. For example, the settling velocity of the flocs can be calculated from a still image, but it can be more easily calculated from a moving image.

[0057] When the flocculation state is poor, the operator or developer can determine whether the cause is poor flocculation due to an insufficient flocculant injection rate, poor flocculation due to an inappropriate pH, or poor flocculation due to an inappropriate alkalinity. To make this determination, the flocculant injection rate, pH, alkalinity, etc. of the treated water may be measured. The inventors have discovered that when the flocculation state is poor, the flocs shown in the photographed image tend to show the following trends depending on the flocculant injection rate, pH, and alkalinity.

[0058] Specifically, when the flocculation state is poor due to an insufficient coagulant injection rate, the average particle size of the flocs tends to be smaller than the target range, the shape of the flocs tends to be round, and the number of flocs tends to be smaller than the target range.

[0059] When the flocs are poorly coagulated due to a high pH, ​​the average particle size of the flocs is within the target range, the shape of the flocs is no longer spherical, and the number of flocs tends to be within the target range. Furthermore, the treated water tends to become cloudy. Furthermore, when the shape of the flocs is no longer spherical, the density of the flocs decreases, making it difficult for the flocs to settle, and the settling rate tends to slow down.

[0060] When the flocs are poorly aggregated due to low pH, the average particle size of the flocs tends to be smaller than the target range, the shape of the flocs tends to be round, and the number of flocs tends to be within the target range.

[0061] When the flocculation state is poor due to low alkalinity, the average particle size of the flocs tends to be smaller than the target range, the shape of the flocs is no longer spherical, and the number of flocs tends to be lower than the target range. Furthermore, the treated water tends to become cloudy.

[0062] Here, the above-mentioned tendency when coagulation failure occurs may be evident or faint depending on the water quality information of the raw water. The inventors discovered that the manifestation of the above-mentioned tendency differs depending on the water quality information of the raw water, for example, on the turbidity. The inventors discovered that, in particular, when the turbidity range is classified into turbidity ranges of 0 degrees or more but less than 2 degrees, 2 degrees or more but less than 10 degrees, and 10 degrees or more, the manifestation of the above-mentioned tendency differs for each classification (for example, the above-mentioned tendency may be evident or faint for each turbidity range). For this reason, in this embodiment, multiple prediction models corresponding to different turbidity ranges are prepared, and each prediction model is trained using training data belonging to the turbidity range corresponding to each prediction model. Therefore, the prediction accuracy of each prediction model can be improved compared to, for example, training a single prediction model using training data.

[0063] The trends for each cause of the above-mentioned coagulation failure can be grasped from the acquired images, and therefore the prediction of the cause of the coagulation failure made by the prediction model is considered to be based on the above-mentioned trends that can be grasped from the images.

[0064] Step S103: The operator or developer generates a single piece of training data (image and correct label) to which water quality information is associated, based on the water quality information (e.g., information indicating turbidity) measured by the water quality information acquisition device 70, the image acquired by the image acquisition device 80, and the floc aggregation state (correct label) of the treated water determined by the operator or developer.

[0065] Operators and developers repeat the above procedure with raw water under different conditions to collect a large amount of training data. Note that the turbidity of raw water, for example, fluctuates due to rainfall, etc.

[0066] To increase the amount of training data used to train a predictive model, developers can conduct batch tests (laboratory tests) or create training data based on the processing of a mini-plant (a miniaturized plant that performs the same processing as a water purification plant). In these cases, the water quality information acquisition device 70 and the image acquisition device 80 are installed in a location different from the actual facility where the floc coagulation state is to be determined. For example, developers can create multiple training data sets by changing one or more of the following conditions: the turbidity of the raw water, the coagulant injection rate of the treated water, the pH, and the alkalinity. Developers can also modify the water quality information acquired by the water quality information acquisition device 70 and the images acquired by the image acquisition device 80 as needed so that they can be applied to the actual facility where the floc coagulation state is to be determined.

[0067] Developers should pay attention to the content and duration of water treatment so that the batch test and mini-plant treatment correspond to the treatment in the actual facility where the floc coagulation state is judged.

[0068] (Learning operation of water purification treatment monitoring device) The operator or developer uses the operation unit 63 to input the training data associated with the water quality information into the learning means 65 of the water purification treatment monitoring device 60.

[0069] The learning means 65 trains the prediction model corresponding to the input water quality information of the raw water using the training data. For example, in this embodiment, when the turbidity as water quality information of the raw water is 3 degrees, the second prediction model is trained using the training data. By training the prediction model corresponding to the input water quality information of the raw water in this way using the training data including images of treated water obtained by treating the raw water, the accuracy of the prediction model can be improved.

[0070] When an image of treated water in a coagulation sedimentation process is input, the prediction model (trained model) that has completed learning outputs a prediction result of the coagulation state of flocs in the treated water. In this embodiment, the output prediction result is one of "normal," "poor coagulation due to insufficient coagulant injection rate," "poor coagulation due to inappropriate pH," and "poor coagulation due to inappropriate alkalinity."

[0071] (Determination operation of water purification treatment monitoring device) Next, the determination operation of the water purification treatment monitoring device 60 will be described with reference to FIG.

[0072] Step S200: The water quality information acquisition device 70 measures the water quality information of the raw water. This step is the same as step S100 described above.

[0073] Step S201: The image acquisition device 80 acquires an image of the treated water in the coagulation and sedimentation process, similar to step S101 described above.

[0074] Step S202: The water quality information acquired by the water quality information acquisition device 70 and the images acquired by the image acquisition device 80 are input to the water purification treatment monitoring device 60. This input can be performed automatically.

[0075] Step S203: The selection means 662 of the water purification treatment monitoring device 60 selects one prediction model from among the multiple prediction models that corresponds to the water quality information acquired by the water quality information acquisition device 70. For example, in this embodiment, when the turbidity of the raw water as the water quality information is 9 degrees, the second prediction model is selected.

[0076] Step S204: The determination means 663 of the water purification treatment monitoring device 60 determines the floc coagulation failure in the treated water based on the selected prediction model and the image acquired by the image acquisition device 80. The determination result is output to the display unit 62 or the like. As an output, the notification means 664 can also notify an external person (e.g., a monitor) of the floc coagulation failure and its cause.

[0077] Specifically, the determination means 663 inputs the image acquired by the image acquisition means 667 into the prediction model. The determination means 663 determines whether flocs are poorly coagulated based on the prediction result of the floc coagulation state output from the prediction model.

[0078] For example, if the prediction result is "normal", the determining means 663 determines that poor flocculation has not occurred (that is, the flocs have normally flocculated).

[0079] Furthermore, if the prediction result is "poor flocculation due to insufficient flocculant injection rate," the determining means 663 determines that poor flocculation of flocs has occurred and that the insufficient flocculant injection rate is the cause.

[0080] Furthermore, if the prediction result is "poor flocculation due to improper pH," the determining means 663 determines that poor flocculation of flocs has occurred and that improper pH is the cause.

[0081] Furthermore, if the prediction result is "poor flocculation due to inappropriate alkalinity," the determining means 663 determines that poor flocculation has occurred and that inappropriate alkalinity is the cause.

[0082] The operator takes action based on the judgment result made by the judgment means 663. For example, if it is judged that the flocculation state is poor due to an insufficient flocculant injection rate, the operator injects additional flocculant into the treated water. Then, the process of steps S200 to S204 is performed again to check whether the flocculation state has returned to normal.

[0083] As described above, according to this embodiment, the water purification treatment monitoring system includes a memory unit 64 that stores multiple prediction models corresponding to water quality information, a water quality information acquisition device 70 that acquires water quality information of raw water, an image acquisition device 80 that acquires images of treated water in the coagulation-sedimentation process, a selection means 662 that selects one prediction model from the multiple prediction models that corresponds to the water quality information acquired by the water quality information acquisition device 70, and a determination means 663 that determines the poor coagulation of flocs in the treated water based on the selected prediction model and images acquired by the image acquisition device 80. Because the water purification treatment monitoring system of this embodiment does not require special sensors, it can be introduced at low cost. Furthermore, by storing multiple prediction models corresponding to water quality information of the raw water, each prediction model can accurately predict the coagulation state of flocs resulting from treating the raw water.

[0084] Although the present invention has been described based on the drawings and examples, it should be noted that those skilled in the art can easily make various modifications and alterations based on the present disclosure. Therefore, it should be noted that these modifications and alterations are included within the scope of the present invention. For example, the functions included in each means, step, etc. can be rearranged so as not to be logically inconsistent, and multiple means, steps, etc. can be combined into one or divided into two or more.

[0085] For example, the functions of the water purification treatment monitoring device 60 according to the above-described embodiment may be divided among multiple information processing devices. At least one of the multiple information processing devices may be implemented as a server connected to a network such as the Internet. For example, among the multiple water purification treatment monitoring devices, a first water purification treatment monitoring device may function as some of the means including the selection means 662, determination means 663, notification means 664, storage means 665, water quality information acquisition means 666, and image acquisition means 667, and a second water purification treatment monitoring device may function as the remaining means. The above-described example of distributed arrangement of multiple means is not limited to this, and the means may be distributed and arranged in any number of water purification treatment monitoring devices.

[0086] In the above-described embodiment, an example has been described in which turbidity is used as water quality information for raw water. However, water quality information other than turbidity may also be used. For example, it is conceivable to use water temperature in addition to raw water turbidity. In such a case, the storage unit 64 can store multiple prediction models corresponding to combinations of turbidity ranges and water temperature ranges. For example, as shown in FIG. 5, the storage unit 64 can store a total of six prediction models corresponding to combinations of turbidity ranges and water temperature ranges.

[0087] In the above-described embodiment, an example has been described in which the prediction model classifies the floc formation state into one of four categories: "normal," "poor coagulation due to insufficient coagulant injection rate," "poor coagulation due to inappropriate pH," and "poor coagulation due to inappropriate alkalinity." However, the number of categories that the prediction model can classify and the content of each category are not limited to this example. For example, "poor coagulation due to inappropriate pH" may be classified into "poor coagulation due to high pH" and "poor coagulation due to low pH." Furthermore, for example, "poor coagulation" may be classified into "slight problem" and "major problem." Furthermore, the output prediction result may be either "normal" or "poor coagulation."

[0088] In the above-described embodiment, an example of the operation of the water purification treatment monitoring device 60 has been described with reference to Figures 2 to 4. However, a configuration in which some steps included in the above-described operation, or some operations included in one step, are omitted as long as it is not logically inconsistent is also possible. Also, a configuration in which the order of multiple steps included in the above-described operation is reversed as long as it is not logically inconsistent is also possible.

[0089] In the above embodiment, the various means realized by the control unit 66 of the water purification treatment monitoring device 60 have been described as software configurations, but at least some of the means may be a concept including software resources and / or hardware resources. For example, the notification means 664 may include one or more display devices or printing devices, etc.

[0090] Furthermore, a device such as a computer or a mobile phone can be used to function as the water purification treatment monitoring device 60 according to the above-described embodiment. This device can be realized by storing a program describing the processing content for realizing each function of the water purification treatment monitoring device 60 according to the embodiment in the memory of the device, and having the processor of the device read and execute the program. [Explanation of symbols]

[0091] 10 Landing well 20 Chemical Mixing Pond 21 Stirrer 30 Flocculation pond 31 First flocculation section 32 Second floc forming section 33 Third flocculation section 34 Stirrer 40 Sedimentation tank 41 Collector 50 Filtration tank 60 Water purification treatment monitoring device 61 Communications Department 62 Display section 63 Operation section 64 Memory section 66 Control Unit 661 Learning Tools 662 Selection Method 663 Judgment means 664 Notification means 665 Storage means 666 Water quality information acquisition means 667 Image Acquisition Method 70 Water quality information acquisition device 80 Image acquisition device 100 Water Treatment Systems

Claims

1. A storage means for storing a plurality of prediction model programs each associated with a different turbidity range; a water quality information acquisition means for acquiring water quality information including the turbidity of raw water; an image acquisition means for acquiring an image of treated water in a coagulation and sedimentation process; a selection means for selecting one prediction model program from the plurality of prediction model programs that corresponds to the turbidity range to which the turbidity of the raw water acquired by the water quality information acquisition means belongs; a determining means for determining the coagulation failure of flocs in the treated water based on the one selected prediction model program and the image acquired by the image acquiring means; A water purification monitoring system comprising:

2. 2. The water purification monitoring system according to claim 1, wherein the determining means determines whether the coagulation failure is due to an insufficient coagulant injection rate, an inappropriate pH, or an inappropriate alkalinity.

3. 3. The water purification monitoring system according to claim 1, wherein the image acquisition means acquires an image of a final stage of the flocculation basin.

4. The water purification treatment monitoring system according to any one of claims 1 to 3, wherein the image is a moving image.

5. 5. The water purification treatment monitoring system according to claim 1, wherein the plurality of prediction model programs are machine-learned using images of treated water in a coagulation-sedimentation process and the coagulation state of flocs as training data.

6. A storage means for storing a plurality of prediction model programs each associated with a different turbidity range; a water quality information acquisition means for acquiring water quality information including the turbidity of raw water; an image acquisition means for acquiring an image of treated water in a coagulation and sedimentation process; a selection means for selecting one prediction model program from the plurality of prediction model programs that corresponds to the turbidity range to which the turbidity of the raw water acquired by the water quality information acquisition means belongs; a determining means for determining the coagulation failure of flocs in the treated water based on the one selected prediction model program and the image acquired by the image acquiring means; A water purification treatment monitoring device comprising:

7. A storage means for storing a plurality of prediction model programs each associated with a different turbidity range; a water quality information acquisition means for acquiring water quality information including the turbidity of raw water; an image acquisition means for acquiring an image of treated water in a coagulation and sedimentation process; a selection means for selecting one prediction model program from the plurality of prediction model programs that corresponds to the turbidity range to which the turbidity of the raw water acquired by the water quality information acquisition means belongs; a determining means for determining the coagulation failure of flocs in the treated water based on the one selected prediction model program and the image acquired by the image acquiring means; To configure a water purification treatment monitoring system comprising: An information processing device comprising at least one of the storage means, the water quality information acquisition means, the image acquisition means, the selection means, and the determination means.

8. The computer that monitors the water purification process a storage means for storing a plurality of prediction model programs each associated with a different turbidity range; a water quality information acquisition means for acquiring water quality information including the turbidity of raw water; an image acquisition means for acquiring an image of treated water in a coagulation and sedimentation process; a selection means for selecting one prediction model program from the plurality of prediction model programs that corresponds to the turbidity range to which the turbidity of the raw water acquired by the water quality information acquisition means belongs; a determining means for determining the coagulation failure of flocs in the treated water based on the one selected prediction model program and the image acquired by the image acquiring means; A program that makes things work.

9. The system is configured by a plurality of information processing devices connected to each other so as to be able to communicate with each other, a storage means for storing a plurality of prediction model programs each associated with a different turbidity range; a water quality information acquisition means for acquiring water quality information including the turbidity of raw water; an image acquisition means for acquiring an image of treated water in a coagulation and sedimentation process; a selection means for selecting one prediction model program from the plurality of prediction model programs that corresponds to the turbidity range to which the turbidity of the raw water acquired by the water quality information acquisition means belongs; a determining means for determining the coagulation failure of flocs in the treated water based on the one selected prediction model program and the image acquired by the image acquiring means; In a water purification treatment monitoring system comprising: one information processing device among the plurality of information processing devices; A program that functions as at least one of the storage means, the water quality information acquisition means, the image acquisition means, the selection means, and the determination means.

10. Storing a plurality of prediction model programs each associated with a different turbidity range; acquiring water quality information including turbidity of raw water; acquiring an image of treated water in a coagulation and sedimentation process; selecting one prediction model program from the plurality of prediction model programs that corresponds to a turbidity range to which the acquired turbidity of the raw water belongs; determining the flocculation failure of the flocs in the treated water based on the selected one of the prediction model programs and the acquired image; A water purification treatment monitoring method executed by an information processing device, comprising:

11. A storage means for storing a plurality of prediction model programs according to turbidity information; a water quality information acquisition means for acquiring water quality information including turbidity information and alkalinity information of raw water; an image acquisition means for acquiring an image of treated water in a coagulation and sedimentation process; a selection means for selecting one prediction model program corresponding to the turbidity information acquired by the water quality information acquisition means from among the plurality of prediction model programs; a determining means for determining whether the flocs in the treated water are coagulating poorly due to insufficient coagulant injection, insufficient coagulation due to inappropriate pH, or insufficient coagulation due to inappropriate alkalinity, based on the one selected prediction model program, the image acquired by the image acquiring means, and the alkalinity information; A water purification monitoring system comprising:

Images