Water treatment condition monitoring system and water treatment condition monitoring method

Abstract

The present invention provides a water treatment condition monitoring system and method, which can calculate information that even a non-expert can easily identify whether the coagulation state of the turbidity substances in the treated water in the image is good. The water treatment condition monitoring system includes first and second imaging devices, and first and second measurement devices. The first and second imaging devices are arranged at a plurality of different places capable of photographing the treated water, and acquire image data including an image of the treated water. The first measurement device measures and acquires treated water data including the turbidity of the treated water. The second measurement device measures and acquires treated water data including the turbidity of the treated water. The information processing device constructs first and second coagulation goodness judgment models using the captured images of the first and second imaging devices respectively, and uses any one of the first and second coagulation goodness judgment models to calculate the predicted turbidity of the treated water according to the input data including the image of the treated water.

Description

Technical Field

[0001] The present invention relates to a water treatment condition monitoring system and a water treatment condition monitoring method. Background Art

[0002] In the drainage treatment systems of waterworks and other factories, various water treatment technologies are used according to the quality of the water to be treated (also referred to as "raw water") to achieve the target water quality of the treated water. In water treatment technologies, coagulation treatment for removing solids in water (turbidity (also referred to as "turbid substances")) requires appropriate chemical injection management. However, when the quality of the water to be treated changes, etc., the chemical injection amount sometimes depends on the experience of skilled operators (skilled persons).

[0003] Based on experience, skilled persons can determine an appropriate chemical injection amount according to the results of beaker tests for confirming the coagulation state of turbid substances, monitoring control data, visually observing the coagulation state of turbid substances at the work site, and confirming images of flocs formed due to coagulation of turbid substances in the monitoring room. On the other hand, the coagulation state of turbid substances visually observed by skilled persons, etc. is not recorded in the monitoring control system (monitoring control data). Therefore, since the experience of skilled persons has become a kind of tacit knowledge, there is a technical problem that knowledge required for chemical injection management is not easily inherited by others other than skilled persons.

[0004] In the future, it is desired that even unskilled persons can determine the coagulation state of turbid substances and have the skill to judge in advance that the coagulation of turbid substances is poor. Thus, a technology for obtaining image data of flocs through sensing technology and using machine learning and artificial intelligence to judge whether the coagulation of turbid substances is good is being promoted for development.

[0005] Patent Document 1 discloses the following information processing device (hereinafter sometimes also referred to as "existing device"). In the existing device, a known amount of coagulant is added to the raw water to be subjected to upper water treatment and stirred, and teacher data is obtained by associating each image data within 400 seconds from this time with a category indicating whether the flocs to be formed in the future will have an adverse effect on the upper water treatment, and deep learning is performed on the teacher data thus obtained.

[0006] After that, when the existing device inputs image data (discrimination target image data), it judges whether the flocs to be formed in the future in the raw water corresponding to the discrimination target image data will have an adverse effect on the upper water treatment according to the learning result of deep learning and the discrimination target image data. Among them, the discrimination target image data is image data within 300 seconds from when a known amount of coagulant is added to the raw water to be subjected to upper water treatment and stirred, and is image data not associated with a category indicating whether the flocs to be formed in the future will have an adverse effect on the upper water treatment.

[0007] Prior art documents

[0008] Patent documents

[0009] Patent Document 1: Japanese Unexamined Patent Application Publication No. 2020-134284 Summary of the invention

[0010] Technical problem to be solved by the invention

[0011] In a facility for treating and purifying treated water to obtain treated water with a target water quality, a technique is needed that enables even an unskilled person to easily determine whether the aggregation state of the turbidity substances in the treated water presented in an image is good. The present invention has been completed to solve the above technical problem. That is, one of the objects of the present invention is to provide a water treatment condition monitoring system and a water treatment condition monitoring method that can calculate information that enables even an unskilled person to easily determine whether the aggregation state of the turbidity substances in the treated water presented in an image is good.

[0012] Technical means for solving the problem

[0013] In order to solve the above technical problems, the water treatment condition monitoring system of the present invention is applied to a facility for generating treated water. The facility generates the treated water obtained by purifying the water to be treated by sequentially transporting the water to be treated to a plurality of treatment sites and performing treatments including coagulation sedimentation treatment for removing turbidity substances contained in the water to be treated. The water treatment condition monitoring system of the present invention includes: a plurality of imaging devices arranged at a plurality of different sites capable of photographing the water to be treated, and obtaining image data of the water to be treated at each site, including images of the water to be treated photographed from each site, by photographing the water to be treated; a water to be treated measuring device for measuring water quality parameters of the water to be treated, including the turbidity of the water to be treated, and obtaining water to be treated data including the water quality parameters of the water to be treated; a treated water measuring device for measuring the turbidity of the treated water and obtaining treated water data including the turbidity of the treated water; and an information processing device for obtaining the image data, the water to be treated data, and the treated water data from the imaging device, the water to be treated measuring device, and the treated water measuring device. The information processing device is configured to use learning input data including the image data of the water to be treated and the treated water data as learning data, and construct a plurality of coagulation quality judgment models through machine learning, wherein the coagulation quality judgment models are used to output information indicating whether the coagulation state of the turbidity substances in the water to be treated is good according to input data including the image of the water to be treated. The information processing device is configured to select at least one of the plurality of coagulation quality judgment models, obtain the image of the water to be treated as judgment image from at least one of the plurality of imaging devices, and use the selected coagulation quality judgment model to calculate information indicating whether the coagulation state of the turbidity substances in the water to be treated presented in the judgment image is good according to judgment input data including the judgment image.

[0014] The water treatment condition monitoring system of the present invention is applied to a facility for generating treated water. The facility generates the treated water obtained by purifying the water to be treated by transporting the water to be treated to a plurality of treatment sites in sequence and performing treatments including coagulation sedimentation treatment for removing turbidity substances contained in the water to be treated. The water treatment condition monitoring system of the present invention includes: a plurality of water sampling devices for sampling the water to be treated at a plurality of different sites; a plurality of imaging devices for obtaining, by respectively photographing the water to be treated at a plurality of sites sampled by the plurality of water sampling devices, water sampling image data of the water to be treated at each site, which includes images of the water to be treated at each site, i.e., water sampling images; a water to be treated measuring device for measuring water quality parameters of the water to be treated, including the turbidity of the water to be treated, and obtaining water to be treated data including the water quality parameters of the water to be treated; a treated water measuring device for measuring the turbidity of the treated water and obtaining treated water data including the turbidity of the treated water; and an information processing device for obtaining the water sampling image data, the water to be treated data, and the treated water data from the imaging device, the water to be treated measuring device, and the treated water measuring device. The information processing device is configured to use the learning input data including the water sampling image data of the water to be treated and the treated water data as learning data, and through machine learning, construct coagulation quality judgment models for the water sampling image data of the water to be treated at each site respectively. Among them, the coagulation quality judgment model is used to output information indicating whether the coagulation state of the turbidity substances in the water to be treated is good according to the input data including the water sampling image of the water to be treated. The information processing device is configured to select at least one of the plurality of coagulation quality judgment models constructed for the water sampling image data of the water to be treated corresponding to each site, obtain the water sampling image of the water to be treated as the judgment image from the imaging device, and use the selected coagulation quality judgment model to calculate information indicating whether the coagulation state of the turbidity substances in the water to be treated presented in the judgment image is good according to the judgment input data including the judgment image.

[0015] The method for monitoring the water treatment status of the present invention is applied to a facility for generating treated water. The facility generates the treated water obtained by purifying the water to be treated by sequentially transporting the water to be treated to a plurality of treatment sites and performing treatments including coagulation sedimentation treatment for removing turbidity substances contained in the water to be treated. The method includes: using a plurality of imaging devices provided at a plurality of different sites capable of photographing the water to be treated, obtaining image data of the water to be treated at each site, including images of the water to be treated taken from each site, by photographing the water to be treated; using a water-to-be-treated measuring device to measure water-to-be-treated water quality parameters regarding the water to be treated, including the turbidity of the water to be treated, and obtaining water-to-be-treated data including the water-to-be-treated water quality parameters; using a treated-water measuring device to measure the turbidity of the treated water and obtaining treated-water data including the turbidity of the treated water; and using an information processing device to obtain the image data, the water-to-be-treated data, and the treated-water data from the imaging device, the water-to-be-treated measuring device, and the treated-water measuring device. Among them, using the information processing device, learning input data including the image data of the water to be treated and the treated-water data are used as learning data, and a plurality of coagulation quality judgment models are constructed through machine learning. The coagulation quality judgment model is used to output information indicating whether the coagulation state of the turbidity substances in the water to be treated is good according to input data including the image of the water to be treated. At least one of the plurality of coagulation quality judgment models is selected, the image of the water to be treated is obtained as judgment-use image from at least one of the plurality of imaging devices, and using the selected coagulation quality judgment model, information indicating whether the coagulation state of the turbidity substances in the water to be treated presented in the judgment-use image is good is calculated according to judgment-use input data including the judgment-use image.

[0016] Advantages of the Invention

[0017] According to the present invention, it is possible to calculate information indicating whether the coagulation state of the turbidity substances in the water to be treated presented in the image is good, which can be easily distinguished even by non-experts. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a schematic structural diagram showing an example of a water treatment status monitoring system to which the first embodiment of the present invention is applied in a water purification plant.

[0019] Figure 2 It is a block diagram showing an example of the hardware structure of an information processing device included in the water treatment status monitoring system.

[0020] Figure 3 It is a table for explaining the test conditions and the treatment results of the water to be treated in a test example.

[0021] Figure 4 It is a coordinate graph for explaining the change over time of the cumulative value of the floc area.

[0022] Figure 5 It is a flowchart showing the processing flow executed by the model generation unit.

[0023] Figure 6 It is a flowchart showing the processing flow executed by the model generation unit.

[0024] Figure 7 It is a flowchart showing the processing flow executed by the flocculation quality judgment unit.

[0025] Figure 8 It is a flowchart showing the processing flow executed by the flocculation quality judgment unit.

[0026] Figure 9 It is a flowchart showing the processing flow executed by the model generation unit of the first modification example.

[0027] Figure 10 It is a flowchart showing the processing flow executed by the model generation unit of the first modification example.

[0028] Figure 11 It is a coordinate graph for explaining an example of the evaluation result of the flocculation quality judgment model.

[0029] Figure 12 It is a flowchart showing the processing flow executed by the flocculation quality judgment unit of the first modification example.

[0030] Figure 13 It is a schematic structural diagram of an example showing the water treatment status monitoring system to which the second embodiment of the present invention is applied in a water purification plant.

[0031] Figure 14 It is a flowchart showing the processing flow executed by the model generation unit of the water treatment status monitoring system according to the third embodiment of the present invention.

[0032] Figure 15 It is a flowchart showing the processing flow executed by the model generation unit of the water treatment status monitoring system according to the third embodiment.

[0033] Figure 16 It is a flowchart showing the processing flow executed by the model generation unit of the water treatment status monitoring system according to the fourth embodiment of the present invention.

[0034] Figure 17 It is a flowchart showing the processing flow executed by the flocculation quality judgment unit of the water treatment status monitoring system according to the fourth embodiment. Detailed implementation manners

[0035] <<Details of the Background Art>>

[0036] First, to facilitate understanding of the present invention, details of the background art will be described. In the maintenance management operations of factories such as waterworks (both upstream and downstream), chemical factories, power plants, and waste treatment plants, there are cases where management depends on the visual observation and experience of operators. When factories operate continuously for 24 hours, currently, in most cases, this is addressed by having operators always present inside the factory. However, the working-age population in Japan (aged 15 to 64) is on a decreasing trend. Therefore, it is necessary to consider sustainable solutions for future labor shortages starting from now.

[0037] In recent years, with the improvement of computer computing power and communication infrastructure environments, it has become possible to easily process large amounts of data. As a result, sensing technologies (IoT technologies) that replace the five senses of operators have been continuously evolving and spreading. In addition, with the evolution of analysis algorithms for machine learning and artificial intelligence (AI technologies), research on using AI technologies is actively underway even for operations that have been difficult to mechanize and automate because they require human judgment. For example, to continue the maintenance management operations of factories even in the event of a labor shortage, solutions have been provided that involve installing cameras in place of operators' vision and processing the captured images and videos to diagnose changes in the appearance of equipment and products.

[0038] As described in the background art, in the drainage treatment systems of waterworks (both upstream and downstream) and other factories, various water treatment technologies are used according to the quality of the water to be treated to achieve the target water quality of the treated water. In coagulation treatment among water treatment technologies, appropriate chemical injection management is required. However, when the quality of the water to be treated changes, etc., the chemical injection amount sometimes depends on the experience of skilled operators. Based on experience, skilled operators can determine the appropriate chemical injection amount according to the results of beaker tests for confirming the coagulation state of turbidity substances, the monitoring control data of the monitoring control system for monitoring the factory, visually observing the coagulation state of turbidity substances at the work site, and confirming the floc images in the monitoring room, etc.

[0039] On the other hand, the coagulation state of turbidity substances visually observed by skilled operators, etc., is not recorded in the monitoring control system (monitoring control data). Therefore, since the experience of skilled operators has become a kind of tacit knowledge, there is a technical problem that the knowledge required for chemical injection management, etc., is not easily inherited by others other than skilled operators.

[0040] In the future, it is hoped that even unskilled personnel will be able to determine the coagulation state of the turbidity-causing substances, and the monitoring and control system (monitoring system) has the function of pre-judging poor coagulation. Therefore, a technology is being developed to obtain image data of the flocs through sensing technology and use machine learning and artificial intelligence to judge whether the coagulation state of the turbidity-causing substances is good (hereinafter also referred to as "whether the coagulation of the turbidity-causing substances is good" or simply "coagulation quality").

[0041] For example, the technology of Patent Document 1 above sets cameras in the rapid mixing tank and the floc formation tank, obtains image data representing the process of coagulation nucleus formation (temporal change) (image data captured during a specified period) in the rapid mixing tank, and obtains image data representing the state in the initial stage of floc formation (image data captured during a specified period) in the floc formation tank. These obtained image data are used as teacher data for learning through deep learning, and based on the learning results, it is judged whether the flocs to be formed in the future are good flocs according to the image data.

[0042] However, it is feasible to photograph the formation process (temporal change) of the coagulation nucleus and flocs (present the formation process of the coagulation nucleus and flocs in the image data during a specified period (image data of 400 seconds or 300 seconds)) in a small experimental device or a batch-type device, but it is difficult to set up cameras at appropriate locations in an actual waterworks. Therefore, depending on the specifications of the waterworks, there is a possibility that the technology of Patent Document 1 cannot properly distinguish whether the flocs are good.

[0043] In contrast, one of the features of the present invention is to conceive of an actual continuously operating factory, and according to the quality of the water to be treated, obtain an image (hereinafter sometimes referred to as a "coagulation image") captured at a certain moment at a place where it is easy to judge the difference in coagulation quality and the coagulation state of the turbidity-causing substances in the water to be treated can be photographed, and use a model with this image as the input and information indicating whether the coagulation state of the turbidity-causing substances in the water to be treated presented in the coagulation image is good as the output, and calculate information indicating whether the coagulation state is good according to the obtained coagulation image.

[0044] Thereby, the present invention can calculate information that even unskilled personnel can simply distinguish whether the coagulation state of the turbidity-causing substances is good based on the coagulation image. Further, according to the present invention, using the information that even unskilled personnel can simply distinguish whether the coagulation state of the turbidity-causing substances is good, it is possible to pre-judge poor coagulation of the turbidity-causing substances.

[0045] <<Embodiment>>

[0046] The following describes each embodiment of the present invention with reference to the accompanying drawings.

[0047] <<First Embodiment>>

[0048] <Structure>

[0049] The water treatment condition monitoring system of the first embodiment of the present invention (hereinafter sometimes referred to as the "first monitoring system") will be described. Figure 1 It is a schematic structural diagram showing an example in which the first monitoring system is applied to a water purification plant 100 as an example of factory facilities.

[0050] The water purification plant 100 is a facility that purifies river water, reservoir water, and groundwater and delivers the treated water to factories, ordinary households, etc., and is composed of multiple facilities. In this example, the water purification plant 100 is as Figure 1 shown, including an intake well 110, a mixing tank 120, a flocculation tank 130, a sedimentation tank 140, a filtration tank 150, etc. The water to be treated Wa1 is sequentially delivered from rivers, reservoirs, etc. to the intake well 110, the mixing tank 120, the flocculation tank 130, the sedimentation tank 140, and the filtration tank 150. By performing purification, disinfection, etc., turbidity substances are removed from the water to be treated Wa1 and the water to be treated Wa1 is sterilized (disinfected). The water to be treated Wa1 after purification and disinfection is called "treated water Wa2". The treated water Wa2 is supplied as tap water to each household, factory, etc.

[0051] The first monitoring system is applied to the water purification plant 100 and includes a first water quality meter 210a, a second water quality meter 210b, a first imaging device 220a, a second imaging device 220b, and an information processing device 230. They are connected by wire or wirelessly so as to be able to communicate with each other. In the following description, the "first water quality meter 210a" and the "second water quality meter 210b" are referred to as "water quality meters 210" when it is not particularly necessary to distinguish them. The "first water quality meter 210a" is sometimes referred to as the "water to be treated measuring device" for the convenience of explanation. The "second water quality meter 210b" is sometimes also referred to as the "treated water measuring device" for the convenience of explanation. The "first imaging device 220a" and the "second imaging device 220b" are referred to as "imaging devices 220" when it is not particularly necessary to distinguish them.

[0052] The water quality meter 210 is a sensor group including multiple sensors that measure water quality items regarding water treatment. The multiple sensors are, for example, a turbidimeter that measures the turbidity of water, a thermometer that measures the water temperature, a pH meter that measures the pH value, an alkalinity meter that measures the alkalinity, a TOC meter that measures organic substances in water, an ultraviolet absorbance meter, and a water volume meter that measures the treatment volume instead of water quality.

[0053] The first water quality meter 210a is installed at a position where the water to be treated Wa1 in the water purification plant 100 can be measured. In this example, the first water quality meter 210a is installed at a position where the water to be treated Wa1 in the intake well 110 can be measured. The second water quality meter 210b is installed at a position where the treated water Wa2 can be measured.

[0054] The first water quality meter 210a measures the water to be treated Wa1 at a prescribed time interval and transmits the measurement values (measurement values of each sensor) to the information processing device 230. Here, regarding the measurement values of each sensor of the water to be treated Wa1, for the sake of convenience of explanation, they are sometimes also referred to as "water quality parameters of the water to be treated". The second water quality meter 210b measures the treated water Wa2 at a prescribed time interval and transmits the measurement values (measurement values of each sensor) to the information processing device 230.

[0055] The installation positions of the first imaging device 220a and the second imaging device 220b are positions where the coagulation state of the turbidity substances in the water to be treated Wa1 can be photographed during the process from chemical injection to the formation of coagulation nuclei and the growth of flocs. Each of the first imaging device 220a and the second imaging device 220b only needs to be an imaging device having a structure capable of shooting a moving image or recording images at a certain time interval. In this example, each of the first imaging device 220a and the second imaging device 220b is a camera that shoots a moving image. In this example, the first imaging device 220a is installed at a position where the water to be treated Wa1 at the entrance of the floc formation tank 130 can be photographed. The second imaging device 220b is installed at a position where the water to be treated Wa1 at the exit of the floc formation tank 130 can be photographed. The installation positions of the first imaging device 220a and the second imaging device 220b are not limited to this. For example, the first imaging device 220a may also be installed at a position where the water to be treated Wa1 in the mixing tank 120 can be photographed.

[0056] The imaging device 220 can photograph the water to be treated Wa1 in a state where the imaging device 220 is directly placed in the water, can also photograph on the water surface, or can photograph the water to be treated Wa1 through a transparent wall surface. That is, the method for the imaging device 220 to photograph the water to be treated Wa1 is not particularly limited as long as the coagulation state of the turbidity substances can be photographed.

[0057] The imaging device 220 uses an image extraction unit (not shown) to extract images (still images) at a prescribed time interval and transmits them to the information processing device 230. As the time interval for extracting images, it is preferably the same as the measurement time interval of the water quality meter 210. Here, the image extraction unit may be included in the imaging device 220 or may be included in the information processing device 230. In the case where the information processing device 230 includes the image extraction unit, the image extraction unit processes the moving image received from the imaging device 220 as described above.

[0058] The information processing device 230 includes an information storage unit 231, a model generation unit 232, a coagulation quality determination unit 233, and a coagulation quality output unit 234.

[0059] The information storage unit 231 attaches time information of the measurement time point of the measurement value (for example, turbidity, water temperature, pH, alkalinity, etc.) sent from the first water quality meter 210a, and holds (stores, saves) multiple measurement values (multiple "sets of each measurement value") as time series data (hereinafter referred to as "raw water data"). The information storage unit 231 attaches time information of the measurement time point of the measurement value (for example, turbidity, water temperature, pH, alkalinity, etc.) sent from the second water quality meter 210b, and holds (stores, saves) the measurement value (multiple "sets of each measurement value") as time series data (hereinafter referred to as "treated water data").

[0060] The information storage unit 231 attaches time information of the shooting time point of each image (hereinafter also referred to as "first image") sent from the first imaging device 220a, and holds (stores, saves) multiple first images at each specified time interval as time series data (hereinafter referred to as "first image data"). The information storage unit 231 attaches time information of the shooting time point of each image (hereinafter also referred to as "second image") sent from the second imaging device 220b, and holds (stores, saves) multiple second images at each specified time interval as time series data (hereinafter referred to as "second image data"). The information storage unit 231 may further attach information indicating the shooting location of each image.

[0061] The information storage unit 231 correlates and holds (stores, saves) the raw water data, the treated water data, the first image data, and the second image data with each other based on the time information. For example, the information storage unit 231 uses the time as the Key, combines (correlates) the raw water data obtained by the first water quality meter 210a, the treated water data obtained by the second water quality meter 210b, and the image data obtained by the first imaging device 220a and the second imaging device 220b at the same time (same time range), and holds (stores, saves) them as a data set.

[0062] In addition, for example, each piece of data can also be associated with data corresponding to the treatment process of the water to be treated Wa1 when the water to be treated Wa1 is treated at the water purification plant 100. In this case, for example, the information storage unit 231 preferably associates the measured value of the water to be treated Wa1 at a certain time t1, the first image at the time t1 + ta, the second image at the time t1 + tb, and the measured value of the treated water at the time t1 + tc with each other based on the time information. ta corresponds to the time during which the water to be treated Wa1 is conveyed from the installation position of the first water quality meter 210a to the first imaging device 220a. tb corresponds to the time during which the water to be treated Wa1 is conveyed from the installation position of the first water quality meter 210a to the installation position of the second imaging device 220b. tc corresponds to the time during which the water to be treated Wa1 is conveyed from the installation position of the first water quality meter 210a to the installation position of the second water quality meter 210b. However, since the water quality of the typical water to be treated Wa1 does not change within a short period of time, as described above, it is also possible to associate each piece of data (each image, each measured value) at the same time, and in this case, data corresponding to the treatment process of the water to be treated Wa1 can also be obtained.

[0063] The model generation unit 232 uses the data stored in the information storage unit 231 to generate (construct) a model for judging the coagulation quality of the turbidity substances presented in the image of the water to be treated Wa1 (a model for outputting information indicating whether the coagulation state of the turbidity substances of the water to be treated Wa1 presented in the image is good). More specifically, the model generation unit 232 uses a data set including the water to be treated data, the treated water data, and the first image data that are associated with each other to generate a first coagulation quality judgment model. The model generation unit 232 uses a data set including the water to be treated data, the treated water data, and the second image data that are associated with each other to generate a second coagulation quality judgment model. The model generation unit 232 saves (stores) the generated first coagulation quality judgment model and the second coagulation quality judgment model.

[0064] The coagulation quality judgment unit 233 uses one of the first coagulation quality judgment model and the second coagulation quality judgment model generated by the model generation unit 232, and calculates information indicating whether the coagulation state of the turbidity substances of the water to be treated Wa1 is good based on the input data including an image (the first image or the second image) representing the coagulation state of the turbidity substances of the water to be treated Wa1 at a certain time (for example, the current time). The coagulation quality judgment unit 233 can select (set) one of the first coagulation quality judgment model and the second coagulation quality judgment model as the model used for calculating the information indicating whether the coagulation state of the turbidity substances is good based on the operation of the user on an operation device (not shown).

[0065] The aggregation quality output unit 234 outputs the information calculated by the aggregation quality determination unit 233.

[0066] Figure 2 It is a schematic structural diagram showing an example of the hardware configuration of the information processing device 230. As Figure 2 shown, the information processing device 230 includes a CPU 241, a ROM 242, a RAM 243, a non-volatile storage device (HDD) 244 capable of reading and writing data, a network interface 245, an input / output interface 246, etc. They are connected via a bus 247 so as to be able to communicate with each other. Among them, the information processing device 230 may be composed of a plurality of information processing devices.

[0067] The CPU 241 loads various programs (not shown) stored in the ROM 242 and / or the HDD 244 into the RAM 243, and realizes various functions by executing the programs loaded into the RAM 243. In the RAM 243, various programs executed by the CPU 241 are loaded as described above, and data used when the CPU 241 executes various programs is temporarily stored. The ROM 242 and / or the HDD 244 are non-volatile storage media and store various programs. The network interface 245 is an interface for connecting the information processing device 230 to a network. The input / output interface 246 is an interface for connecting to operation devices such as a keyboard and a mouse and a display.

[0068] The information storage unit 231 is composed of the HDD 244, the network interface 245, and / or the input / output interface 246. The model generation unit 232 is composed of programs stored in the ROM 242 and / or the HDD 244 executed by the CPU 241 of the information processing device 230 and the HDD 244. The aggregation quality determination unit 233 is composed of programs stored in the ROM 242 and / or the HDD 244 executed by the CPU 241 of the information processing device 230. The aggregation quality output unit 234 is composed of the network interface 245 and / or the input / output interface 246.

[0069] The information processing device 230 is connected to the monitoring and control device 310. The monitoring and control device 310 is connected to the chemical agent pump 320. The monitoring and control device 310 is configured to be able to control the injection amount of the flocculant injected into the water to be treated Wa1 in the mixing tank 120 by controlling the chemical agent pump 320. Among them, the monitoring and control device 310 and the chemical agent pump 320 may be included in the first monitoring system. The function of the monitoring and control device 310 may be included in the information processing device 230.

[0070] <Test Example>

[0071] In the water purification plant 100, the process of removing turbidity substances from the water to be treated Wa1 is called "coagulation sedimentation treatment". In the coagulation sedimentation treatment of a general water purification plant 100, chemicals called coagulants such as polyaluminum chloride and aluminum sulfate are injected into the water to be treated Wa1 flowing into the facility. One of the purposes of injecting the chemicals is to remove insoluble substances (turbidity substances) suspended in the water. Turbidity substances are negatively charged in water, and charge neutralization can be achieved by adding chemicals. In addition, another purpose of injecting the chemicals is to utilize the cross-linking effect between the chemicals to make it easier for the turbidity substances to coagulate when the neutralized turbidity substances collide with each other.

[0072] The chemical is injected into the water to be treated Wa1 in the mixing tank 120 (rapid mixing tank), and the turbidity substances are made to collide with each other by rapid stirring to form coagulation nuclei in the water to be treated Wa1. Then, in the floc formation tank 130 (floc formation tank), the flocs are grown by gently stirring at a speed that does not damage the strength of the flocs. Then, the grown flocs are precipitated and removed in the sedimentation tank 140 (sedimentation tank).

[0073] As factors of the floc coagulation mechanism, several examples are listed. When the injection amount of the chemical increases, the cross-linking effect becomes stronger, so the growth of the flocs becomes faster, the coagulation state of the turbidity substances becomes better, and the turbidity of the treated water (hereinafter also referred to as "treated water turbidity") can be reduced. However, when the injection amount of the chemical is too large, the turbidity substances will become positively charged, the coagulation state of the turbidity substances will become poor, and the treated water turbidity may increase.

[0074] When the stirring intensity increases, since the collision frequency between the turbidity substances can be increased, the greater the stirring intensity, the faster the flocs grow. However, when the stirring intensity is too large, it will become the main reason for the formed flocs to be damaged.

[0075] The larger the amount of turbidity in the water (the turbidity of the water to be treated (hereinafter also referred to as "water to be treated turbidity")), the greater the collision frequency of the turbidity substances (or coagulation nuclei, flocs) can be increased, so the growth of the flocs becomes faster.

[0076] As described above, the growth of the flocs is affected by the operating conditions of the plant such as the injection amount of the chemical and the stirring intensity, as well as the water quality conditions.

[0077] Skilled workers visually observe on-site or view images of the coagulation state of the turbidity substances captured in the monitoring room. When phenomena such as the size of the flocs being smaller than usual are confirmed based on experience, it can be judged that the coagulation is poor, and countermeasures are taken to achieve appropriate water purification treatment.

[0078] However, as described above, the working-age population (15 to 64 years old) in Japan is on a decreasing trend. Therefore, sustainable solutions need to be considered now in case of a future shortage of manpower.

[0079] In response to this, the information processing device 230 of the first monitoring system generates a model that can calculate information indicating whether the coagulation state of the turbidity substances in the treated water Wa1 is good based on the coagulated image, so that even a non-expert can determine whether the coagulation state of the flocs (turbidity substances) presented in the image (screen) is good. That is, the information processing device 230 of the first monitoring system generates a model that takes the data of the image at a certain moment as input data, detects the change in the coagulation state (the difference in the coagulation state presented in the image at a certain moment), and can calculate information indicating whether the coagulation state of the turbidity substances in the treated water Wa1 is good based on the input data. Further, the first monitoring system uses the model to calculate information indicating whether the coagulation state of the turbidity substances in the treated water Wa1 is good (in this example, the predicted value of the treated water turbidity (sometimes also referred to as "predicted treated water turbidity")) based on the coagulated image. Here, the smaller the value of the predicted treated water turbidity, the better the coagulation state of the treated water Wa1 (good), and the larger the value, the worse the coagulation state of the treated water Wa1 (bad). In order to confirm whether the change in the coagulation state related to the treated water turbidity can be detected by the image, a coagulated image acquisition experiment was conducted. Figure 3 Table showing the test conditions of the coagulated image acquisition experiment and the treatment results (treated water turbidity) of the treated water. In the coagulated image acquisition experiment, a stirrer and a blackboard (background board) used as the background when shooting the coagulation state were set at a position 15 mm away from the inner wall of the rectangular transparent water tank, and a camera was set on the outer wall side of the water tank so as to be able to shoot the background board. In addition, the light source was located above the water tank. The treated water was adjusted by adding kaolin reagent as a simulated turbidity substance to tap water. The water quality of each test condition at this time is as Figure 3 shown. The coagulation conditions were changed by changing the amount of the medicament injected into the treated water adjusted to turbidities of 25 degrees, 81 degrees, and 85 degrees. After the camera started shooting, a coagulant with a preset medicament injection amount was added, and rapid stirring was performed for 3 minutes, slow stirring was performed for 10 minutes, and then standing was performed for 10 minutes. Then, the clarified water in the water tank was collected, and the treated water turbidity was measured. Here, the rapid stirring corresponding to the treatment in the rapid mixing tank (mixing tank 120) was performed under the condition of a stirring intensity of 148.2 s -1 and the slow stirring corresponding to the treatment in the floc formation tank (floc formation tank 130) was performed under the condition of a stirring intensity of 64.0 s -1 . In addition, after 10 minutes of standing was completed, the shooting of the camera was ended. The camera photographed the flocs between the inner wall of the water tank and the background board during the experiment.

[0080] Extract images from the captured video for image processing, and calculate the area value (in pixels) of the flocs in the image. In this test example, the coagulation images are extracted from the captured video in units of 0.5 seconds. In addition, the image processing includes grayscale conversion, binarization, and object recognition (floc detection). When calculating based on only one image, the results may fluctuate. Therefore, starting from the beginning of slow stirring (0 minutes), coagulation images are extracted in units of 0.5 seconds to calculate the area value of the flocs, and then the values for 10 seconds are accumulated to obtain the cumulative floc area value. The calculation interval for the cumulative floc area value is 30 seconds (0 minutes, 0.5 minutes, 1 minute...).

[0081] Figure 4 This is the time history change of the cumulative floc area value. The cumulative floc area value increases with the passage of treatment time. This is due to the growth of the flocs. Compared with Tests 3 and 4, the cumulative floc area values in Tests 1 and 2 are smaller. This is because the turbidity of the water to be treated is about 25 degrees, which is smaller than about 80 degrees in Tests 3 and 4, so the amount of flocs generated is less. In addition, compared with Tests 3 and 4, the increase in the cumulative floc area value in Tests 1 and 2 is relatively slow.

[0082] This is determined by the above-mentioned floc coagulation mechanism. The higher the turbidity of the water to be treated, the higher the collision frequency of the turbidity substances (or coagulation nuclei, flocs) can be increased, so the growth of the flocs becomes faster. As described above, in Tests 1 and 2 where the turbidity of the water to be treated is smaller than that in Tests 3 and 4, the growth of the flocs is relatively slow, and the increase in the floc area value slows down. Further, in Tests 1 and 2, the higher the coagulant injection amount, the earlier the cumulative floc area value starts to increase. This is because there is also a cross-linking effect between the coagulants, and the coagulation probability when the turbidity substances (or coagulation nuclei, flocs) collide is increased.

[0083] Comparing Test 2 and Test 4 shows that although the final turbidity of the treated water is at the same level, the trend and the maximum value of the cumulative floc area value are different.

[0084] For the reasons described above, when judging whether the coagulation state of the turbidity substances is good based on the image, it is necessary to consider the water quality, especially the turbidity of the water to be treated.

[0085] Specifically, when determining whether coagulation is good based on the images of the treated water in Test 1 and Test 2 (calculating information indicating whether the coagulation state is good (predicting the turbidity of the treated water)), it is preferable to use images of the treatment time that can produce a difference in the cumulative value of the floc area. In the cases of Test 1 and Test 2, it is 1 minute to 7 minutes, and particularly preferably around 3 to 5 minutes. In addition, in the cases of Test 3 and Test 4, since a difference appears immediately after the slow stirring starts, even if an image of the treated water at 0 minutes is used, it is possible to judge the goodness or badness of coagulation (calculate information indicating whether the coagulation state is good (predicting the turbidity of the treated water)).

[0086] As described above, it can be known that based on the images at specific times (moments), the change in the coagulation state of the turbidity-causing substances that has a correlation with the turbidity of the treated water (the difference in the coagulation state of the turbidity-causing substances presented in the images) can be detected. Further, it can be known that corresponding to the quality of the treated water (especially the turbidity of the treated water), there are differences in the coagulation formation behavior (floc formation behavior) of the turbidity-causing substances presented in the images. And it can be known that depending on the quality of the treated water, the elapsed time required for the difference in the coagulation state of the turbidity-causing substances presented in the coagulation images that has a correlation with the turbidity of the treated water to appear larger (or smaller) is different. Therefore, it can be known that if an image at an appropriate specific elapsed time (moment) based on the quality of the treated water (that is, an image taken at a location corresponding to the treatment elapsed time suitable for obtaining an image of the treated water) is used as input data, since the difference in the coagulation state of the turbidity-causing substances presented in the images that has a correlation with the turbidity of the treated water is more prominent, it is possible to distinguish whether the coagulation state of the turbidity-causing substances in the treated water is good with higher accuracy (calculate information indicating whether the coagulation state is good). Among them, if a parameter indicating the quality of the treated water that has a correlation with the turbidity of the treated water is also used as input data, it is possible to distinguish whether the coagulation state of the turbidity-causing substances in the treated water is good with higher accuracy (calculate information indicating whether the coagulation state is good).

[0087] <Summary of the operation>

[0088] As described above, the information processing device 230 generates a first coagulation quality determination model and a second coagulation quality determination model. When the information processing device 230 calculates information indicating whether the coagulation state of the turbidity substances in the water to be treated Wa1 is good (in this example, the predicted value of the turbidity of the treated water Wa2 (predicted treated water turbidity)), it selects either the first coagulation quality determination model or the second coagulation quality determination model as the model for calculating information indicating whether the coagulation state of the turbidity substances is good. For example, the information processing device 230 is connected to an operation device (not shown), and based on the operation performed by the user on the operation device, selects either the first coagulation quality determination model or the second coagulation quality determination model as the model for calculating information indicating whether the coagulation state of the turbidity substances is good.

[0089] The information processing device 230 uses either the selected first coagulation quality determination model or the second coagulation quality determination model to calculate information indicating whether the coagulation state of the turbidity substances in the water to be treated Wa1 is good. For example, the user selects the most suitable coagulation quality determination model through the operation device (not shown) according to the previously measured turbidity of the water to be treated Wa1. Thus, the information processing device 230 uses the most suitable coagulation quality determination model corresponding to the water quality (turbidity of the water to be treated) of the water to be treated Wa1 to calculate information indicating whether the coagulation state of the turbidity substances in the water to be treated Wa1 is good. Thereby, the information processing device 230 can appropriately use the most suitable coagulation quality determination model to calculate information that can simply distinguish whether the coagulation state of the turbidity substances in the water to be treated Wa1 is good.

[0090] <Specific Operations>

[0091] Figure 5 is a flowchart showing the processing flow executed by the model generation unit 232. The model generation unit 232 executes Figure 5 the flowchart shown. Therefore, the model generation unit 232 starts processing from Figure 5 step 500, and sequentially executes the processing of steps 505 to 520 described below. After that, the model generation unit 232 proceeds to step 595 to temporarily end this processing flow.

[0092] Step 505: The model generation unit 232 obtains a data set of first image data, water-to-be-treated data, and treated-water data that are associated with each other from the information storage unit 231. In this example, the water-to-be-treated data includes, for example, the turbidity of the water to be treated, water temperature, pH, alkalinity, TOC (Total Organic Carbon), and coagulant injection concentration.

[0093] Step 510: The model generation unit 232 performs image processing on each first image of the first image data. Specifically, the model generation unit 232 grayscales each first image of the first image data and then performs brightness correction.

[0094] Step 515: The model generation unit 232 normalizes (standardizes) the processed water data (each measured value of the processed water data).

[0095] Step 520: The model generation unit 232 uses the processed first image data, the processed processed water data, and the processed water data as learning data, and through deep learning, generates a learned model (first coagulation pass / fail judgment model) that takes the first image (the first image after image processing) and each measured value of the processed water data (each measured value after normalization processing) as inputs and the predicted processed water turbidity as the output. The model generation unit 232 saves (stores) the generated first coagulation pass / fail judgment model.

[0096] Among them, in deep learning, for example, the processed first image and each measured value of the processed water data (each measured value after normalization processing) that are associated with each other with time as the key are used as inputs, and the processed water turbidity of the processed water data associated with the first image and each measured value is used as the correct answer value to perform one learning. This learning is performed separately for each first image, each measured value of the processed water data, and the processed water turbidity that are associated with each other based on time in the learning data set. That is, multiple repeated learnings are performed.

[0097] In deep learning, for the input of the first image, a known CNN is used to output the first output neuron before the final output, and for the input of each measured value, a known NN is used to output the second output neuron before the final output. In the final stage, the outputs from the first output neuron and the second output neuron are input to the combination layer, and then the final predicted value (predicted processed water turbidity) is output through a function. The error between the predicted value and the correct answer value is evaluated using the function, and learning is performed in such a way that the function is minimized.

[0098] Figure 6 is a flowchart showing the processing flow executed by the model generation unit 232. The model generation unit 232 executes Figure 6 the flowchart shown. Therefore, the model generation unit 232 starts processing from Figure 6 step 600 as described below, and sequentially executes the processing of step 605 to step 620. After that, the model generation unit 232 proceeds to step 695 to temporarily end this processing flow.

[0099] Step 605: The model generation unit 232 obtains a data set of second image data, treated water data, and treated water data that are associated with each other from the information storage unit 231.

[0100] Step 610: The model generation unit 232 performs image processing on each second image of the second image data. Specifically, the model generation unit 232 grayscales each second image of the second image data and then performs brightness correction.

[0101] Step 615: The model generation unit 232 normalizes the treated water data (each measured value of the treated water data).

[0102] Step 620: The model generation unit 232 uses the processed second image data, the processed treated water data, and the treated water data as learning data, and generates a learned complete model (second coagulation quality judgment model) that takes the second image (the processed second image) and each measured value of the treated water data (each normalized measured value) as inputs and predicts the turbidity of the treated water as an output through the same deep learning as above. The model generation unit 232 saves (stores) the generated second coagulation quality judgment model.

[0103] Figure 7 is a flowchart showing the processing flow executed by the coagulation quality judgment unit 233. This processing flow is the processing flow executed by the coagulation quality judgment unit 233 when the first coagulation quality judgment model is selected as the model for calculating information indicating whether the coagulation state of the turbidity-causing substance is good. When the first coagulation quality judgment model is selected, the coagulation quality judgment unit 233 executes Figure 7 the flowchart shown. Therefore, the coagulation quality judgment unit 233 starts processing from Figure 7 step 700 and sequentially executes the processing of steps 705 to 720 described below. After that, the coagulation quality judgment unit 233 proceeds to step 795 to temporarily end this processing flow.

[0104] Step 705: The coagulation quality judgment unit 233 obtains the first image at the current time from the first imaging device 220a and obtains the treated water data (each measured value at the current time) from the first water quality meter 210a.

[0105] Step 710: The coagulation quality judgment unit 233 performs image processing on the first image. Specifically, the coagulation quality judgment unit 233 grayscales the first image and then performs brightness correction.

[0106] Step 715: The coagulation quality judgment unit 233 normalizes the treated water data (each measured value at the current time).

[0107] Step 720: The coagulation quality determination unit 233 obtains the first coagulation quality determination model from the model generation unit 232, inputs the first image after image processing and the processed water data after normalization processing into the first coagulation quality determination model, and calculates and obtains the predicted value (predicted processed water turbidity) of the processed water turbidity as the output of the first coagulation quality determination model.

[0108] Figure 8 is a flowchart showing the processing flow executed by the coagulation quality determination unit 233. This processing flow is the processing flow executed by the coagulation quality determination unit 233 when the second coagulation quality determination model is selected as the model for calculating information indicating whether the coagulation state of the turbidity-causing substances is good. Among them, this processing flow is the same as Figure 7 except that the first image is replaced with the second image and the first coagulation quality determination model is replaced with the second coagulation quality determination model, Figure 7 so detailed description is omitted.

[0109] <Effect>

[0110] As described above, the first monitoring system can appropriately use the most suitable coagulation quality determination model to calculate information (predicted processed water turbidity) indicating whether the coagulation state of the turbidity-causing substances in the processed water Wa1 presented in the image is good. In the first monitoring system, by converting the knowledge about the coagulation state of the turbidity-causing substances, which was previously the tacit knowledge of skilled workers and relied on visual observation, into explicit knowledge, even an inexperienced operator (unskilled worker) can simply and early determine whether the coagulation state of the turbidity-causing substances presented in the image is good by using this first monitoring system. Furthermore, in the past, when the water quality changed, the coagulant injection amount was mostly increased for safe operation. However, according to the first monitoring system, by quickly determining whether the coagulation state of the turbidity-causing substances in the processed water Wa1 presented in the image is good and operating with the most suitable coagulant injection amount, the coagulant injection amount can be reduced, the unnecessary coagulant cost can be reduced, and thus the cost can be saved.

[0111] <<First Variation Example>>

[0112] A first modification example of the first monitoring system will be described. The first modification example generates a first aggregation quality determination model and a second aggregation quality determination model using a part of the data stored in the information storage unit 231. The first modification example obtains a threshold turbidity using at least a part (in this example, all) of the remaining data stored in the information storage unit 231. The threshold turbidity is used to determine which of the first aggregation quality determination model and the second aggregation quality determination model has a smaller error in predicting the turbidity of the treated water (predicted treated water turbidity). The first modification example compares the turbidity of the water to be treated Wa1 with the threshold turbidity and selects either the first aggregation quality determination model or the second aggregation quality determination model based on the comparison result. The first modification example calculates the predicted treated water turbidity of the treated water Wa2 using the selected model. Except for the above aspects, it is the same as the first monitoring system.

[0113] Next, the description will focus on the differences.

[0114] <Specific operations>

[0115] Figure 9 is a flowchart showing the processing flow executed by the model generation unit 232 of the first modification example. The model generation unit 232 executes Figure 9 the flowchart shown. Therefore, the model generation unit 232 starts processing from Figure 9 step 900 and sequentially executes the processing of steps 905 to 935 described below.

[0116] Step 905: The model generation unit 232 obtains a data set of first image data, second image data, water-to-be-treated data, and treated water data that are associated with each other from the information storage unit 231. At this time, the model generation unit 232 obtains a part (in this example, 70%) of the entire data set stored in the information storage unit 231. The model generation unit 232 uses the obtained data set to execute the processing of steps 910 to 930 described below.

[0117] Step 910: The model generation unit 232 performs image processing on each first image of the first image data. Specifically, the model generation unit 232 grayscales each first image of the first image data and then performs brightness correction.

[0118] Step 915: The model generation unit 232 performs normalization processing on the water-to-be-treated data (each measured value of the water-to-be-treated data).

[0119] Step 920: The model generation unit 232 uses the processed first image data, the processed treated water data, and the treated water data as learning data, and through deep learning, generates a learned complete model (the first coagulation quality judgment model) with the first image (the first image after image processing) and each measured value of the treated water data (each measured value after normalization processing) as inputs and the predicted turbidity of the treated water as the output. The model generation unit 232 saves (stores) the generated first coagulation quality judgment model.

[0120] Step 925: The model generation unit 232 performs image processing on each second image of the second image data. Specifically, the model generation unit 232 grayscales each second image of the second image data and then performs brightness correction.

[0121] Step 930: The model generation unit 232 uses the processed second image data, the processed treated water data, and the treated water data as learning data, and through deep learning, generates a learned complete model (the second coagulation quality judgment model) with the second image (the second image after image processing) and each measured value of the treated water data (each measured value after normalization processing) as inputs and the predicted turbidity of the treated water as the output. The model generation unit 232 saves (stores) the generated second coagulation quality judgment model.

[0122] Step 935: The model generation unit 232 executes Figure 10 the evaluation process shown in the flowchart of

[0123] Figure 10 is a flowchart showing the processing flow executed by the model generation unit 232 of the first modification example. When the model generation unit 232 advances to Figure 9 step 935 of Figure 10 it starts processing from

[0124] step 1000 of

[0125] and sequentially executes the processing of steps 1005 to 1035 as described below. Step 1005: The model generation unit 232 obtains a data set of the first image data, the second image data, the treated water data, and the treated water data that are associated with each other from the information storage unit 231. At this time, the model generation unit 232 obtains all the remaining (30% in this example) data sets in the entire data set stored in the information storage unit 231. The model generation unit 232 uses the obtained data set and sequentially executes the processing of steps 1010 to 1035 as described below.

[0125] Step 1010: The model generation unit 232 performs image processing on each first image of the first image data. Specifically, the model generation unit 232 grayscales each first image of the first image data and then performs brightness correction.

[0126] Step 1015: The model generation unit 232 normalizes the processed water data (each measured value of the processed water data).

[0127] Step 1020: The model generation unit 232 uses the first image data after image processing, the processed water data after normalization processing, and the first coagulation pass / fail judgment model to calculate the predicted value of the treated water turbidity (predict the treated water turbidity). The model generation unit 232 uses the predicted value of the treated water turbidity (predict the treated water turbidity) and the treated water turbidity of the treated water data to calculate the error of the predicted value (= |predicted treated water turbidity - treated water turbidity (actual value) of the treated water data|).

[0128] Step 1025: The model generation unit 232 performs image processing on the second image data. Specifically, the model generation unit 232 grayscales each second image of the second image data and then performs brightness correction.

[0129] Step 1030: The model generation unit 232 uses the second image data after image processing, the processed water data after normalization processing, and the second coagulation pass / fail judgment model to calculate the predicted value of the treated water turbidity (predict the treated water turbidity). The model generation unit 232 uses the predicted value of the treated water turbidity (predict the treated water turbidity) and the treated water turbidity of the treated water data to calculate the error of the predicted value (= |predicted treated water turbidity - treated water turbidity (actual value) of the treated water data|).

[0130] Step 1035: The model generation unit 232 calculates the threshold turbidity based on the treated water turbidity of the processed water data, the error of the predicted value calculated in step 1020, and the error of the predicted value calculated in step 1030. For example, the threshold turbidity is the turbidity of the intersection point P10 of line a1 and line a2 when making the Figure 11 coordinate graph shown. Line a1 is a line drawn based on the treated water turbidity of the processed water data and the error of the predicted value calculated in step 1020, representing the relationship between the treated water turbidity of the processed water data and the average error of the predicted value in the case of using the first coagulation pass / fail model. Line a2 is a line drawn based on the treated water turbidity of the processed water data and the error of the predicted value calculated in step 1030, representing the relationship between the treated water turbidity of the processed water data and the average error of the predicted value in the case of using the second coagulation pass / fail model.

[0131] According to Figure 11From the coordinate diagram, it can be seen that when the turbidity of the water to be treated Wa1 is greater than the threshold turbidity, compared with using the second coagulation quality judgment model, the average error is smaller when using the first coagulation quality judgment model. Therefore, the accuracy of the predicted value of the treated water turbidity (predicted treated water turbidity) is high. In addition, it can be seen that when the turbidity of the water to be treated Wa1 is below the threshold turbidity, compared with using the first coagulation quality judgment model, the average error is smaller when using the second coagulation quality judgment model. Therefore, the accuracy of the predicted value (predicted treated water turbidity) is high.

[0132] After that, the model generation unit 232 proceeds to step 1095 to temporarily end this processing flow, and then proceeds to step 995 to temporarily end Figure 9 the processing flow.

[0133] Figure 12 is a flowchart showing the processing flow executed by the coagulation quality judgment unit 233 of the first modification example. The coagulation quality judgment unit 233 executes Figure 12 the flowchart shown. Therefore, the coagulation quality judgment unit 233 starts processing from Figure 12 step 1200, proceeds to step 1205, and obtains the water data to be treated at the current time (each measured value at the current time) from the first water quality meter 210a.

[0134] After that, the coagulation quality judgment unit 233 proceeds to step 1210 to judge whether the turbidity of the water to be treated in the water data to be treated is greater than the threshold turbidity.

[0135] When the turbidity of the water to be treated is greater than the threshold turbidity, the coagulation quality judgment unit 233 judges "yes" in step 1210, and sequentially executes the processing of steps 1215 to 1230 described below, and then proceeds to step 1295 to temporarily end this processing flow.

[0136] Step 1215: The coagulation quality judgment unit 233 obtains the first image at the current time from the first imaging device 220a.

[0137] Step 1220: The coagulation quality judgment unit 233 performs image processing on the first image. Specifically, the coagulation quality judgment unit 233 grayscales the first image and then performs brightness correction.

[0138] Step 1225: The coagulation quality judgment unit 233 performs normalization processing on the water data to be treated (each measured value at the current time).

[0139] Step 1230: The coagulation quality determination unit 233 obtains the first coagulation quality determination model from the model generation unit 232, inputs the first image after image processing and the treated water data after normalization processing into the first coagulation quality determination model, and calculates and obtains the predicted value of the turbidity of the treated water (predicted treated water turbidity) as the output of the first coagulation quality determination model.

[0140] When the turbidity of the treated water is below the threshold turbidity, the coagulation quality determination unit 233 determines "no" in step 1210, and sequentially executes the processes of steps 1235 to 1250 described below, and then advances to step 1295 to temporarily end this processing flow.

[0141] Step 1235: The coagulation quality determination unit 233 obtains the second image at the current time from the second imaging device 220b.

[0142] Step 1240: The coagulation quality determination unit 233 performs image processing on the second image. Specifically, the coagulation quality determination unit 233 grayscales the second image and then performs brightness correction.

[0143] Step 1245: The coagulation quality determination unit 233 normalizes the treated water data (each measured value at the current time).

[0144] Step 1250: The coagulation quality determination unit 233 obtains the second coagulation quality determination model from the model generation unit 232, inputs the second image after image processing and the treated water data after normalization processing into the second coagulation quality determination model, and calculates and obtains the predicted value of the turbidity of the treated water (predicted treated water turbidity) as the output of the second coagulation quality determination model.

[0145] <Effect>

[0146] In the first modification example, according to the turbidity of the treated water of the treated water data, the most suitable coagulation quality determination model is selected, and by using the selected most suitable coagulation quality determination model, it is possible to calculate information (predicted treated water turbidity) with higher accuracy indicating whether the coagulation state of the turbidity-causing substances is good. Among them, the features of the first modification example can also be applied to the second and third embodiments described later.

[0147] <<Second Embodiment>>

[0148] A water treatment condition monitoring system (hereinafter, sometimes also referred to as "second monitoring system") according to a second embodiment of the present invention will be described.

[0149] Figure 13 It is a schematic structural diagram showing an example in which the second monitoring system is applied to the water purification plant 100. As Figure 13As shown, the second monitoring system includes a first water extraction pump 1310a, a second water extraction pump 1310b, a first valve 1320a, a second valve 1320b, a flow-through cell 1330, a pipe 1340, and a valve control device 1350. In the second monitoring system, the second imaging device 220b of the first monitoring system is omitted, and the first imaging device 220a is disposed at a position capable of photographing the flow-through cell 1330. Except for the above aspects, it is the same as Figure 1 the structure of the first monitoring system shown.

[0150] The first water extraction pump 1310a is disposed at the inlet of the flocculation tank 130 and collects the water to be treated Wa1 at the inlet of the flocculation tank 130. The collected water to be treated Wa1 flows into the flow-through cell 1330 through the pipe 1340. It can also be configured such that the first water extraction pump 1110a is disposed in the mixing tank 120 and collects the water to be treated Wa1 in the mixing tank 120.

[0151] The second water extraction pump 1310b is disposed at the outlet of the flocculation tank 130 and collects the water to be treated Wa1 at the outlet of the flocculation tank 130. The collected water to be treated Wa1 flows into the flow-through cell 1330 through the pipe 1340.

[0152] The flow-through cell 1330 is a transparent container through which the collected water to be treated Wa1 can pass.

[0153] The first valve 1320a is an electromagnetic valve and is disposed on the pipe 1340 between the branch portion Pt1 of the first water extraction pump 1310a and the pipe 1340. By controlling the energization of the first valve 1320a, the first valve 1320a is set to either an open state that allows the water to be treated Wa1 to flow or a closed state that blocks the flow of the water to be treated Wa1.

[0154] The second valve 1320b is an electromagnetic valve and is disposed on the pipe 1340 between the branch portion Pt1 of the second water extraction pump 1310b and the pipe 1340. By controlling the energization of the second valve 1320b, the second valve 1320b is set to either an open state that allows the water to be treated Wa1 to flow or a closed state that blocks the flow of the water to be treated Wa1.

[0155] The valve control device 1350 controls the first valve 1320a to be either in an open state or a closed state, and controls the second valve 1320b to be either in an open state or a closed state. The valve control device 1350 controls the water to be treated Wa1 flowing into the flow cell 1330 by controlling the open / closed states of the first valve 1320a and the second valve 1320b respectively, such that the water to be treated Wa1 collected by the first water pump 1310a and the water to be treated Wa1 collected by the second water pump 1310b flow into the flow cell 1330 alternately every predetermined time. That is, the valve control device 1350 controls the water to be treated Wa1 flowing into the flow cell 1330 in such a way that the water to be treated Wa1 collected by the first water pump 1310a flows into the flow cell 1330 repeatedly at a predetermined time, and then the water to be treated Wa1 collected by the second water pump 1310b flows into the flow cell 1330 at a predetermined time. Furthermore, the valve control device 1350 sends the timing data of the control signals of the first valve 1320a and the second valve 1320b to the information processing device 230.

[0156] The first imaging device 220a captures the water to be treated Wa1 passing through the inside of the flow cell 1330. The first imaging device 220a extracts images at a predetermined time interval through an image extraction unit (not shown) and sends them to the information processing device 230.

[0157] Based on the images received from the first imaging device 220a and the timing data of the control signals of the electromagnetic valves, the information processing device 230 differentiates whether the image is an image of the water to be treated Wa1 collected by the first water pump (hereinafter referred to as "first water collection image") or an image of the water to be treated Wa1 collected by the second water pump (hereinafter referred to as "second water collection image").

[0158] Similar to the first monitoring system, the information storage unit 231 associates and stores (stores, saves) the water to be treated data, the treated water data, the first water collection image data including a plurality of first water collection images at each predetermined time interval, and the second water collection image data including a plurality of second water collection images at each predetermined time interval based on the time information.

[0159] <Summary of the operation>

[0160] Similar to the first monitoring system, the model generation unit 232 of the second monitoring system uses the data stored in the information storage unit 231 to generate a model for judging the coagulation quality (a model that outputs information indicating whether the coagulation state of the turbidity substances in the water to be treated Wa1 presented in the image is good).

[0161] The model generation unit 232 uses the first water sampling image data to replace the first image data and the second water sampling image data to replace the second image data, and generates the first coagulation quality determination model and the second coagulation quality determination model in the same manner as the first monitoring system except for this.

[0162] The coagulation quality determination unit 233 uses either the first coagulation quality determination model or the second coagulation quality determination model generated by the model generation unit 232, and calculates information indicating whether the coagulation state of the turbidity substances in the water to be treated Wa1 at the current time is good based on the input data including the image (the first water sampling image or the second water sampling image) representing the coagulation state of the turbidity substances in the water to be treated Wa1. The coagulation quality determination unit 233 can set which one of the first coagulation quality determination model and the second coagulation quality determination model to use based on the operation input of the user to an operation device (not shown).

[0163] <Specific operations>

[0164] In the processing flow of Figure 5 , the model generation unit 232 replaces the first image data with the first water sampling image data, and performs the same processing flow as the processing flow of Figure 5 except for this. In the processing flow of Figure 6 , the model generation unit 232 replaces the second image data with the second water sampling image data, and performs the same processing flow as the processing flow of Figure 6 except for this. In the processing flow of Figure 7 , the coagulation quality determination unit 233 replaces the first image with the first water sampling image, and performs the same processing flow as the processing flow of Figure 7 except for this. In the processing flow of Figure 8 , the coagulation quality determination unit 233 replaces the second image with the second water sampling image, and performs the same processing flow as the processing flow of Figure 8 except for this. These processing flows are the same as the processing flow of Figures 5 - 8 except for the above aspects, so detailed descriptions are omitted.

[0165] <Effects>

[0166] As described above, the second monitoring system can use one first imaging device 220a to photograph the coagulation states of the turbidity substances in the water to be treated Wa1 at two (multiple) places in the water treatment plant where the water to be treated Wa1 is in different treatment processes. In addition, since the installation location of the flow-through cell 1330 can be freely selected to a certain extent in the second monitoring system, the flow-through cell 1330 can be arranged in a stable environment such as indoors. Therefore, compared with the first monitoring system, the second monitoring system can improve the degree of freedom of the place and shooting conditions of the water to be treated Wa1 to be photographed by the first imaging device 220a.

[0167] <<Third Embodiment>>

[0168] A water treatment condition monitoring system according to the third embodiment of the present invention (hereinafter sometimes referred to as "third monitoring system") will be described. The third monitoring system is different from the first monitoring system in the following aspects.

[0169] In the third monitoring system, the model generation unit 232 extracts information on flocs (hereinafter referred to as "floc information") from the image data, and also uses the floc information to generate a coagulation quality judgment model. The floc information is, for example, the number of flocs in the image, the size of the flocs (particle diameter, area, etc.), the brightness of the non-floc part, etc.

[0170] The following description will focus on the differences.

[0171] <Specific Operations>

[0172] The model generation unit 232 of the third monitoring system executes Figure 14 the processing flow shown in Figure 5 instead of the processing flow shown in. Therefore, after the model generation unit 232 sequentially executes the processing of steps 1405 to 1430 described below, it proceeds to step 1495 to end this processing flow.

[0173] Step 1405: The model generation unit 232 obtains a data set of first image data, raw water data, and treated water data that are associated with each other from the information storage unit 231.

[0174] Step 1410: The model generation unit 232 performs image processing on each first image of the first image data. Specifically, the model generation unit 232 grayscales each first image of the first image data and then performs brightness correction.

[0175] Step 1415: The model generation unit 232 binarizes each first image of the first image data.

[0176] Step 1420: The model generation unit 232 extracts flocs from each binarized first image of the first image data and obtains the floc information of each first image.

[0177] Step 1425: The model generation unit 232 performs normalization processing on the floc information of each first image and the raw water data.

[0178] Step 1430: The model generation unit 232 uses the processed first image data, the processed water data to be processed, the processed floc information, and the treated water data as learning data, and through deep learning, generates a learned complete model (first coagulation quality judgment model) with the first image (the first image after image processing), each measured value of the water data to be processed (each measured value after normalization processing), and the floc information (the floc information after normalization processing) as inputs and the predicted turbidity of the treated water as the output. The model generation unit 232 saves (stores) the generated first coagulation quality judgment model.

[0179] The model generation unit 232 of the third monitoring system executes Figure 15 the processing flow shown to replace Figure 6 the processing flow shown. Therefore, after the model generation unit 232 sequentially executes the processes of steps 1505 to 1530 described below, it proceeds to step 1595 to end this processing flow.

[0180] Step 1505: The model generation unit 232 obtains a data set of the second image data, the water data to be processed, and the treated water data that are associated with each other from the information storage unit 231.

[0181] Step 1510: The model generation unit 232 performs image processing on each second image of the second image data. Specifically, the model generation unit grayscales each second image of the second image data and then performs brightness correction.

[0182] Step 1515: The model generation unit 232 binarizes each second image of the second image data.

[0183] Step 1520: The model generation unit 232 extracts flocs from each second image of the binarized second image data and obtains the floc information of each second image.

[0184] Step 1525: The model generation unit 232 performs normalization processing on the floc information of each second image and the water data to be processed.

[0185] Step 1530: The model generation unit 232 uses the processed second image data, the processed water data to be processed, the processed floc information, and the treated water data as learning data, and through deep learning, generates a learned complete model (second coagulation quality judgment model) with the second image (the second image after image processing), each measured value of the water data to be processed (each measured value after normalization processing), and the floc information (the floc information after normalization processing) as inputs and the predicted turbidity of the treated water as the output. The model generation unit 232 saves (stores) the generated second coagulation quality judgment model.

[0186] The coagulation quality determination unit 233 executes a processing flow that is different only in the following aspects from Figure 7 the processing flow. Among them, similar to the first monitoring system, this processing flow is the processing flow executed by the coagulation quality determination unit 233 when the first coagulation quality determination model is selected as the model for calculating information indicating whether the coagulation state of the turbidity-causing substance is good.

[0187] · Between step 710 and step 715, the following steps A1 and B1 are added.

[0188] Step A1: The coagulation quality determination unit 233 binarizes the first image.

[0189] Step B1: The coagulation quality determination unit 233 extracts flocs from the binarized first image and obtains the floc information of the first image.

[0190] · Instead of step 715, the following step C1 is executed, and instead of step 720, the following step D1 is executed.

[0191] Step C1: The coagulation quality determination unit 233 normalizes the treated water data (each measured value at the current moment) and the floc information.

[0192] Step D1: The coagulation quality determination unit 233 obtains the first coagulation quality determination model from the model generation unit 232, inputs the first image after image processing, the normalized treated water data (each normalized measured value), and the floc information (normalized floc information) into the first coagulation quality determination model, and calculates and obtains the predicted value of the turbidity of the treated water (predicted turbidity of the treated water) as the output of the first coagulation quality determination model.

[0193] The coagulation quality determination unit 233 executes a processing flow that is different only in the following aspects from Figure 8 the processing flow.

[0194] · Between step 810 and step 815, the following steps A2 and B2 are added. Among them, similar to the second monitoring system, this processing flow is the processing flow executed by the coagulation quality determination unit 233 when the second coagulation quality determination model is selected as the model for calculating information indicating whether the coagulation state of the turbidity-causing substance is good.

[0195] Step A2: The coagulation quality determination unit 233 binarizes the second image.

[0196] Step B2: The coagulation quality determination unit 233 extracts flocs from the binarized second image and obtains the floc information of the second image.

[0197] ·Instead of performing step 815, perform the following step C2, and instead of performing step 820, perform the following step D2.

[0198] Step C2: The flocculation quality judgment unit 233 normalizes the treated water data (each measured value at the current moment) and the floc information.

[0199] Step D2: The flocculation quality judgment unit 233 obtains the second flocculation quality judgment model from the model generation unit 232, inputs the second image after image processing, the normalized treated water data (each normalized measured value), and the floc information (normalized floc information) into the second flocculation quality judgment model, and calculates and obtains the predicted value of the treated water turbidity (predicted treated water turbidity) as the output of the second flocculation quality judgment model.

[0200] <Effect>

[0201] As described above, in the third monitoring system, the floc information that can affect whether the coagulation state of the turbidity substances in the treated water Wa1 is good is also used to generate the first flocculation quality judgment model and the second flocculation quality judgment model. Using the generated flocculation quality judgment model (the first flocculation quality judgment model or the second flocculation quality judgment model), it is possible to calculate with higher accuracy the information indicating whether the coagulation state of the turbidity substances in the treated water Wa1 presented in the image is good.

[0202] <<Fourth Embodiment>>

[0203] The water treatment condition monitoring system of the fourth embodiment of the present invention (hereinafter, sometimes also referred to as the "fourth monitoring system") will be described. The fourth monitoring system is different from the first monitoring system only in the following aspects.

[0204] The fourth monitoring system generates one flocculation quality judgment model using the first image data and the second image data.

[0205] The following will mainly describe this difference.

[0206] <Specific Operations>

[0207] The model generation unit 232 of the fourth monitoring system executes Figure 16 the processing flow shown. Therefore, the model generation unit 232 starts processing from Figure 16 step 600, and after sequentially executing the processing of steps 1605 to 1625 described below, proceeds to step 1695 to end this processing flow.

[0208] Step 1605: The model generation unit 232 obtains a data set of the first image data, the second image data, the treated water data, and the treated water data that are associated with each other from the information storage unit 231.

[0209] Step 1610: The model generation unit 232 performs image processing on each first image of the first image data. Specifically, the model generation unit 232 grayscales each first image of the first image data and then performs brightness correction.

[0210] Step 1615: The model generation unit 232 performs image processing on each second image of the second image data. Specifically, the model generation unit 232 grayscales each second image of the second image data and then performs brightness correction.

[0211] Step 1620: The model generation unit 232 normalizes the processed water data (each measured value of the processed water data).

[0212] Step 1625: The model generation unit 232 uses the processed first image data, the processed second image data, the processed processed water data, and the processed water data as learning data, and through deep learning, generates a learned complete model (coagulation quality judgment model) that takes the first image (the processed first image), the second image (the processed second image data), and each measured value of the processed water data (each normalized measured value) as inputs and the predicted turbidity of the processed water as the output. The model generation unit 232 saves (stores) the generated coagulation quality judgment model.

[0213] Figure 17 It is a flowchart showing the processing flow executed by the coagulation quality judgment unit 233. The coagulation quality judgment unit 233 executes Figure 17 the flowchart shown. Therefore, the coagulation quality judgment unit 233 starts processing from Figure 17 step 1700 and sequentially executes the processing of steps 1705 to 1720 as described below. After that, the coagulation quality judgment unit 233 proceeds to step 1795 to temporarily end this processing flow.

[0214] Step 1705: The coagulation quality judgment unit 233 obtains the first image at the current time from the first imaging device 220a, obtains the second image at the current time from the second imaging device 220b, and obtains the processed water data (each measured value at the current time) from the first water quality meter 210a.

[0215] Step 1710: The coagulation quality judgment unit 233 performs image processing on the first image. Specifically, the coagulation quality judgment unit 233 grayscales the first image and then performs brightness correction.

[0216] Step 1715: The coagulation quality judgment unit 233 performs image processing on the second image. Specifically, the coagulation quality judgment unit 233 grayscales the second image and then performs brightness correction.

[0217] Step 1720: The coagulation quality determination unit 233 normalizes the treated water data (each measured value at the current moment).

[0218] Step 1725: The coagulation quality determination unit 233 obtains the coagulation quality determination model from the model generation unit 232, inputs the first image after image processing and the normalized treated water data into the coagulation quality determination model, and calculates and obtains the predicted value of the turbidity of the treated water (predicted treated water turbidity) as the output of the coagulation quality determination model.

[0219] <Effect>

[0220] As described above, the fourth monitoring system uses the first image data and the second image data obtained by photographing the treated water Wa1 located at different places, which reflect the formation process (temporal change) of coagulation nuclei and flocs, to generate a coagulation quality determination model, and uses the generated coagulation quality determination model to calculate information indicating whether the coagulation state of the turbidity substances is good. Thus, the fourth monitoring system can accurately calculate information indicating whether the coagulation state of the turbidity substances in the treated water Wa1 presented in the image is good. Furthermore, compared with the first monitoring system, the fourth monitoring system can reduce the parameters regarding model selection and simplify the system structure.

[0221] According to the above-described embodiments and the first modification example, for example, even in a developing country where there are no experts, water treatment can be appropriately performed, thereby improving the hygiene status of the people living there, and environmental protection achieved through appropriate chemical agent specifications also contributes to the realization of a sustainable society.

[0222] <<Other Modification Examples>>

[0223] The present invention is not limited to the above-described embodiments and the first modification example, and various modification examples can be adopted within the scope of the present invention. The above-described embodiments and the first modification example can be combined with each other within the scope of the present invention.

[0224] For example, the first embodiment and the fourth embodiment can be combined. In this case, for example, the model generation unit 232 generates a plurality of coagulation quality determination models (in this example, three: the first coagulation quality determination model, the second coagulation quality determination model, and the coagulation quality model). The coagulation quality determination unit 233 uses any one of the plurality of coagulation quality determination models (in this example, the first coagulation quality determination model, the second coagulation quality determination model, and the coagulation quality determination model) generated by the model generation unit 232, and calculates information indicating whether the coagulation state of the turbidity substances in the treated water Wa1 is good based on the input data including the image indicating the coagulation state of the turbidity substances in the treated water Wa1.

[0225] For example, in each of the above-described embodiments and the first modification, as a model for judging the coagulation quality of the turbidity substances in the water to be treated Wa1, the model generation unit 232 generates a learned complete model through deep learning. However, as a model for judging the coagulation quality, an empirical formula, a statistical model, etc. may also be generated as long as it can represent the coagulation quality.

[0226] For example, in each of the above-described embodiments and the first modification, the shooting conditions of the image of the water to be treated Wa1 are not particularly limited as long as the coagulation state of the turbidity substances in the water to be treated Wa1 can be captured. However, when the light source is only sunlight and the brightness between day and night varies greatly, it is preferable to provide a light source that can illuminate the shooting location of the imaging device 220. Furthermore, by recording the change in brightness and using it for pre-processing (brightness correction) of the image of the water to be treated Wa1, the judgment accuracy of whether the coagulation state of the turbidity substances is good can also be improved. Therefore, an illuminometer or the like for recording the change in brightness of the shooting location may also be provided.

[0227] In the above-described first embodiment to the above-described third embodiment and the first modification, for example, when the water quality of the water to be treated Wa1 is relatively unlikely to change and remains constant, it is preferable to investigate in advance the treatment time (or location) where the coagulation state of the turbidity substances is likely to differ in the water purification plant, and to provide the imaging device 220 (the water sampling pump in the case of the second embodiment) at a location where the coagulation state of the turbidity substances in the water to be treated Wa1 can be captured (the location where water can be sampled in the case of the second embodiment).

[0228] In the above-described first embodiment to the third embodiment and the first modification, for example, when the water quality of the water to be treated Wa1 has changed, the image (input data) and the coagulation quality judgment model used by the information processing device 230 for judging the coagulation quality can be manually or automatically and appropriately switched according to the water quality of the water to be treated Wa1 (based on the water to be treated data).

[0229] In this case, for example, it can be configured such that the information processing device 230 judges whether the water to be treated data conforms to the conditions for promoting floc growth based on the water to be treated data. Based on the judgment result, when the water to be treated data conforms to the conditions for promoting floc growth, compared with the case where the water to be treated Wa1 does not conform to the conditions for promoting floc growth, a coagulation quality judgment model generated from the image data captured by the imaging device provided at the upstream location is selected and used, and the selected coagulation quality judgment model is used to output the predicted turbidity of the treated water according to the input data including the image of the water to be treated Wa1.

[0230] For example, it can be configured such that when the turbidity of the water to be treated Wa1 is approximately 80 degrees, the data of the water to be treated meets the conditions for promoting floc growth. An image captured by the imaging device 220 disposed at a position where the water to be treated Wa1 corresponding to the upstream side (for example, the treatment time of slow stirring is 0 minutes) can be used to generate a coagulation quality judgment model, and the generated coagulation quality judgment model can be used to output the predicted turbidity of the treated water based on this image.

[0231] For example, it can be configured such that when the turbidity of the water to be treated Wa1 is approximately 35 degrees, the data of the water to be treated does not meet the conditions for promoting floc growth. An image captured by the imaging device 220 disposed at a position where the water to be treated Wa1 corresponding to the downstream side (for example, the treatment time of slow stirring is 4 minutes) can be used to generate a coagulation quality judgment model, and the generated coagulation quality judgment model can be used to output the predicted turbidity of the treated water based on this image.

[0232] In each of the above-described embodiments and the first modification, it can be configured to dispose three or more imaging devices 220 at different locations, and use the image data captured by each imaging device 220 to generate a plurality of coagulation quality judgment models. In this case, similar to the case where there are two coagulation quality judgment models in each of the above-described first to third embodiments and the first modification, one coagulation quality judgment model is selected from the plurality of coagulation quality judgment models, and the selected coagulation quality judgment model is used to output the predicted turbidity of the treated water based on the image of the water to be treated Wa1.

[0233] In the above-described fourth embodiment, it can be configured to dispose three or more imaging devices 220 at different locations, and the information processing device 230 uses the image data captured by each imaging device 220 to generate one coagulation quality judgment model.

[0234] In each of the above-described embodiments and the first modification, it can be configured that as the output of the coagulation quality judgment model, the information processing device 230 outputs either information indicating that the coagulation state of the turbidity-causing substances is good or information indicating that the coagulation state is bad, and performs binary classification.

[0235] In each of the above-described embodiments and the first modification, it can be configured that the information processing device 230 compares the predicted turbidity of the treated water output from the coagulation quality judgment model with a specified threshold turbidity of the treated water (for example, 1 degree) to determine whether the coagulation state of the turbidity-causing substances is good or bad, and outputs the determination result.

[0236] In each of the above-described embodiments and the first modification, it may be configured that the information processing device 230 predicts the turbidity of the treated water and adjusts the injection amount of the flocculant from the chemical pump through the monitoring and control device. For example, when the flocculation state is poor (defective), the information processing device 230 adjusts the injection amount of the flocculant to make the flocculation state appropriate, and performs adjustment so that the turbidity of the treated water Wa2 becomes clear.

[0237] In each of the above-described embodiments and the first modification, it may be configured that the information processing device 230 generates a flocculation quality determination model using only the image data.

[0238] Each of the above-described embodiments and the first modification is applied to the water purification plant 100, but may also be applied to other plants that purify water other than the water purification plant 100.

[0239] In the above-described first embodiment, it may be configured that, as a model for calculating information indicating whether the flocculation state of the turbidity-causing substances is good, the information processing device 230 selects at least two models (in this example, two models: the first flocculation quality determination model and the second flocculation quality determination model), attaches weights set in advance according to the turbidity of the water to be treated to the predicted turbidity of the treated water output from at least two flocculation quality determination models (in this example, the first flocculation quality determination model and the second flocculation quality determination model), and calculates the final predicted turbidity of the treated water by summation.

[0240] In each of the above-described embodiments and the first modification, it may be configured that the information processing device 230 outputs the determination result of the flocculation quality determination unit 233 to a monitor or the like connected to the information processing device 230. In this case, for example, it may be configured that the information processing device 230 is connected to a device that can issue an alarm, and when the predicted turbidity of the treated water exceeds a threshold value (1 degree), the information processing device 230 causes the alarm device to issue an alarm.

[0241] When there is one imaging device 220 but the imaging device 220 can move along the flow direction of the water treatment (such as in a track type or a drone), even one imaging device can take pictures at multiple locations. At this time, by recording the image data together with the information that can determine the shooting location, the present invention can be realized. As the information that can determine the shooting location, for example, there is a method of attaching a GPS (Global Positioning System) function to the imaging device 220 to obtain GPS information.

[0242] In each of the above-described embodiments and the first modification, it may be configured that the information processing device 230 generates a flocculation quality determination model using an image and / or the water data to be treated (each measured value) at a certain moment as input data instead of an image and / or the water treatment data within a certain time range.

[0243] In each of the above-described embodiments and the first modification, it may be configured such that when the water to be treated is not transported to a plurality of treatment sites and is treated sequentially at one site, the change in the treatment status (at multiple locations with different treatment times) can be captured by one imaging device. In this case, the model generation unit 232 may generate the aggregation pass / fail judgment model for multiple locations with different treatment times.

[0244] Description of Reference Numerals

[0245] 100... water purification plant, 210a... first water quality meter, 210b... second water quality meter, 220a... first imaging device, 220b... second imaging device, 230... information processing device, 231... information storage unit, 232... model generation unit, 233... aggregation pass / fail judgment unit, 234... aggregation pass / fail output unit.

Claims

1. A water treatment condition monitoring system, which is applied to a facility for generating treated water. The facility generates the treated water obtained by purifying the water to be treated by sequentially transporting the water to be treated to a plurality of treatment sites and performing treatments including coagulation sedimentation treatment for removing turbidity substances contained in the water to be treated. It is characterized in that, Comprising: A plurality of imaging devices, which are arranged at a plurality of locations capable of photographing the water to be treated and having different treatment elapsed times of the water to be treated, and obtain image data of the water to be treated at each location by photographing the water to be treated, wherein the image data includes images of the water to be treated photographed from each location; A water-to-be-treated measuring device, which measures water-to-be-treated quality parameters including the turbidity of the water to be treated with respect to the water to be treated, and obtains water-to-be-treated data including the water-to-be-treated quality parameters; A treated-water measuring device, which measures the turbidity of the treated water and obtains treated-water data including the turbidity of the treated water; And An information processing device, which obtains the image data, the water-to-be-treated data, and the treated-water data from the imaging device, the water-to-be-treated measuring device, and the treated-water measuring device, wherein, The information processing device is configured to use learning input data including the image data of the water to be treated and the treated-water data as learning data, and construct a plurality of coagulation quality judgment models through machine learning, wherein the coagulation quality judgment models are used to output information indicating whether the coagulation state of the turbidity substances of the water to be treated is good based on the input data including the image of the water to be treated; The information processing device is configured to select at least one of the plurality of coagulation quality judgment models based on the turbidity of the water to be treated, obtain the image of the water to be treated as a judgment image from at least one of the plurality of imaging devices, use the selected coagulation quality judgment model, and calculate information indicating whether the coagulation state of the turbidity substances of the water to be treated presented in the judgment image is good based on the judgment input data including the judgment image.

2. The water treatment condition monitoring system according to claim 1, characterized in that: The information processing device is configured to construct one coagulation quality judgment model based on the image data photographed by one of the imaging devices.

3. The water treatment condition monitoring system according to claim 1, characterized in that: The information processing device is configured to construct one coagulation quality judgment model based on the image data photographed by a plurality of the imaging devices.

4. The water treatment condition monitoring system according to claim 1, characterized in that: The information processing device is configured to compare the turbidity of the water to be treated with a threshold turbidity, and select any one of the plurality of coagulation quality judgment models based on the comparison result.

5. The water treatment condition monitoring system according to claim 4, characterized in that: The information processing device has a storage unit, which stores a data set including the image data, the water-to-be-treated data, and the treated-water data, The information processing device is configured to use a part of the data set stored in the storage unit as the learning data, construct a plurality of the coagulation quality judgment models, and use at least a part of the other part except for the part of the data set to evaluate the accuracy corresponding to the turbidity of the water to be treated of the information indicating whether the coagulation state of the turbidity substances of the water to be treated output by the plurality of constructed coagulation quality judgment models, thereby calculating the threshold turbidity.

6. The water treatment condition monitoring system according to claim 1, characterized in that: The information processing device is configured to calculate a predicted turbidity of the treated water as information indicating whether the aggregation state of the turbidity-causing substances in the treated water presented in the determination image is good.

7. The water treatment condition monitoring system according to claim 1, characterized in that: The information processing device is configured to calculate either a good or a bad determination result as information indicating whether the aggregation state of the turbidity-causing substances in the treated water presented in the determination image is good.

8. The water treatment condition monitoring system according to claim 1, characterized in that: The learning input data only includes the image data of the treated water. The determination input data only includes the determination image.

9. The water treatment condition monitoring system according to claim 1, characterized in that: The information processing device obtains the treated water quality parameter from the treated water measuring device as the determination treated water quality parameter. The learning input data further includes the treated water data. The determination input data further includes the determination treated water quality parameter.

10. The water treatment condition monitoring system according to claim 1, characterized in that: The information processing device is configured to obtain learning floc data including information on flocs formed by aggregates of the turbidity-causing substances, i.e., floc information, in each image including the image data of the treated water, and obtain determination floc information including information on flocs formed by aggregates of the turbidity-causing substances from the determination image. The learning input data further includes the learning floc data. The determination input data further includes the determination floc information.

11. The water treatment condition monitoring system according to claim 1, characterized in that: The information processing device is configured to Based on the treated water quality parameter, determine whether the treated water data meets the conditions for promoting the growth of aggregates of the turbidity-causing substances, i.e., flocs. When the treated water quality parameter meets the conditions for promoting the growth of the flocs, compared with the case where the treated water quality parameter does not meet the conditions for promoting the growth of the flocs, select the aggregation quality determination model constructed using the image data captured at an upstream location.

12. A water treatment condition monitoring system, which is applied to a facility for generating treated water. The facility generates the treated water obtained by purifying the water to be treated by sequentially transporting the water to be treated to a plurality of treatment sites and performing a treatment including a coagulation and sedimentation treatment for removing turbidity substances contained in the water to be treated. It is characterized in that, Comprising: A plurality of water sampling devices that sample the treated water at a plurality of locations where the treatment elapsed time of the treated water is different from each other. A plurality of imaging devices that obtain water sampling image data of the treated water at a plurality of locations by respectively imaging the treated water sampled by the plurality of water sampling devices, wherein the water sampling image data includes an image of the treated water at each location, i.e., a water sampling image. A treated water measuring device that measures the treated water quality parameter including the treated water turbidity of the treated water and obtains the treated water data including the treated water quality parameter. A treated water measuring device that measures the treated water turbidity of the treated water and obtains the treated water data including the treated water turbidity. And An information processing device that obtains the water sampling image data, the treated water data, and the treated water data from the imaging device, the treated water measuring device, and the treated water measuring device, wherein The information processing device is configured to use the learning input data including the water sampling image data of the water to be treated and the treated water data as learning data, and through machine learning, construct a coagulation quality determination model for the water sampling image data of the water to be treated at each site respectively, wherein the coagulation quality determination model is used to output information indicating whether the coagulation state of the turbidity substances in the water to be treated is good based on the input data including the water sampling image of the water to be treated. The information processing device is configured to select at least one of the multiple coagulation quality determination models respectively constructed for the water sampling image data of the water to be treated corresponding to each site based on the turbidity of the water to be treated, obtain the water sampling image of the water to be treated as a judgment image from the imaging device, and use the selected coagulation quality determination model to calculate information indicating whether the coagulation state of the turbidity substances in the water to be treated presented in the judgment image is good according to the judgment input data including the judgment image.

13. A water treatment condition monitoring method, which is applied to a facility for generating treated water. The facility generates the treated water obtained by purifying the water to be treated by sequentially transporting the water to be treated to a plurality of treatment sites and performing a treatment including a coagulation and sedimentation treatment for removing turbidity substances contained in the water to be treated. It is characterized in that, Comprising: Using a plurality of imaging devices provided at a plurality of sites where the water to be treated can be photographed and the treatment elapsed times of the water to be treated are different from each other, photographing the water to be treated to obtain the image data of the water to be treated at each site, wherein the image data includes the images of the water to be treated photographed from each site; Using a water to be treated measuring device to measure the water quality parameters of the water to be treated including the turbidity of the water to be treated, and obtaining the water to be treated data including the water quality parameters of the water to be treated; Using a treated water measuring device to measure the turbidity of the treated water, and obtaining the treated water data including the turbidity of the treated water; And Using an information processing device to obtain the image data, the water to be treated data, and the treated water data from the imaging device, the water to be treated measuring device, and the treated water measuring device, wherein, Using the information processing device, Using the learning input data including the image data of the water to be treated and the treated water data as learning data, and constructing a plurality of coagulation quality determination models through machine learning, wherein the coagulation quality determination model is used to output information indicating whether the coagulation state of the turbidity substances in the water to be treated is good based on the input data including the image of the water to be treated. Select at least one of the multiple coagulation quality determination models based on the turbidity of the water to be treated, obtain the image of the water to be treated as a judgment image from at least one of the multiple imaging devices, and use the selected coagulation quality determination model to calculate information indicating whether the coagulation state of the turbidity substances in the water to be treated presented in the judgment image is good according to the judgment input data including the judgment image.

Images