WO2022025360A1

WO2022025360A1 - Apparatus and system for measuring water quality on basis of ai learning scheme and floating matter vector information

Info

Publication number: WO2022025360A1
Application number: PCT/KR2020/017970
Authority: WO
Inventors: 이동욱
Original assignee: 주식회사 에스엠티
Priority date: 2020-07-28
Filing date: 2020-12-09
Publication date: 2022-02-03

Abstract

Provided are an apparatus and a system for measuring water quality on the basis of an AI learning scheme and floating matter vector information. The water quality measurement system comprises a water quality measurement apparatus and a deep learning server. The water quality measurement apparatus comprises: an image capturing unit for capturing an image of fluid flowing therein; and a control unit which is electrically connected to the image capturing unit and which acquires information about at least one from among a turbidity level, pH, and floating matter of the fluid from the image on the basis of a pre-trained machine learning algorithm, wherein the control unit transmits the information about the at least one from among a turbidity level, pH, and floating matter to the deep learning server, and the deep learning server updates the pre-trained machine learning algorithm on the basis of the received training information.

Description

Water quality measurement device and system based on AI learning method and floating vector information

The present invention relates to an apparatus and system for measuring water quality, and more particularly, to an apparatus and system for determining water quality based on AI learning method and floating vector information.

Water is an essential element for human life. In particular, the act of drinking water or washing with water is an essential act for human survival or social life.

Therefore, if the quality of water that humans drink but come into direct contact with is excessively degraded, it may have a detrimental effect on humans as well.

However, when viewed with the human eye, the actual water quality may not be good even for normal water, and it is difficult to visually check the pipe through which the fluid flows because it is sealed.

Accordingly, there is a need for a device or system capable of providing information on whether water to be directly touched or to be drunk is safe.

The prior art document (Republic of Korea Patent Publication No. 10-2120427) discloses a pipe-attached water quality sensor having a structure that measures water quality through residual chlorine concentration, electrical conductivity, and water temperature information, and then provides it to a user.

However, since residual chlorine concentration, electrical conductivity, and water temperature information are not information directly indicating water quality, and information on floating matter cannot be identified, a problem of reduced reliability of measurement may occur.

An object of the present invention is to provide an apparatus and system for measuring water quality having a structure capable of solving the above-described problems.

Another object of the present invention is to provide a water quality measuring device and a water quality measuring system including the same, which are installed at a place where the fluid flows and can provide information on the water quality of the fluid to a user.

In addition, an object of the present invention is to provide a water quality measuring device and a water quality measuring system including the same, in which information on the water quality of a fluid is analyzed by a machine learning algorithm and provided to a user.

In addition, an object of the present invention is to provide a water quality measurement device in which a machine learning algorithm for analyzing the water quality of a fluid can be continuously updated by learning information of the deep learning algorithm, and a water quality measurement system including the same.

Another object of the present invention is to provide a water quality measuring device that is switched to and driven in a low power mode when the fluid does not flow, and a water quality measuring system including the same.

Another object of the present invention is to provide a water quality measuring apparatus capable of providing water quality information in a visually recognizable form to a user or a user terminal, and a water quality measuring system including the same.

In addition, an object of the present invention is to provide a water quality measuring device that can be driven through self-generation using the kinetic energy of a fluid and a water quality measuring system including the same.

In addition, the water quality is primarily determined by the machine learning algorithm, and the water quality is secondarily determined as the acquired image data is analyzed based on the deep learning algorithm, and the determined water quality can be provided to the user. Its purpose is to provide

Another object of the present invention is to provide a water quality measurement system that learns based on data received from a plurality of water quality measurement devices and updates a machine learning algorithm of the water quality measurement device based on the learned data.

Another object of the present invention is to provide a water quality measurement system that updates a machine learning algorithm of a water quality measurement device at preset time intervals.

The problems to be solved by the present invention are not limited to the problems mentioned above, and other problems not mentioned will be clearly understood by those skilled in the art from the following description.

In order to solve the above problems, the water quality measuring device according to an embodiment of the present invention is configured to measure the water quality of a fluid flowing in an internal space, and is a water quality measuring device energably connected to a deep learning server, The water quality measuring device includes: an image capturing unit for capturing an image of a fluid flowing therein; and and a control unit connected to the image capturing unit to be energized and configured to acquire information about at least one of turbidity, pH, and floating matter of the fluid from the image based on a previously learned machine learning algorithm.

In addition, the control unit determines the water quality based on the information on at least one of the turbidity, pH, and float, and transmits information on at least one of the turbidity, pH, and float to the deep learning server, the deep learning The previously learned machine learning algorithm is updated based on the learning information received from the server.

In addition, it is connected to the energization of the control unit, further comprising a flow sensor for detecting the flow of the fluid in the inner space, wherein the control unit, when the flow of the fluid is not detected by the flow sensor, the Control the water quality measurement device to switch to the low power mode.

In addition, the water quality measuring device is energably connected to the user terminal, and when the connection with the deep learning server is blocked, the control unit transmits the determined information on the water quality to the user terminal.

In addition, it may further include a notification unit connected to the control unit energized and providing a visually recognizable danger signal to the user.

Also, when the determined water quality is lower than a preset reference water quality, the controller may control the notification unit to provide the danger signal.

In addition, the water quality measurement device may further include a power generation unit configured to convert the kinetic energy of the flowing fluid into electrical energy to supply the electrical energy to the water quality measurement device.

In addition, the water quality measuring system according to an embodiment of the present invention, the water quality measuring device according to the embodiment of the present invention; and a deep learning server energably connected to the water quality measurement device.

In addition, the deep learning server, based on a Convolutional Neural Network (CNN) algorithm, analyzes information on at least one of the turbidity and pH, and based on a Recurrent Nueral Network (RNN) algorithm, information on the float Analyze and learn by generating learning data based on the analyzed results.

In addition, the learning information may be information learned by the deep learning server by the learning data.

In addition, the deep learning server determines the water quality based on the analyzed result, and when the connection between the water quality measuring device and the deep learning server is maintained, the deep learning server determines the water quality based on the analyzed result transmits information on to the user terminal, and when the connection between the water quality measuring device and the deep learning server is blocked, the water quality measuring device is information is transmitted to the user terminal.

In addition, the Recurrent Nueral Network (RNN) algorithm may be an algorithm based on vector information of both ends of the recognized floating object and a midpoint between the two end points.

In addition, the deep learning server may transmit the learning information to the water quality measuring device at preset time intervals.

Other specific details of the invention are included in the detailed description and drawings.

According to the above-described solution, the following effects can be derived.

First, water quality information about a fluid to drink or contact may be provided to a user.

In addition, since the water quality determination of the water quality measuring device is performed by a machine learning algorithm previously learned, relatively little power is consumed and the size of the water quality measuring device can be miniaturized. Accordingly, the restriction of the installation place is reduced, and further, it can be installed in various places of the flow path used by the user in real life.

In addition, since the machine learning algorithm is continuously updated by the learning information of the deep learning algorithm, the accuracy of water quality determination can be gradually increased.

In addition, when the fluid does not flow, it is switched to a low power mode and driven, so that electrical energy consumed for driving the water quality measuring device is reduced, thereby reducing the size of the water quality measuring device. In addition, since consumed electrical energy is reduced, the water quality measuring device can be driven only by self-generation using the kinetic energy of the fluid.

In addition, regardless of whether the connection to the deep learning server is connected, information on water quality judgment can be continuously provided to the user.

In addition, since the water quality is primarily determined by the machine learning algorithm and secondarily determined by the deep learning algorithm, accurate water quality information can be provided to the user.

In addition, since the deep learning server is learned based on data received from a plurality of water quality measurement devices, the accuracy of water quality determination can be improved as the number of users and usage time increase.

In addition, since the update of the machine learning algorithm is performed at preset time intervals rather than in real time, electrical energy consumed for driving the water quality measuring device can be reduced.

Effects of the present invention are not limited to the effects mentioned above, and other effects not mentioned will be clearly understood by those skilled in the art from the following description.

1 is a block diagram showing the configuration of a water quality measurement system according to an embodiment of the present invention.

FIG. 2 is a block diagram illustrating a specific configuration of the water quality measuring device and the deep learning server according to FIG. 1 .

3 is a perspective view illustrating an embodiment of the water quality measuring device according to FIG. 1 .

4 is a perspective view illustrating another embodiment of the water quality measuring device according to FIG. 1 .

5 is a flowchart specifically illustrating an operation process of the water quality measuring device according to FIG. 1 .

6 is a flowchart specifically illustrating an operation process of the deep learning server according to FIG. 1 .

7 is a flowchart specifically illustrating an operation process of the water quality measurement system according to FIG. 1 .

8 is a flowchart illustrating an example of an RNN algorithm according to the present invention.

Advantages and features of the present invention and methods of achieving them will become apparent with reference to the embodiments described below in detail in conjunction with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below, but may be implemented in various different forms, and only the present embodiments allow the disclosure of the present invention to be complete, and those of ordinary skill in the art to which the present invention pertains. It is provided to fully understand the scope of the present invention to those skilled in the art, and the present invention is only defined by the scope of the claims.

The terminology used herein is for the purpose of describing the embodiments and is not intended to limit the present invention. In this specification, the singular also includes the plural unless specifically stated otherwise in the phrase. As used herein, “comprises” and/or “comprising” does not exclude the presence or addition of one or more other components in addition to the stated components. Like reference numerals refer to like elements throughout, and "and/or" includes each and every combination of one or more of the recited elements. Although "first", "second", etc. are used to describe various elements, these elements are not limited by these terms, of course. These terms are only used to distinguish one component from another. Accordingly, it goes without saying that the first component mentioned below may be the second component within the spirit of the present invention.

Unless otherwise defined, all terms (including technical and scientific terms) used herein will have the meaning commonly understood by those of ordinary skill in the art to which this invention belongs. In addition, terms defined in a commonly used dictionary are not to be interpreted ideally or excessively unless specifically defined explicitly.

The term “energized” used below means electrically connected to each other or connected to enable information communication.

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

Referring to FIG. 1 , the components of a water quality measurement system according to an embodiment of the present invention are shown.

The water quality measurement system according to the present embodiment includes a water quality measurement device configured to measure water quality, and a deep learning server 200 and a user terminal 300 that receives information about the measured water quality.

The user terminal 300 is communicatively connected to the water quality measuring device 100 and the deep learning server 200, and receives information on water quality from them.

In one embodiment, the user terminal 300 is a computer, UMPC (Ultra Mobile PC), workstation, net-book (net-book), PDA (Personal Digital Assistants), portable (portable) computer, web tablet (web tablet), It may be one of electronic devices such as a wireless phone, a mobile phone, a smart phone, and a portable multimedia player (PMP).

However, the present invention is not limited thereto, and various electronic devices capable of receiving information from the water quality measuring apparatus 100 and the deep learning server 200 and providing the received information in a visually recognizable form to the user may be employed. .

Hereinafter, detailed configurations of the water quality measuring apparatus 100 and the deep learning server 200 will be described with reference to FIGS. 2 to 4 .

1. 본 발명의 실시 예에 따른 수질 측정 장치(100)의 설명1. Description of the water quality measuring device 100 according to an embodiment of the present invention

The water quality measuring apparatus 100 has a predetermined space provided therein, and acquires image information about a fluid flowing in the predetermined space.

In addition, the water quality measuring apparatus 100 is configured to determine the water quality of the fluid by using the obtained image information.

To this end, the water quality measuring device 100 includes a control unit 110 , an image capturing unit 120 , a flow sensor 130 , a notification unit 140 , a database unit 150 , a communication unit 160 , and a power supply unit 170 . ) and a power generation unit 180 may be included.

The components of the water quality measuring device 100 are electrically connected to each other. For example, communication between the control unit 110 and the database unit 150 may be performed through a wired/wireless network. A wired/wireless network may use standard communication technologies and/or protocols.

In addition, various functions of each configuration described below may be controlled by the controller 110 .

The control unit 110 may also be referred to as a processor, a controller, a microcontroller, a microprocessor, a microcomputer, etc., and the control unit 110 is hardware or It may be implemented by firmware, software, or a combination thereof.

In addition, information obtained or calculated by the components of the water quality measuring apparatus 100 described below may be stored in the database unit 150 and then transferred to other components.

In one embodiment, the database unit 190 is a flash memory type (flash memory type), hard disk type (hard disk type), multimedia card micro type (multimedia card micro type), card type memory (for example, SD or XD memory, etc.), RAM (RAM, Random Access Memory) SRAM (Static Random Access Memory), ROM (Read-Only Memory), EEPROM (Electrically Erasable Programmable Read-Only Memory), PROM (Programmable Read-Only Memory) It may include at least one type of storage medium among a magnetic memory, a magnetic disk, and an optical disk.

Hereinafter, the components of the water quality measuring device 100 for measuring the water quality will be described in detail.

(1) Description of the image capturing unit 120

The image capturing unit 120 is configured to detect image information about a fluid flowing in the space inside the water quality measuring device 100 .

The image capturing unit 120 may be implemented as a variety of devices capable of continuously sensing preset image information on a flowing fluid. In an embodiment, the image capturing unit 120 may be implemented as a small image camera.

In an embodiment, in order to obtain a clear image of a flowing fluid, the image capturing unit 120 may include a light source module (not shown).

The image information detected by the image capturing unit 120 is analyzed by a pre-learned machine learning algorithm provided in the control unit 110 . In this regard, it will be described in detail later.

(2) Description of the flow sensor 130

The flow sensor 130 detects the flow of fluid in the space inside the water quality measuring device 100 . The flow sensor 130 may be implemented as a mechanical type, an electronic type, or a combination thereof. In an embodiment, the flow sensor 130 may be configured as a sensor for detecting the water level.

When the flow of the fluid is not detected by the flow sensor 130 , the control unit 110 may convert the water quality measuring apparatus 100 to a low power mode to reduce power consumption. In this regard, it will be described in detail later.

Since the power consumed by the water quality measuring device 100 is reduced, the water quality measuring device 100 can be downsized. Accordingly, the fluid pipe provided at home, the water quality measuring device 100 may be coupled to the end of the fluid pipe.

(3) Description of the notification unit 140

The notification unit 140 is configured to notify the user of the water quality in a visually recognizable form when the water quality is lower than a preset reference water quality.

In one embodiment, the notification unit 140 may be in the form of notifying that the water quality is lower than a preset reference water quality using LED light emission.

The notification unit 140 may include a light source module that emits light when the control unit 110 determines that the water quality is lower than the preset reference water quality. The light source module may be provided in plurality, and as water quality deteriorates step by step, each light source module may be configured to emit light corresponding to each step.

Through this, information on water quality can be recognized immediately.

Furthermore, it can be prevented that the user comes into contact with or drinks a fluid harmful to the human body.

Also, the notification unit 140 may include a notification unit in the form of a display device. In this case, when the water quality is lower than the preset reference water quality, the notification unit 140 may display visual information indicating that the water quality is dangerous under the control of the controller 110 . The visual information may include at least one of a numerical value, text, and image indicating the degree of water quality, and may include a current image of the fluid captured by the image capturing unit 120 .

However, as will be described later, since information on water quality is directly provided to the user terminal 300 , the water quality measuring apparatus 100 may be configured except for the notification unit 140 .

(3) Description of the power supply unit 170 and the power generation unit 180

The power supply unit 170 is configured to supply the stored electrical energy to each component of the water quality measurement device 100 . In an embodiment, the power supply unit 170 may be implemented in the form of a secondary battery (battery). However, the present invention is not limited thereto.

In addition, the power generation unit 180 is configured to convert the kinetic energy of the flowing fluid into electrical energy and supply it to each component of the water quality measuring device 100 . In one embodiment, the power generation unit 180 may be implemented in various forms for generating electromotive force by electromagnetic induction by using the kinetic energy of the fluid.

In an embodiment, the electrical energy generated by the power generation unit 180 may be stored in the power supply unit 170 .

In an embodiment, the power generation unit 180 has its own electrical energy storage module (not shown), and the electrical energy generated by the power generation unit 180 may be stored in the electrical energy storage module.

The water quality measuring apparatus 100 according to the present embodiment determines the water quality by a pre-learned machine learning algorithm that consumes relatively little power, and is converted to a low power mode when the flow of the fluid is not detected.

Accordingly, the water quality measuring apparatus 100 may be driven by the power supply unit 170 and the power generation unit 180 having a relatively small capacity.

That is, the water quality measuring apparatus 100 may be driven by itself without external power supply.

In an embodiment, the water quality measuring device 100 may be driven by the power generation unit 180 .

(4) Description of the control unit 110

The control unit 110 is configured to receive image information from the image capturing unit 120 to determine water quality.

The water quality may be determined through the machine learning module 111 and the water quality determination module 114 .

The machine learning module 111 includes a turbidity analysis unit 1111 , a pH analysis unit 1112 , and a floating matter analysis unit 1113 that analyze image information based on a previously learned machine learning algorithm.

The turbidity analysis unit 1111 analyzes the turbidity of the fluid included in the received image.

In an embodiment, the turbidity analysis unit 1111 may analyze the turbidity by obtaining values for the three primary colors of light, Red, Green, and Blue, from pixels included in the image information. have. For example, if the value for the red/green/blue color is increased, it is determined that the turbidity is high, and the value for the red/green/blue color is When this is lowered, it can be determined that the turbidity is low. However, the present invention is not limited thereto.

The pH analysis unit 1112 analyzes the pH of the fluid included in the received image.

In an embodiment, the pH analysis unit 1112 may perform pH analysis based on a machine learning algorithm learned by analyzing RGB patterns included in image information of fluids having different pH values. However, the present invention is not limited thereto.

The floating matter analysis unit 1113 analyzes the floating matter of the fluid included in the received image.

In an embodiment, the floating material analysis unit 1113 may analyze the floating material based on the machine learning algorithm learned by analyzing the pattern of the floating material included in the image. For example, floating material analysis may be performed based on the learned algorithm by separating the patterns in the case of live larvae, dead larvae, and sludge. That is, information on the number of larvae and sludge included in the image can be obtained. However, the present invention is not limited thereto.

The information analyzed by the machine learning module 111 is transmitted to the deep learning server 200 through the communication unit 160 .

The deep learning server 200 receives the information analyzed from the plurality of water quality measurement devices 100 and learns by the deep learning algorithm, and based on the learned data, the machine learning algorithm of the machine learning module 111 can be updated. have.

Accordingly, as the number of users who use the water quality measuring apparatus 100 increases and the period of use thereof increases, the accuracy of the machine learning algorithm applied to the water quality measuring apparatus 100 may be improved.

The water quality determination module 114 determines water quality using the analysis information received from the machine learning module 111 .

In an embodiment, when the analyzed turbidity value is lower than a preset reference turbidity value, the water quality determination module 114 may determine that the fluid is harmful to the user (not drinkable or unusable).

In one embodiment, when the analyzed pH value is lower than or higher than the preset reference turbidity value, the water quality determination module 114 may determine that the fluid is harmful to the user (not drinkable or unusable).

In an embodiment, when the analyzed floating material information is higher than the preset floating material standard, the water quality determination module 114 may determine that the fluid is harmful to the user (not drinkable or not usable). The preset floating object standard may be defined as the number of inanimate objects/living matter.

In an embodiment, the water quality determination module 114 may be included in the machine learning module 111 .

In addition, the control unit 110 transmits information on water quality to the deep learning server 200, and the deep learning server 200 uses the received information to determine the water quality secondarily and then transmits it to the user.

The pre-processing module 112 pre-processes image information, information analyzed by the machine learning module 111, and information determined by the water quality determination module 114 in order to transmit data to the deep learning server 200 . The pre-processed information may be used for analysis and learning of the deep learning server 200 .

When communication with the deep learning server 200 is blocked, the control unit 110 provides information about the water quality determined by the water quality determination module 114 to the user.

In an embodiment, the control unit 110 may control the notification unit 140 to transmit information on water quality to the user in a visually recognizable form. For example, the notification unit 140 may be controlled to emit light of a preset color.

In an embodiment, the controller 110 may transmit information on water quality to the user terminal 300 . The user terminal 300 receives information on water quality and provides it to the user in a visually recognizable form. In an embodiment, information on water quality may be displayed in the user terminal 300 .

Regardless of the connection state between the deep learning server 200 and the water quality measurement device 100 , information on water quality may be provided to the user. As a result, the user may be provided with information on water quality as long as the water quality measuring apparatus 100 is operating.

In addition, the controller 110 controls the water quality measuring device 100 to switch to the low power mode when the flow of the fluid is not detected by the flow sensor 130 .

When the flow of the fluid is not detected by the flow sensor 130 , the mode conversion module 113 may convert the water quality measuring apparatus 100 to a low power mode. As described above, since the amount of power consumed to drive the water quality measuring device 100 is reduced, the water quality measuring device 100 can be downsized. In addition, it may be continuously driven by the relatively small power supply unit 170 and the power generation unit 180 .

Information sensed by the water quality measuring device 100 or calculated by the controller 110 may be transmitted to and stored in the database 150 .

In an embodiment, the control unit 110 may pre-process the analysis information and determination information stored in the database unit 150 for a preset time and transmit the pre-processing to the deep learning server 200 .

In an embodiment, the control unit 110 may transmit the analysis information and the determination information to the deep learning server 200 in real time.

(5) Description of the structure of the water quality measuring device 100

3 and 4 , an embodiment of the structure of the water quality measuring apparatus 100 is shown. However, the present invention is not limited thereto, and the water quality measuring apparatus 100 may be implemented in various shapes.

Referring to FIG. 3 , the water quality measuring device 100 is coupled to the end of the fluid pipe through which the fluid flows and is configured to measure the water quality of the fluid immediately before being supplied to the user.

The water quality measuring device 100 according to FIG. 3 includes a body part 105 constituting the outer shape of the water quality measuring device 100 .

The body portion 105 has one end coupled to the end of the fluid pipe, and includes a columnar gripper extending in the only direction.

The grip part is formed through the inside so that the fluid can pass through the grip part, and a filter for filtering the fluid passing through the grip part may be provided in the inner space.

At the other end of the gripping portion, a head portion having a predetermined internal space is formed. That is, the body portion 105 includes a grip portion and a head portion. The inner space 105a of the head part communicates with the space of the grip part.

In addition, an injection hole for injecting the fluid in the internal space to the outside is formed on one side of the head part. The injection hole is formed through the inner space of the head toward the outside. The injection hole is provided in plurality, and various patterns may be formed.

The internal space 105a of the head part is divided into a first space in which a fluid flows and a second space in which the fluid does not come into contact. The second space may include an image capture unit 120 , a power supply unit 170 , and a control unit 110 . In an embodiment not shown, the communication unit 160 , the database unit 150 , and the power generation unit 180 may be disposed in the second space.

Also, in an embodiment not shown, the flow sensor 130 may be disposed in the first space.

Also, in an embodiment not shown, the notification unit 140 may be disposed at a position that can be visually recognized by the user. That is, at least a part of the notification unit 140 may be disposed to protrude from the outer surface of the head unit or the filter unit.

Referring to FIG. 4 , the water quality measuring device 100 is coupled to the middle of the fluid pipe through which the fluid flows and is configured to measure the water quality of the fluid before being supplied to the user.

The water quality measuring device 100 according to FIG. 4 includes a body portion 105 that forms an outer shape of the water quality measuring device 100 .

A predetermined space 105a through which a fluid can flow is formed in the body 105 .

In addition, in the direction in which the fluid flows, fluid pipes are respectively coupled to both sides of the body 105 . Each fluid pipe and the internal space 105a of the body 105 are communicatively connected to each other, whereby the fluid pipes on both sides communicate with each other through the internal space 105a of the body 105 .

One side of the body 105 is formed to transmit light, and the image capturing unit 120 is disposed at a position facing the one side of the body 105 .

In the illustrated embodiment, the image capturing unit 120 is formed to protrude from the outer surface of the body portion 105 in a “L” shape. An image of a fluid flowing through a portion of the image capturing unit 120 facing one side of the body 105 may be detected.

In addition, the communication unit 160 is formed to protrude from the outer surface of the body portion (105).

In an embodiment not shown, the power supply unit 170 and the power generation unit 180 may be disposed at positions that do not come into contact with the fluid.

In an embodiment not shown, the flow sensor 130 may be disposed in the inner space 105a of the body 105 .

In an embodiment not shown, the notification unit 140 may be disposed at a location that is visually recognizable to the user. For example, the notification unit 140 may be disposed such that at least a part thereof is exposed to the outside of the body unit 105 .

2. 본 발명의 실시 예에 따른 딥러닝 서버(200)의 설명2. Description of the deep learning server 200 according to an embodiment of the present invention

The deep learning server 200 receives information pre-processed by the control unit 110 of the water quality measurement device 100 . The deep learning server 200 secondaryly analyzes the received information based on the deep learning algorithm and determines the water quality.

Information including determination of water quality may be provided to the user through the user terminal 300 .

Referring to FIG. 2 , the deep learning server 200 may include a control unit 210 , a communication unit 220 , and a database unit 230 .

The components of the deep learning server 200 are electrically connected to each other. For example, communication between the control unit 210 and the database unit 230 may be performed through a wired/wireless network. A wired/wireless network may use standard communication technologies and/or protocols.

In addition, various functions of each configuration described below may be controlled by the controller 210 .

The control unit 210 may also be referred to as a processor, a controller, a microcontroller, a microprocessor, a microcomputer, etc., and the control unit 110 is hardware or It may be implemented by firmware, software, or a combination thereof.

In addition, information obtained or calculated by the components of the deep learning server 200 described below may be stored in the database unit 230 and then transferred to other components.

In one embodiment, the database unit 230 is a flash memory type (flash memory type), a hard disk type (hard disk type), a multimedia card micro type (multimedia card micro type), a card type memory (for example, SD or XD memory, etc.), RAM (RAM, Random Access Memory) SRAM (Static Random Access Memory), ROM (Read-Only Memory), EEPROM (Electrically Erasable Programmable Read-Only Memory), PROM (Programmable Read-Only Memory) It may include at least one type of storage medium among a magnetic memory, a magnetic disk, and an optical disk.

Below, the control unit 210 of the deep learning server 200 will be described in detail.

The control unit 210 includes a deep learning module 211 that analyzes the information transmitted and pre-processed by the water quality measurement device 100 .

The deep learning module 211 includes a CNN analysis unit 2111 configured to analyze information on at least one of turbidity and pH, based on a Convolutional Neural Network (CNN) algorithm.

In an embodiment, the CNN algorithm may include a convolution layer, a pooling layer, a ReLu layer, and a fully connected layer.

In an embodiment, the CNN analysis unit 2111 uses at least one of RGB information included in the image information of the fluid, information analyzed by the water quality measurement device 100, and determination information on water quality for turbidity and pH values. analysis can be performed.

Further, the deep learning module 211 includes an RNN analysis unit 2112, configured to analyze information about the float, based on a Recurrent Nueral Network (RNN) algorithm.

In an embodiment, the RNN analysis module 2112 may analyze information on the floating object based on the skeleton vector of both ends and the middle point of the floating object.

In the embodiment, the RNN analysis module 2112 extracts the floating object from the image, and then sets both ends of the floating object and a midpoint between the two end points.

When both end points and midpoints are set, the RNN analysis module 2112 determines whether the vectors of both end points are simultaneously increased or decreased in the same or opposite directions.

When the vectors of both ends are simultaneously increased or decreased in the same or opposite directions, the RNN analysis module 2112 determines whether the vectors of the midpoints are changed. If the vector at the midpoint is changed, the extracted float can be classified as a living larva. On the other hand, if the vector at the midpoint does not change, the extracted suspension may be classified as sludge.

When the vectors of both end points are not simultaneously increased or decreased in the same or opposite directions, the RNN analysis module 2112 determines whether the distance between the both end points is equal to or greater than a preset length. For example, the preset length may be 3 mm. When the distance between the both ends is less than a preset length, the extracted suspended matter may be classified as sludge.

When the distance between the two end points is equal to or greater than a preset length, the RNN analysis module 2112 determines whether the vectors of both end points point in the same direction. If not directed in the same direction, the extracted suspension may be classified as sludge. On the other hand, if they face the same direction, the extracted float can be classified as a dead larva. Since the information sensed by the image capturing unit 120 is secondaryly analyzed by the deep learning module 211, more precisely determined information on turbidity, pH, and suspended matter can be provided to the user.

Further, the deep learning module 211 may include a learning data generating unit 2113 that generates learning data based on the information analyzed by the CNN analysis unit 2111 and the RNN analysis unit 2112 .

The deep learning algorithm of the deep learning module 211 is learned based on the generated learning data. In an embodiment, the weights of the CNN algorithm and the RNN algorithm are corrected by the generated learning data, whereby the algorithms for turbidity analysis, pH analysis, and floating matter analysis may be modified.

The update module 212 transmits the learning information learned by the learning data to the machine learning module 111 of the water quality measurement device 100, and each machine learning algorithm of the machine learning module 111 is based on the received learning information. can be updated.

Accordingly, the accuracy of the analysis of the water quality measuring apparatus 100 may be continuously improved.

When the learning information is received by the water quality measuring device 100 , the mode conversion module 113 controls the water quality measuring device 100 to be switched from the low power mode to the normal state.

In an embodiment, the update module 212 may transmit learning information at preset time intervals.

Through this, the time for which the water quality measuring apparatus 100 is driven in the low power mode may be increased, and thus, the amount of power consumed for driving the water quality measuring apparatus 100 may be reduced.

In addition, the control unit 210 may include a water quality determination module 213 that determines the water quality by using the information analyzed by the deep learning module 211 .

The water quality determination module 213 determines that the fluid is harmful to the user when the analyzed turbidity, pH, and suspended matter do not meet preset criteria. Specific details may be understood with reference to the description of the water quality determination module 114 of the water quality measuring apparatus 100 .

In an embodiment, the deep learning module 211 may include a water quality determination module 213 .

In an embodiment, the control unit 210 may transmit information on water quality to the user terminal 300 . The user terminal 300 receives information on water quality and provides it to the user in a visually recognizable form. In an embodiment, information on water quality may be displayed in the user terminal 300 .

3. 본 발명의 실시 예에 따른 수질 측정 시스템의 작동과정에 대한 설명3. Description of the operation process of the water quality measurement system according to an embodiment of the present invention

5 to 7 , a detailed flow of the operation of the water quality measurement system according to an embodiment of the present invention is shown.

Referring to FIG. 5 , a detailed flow S100 of the operation process of the water quality measuring device 100 is illustrated.

First, the control unit 110 determines whether the flow is detected by the flow sensor 130 (S110).

When the flow is not detected, the mode conversion module 113 of the control unit 110 controls the water quality measuring apparatus 100 to be switched to the low power mode (S190).

When a flow is detected, the controller 110 activates the image capturing unit 120 , and based on the image information of the fluid flowing in the water quality measuring device 100 captured through the activated image capturing unit 120 . , the fluid is sensed (S120). That is, in the present invention, unnecessary power consumption can be prevented by activating the image capturing unit 120 such as a camera only when a flow is detected.

When the image information is detected, the control unit 110 performs a primary analysis on turbidity, pH, and suspended matter based on a machine learning algorithm previously learned through the machine learning module 111 (S130).

When the analysis is completed, the control unit 110 determines the water quality based on the primary analysis result through the water quality determination module 114 (S140).

In addition, the control unit 110 checks whether the water quality measurement device 100 and the deep learning server 200 are communicatively connected (S150), and when the connection is blocked, the water quality determination result is visually recognizable to the user. provided (S160).

The water quality determination result may be provided to the user through the notification unit 140 or may be provided to the user through the user terminal 300 .

When the deep learning server 200 and the water quality measurement device 100 are communicatively connected, the control unit 110 pre-processes image information and information analyzed and determined by the control unit 110 (S170).

In addition, the control unit 110 transmits the pre-processed information to the deep learning server 200 through the communication unit 160 (S180).

Referring to FIG. 6 , a detailed flow S200 of the operation process of the deep learning server 200 is shown.

First, the deep learning server 200 receives the pre-processed data from the water quality measurement device (S210).

When data is received, the deep learning server 200 determines what type of data the preprocessed data is (S220).

In the case of turbidity or pH data, the deep learning server 200 secondaryly analyzes it based on the CNN algorithm (S224). In addition, in the case of floating data, the deep learning server 200 secondaryly analyzes it based on the RNN algorithm (S226).

When the analysis is completed, the deep learning server 200 generates training data based on the analyzed data, and uses this to revise the algorithm (S228). That is, the deep learning server 200 is learned based on the learning data.

When the learning of the deep learning server 200 is completed, the machine learning module 111 of the water quality measuring apparatus 100 is updated based on the learning information (S230).

In addition, the deep learning server 200 determines the water quality based on the secondary analysis result and provides it to the user (S240). The water quality determination result may be provided to the user through the user terminal 300 .

Referring to FIG. 7 , a detailed flow of the operation process of the water quality measuring device 100 and the deep learning server 200 is shown.

Referring to FIG. 8 , an example of a classification method by the RNN analysis module 2112 is illustrated.

In the illustrated embodiment, the RNN analysis module 2112 extracts a floating object from the image (S2261), and sets both ends of the floating object and a midpoint between the two end points (S2262).

When both end points and the midpoint are set, the RNN analysis module 2112 determines whether the vectors of both end points are simultaneously increased or decreased in the same or opposite directions (S2263).

When the vectors at both ends are simultaneously increased or decreased in the same or opposite directions, the RNN analysis module 2112 determines whether the vectors at the midpoints are changed ( S2264 ). When the vector of the intermediate point is changed, the extracted float may be classified as a living larva (S2267). On the other hand, when the vector of the intermediate point does not change, the extracted suspended matter may be classified as sludge (S2268).

When the vectors of both end points are not simultaneously increased or decreased in the same or opposite directions, the RNN analysis module 2112 determines whether the distance between the both end points is equal to or greater than a preset length (S2265). For example, the preset length may be 3 mm. When the distance between the two ends is less than a preset length, the extracted floating matter may be classified as sludge (S2268).

When the distance between the two end points is equal to or greater than a preset length, the RNN analysis module 2112 determines whether the vectors of both end points point in the same direction (S2266). If they do not face the same direction, the extracted suspended matter may be classified as sludge (S2268). On the other hand, when facing the same direction, the extracted float may be classified as a dead larva (S2269).

The steps of a method or algorithm described in connection with an embodiment of the present invention may be implemented directly in hardware, as a software module executed by hardware, or by a combination thereof. A software module may include random access memory (RAM), read only memory (ROM), erasable programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), flash memory, hard disk, removable disk, CD-ROM, or It may reside in any type of computer-readable recording medium well known in the art to which the present invention pertains.

In the above, embodiments of the present invention have been described with reference to the accompanying drawings, but those of ordinary skill in the art to which the present invention pertains can realize that the present invention can be embodied in other specific forms without changing the technical spirit or essential features thereof. you will be able to understand Therefore, it should be understood that the embodiments described above are illustrative in all respects and not restrictive.

[Explanation of code]

100: water quality measurement device

110: control unit

111: machine learning module

1111: turbidity analysis unit

1112: pH analysis unit

1113: floating material analysis unit

112: preprocessing module

113: mode switching module

114: water quality judgment module

120: video recording unit

130: flow detection sensor

140: notification unit

150: database unit

160: communication department

170: power supply

180: power generation unit

200: deep learning server

210: control unit

211: deep learning module

2111: CNN analysis unit

2112: RNN analysis unit

2113: training data generating unit

212: update module

213: water quality determination module

220: communication department

230: database unit

Claims

As a water quality measuring device configured to measure the water quality of a fluid flowing in the internal space and capable of communicating with a deep learning server,

The water quality measurement device,

a communication unit configured to communicate with the deep learning server;

an image capturing unit for capturing an image of a fluid flowing therein; and

A control unit for acquiring information on at least one of turbidity, pH, and floating matter of the fluid from the image captured through the image capturing unit based on a previously learned machine learning algorithm;

The control unit is

Determining the water quality based on the information on at least one of the turbidity, pH and suspended matter,

Controls the communication unit so that information on at least one of the turbidity, pH, and float is transmitted to the deep learning server,

Updating the machine learning algorithm based on the learning information received from the deep learning server through the communication unit,

Water quality measuring device.
According to claim 1,

Further comprising; a flow sensor for detecting the flow of the fluid in the inner space;

The control unit is

When the flow of the fluid is not detected through the flow sensor, controlling the water quality measuring device to switch to a low power mode,

Water quality measuring device.
According to claim 1,

The control unit is

When the connection with the deep learning server is blocked, controlling the communication unit to transmit information on the determined water quality to an external user terminal,

Water quality measuring device.
According to claim 1,

Further comprising a notification unit that provides a visually recognizable danger signal to the user,

The control unit is

When the determined water quality is lower than a preset reference water quality, controlling information indicating that the determined water quality is dangerous to be displayed on the notification unit,

Water quality measuring device.
According to claim 1,

Further comprising a power generation unit configured to convert the kinetic energy of the flowing fluid into electrical energy to supply the electrical energy to the water quality measuring device,

Water quality measuring device.
The water quality measuring device according to any one of claims 1 to 5; and

A deep learning server operably connected to the water quality measurement device, wherein the deep learning server,

Based on a CNN (Convolutional Neural Network) algorithm, analyzing information about at least one of the turbidity and pH,

Based on the RNN (Recurrent Nueral Network) algorithm, the information on the floating object is analyzed,

To learn by generating learning data based on the analyzed at least one information and information on the floating object,

The learning information is information learned by the deep learning server by the learning data,

Water quality measurement system.
7. The method of claim 6,

The deep learning server determines the water quality based on the analyzed at least one piece of information and information on the floating matter, and when the communication connection with the water quality measuring device is maintained, the determined information on the water quality to the user terminal transmit,

The water quality measuring device, when the communication connection with the water quality measuring device is cut off, transmitting information about the water quality determined based on information on at least one of the turbidity, pH, and floating matter to the user terminal,

Water quality measurement system.
7. The method of claim 6,

The RNN (Recurrent Nueral Network) algorithm is an algorithm based on vector information of both ends of the recognized floating object and the midpoint between the two ends,

Water quality measurement system.
7. The method of claim 6,

The deep learning server transmits the learning information to the water quality measurement device at preset time intervals,

Water quality measurement system.