Water quality determination method and water quality determination apparatus

The method uses image analysis and machine learning to determine treated water quality by assessing sedimentation properties and SS concentration, addressing the challenge of detecting water quality deterioration in wastewater treatment systems.

JP2024165756APending Publication Date: 2024-11-28JFE ENGINEERING CORP
View PDF 7 Cites 0 Cited by

Patent Information

Application Number
JP2023082237
Authority / Receiving Office
JP · JP
Patent Type
Applications
Current Assignee / Owner
Filing Date
2023-05-18
Publication Date
2024-11-28

AI Technical Summary

Technical Problem

Existing methods struggle to accurately determine the quality of treated water in wastewater treatment systems due to difficulties in detecting signs of deterioration, such as sludge rising to the surface caused by filamentous bacteria, using limited plant data.

Method used

A water quality determination method and device that utilize image analysis of activated sludge through a convolutional neural network to assess sedimentation properties and combine this with plant data to predict suspended solids (SS) concentration, employing machine learning models to enhance accuracy.

Benefits of technology

Enables high-accuracy determination of treated water quality by detecting early signs of deterioration, improving the reliability of wastewater treatment systems.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 2024165756000001_ABST
    Figure 2024165756000001_ABST
Patent Text Reader

Abstract

To provide a highly accurate quality determination of a quality of water to be treated by image analysis using an image of the water to be treated.SOLUTION: A method for determining a quality of treated water containing activated sludge includes: an imaging step for capturing images of treated water to obtain a plurality of sludge image data as a first input parameter; a plant data acquisition step for obtaining plant data of a treatment facility for treated water as a second input parameter; a sedimentary property determination step for inputting the plurality of sludge image data to a sedimentary property determination model as the first input parameter and outputting the quality of the treated water as the first output parameter; and a treated water SS prediction step in which the plant data are input to a treated water SS prediction model as the second input parameter and the quality of the treated water is output as the second output parameter.SELECTED DRAWING: Figure 8
Need to check novelty before this filing date? Find Prior Art

Description

[Technical field]

[0001] The present invention relates to a water quality determination method and a water quality determination device. [Background technology]

[0002] Conventionally, biological treatment using microorganisms has been used in many wastewater treatment systems to treat wastewater containing organic matter. In biological treatment, a collection of microorganisms artificially cultivated using the organic matter contained in sewage or wastewater as a substrate is called activated sludge. In activated sludge, various species of organisms, such as bacteria (heterotrophs), filamentous fungi, yeasts, protozoans, and metazoans, interact with each other to remove organic matter and nitrogen (N 2 ), and phosphorus (P), etc. are processed.

[0003] As an operational support tool for this wastewater treatment system, a method has been proposed in which various plant data are used as input values, and empirical formulas and artificial intelligence are used to predict the concentration of suspended solids (SS), which is one of the water quality indicators of treated water (see Patent Documents 1 and 2). [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Patent No. 7122989 [Patent Document 2] Japanese Patent Application Publication No. 6-328091 Summary of the Invention [Problem to be solved by the invention]

[0005] One of the causes of deterioration of the SS concentration in treated water such as wastewater is the dominance of filamentous bacteria, which causes sludge to float. However, it can be difficult to comprehensively detect signs of deterioration in treated water using only limited plant data. Therefore, there has been a demand for a method that can accurately determine the quality of treated water by image analysis using images of the treated water.

[0006] The present invention has been made in consideration of the above, and its object is to provide a water quality determination method and a water quality determination device that can determine with high accuracy whether the water quality of treated water is good or bad by image analysis using images of the treated water. [Means for solving the problem]

[0007] In order to solve the above-mentioned problems and achieve the object, a water quality determination method according to one aspect of the present invention is a water quality determination method for determining the water quality of acquired treated water containing activated sludge, comprising an imaging step of acquiring a plurality of sludge image data as a first input parameter by sampling and imaging the treated water containing activated sludge, a plant data acquisition step of acquiring plant data of a treatment facility that treats the treated water containing activated sludge as a second input parameter, a settleability determination step of inputting the plurality of sludge image data as the first input parameter into a settleability determination model and outputting the quality of the treated water as a first output parameter, and a settleability determination step of inputting the plant data as the second input parameter into a treated water SS prediction model and outputting the quality of the treated water as a second output parameter. and a treated water SS prediction step of outputting the result as data for the settling property judgment model, wherein the settling property judgment model is a learning model generated in advance by machine learning using sludge image data of activated sludge collected in advance as learning input parameters and a judgment result of the settling property of the sludge corresponding to the sludge image data and the quality of the treated water judged based on the settling property of the sludge as learning output parameters. The treated water SS prediction model is a learning model generated in advance by machine learning using plant data acquired in advance as learning input parameters and a SS concentration or SS class of the treated water corresponding to the plant data and the quality of the treated water based on the SS concentration or SS class as learning output parameters.

[0008] In one aspect of the water quality determination method of the present invention, in the above invention, the machine learning neural network used to generate the sedimentation determination model is a convolutional neural network.

[0009] In the water quality determining method according to one aspect of the present invention, in the above invention, the treated water SS predicting step is executed when the settling property of the activated sludge is determined to be good in the settling property determining step.

[0010] In one embodiment of the water quality determination method of the present invention, in the above invention, the plant data includes at least one data selected from nitrate concentration, MLSS concentration, inflow water volume, excess sludge withdrawal volume, treated water temperature, total nitrogen concentration, ammonia concentration, organic matter concentration, and air supply volume.

[0011] In one embodiment of the water quality determination method of the present invention, in the above invention, the plant data includes data on nitrate concentration, MLSS concentration, inflow water volume, excess sludge withdrawal volume, treated water temperature, total nitrogen concentration, ammonia concentration, organic matter concentration, and air supply volume.

[0012] In one embodiment of the water quality determination method of the present invention, in the above invention, the sludge image data used as the learning input parameters in the settling property determination model is used as a learning output parameter, and the water quality determination result for sludge image data that is 80% or less of the standard value for the treated water corresponding to the sludge image data is determined to be good, and the water quality determination result for sludge image data that is 200% or more of the standard value for the treated water is determined to be poor.

[0013] In one embodiment of the water quality assessment method of the present invention, in the above invention, the settleability assessment model defines the degree of defect as the ratio of the number of collected sludge image data data whose class belonging probability is equal to or greater than a predetermined first threshold to the sum of the number of data whose class belonging probability is equal to or greater than a predetermined first threshold and the number of data whose class belonging probability is equal to or less than a predetermined second threshold, and assesses the settleability of the treated water based on the degree of defect.

[0014] A water quality determination device according to one aspect of the present invention is a water quality determination device having a control unit that determines the water quality of treated water containing activated sludge, wherein the control unit is capable of obtaining a plurality of sludge image data as first input parameters by sampling and photographing the treated water containing activated sludge, inputting the plurality of sludge image data as first input parameters into a settling property determination model, and outputting the quality of the treated water as a first output parameter, and is capable of obtaining plant data in a treatment facility that treats the treated water containing activated sludge as a second input parameter, inputting the plant data as the second input parameter into a treated water SS prediction model, and outputting the quality of the treated water as a second output parameter. The settling property judgment model is a learning model generated in advance by machine learning, using sludge image data of activated sludge collected in advance as learning input parameters, and using the settling property of the sludge corresponding to the sludge image data and a judgment result of the quality of the treated water based on the settling property of the sludge as learning output parameters.The treated water SS prediction model is a learning model generated in advance by machine learning, using plant data acquired in advance as learning input parameters, and using the SS concentration or SS class of the treated water corresponding to the plant data and the quality of the water quality of the treated water based on the SS concentration or SS class as learning output parameters.

[0015] In one aspect of the water quality determination device of the present invention, in the above invention, the machine learning neural network used to generate the sedimentation determination model is a convolutional neural network.

[0016] In one embodiment of the water quality determination device of the present invention, in the above invention, the control unit outputs, from the settling property determination model, as the first output parameter, an indication of whether the water quality of the treated water is good or bad, from the treated water SS prediction model.

[0017] In one embodiment of the water quality determination device of the present invention, in the above invention, the plant data includes at least one data selected from nitrate concentration, MLSS concentration, inflow water volume, excess sludge withdrawal volume, treated water temperature, total nitrogen concentration, ammonia concentration, organic matter concentration, and air supply volume.

[0018] In one embodiment of the water quality determination device of the present invention, in the above invention, the plant data includes data on nitrate concentration, MLSS concentration, inflow water volume, excess sludge extraction volume, treated water temperature, total nitrogen concentration, ammonia concentration, organic matter concentration, and air supply volume.

[0019] In one embodiment of the water quality assessment device of the present invention, in the above invention, for the sludge image data used as the learning input parameters in the settling property assessment model, the water quality assessment result for sludge image data that is 80% or less of the standard value for the treated water corresponding to the sludge image data is determined to be good as a learning output parameter, and the water quality assessment result for sludge image data that is 200% or more of the standard value for the treated water is determined to be poor.

[0020] In one embodiment of the water quality assessment device of the present invention, in the above invention, the settleability assessment model defines the degree of defect as the ratio of the number of collected sludge image data data whose class belonging probability is equal to or greater than a predetermined first threshold to the sum of the number of data whose class belonging probability is equal to or greater than a predetermined first threshold and the number of data whose class belonging probability is equal to or less than a predetermined second threshold, and assesses the settleability of the treated water based on the degree of defect. Effect of the Invention

[0021] According to the water quality determining method and water quality determining device of the present invention, it is possible to determine with high accuracy whether the water quality of treated water is good or bad by image analysis using images of the treated water. [Brief description of the drawings]

[0022] [Figure 1] FIG. 1 is a diagram showing an example of a sewage treatment flow in a sewage treatment facility according to an embodiment of the present invention. [Diagram 2]FIG. 2 is a diagram showing an overall configuration of an example of a sewage treatment facility according to an embodiment of the present invention. [Diagram 3] FIG. 3 is a block diagram including a partial flow chart illustrating the concept of the model construction unit and the sludge settling property determination unit in the information processing device according to one embodiment of the present invention. [Figure 4A] FIG. 4A is a diagram showing an example of sludge with poor settling properties in captured image data of a water quality determination process according to an embodiment of the present invention. [Figure 4B] FIG. 4B is a diagram showing an example of sludge with good settling properties in captured image data of the water quality determination process according to one embodiment of the present invention. [Diagram 5] FIG. 5 is a graph showing an example of a class membership probability distribution of an image obtained by statistical processing of the image analysis results in the water quality determination process according to one embodiment of the present invention. [Figure 6] FIG. 6 is a graph showing an example of the measurement results of the degree of defect determined by the water quality determining method according to one embodiment of the present invention and the SS concentration of treated water on a daily basis. [Figure 7] FIG. 7 is a block diagram including a partial flow chart showing the concept of the model construction unit and the treated water SS prediction unit in the information processing device according to one embodiment of the present invention. [Figure 8] FIG. 8 is a flowchart illustrating a water quality determination method according to an embodiment of the present invention. [Figure 9] FIG. 9 is a flowchart for explaining a water quality determination method according to a first modified example of an embodiment of the present invention. [Figure 10] FIG. 10 is a block diagram including a partial flow chart for explaining a water quality determination method according to a second modified example of an embodiment of the present invention. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0023] Hereinafter, an embodiment of the present invention will be described with reference to the drawings. In addition, in all the drawings of the embodiment below, the same or corresponding parts are given the same reference numerals. Furthermore, the present invention is not limited to the embodiment described below. In addition, one embodiment of the present invention can be applied to, for example, a prediction method and a prediction device for predicting treated water quality in a sewage treatment facility, and is particularly suitable for application to a treatment method for predicting treated water quality in a sewage treatment facility with high accuracy by image analysis using activated sludge images and prediction of SS concentration in the treated water using plant data.

[0024] FIG. 1 is a diagram showing an example of a basic sewage treatment flow of a sewage treatment facility according to an embodiment of the present invention. As shown in FIG. 1, the sewage treatment facility 1 according to an embodiment includes a primary sedimentation tank 10, a biological reaction tank 20, and a final sedimentation tank 30. The primary sedimentation tank 10 is a facility for removing solid contaminants such as toilet paper and sand from sewage flowing in from a sewage line. The biological reaction tank 20 is a facility for removing nitrogen, phosphoric acid, organic matter, and the like contained in ammonia by the action of a group of microorganisms. The final sedimentation tank 30 is a facility for performing solid-liquid separation between activated sludge and treated water by, for example, natural sedimentation.

[0025] The main treatment targets and items in the sewage treatment facility 1 according to one embodiment include the following: For organic matter, biological oxygen demand (BOD), chemical oxygen demand (COD), and suspended solids (SS). For nitrogen, total nitrogen (organic nitrogen + inorganic nitrogen) (TN), ammoniacal nitrogen (NH 4 -N), nitrate nitrogen (NO 3 -N), nitrite nitrogen (NO 2 Phosphorus (P) is divided into total phosphorus (organic phosphorus + inorganic phosphorus) TP, phosphate phosphorus (PO 4 -P), etc.

[0026] In addition, in sewage treatment using activated sludge, bacteria such as those of the genus Zoogloea, Bacillus, Pseudomonas, and Flavobacterium are the ones that contribute most to the treatment of pollutants such as organic matter, nitrogen, and phosphorus. These bacteria are observed as flocs in optical observations such as those using microscopes used in sewage treatment facilities. These flocs have complex shapes with various characteristics such as size, shade, and color due to the presence of extracellular polymeric substances and filamentous fungi that the bacteria release outside their bodies. The performance of activated sludge can be quantitatively evaluated by performing image analysis on the flocs and filamentous fungi, which are bacterial aggregates that account for the majority of the performance of activated sludge.

[0027] In the sewage treatment facility using the above-mentioned activated sludge, monitoring of the activated sludge is required for stable operation. Therefore, in the sewage treatment facility, the activated sludge in the biological reaction tank is observed using a microscope or the like. Here, a judgment based on information obtained from image data of the activated sludge obtained by observing the treated water or information obtained from captured image data often depends on the skill of the observer who observes the captured image data. In addition, in optical observation using a microscope, it is difficult to quantitatively judge biological reactions in which various biological species are intricately related. Therefore, in order to judge the quality of the treated water, a method for judging the quality of the treated water with high accuracy by image analysis using images of the treated water has been required.

[0028] Therefore, the present inventor has conducted various studies based on the above-mentioned requirements, and has devised a method using a treated water SS prediction model that predicts the SS concentration of the treated water based on plant data such as various sensor values ​​by machine learning, in addition to a settleability determination model that determines the quality of sludge settleability from an image of activated sludge using a neural network based on image data of the activated sludge. By analyzing the treated water at a sewage treatment facility using output parameters from the settleability determination model and output parameters from the treated water SS prediction model, it becomes possible to predict deterioration of the SS concentration of the treated water for a wide range of deterioration factors and predict signs of deterioration. An embodiment described below has been devised by the present inventor through the above-mentioned intensive studies.

[0029] FIG. 2 is a configuration diagram showing an example of the overall configuration of a sewage treatment facility according to this embodiment. As shown in FIG. 2, the sewage treatment facility 1 includes a primary sedimentation tank 10, a biological reaction tank 20, and a final sedimentation tank 30. Sewage flows into the primary sedimentation tank 10 from a sewage line 8. The biological reaction tank 20 is connected to the primary sedimentation tank 10 via an inlet line 12. A water volume sensor 14 is provided in the inlet line 12. Measurement data of the water volume sensor 14 is output to an information processing device 60. The final sedimentation tank 30 separates the activated sludge and the treated water into solid and liquid. A grit tank (not shown) may be provided on the inlet side of the primary sedimentation tank 10. The grit tank removes large and heavy impurities to protect pumps and the like (not shown).

[0030] The biological reaction tank 20 includes an anaerobic tank 22, an anoxic tank 24, and an aerobic tank 26. In the anaerobic tank 22, phosphorus (P) is released from the cells of microorganisms contained in the sewage, and polyhydroxyalkanoate (PHA) is synthesized within the cells. In the anoxic tank 24, nitric acid (NO 3 The oxygen (O) in the water causes the organic matter to be oxidized and decomposed, producing nitric acid (NO 3 ) is nitrogen (N 2 In the aerobic tank 26, the organic matter is oxidized and decomposed by dissolved oxygen to produce ammonia (NH 3 ) is converted to nitrate to break down PHA, and phosphoric acid (H 3 PO 4Phosphorus (P) is removed by excessive accumulation of .

[0031] At the outflow side of the final settling tank 30, there are provided a treated water line 32, a sludge return line 34, and an excess sludge line 36. The treated water line 32 is a line for disinfecting the treated water before discharging it into a river or the like. The sludge return line 34 is a line for returning a portion of the sludge to the inlet side of the biological reaction tank 20. The excess sludge line 36 is a line for dehydrating and discharging the excess sludge.

[0032] The sludge return line 34 is provided with a return pump 38 and a flow meter 39. A portion of the sludge accumulated in the final settling tank 30 is returned to the inlet side of the biological reaction tank 20 by the sludge return line 34. The amount of returned sludge is adjusted by the return pump 38 to maintain or adjust the MLSS concentration (sludge concentration).

[0033] The excess sludge line 36 is provided with an excess sludge pump 40 and a flow meter 41. Excess sludge is dehydrated and discharged in the excess sludge line 36. The amount of excess sludge is adjusted by the excess sludge pump 40, thereby adjusting the amount of sludge in the system and regulating the metabolism of the sludge, and thereby controlling the sludge retention time. Data on the amount of excess sludge extracted (excess sludge extracted amount) measured by the flow meter 41 is supplied to an information processing device 60.

[0034] The aerobic tank 26 is provided with an air blower 42 and an air volume meter 43. The aerobic tank 26 is configured so that air can be blown into the treatment water from the outside. The amount of air from the air blower 42 is used to maintain and control DO (dissolved oxygen concentration).

[0035] A nitrification liquid circulation line 44 is provided on the outlet side of the aerobic tank 26 of the biological reaction tank 20. The nitrification liquid circulation line 44 is a line for circulating the nitrification liquid to the anoxic tank 24. A nitrification liquid circulation pump 46 is provided on the nitrification liquid circulation line 44. The amount of nitrification liquid circulated by the nitrification liquid circulation pump 46 is used to adjust the amount of nitrate nitrogen supplied to the anoxic tank 24 to remove nitrogen.

[0036] The anaerobic tank 22 is provided with a sensor group 52 including, for example, a dissolved oxygen concentration sensor (DO sensor), a pH sensor, a temperature sensor, and an oxidation-reduction potential sensor (ORP sensor). The anoxic tank 24 is provided with a sensor group 54 including, for example, a DO sensor, a pH sensor, an ORP sensor, a temperature sensor, and a nitrate concentration sensor. The aerobic tank 26 is provided with a sensor group 56 including, for example, a DO sensor, an MLSS concentration (sludge concentration) sensor, a pH sensor, an ORP sensor, and a temperature sensor. The treated water line 32 is provided with a water quality sensor 58. Data on measurements by the respective sensor groups 52, 54, 56 and the water quality sensor 58 are input to an information processing device 60 as plant data.

[0037] Here, the plant data includes at least one data selected from nitric acid concentration, MLSS concentration, inflow water volume, excess sludge extraction volume, temperature of treated water, total nitrogen concentration, ammonia concentration, organic matter concentration, and airflow. The plant data preferably includes all data of nitric acid concentration, MLSS concentration, inflow water volume, excess sludge extraction volume, temperature of treated water, total nitrogen concentration, ammonia concentration, organic matter concentration, and airflow. The organic matter concentration includes chemical oxygen demand (CODmn, CODcr), biochemical oxygen demand (BOD), etc. The plant data can be not only plant data at the position where activated sludge is collected, but also plant data measured at other positions. For example, the airflow is measured in the aerobic tank 26. The information processing device 60 performs image processing and calculation processing based on the acquired plant data, and controls actuators such as pumps and blowers based on the processing results.

[0038] For example, the pH measured by the pH sensor is used to manage the environmental conditions. The information processing device 60 controls the anaerobic tank 22 and the anoxic tank 24 to maintain, for example, a pH of 7 to 8, and the aerobic tank 26 to maintain, for example, a pH of 6 to 7. The oxidation-reduction potential (ORP) is also used to manage the environmental conditions. The ORP is maintained below a predetermined potential, specifically, for example, below -300 mV, preferably below -400 mV, in the anaerobic tank 22, and below a predetermined potential, specifically, for example, below 0 to -200 mV, in the anoxic tank 24. The dissolved oxygen concentration (DO) is also used to manage the environmental conditions. The DO is maintained, for example, below 0.2 mg / L in the anaerobic tank 22 and the anoxic tank 24, and for example, 1.0 to 2.0 mg / L in the aerobic tank 26. The sludge concentration (MLSS) is also used to manage the amount of microorganisms in the activated sludge. The MLSS is maintained at, for example, 2000-3000 mg / L overall in terms of dry weight of the centrifuged sludge. Sludge properties are analyzed, for example, for MLSS, activated sludge settling rate SV30, which is the scale of the sludge interface when settling for 30 minutes in a 1-L graduated cylinder, and sludge settling target SVI (=SV30×10000 / MLSS), which is the volume occupied by 1 g of sludge after 30 minutes of settling.

[0039] (Information processing device) The information processing device 60 is composed of a normal computer or the like. That is, the information processing device 60 includes a control unit, a storage unit, an input / output unit, and a communication unit (none of which are shown). The control unit specifically includes a graphics processing unit (GPU), and includes a processor such as a central processing unit (CPU), a digital signal processor (DSP), and a field-programmable gate array (FPGA), and a main storage unit such as a random access memory (RAM) and a read only memory (ROM) (none of which are shown). The storage unit is composed of a storage medium selected from a volatile memory such as a RAM, a non-volatile memory such as a ROM, an erasable programmable ROM (EPROM), a hard disk drive (HDD, Hard Disk Drive), and a removable medium. The removable medium is, for example, a universal serial bus (USB) memory, or a disk recording medium such as a compact disc (CD), a digital versatile disc (DVD), or a Blu-ray (registered trademark) disc (BD). The storage unit may also be configured using a computer-readable recording medium such as an externally mountable memory card. The storage unit can store an operating system (OS), various programs, various tables, various databases, and the like for executing the operation of the information processing device 60. Here, the various programs include an automatic determination program for realizing processing based on the learning model according to this embodiment. These various programs can also be recorded on computer-readable recording media such as a hard disk, a flash memory, a CD-ROM, a DVD-ROM, and a flexible disk, and widely distributed.

[0040] The control unit loads the program stored in the storage unit into the working area of ​​the main storage unit and executes it, and through execution of the program, functions of model construction units 100, 100A, sludge settling property determination unit 200, and treated water SS prediction unit 300, which will be described later, can be realized. The program also includes learning models, such as a settling property determination model and a treated water SS prediction model, generated by the model construction unit through machine learning such as a convolutional neural network (CNN). The learning models are also referred to as trained models or models. The learning models can be generated by machine learning such as deep learning using a neural network, using an input / output data set of predetermined input parameters and output parameters as teacher data.

[0041] The input / output unit may be configured, for example, from a touch panel display, a speaker microphone, a button, a switch, a jog dial, and the like. The input / output unit as an output unit is configured to notify the outside of predetermined information by displaying characters, figures, and the like on a screen of a display such as a liquid crystal display, an organic EL display, or a plasma display, or by outputting sound from a speaker, according to the control by the control unit. The input / output unit includes a printer that outputs predetermined information by printing it on a printing paper or the like. The various information stored in the memory unit can be confirmed, for example, on a display of the input / output unit installed in a predetermined office, and the like. The input / output unit as an input unit is configured, for example, from a keyboard or a touch panel keyboard incorporated inside the input / output unit to detect touch operations on a display panel, a voice input device that enables calls with the outside, a switch, or a jog dial, and the like. By inputting predetermined information from the input / output unit of the information processing device 60, actuators such as pumps and blowers in the sewage treatment facility 1 can be controlled.

[0042] The communication unit is, for example, a LAN (Local Area Network) interface board, a wireless communication circuit for wireless communication, or a wired communication circuit capable of communication via a dedicated line or cable. The LAN interface board or wireless communication circuit can be connected to a network such as the Internet, which is a public communication network. The communication unit can connect to the network and communicate with the sewage treatment facility 1, transmitting and receiving signals such as various control signals, as well as information such as image data and plant data.

[0043] Next, an activated sludge determination method using image analysis according to this embodiment by the information processing device 60 will be described. Fig. 3 is a block diagram including a partial flow diagram showing the concept of the water quality determination process according to this embodiment. As shown in Fig. 3, the information processing device 60 is configured with a model construction unit 100 and a sludge settling property determination unit 200.

[0044] In the model construction unit 100, activated sludge is collected (step ST11), an image is captured (step ST12), the image is labeled (step ST13), supervised learning is performed (step ST14), and a model is constructed (step ST15). In the supervised learning in step ST14, a convolutional neural network (CNN) having an input image, a repetition of a convolutional layer and a pooling layer, a fully connected layer, and an output layer is used.

[0045] Meanwhile, in the sludge settling property judgment unit 200, new activated sludge is collected (step ST21), images of the collected activated sludge are taken (step ST22), the images are processed using the learning model constructed in step ST15 of the model construction unit 100 (step ST23), the class belonging probability of each image is output (step ST24), statistical processing is performed on each output result (step ST25), and a judgment result for, for example, the settling property of the collected sludge is output (step ST26).

[0046] Here, the method of image labeling by steps ST11 to ST13 of the model construction unit 100 will be described. That is, activated sludge is collected (step ST11), and a sludge image is captured (step ST12). The activated sludge is retained in the treatment facility for a set sludge retention time (SRT). Therefore, using the sludge collection time as a reference, treated water is acquired within the SRT, and the SS concentration X of the acquired treated water is linked to the sludge image. Since activated sludge is constantly replaced, the SS concentration of the treated water to be linked is preferably the SS concentration of the treated water collected near the sludge collection time, and more preferably the SS concentration of the treated water collected within a time period equivalent to the hydraulic retention time (HRT) starting from the sludge collection time.

[0047] Next, in step ST13, the image is labeled based on the treated water quality associated with the image. Here, the number of classes of teacher images in labeling may be two or more classes, and may be set to various values, such as two classes, "good" and "bad," or four classes, A, B, C, and D.

[0048] Here, when labeling an image based on the water quality (SS concentration X) of the treated water, the image can be classified into classes based on the relationship with the standard value T of the SS concentration of the treated water associated with the image. Specifically, for example, if the classification is into four classes A, B, C, and D, the classes can be classified as shown in the following formulas (1-1), (1-2), (1-3), and (1-4). The standard value T of the SS concentration of the treated water is, for example, 100 mg / L. Class A: X≦0.8T ………(1-1) Class B: 0.8T <X≦T …(1-2) Class C:T <X<2T ………(1-3) Class D: 2T≦X ……………(1-4)

[0049] In this embodiment, among the image data of activated sludge, image data in class A state is regarded as image data with "good" settling property, and image data in class D state is regarded as image data with "poor" settling property. Note that image data of classes B and C are not adopted as training data. Image analysis using CNN is used for class A and class D classification.

[0050] That is, in the analysis procedure using CNN in this embodiment, first, an image of class A with good settling property and an image of class D with poor settling property are obtained and used as a teacher image for learning. Here, FIG. 4A and FIG. 4B are diagrams showing examples of sludge with "poor" (class D) and "good" (class A) settling property in the captured image data of the water quality determination process according to this embodiment, respectively. As shown in FIG. 4A, sludge with poor settling property has relatively small flocs 2, is not dense, has a large amount of filamentous bacteria 4, and has a large amount of microparticles 6. On the other hand, as shown in FIG. 4B, sludge with good settling property has relatively large flocs 2, is dense, has a small amount of filamentous bacteria 4, and has a small amount of microparticles 6.

[0051] The information processing device 60 uses the image data of the above-mentioned classes A and D (see Figs. 4A and 4B) as training data and uses the quality of the settling property of the sludge as a training output parameter, and learns the training image data by, for example, CNN. This generates a settling property determination model, which is a trained model. The information processing device 60 inputs sludge image data obtained by capturing an image of the activated sludge discharged from the biological reaction tank 20 using, for example, a microscope, as input parameters to the settling property determination model, and outputs a judgment of the quality of the settling property as an output parameter (step ST26).

[0052] In the statistical processing step of the sludge settling property determination unit 200, image analysis is performed on multiple sludge images captured from the same sludge sample using a trained CNN model, and the class belonging probability of each image is output. Fig. 5 is a graph showing an example of the class belonging probability distribution of images obtained by statistical processing of the image analysis results in the water quality determination process according to this embodiment.

[0053] Normally, activated sludge contains sludge with various microscopic characteristics. Therefore, the probability of a sludge belonging to a class has a range. For example, the following indexes can be used for the macroscopic evaluation of a sludge sample. (I) First evaluation value = median of class membership probability for each image (II) The second evaluation value (degree of defect) is the ratio of the number A whose class belonging probability is a first threshold value (e.g., 70%) or more to the total number (A+B) of the number A whose class belonging probability is a first threshold value y% (e.g., 70%) or more and the number B whose class belonging probability is a second threshold value x% (e.g., 30%) or less among the number of collected sludge image data. Specifically, it is expressed by the following formula (2). Second evaluation value = Defectiveness [x%, y%] = A / (A+B) × 100(%)…(2)

[0054] Specifically, the example shown in Fig. 5 is a class membership probability distribution when the number of sludge image data is 500. In Fig. 5, the degree of poor settling is output from the settling determination model for each of the 500 pieces of sludge image data, and the result of classifying the degree of poor settling of each piece of image data in increments of 10%, for example, is shown in a histogram. Note that the number of pieces of sludge image data is just one example, and various numbers can be selected as long as they are the required number.

[0055] In Fig. 5, the number B of sludge image data whose settling property can be judged as "good" when the class belonging probability is 30% or less is 91 (left side of the thick vertical line in Fig. 5). The number A of sludge image data whose class belonging probability is 70% or more is judged as "poor" when the settling property can be judged as "poor" is 315 (right side of the thick vertical line in Fig. 5). When the class belonging probability (score) is higher than 30% and less than 70%, it is difficult to judge the settling property as either "good" or "poor", and the number of sludge image data that cannot be judged and are not adopted is 194 (500-AB=). In this case, the degree of failure [30%, 70%] as the second evaluation value can be derived from formula (2) as 315 / (91+315)×100=77.6%.

[0056] The reason why statistical processing is required for the class membership probability here is because the performance of activated sludge is determined by the activity of the microbial population. From a microscopic perspective, such as an image observed under a microscope, sludge with good settling properties (settling property "good") and sludge with poor settling properties (settling property "poor") are mixed. The properties of the activated sludge as a whole depend on which sludge occupies it. Therefore, by calculating the class membership probability for each piece of sludge image data captured on the same sample, it is possible to judge the quality of the sludge from a microscopic perspective. By performing statistical processing on this, it is possible to judge the quality of the settling property of the entire sludge.

[0057] Fig. 6 is a graph showing an example of the daily measurement results regarding the degree of defectiveness (0.7) of activated sludge and the actual measured value of the SS concentration of treated water in the water quality determination method according to this embodiment. Note that in Fig. 6, "degree of defectiveness (0.7)" means the degree of defectiveness [30%, 70%] as the second evaluation value in the above-mentioned formula (2).

[0058] From FIG. 6, for example, in the portion from August to September of year N (box S1 in FIG. 6), it can be seen that the degree of defect (0.7) increased significantly, followed by a large increase in the SS concentration of the treated water. Specifically, from FIG. 6, it can be seen that the degree of defect increased from about 5% to more than 30% in August to September. It can then be seen that the SS concentration of the treated water increased to about 120 mg / L, i.e., to a degree exceeding the standard value T (100 mg / L). Similarly, for example, in the portion from October of year N (box S2 in FIG. 6), it can be seen that after the degree of defect increased from about 30% to about 50%, the SS concentration of the treated water temporarily decreased, but then increased to about 360 mg / L, which significantly exceeded the standard value T. In other words, it can be seen that the deterioration of the water quality of the treated water, which cannot be predicted only by measuring the SS concentration of the treated water, can be judged by deriving the degree of defect based on statistical classification using the sedimentation judgment model. As described above, it is clear that by performing an evaluation using the degree of defect, it is possible to predict the deterioration of the SS concentration of treated water early and sensitively.

[0059] (Method for predicting treated water SS concentration) Next, a description will be given of a method for predicting the SS concentration value or SS concentration class of the treated water from plant data using the information processing device 60. Fig. 7 is a block diagram including a partial flow chart showing the concept of the model construction unit and the treated water SS prediction unit in the information processing device according to this embodiment.

[0060] As shown in Fig. 7, the information processing device 60 has a model construction unit 100A similar to the model construction unit 100 shown in Fig. 3, and a treated water SS prediction unit 300. The model construction unit 100A acquires plant data from various sensors (step ST111), acquires the SS concentration of the treated water (step ST112), associates the SS concentration of the treated water with the plant data to label classes (step ST113), performs supervised learning (step ST114), and constructs a model (step ST115). Meanwhile, the treated water SS prediction unit 300 acquires new plant data from various sensors (step ST121), processes the data using the learning model constructed by the model construction unit 100 in step ST115 (step ST122), and outputs the SS concentration value (SS value) or SS concentration class (SS class) of the treated water (step ST123).

[0061] (Water quality determination method) Next, a description will be given of a water quality determining method performed by the sludge settling property determining unit 200 and the treated water SS predicting unit 300 in the above-mentioned information processing device 60. Fig. 8 is a flowchart for explaining the water quality determining method according to this embodiment.

[0062] As shown in Fig. 8, the sludge settleability determination unit 200 in the information processing device 60 acquires a plurality of data, for example, 500 or more pieces of sludge image data, by capturing images of the collected activated sludge as an imaging step (step ST31). The sludge settleability determination unit 200 inputs the acquired sludge image data to a settleability determination model as a first input parameter (step ST32). The settleability determination model derives the above-mentioned degree of defect [x%, y%] to determine whether the water quality of the treated water containing activated sludge is "good" or "poor", and outputs the determination result as a first output parameter (step ST33). This executes the settleability determination step.

[0063] If the sludge settling property determination unit 200 determines that the water quality of the treated water is "poor" (step ST33: No), the control unit of the information processing device 60 makes a final determination that the water quality of the treated water is poor (step ST39). With the above, the water quality determination process according to this embodiment is completed.

[0064] On the other hand, if the sludge settling property determining unit 200 determines that the water quality of the treated water is "good" (step ST33: Yes), the treated water SS predicting unit 300 acquires plant data from sensors installed at various locations in the sewage treatment facility 1 as a plant data acquisition step ST (step ST34). The treated water SS predicting unit 300 can acquire plant data in parallel with other processes, or at any time, and the timing of acquiring the plant data is not limited. The plant data includes at least one data selected from the nitrate concentration, MLSS concentration, inflow water volume, excess sludge extraction volume, treated water temperature, total nitrogen concentration, ammonia concentration, organic matter concentration, and air flow rate, and preferably includes all of these data. Furthermore, the plant data may include data other than the exemplified data.

[0065] The treated water SS prediction unit 300 inputs the acquired plant data as a second input parameter to the treated water SS prediction model (step ST35). The treated water SS prediction model derives the SS concentration or SS concentration class (hereinafter, SS class) of the treated water (step ST36). The treated water SS prediction unit 300 judges the water quality of the treated water as "good" or "poor" based on the SS concentration or SS class of the treated water, and outputs the judgment result as a second output parameter (step ST37). In this manner, the treated water SS prediction step is executed.

[0066] If the treated water SS prediction unit 300 judges that the quality of the treated water is "poor" (step ST37: No), the control unit of the information processing device 60 makes a final judgment that the quality of the treated water is poor (step ST39). On the other hand, if the treated water SS prediction unit 300 judges that the quality of the treated water is "good" (step ST37: Yes), the control unit of the information processing device 60 makes a final judgment that the quality of the treated water is good (step ST38). This completes the water quality judgment process according to this embodiment.

[0067] (First modified example of the water quality determination method) Next, a description will be given of a first modified example of the water quality determining method by the sludge settling property determining unit 200 and the treated water SS predicting unit 300 in the above-mentioned information processing device 60. Fig. 9 is a flowchart for explaining the water quality determining method according to the modified example of this embodiment.

[0068] 9, the sludge settleability determination unit 200 in the information processing device 60 captures an image of the collected activated sludge to obtain a plurality of sludge image data (step ST41). The sludge settleability determination unit 200 inputs the obtained sludge image data as input parameters to a settleability determination model (step ST42). The settleability determination model derives the above-mentioned degree of defect [x%, y%] to determine whether the water quality of the treated water containing activated sludge is "good" or "poor", and outputs the determination result as an output parameter (step ST43).

[0069] Meanwhile, the treated water SS prediction unit 300 acquires plant data from sensors installed at various locations in the sewage treatment facility 1 in parallel (step ST44). The timing of acquiring the plant data is not limited. The plant data includes at least one data selected from the following: nitric acid concentration, MLSS concentration, inflow water volume, excess sludge extraction volume, treated water temperature, total nitrogen concentration, ammonia concentration, organic matter concentration, and air supply volume, and preferably includes all of the data. Furthermore, the plant data may include other data. The treated water SS prediction unit 300 inputs the acquired plant data as input parameters to the treated water SS prediction model (step ST45). The treated water SS prediction model derives the SS concentration or SS class of the treated water, judges the water quality of the treated water as "good" or "poor", and outputs the judgment result as an output parameter (step ST46).

[0070] When the sludge settling property determining unit 200 determines that the water quality of the treated water is "good" and the treated water SS predicting unit 300 determines that the water quality of the treated water is "good" (step ST47), the control unit of the information processing device 60 makes a final determination that the water quality of the treated water is good and outputs the result (step ST48). On the other hand, when at least one of the determinations of the sludge settling property determining unit 200 determining that the water quality of the treated water is "poor" and the determination of the treated water SS predicting unit 300 determining that the water quality of the treated water is "poor" (step ST47: No), the control unit of the information processing device 60 makes a final determination that the water quality of the treated water is poor (step ST49). With the above, the water quality determination process according to the modified example of this embodiment is completed.

[0071] (Second modified example of the water quality determination method) Next, a second modified example of the water quality determining method by the sludge settling property determining unit 200 and the treated water SS predicting unit 300 in the information processing device 60 will be described. Fig. 10 is a block diagram including a partial flow chart for explaining the water quality determining method according to the modified example of this embodiment.

[0072] As shown in Fig. 10, at least one of the output value (%) of the degree of defect output from the settling property determination model of the sludge settling property determination unit 200 in the information processing device 60 and the determination result of the water quality of the treated water determined by the sludge settling property determination unit 200 (good or bad) is input as one of the input parameters to the treated water SS prediction model of the treated water SS prediction unit 300 together with the plant data. The treated water SS prediction unit 300 inputs the acquired plant data and at least one of the data of the degree of defect and the determination result input from the sludge settling property determination unit 200 as input parameters to the treated water SS prediction model. The treated water SS prediction model derives the SS value or SS class of the treated water and outputs it as an output parameter. The other configurations are the same as those of the above-mentioned embodiment. With the above, the water quality determination process according to the second modified example of this embodiment is completed.

[0073] According to the embodiment described above, sludge image data obtained by capturing an image of the microscopic state of activated sludge using a microscope or the like is input to a settling property determination model of a neural network to output the quality of sludge settling property, while the acquired plant data is input to a treated water SS prediction model generated by machine learning to output the SS concentration or SS class of the treated water to determine the quality of the treated water, thereby making it possible to detect signs of deterioration of SS in the treated water. Furthermore, compared to a case where the quality of the treated water is determined using only the settling property determination model or a case where the quality of the treated water is determined using only the treated water SS prediction model, the state of the water quality of the treated water can be determined more accurately, and furthermore, the presence or absence of deterioration of the water quality of the treated water, particularly the SS concentration, can be accurately predicted.

[0074] (Recording medium) In the above-mentioned embodiment, the program to be executed by the information processing device can be recorded on a recording medium readable by a computer or other machine or device (hereinafter referred to as a computer, etc.). By making a computer, etc. read and execute the program from the recording medium, the computer, etc. functions as a mobile object control device. Here, a recording medium readable by a computer, etc. refers to a non-transient recording medium that accumulates information such as data and programs by electrical, magnetic, optical, mechanical, or chemical action and can be read from a computer, etc. Among such recording media, those that can be removed from a computer, etc. include, for example, a flexible disk, a magneto-optical disk, a CD-ROM, a CD-R / W, a DVD, a BD, a DAT, a magnetic tape, and a memory card such as a flash memory. In addition, a hard disk, a ROM, etc. are examples of recording media fixed to a computer, etc. Furthermore, an SSD can be used as a recording medium that can be removed from a computer, etc., and as a recording medium fixed to a computer, etc.

[0075] Furthermore, the program executed by the information processing device according to the embodiment may be configured to be stored on a computer connected to a network such as the Internet and provided by being downloaded via the network.

[0076] Although one embodiment of the present disclosure has been specifically described above, the present disclosure is not limited to the above-described embodiment, and various modifications based on the technical ideas of the present disclosure are possible. For example, the virtual images and operations given in the above-described embodiment are merely examples, and different virtual images and operations may be used.

[0077] (Other embodiments) In the embodiment described above, the "unit" can be read as a "circuit" etc. For example, a control unit can be read as a control circuit.

[0078] In the description of the flowcharts in this specification, the order of processing between steps is clearly indicated using expressions such as "first," "then," and "continue," but the order of processing required to implement the above-described embodiment is not uniquely determined by these expressions. In other words, the order of processing in the flowcharts described in this specification can be changed as long as there is no contradiction.

[0079] Although one embodiment of the present invention has been specifically described above, the present invention is not limited to the above embodiment, and various modifications based on the technical idea of ​​the present invention are possible. The present invention also includes a configuration in which the above-mentioned components are appropriately combined. Further effects and modifications can be easily derived by those skilled in the art. The broader aspects of the present invention are not limited to the specific details and representative embodiments shown and described above. Thus, various modifications are possible without departing from the spirit or scope of the general inventive concept defined by the appended claims and their equivalents. For example, the numerical values ​​and types of information given in the above embodiment are merely examples, and different numerical values ​​and types of information may be used as necessary, and the present invention is not limited by the description and drawings that form a part of the disclosure of the present invention according to the above embodiment.

[0080] In the embodiment described above, the treated water SS was the subject of prediction, but other water qualities of treated water managed in a sewage treatment facility can also be predicted in a similar manner. Examples of other water qualities of treated water include chemical oxygen demand (COD), ammonia nitrogen, nitrate nitrogen, total nitrogen, and total phosphorus.

[0081] In the above-described embodiment, the present invention is applied to a sewage treatment facility 1 in which the biological reaction tank 20 for removing organic matter, nitrogen, and phosphorus (P) by an anaerobic-anoxic-aerobic process is provided with an anaerobic tank 22, an anoxic tank 24, and an aerobic tank 26. However, the application of the present invention is not limited to this, and it is clear that the present invention can also be applied to, for example, a sewage treatment facility in which the biological reaction tank does not include an anaerobic tank for removing organic matter and nitrogen by an anoxic-aerobic process (circulating nitrification denitrification process), a sewage treatment facility in which the biological reaction tank does not include an anoxic tank but includes an anaerobic tank and an aerobic tank for removing organic matter and phosphorus by an anaerobic-aerobic process, and a sewage treatment facility in which the biological reaction tank consists of only an aerobic tank for removing organic matter by a standard activated sludge process. [Explanation of symbols]

[0082] 1. Sewage treatment facilities 2. Flock 4. Filamentous fungi 6 Microparticles 8. Sewer Line 10 Primary Sedimentation Tank 12 Inlet line 14 Water sensor 20 Biological Reactor 22 Anaerobic tank 24 Anoxic tank 26 Aerobic tank 30 Final Sedimentation Tank 32 Treated Water Line 34 Sludge return line 36 Excess sludge line 38 Return Pump 39,41 Flow meter 40 Excess sludge pump 42 Air Blower 43 Air volume gauge 44 Nitrification liquid circulation line 46 Nitrification liquid circulation pump 52, 54, 56 Sensor group 58 Water quality sensor 60 Information processing equipment 100,100A Model Construction Section 200 Sludge Settling Determination Section 300 Treated Water SS Prediction Department

Claims

1. A water quality determination method for determining the water quality of treated water containing activated sludge obtained, comprising: an imaging step of acquiring a plurality of sludge image data as a first input parameter by sampling and imaging treated water containing activated sludge; a plant data acquisition step of acquiring plant data of a treatment facility that treats treated water containing activated sludge as a second input parameter; a settleability determination step of inputting the plurality of sludge image data as a first input parameter into a settleability determination model and outputting the quality of the treated water as a first output parameter; and a treated water SS prediction step of inputting the plant data as a second input parameter into a treated water SS prediction model and outputting the quality of the treated water as a second output parameter, the settling property determination model is a learning model that is generated in advance by machine learning using sludge image data of activated sludge collected in advance as learning input parameters, and a determination result of the settling property of the sludge corresponding to the sludge image data and the quality of the treated water determined in advance based on the settling property of the sludge as learning output parameters, The treated water SS prediction model is a learning model that is generated in advance by machine learning, with previously acquired plant data as learning input parameters, and with the SS concentration or SS class of the treated water corresponding to the plant data and the quality of the treated water based on the SS concentration or SS class as learning output parameters. Water quality determination method.

2. The machine learning neural network used to generate the sedimentation determination model is a convolutional neural network. The water quality determination method according to claim 1.

3. When the settleability of the activated sludge is determined to be good in the settleability determination step, the treated water SS prediction step is executed. The water quality determination method according to claim 1.

4. The plant data includes at least one data selected from a nitric acid concentration, an MLSS concentration, an inflow water volume, an excess sludge extraction volume, a temperature of the treated water, a total nitrogen concentration, an ammonia concentration, an organic matter concentration, and an air blowing volume. The water quality determination method according to claim 1.

5. The plant data includes data on nitric acid concentration, MLSS concentration, inflow water volume, excess sludge extraction volume, treated water temperature, total nitrogen concentration, ammonia concentration, organic matter concentration, and air flow rate. The water quality determination method according to claim 1.

6. For the sludge image data used as the learning input parameters in the settling property determination model, as learning output parameters, a water quality determination result in sludge image data that is 80% or less of a standard value for treated water corresponding to the sludge image data is determined as good, and a water quality determination result in sludge image data that is 200% or more of the standard value for treated water is determined as bad. The water quality determination method according to claim 1.

7. The settleability determination model determines the degree of defect as the ratio of the number of samples of collected sludge image data whose class belonging probability is equal to or greater than a first threshold to the total number of samples whose class belonging probability is equal to or greater than a second threshold, and determines the settleability of treated water based on the degree of defect. The water quality determination method according to claim 1.

8. A water quality determination device having a control unit for determining the water quality of treated water containing the obtained activated sludge, The control unit is a plurality of sludge image data are acquired as first input parameters by collecting and imaging treated water containing activated sludge, the plurality of sludge image data are input as first input parameters to a settling property determination model, and the quality of the treated water is output as a first output parameter; Plant data of a treatment facility that treats treated water containing activated sludge is acquired as a second input parameter, the plant data is input as the second input parameter to the treated water SS prediction model, and the quality of the treated water is output as a second output parameter, the settling property determination model is a learning model that is generated in advance by machine learning using sludge image data of activated sludge collected in advance as learning input parameters, and a determination result of the settling property of the sludge corresponding to the sludge image data and the quality of the treated water determined in advance based on the settling property of the sludge as learning output parameters, The treated water SS prediction model is a learning model that is generated in advance by machine learning, with previously acquired plant data as learning input parameters, and with the SS concentration or SS class of the treated water corresponding to the plant data and the quality of the treated water based on the SS concentration or SS class as learning output parameters. Water quality determination device.

9. The machine learning neural network used to generate the sedimentation determination model is a convolutional neural network. The water quality determining device according to claim 8.

10. The control unit outputs, when the settleability of the activated sludge is output as the first output parameter from the settleability determination model, a quality of the water of the treated water as the second output parameter from the treated water SS prediction model. The water quality determining device according to claim 8.

11. The plant data includes at least one data selected from a nitric acid concentration, an MLSS concentration, an inflow water volume, an excess sludge extraction volume, a temperature of the treated water, a total nitrogen concentration, an ammonia concentration, an organic matter concentration, and an air blowing volume. The water quality determining device according to claim 8.

12. The plant data includes data on nitric acid concentration, MLSS concentration, inflow water volume, excess sludge extraction volume, treated water temperature, total nitrogen concentration, ammonia concentration, organic matter concentration, and air flow rate. The water quality determining device according to claim 8.

13. For the sludge image data used as the learning input parameters in the settling property determination model, as learning output parameters, a water quality determination result in sludge image data that is 80% or less of a standard value for treated water corresponding to the sludge image data is determined as good, and a water quality determination result in sludge image data that is 200% or more of the standard value for treated water is determined as bad. The water quality determining device according to claim 8.

14. The settleability determination model determines the degree of defect as the ratio of the number of samples of collected sludge image data whose class belonging probability is equal to or greater than a first threshold to the total number of samples whose class belonging probability is equal to or greater than a second threshold, and determines the settleability of treated water based on the degree of defect. The water quality determining device according to claim 8.

Citation Information

Patent Citations

  • Neural net applied operation supporting system

    JP1992133164A

  • Water quality information operational processor

    JP2005034717A

  • Prediction method and apparatus for water quality of treated water in sewage treatment plant

    JP2022046280A

  • Method and apparatus for predicting parameter value of wastewater treatment simulator and control method and apparatus for sewage treatment plant

    JP2022046281A

  • Sewage treatment plant operating condition evaluation method and device, and operating condition control method and device

    JP2022046430A