KR102419614B1 - Providing method of aquaculture facilities - Google Patents

Abstract

The present invention relates to a method for providing an aquaculture facility using artificial intelligence deep learning. The method includes the steps of: reviewing and selecting a location for an aquaculture facility; installing an aquaculture facility at the selected location; and performing aquaculture by operating the installed aquaculture facility. The aquaculture facility includes a root-shaped farm formed so that fish can be scattered into each room when released. In the farm, management is performed on facilities and equipment through fixed cameras equipped with sensors installed in each room and passage installed in the farm to measure and transmit water temperature, pH, and dissolved oxygen (DO). Wind or solar energy is stored as electricity in an energy storage system and is used in combined heat and power generation. Thus, backup power is provided in case of power shortage. According to the present invention, it is possible to provide accuracy higher than that of Scale Invariant Feature Transform (SIFT) or Histogram of Oriented Gradient (HOG), a kind of image recognition technology.

Description

A method of providing aquaculture facilities using artificial intelligence deep learning {Providing method of aquaculture facilities}

The present invention relates to a method for providing aquaculture facilities using artificial intelligence deep learning, and more particularly, to manage a smart farm by using biofloc technology as a follow-up measure while at the same time being able to respond to a disease situation of fish that may occur in the farm. It relates to a method of providing aquaculture facilities using deep learning of artificial intelligence.

Technology based on artificial intelligence is being developed in the operation and management of aquaculture facilities. However, there are still insufficient areas to be optimized for effective operation management of these farms. And, above all, it is necessary to understand the situation of each culture process through observation for each area as a whole. Nevertheless, this aspect has been neglected.

Korean Patent Publication No. 10-2021-0003549

The problem to be solved by the present invention is that by applying CNN (Convolution Neural Network), which is one of the AI deep learning functions, it is possible to determine the state of the fish even without direct observation of the fish, and as a countermeasure to this, the bio It is to provide a method of providing aquaculture facilities using artificial intelligence deep learning that can supply flocks.

In addition, it is a kind of image recognition technology using a deep learning system using image pattern information based on a convolutional neural network (CNN) that can improve the quality of image learning information that is vulnerable to various environmental problems (shake, illuminance, noise, lower recognition rate, etc.) It is to provide an operation method of a smart farm management FEMS system using artificial intelligence deep learning that shows accuracy that exceeds the SIFT (Scale Invariant Figure trensform) or HOG (Histogram of Oriented Gradient).

In addition, in the operation and management of these farms, it is possible to observe the right place at each site to prevent or deal with negative factors in the shortest time in terms of climatic conditions, the occurrence of artificial or natural physical situations, and security factors. It is to provide an operation method of a smart farm management FEMS system using artificial intelligence deep learning that enables

In addition, in performing these observations, it is possible to provide a more stable and reliable structural means in response to various factors, so it is possible to respond to the events of the entire farm based on observation and to prevent cost loss and damage through this. It is to provide an operation method of a smart farm management FEMS system using deep learning.

The problems of 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.

The present invention provides a method for providing aquaculture facilities using artificial intelligence deep learning, comprising the steps of reviewing and selecting a location of aquaculture facilities; installing the aquaculture facility at a selected location; and operating the installed aquaculture facility to perform aquaculture, wherein the aquaculture facility includes a root-shaped aquaculture farm formed so as to be scattered into each room when the fish is released, and the aquaculture farm includes each room installed in the aquaculture farm; It is installed in the passage and manages facilities and equipment through a fixed camera with a sensor that measures and transmits the temperature, pH, and DO (Dissolved Oxygen) of water, and uses wind and solar energy as an energy storage system. It provides a method of providing aquaculture facilities using artificial intelligence deep learning, which is stored in wjsrfmf and used in cogeneration to provide reserve power when power is insufficient.

According to the present invention as described above, there are one or more of the following effects.

In the present invention, by applying CNN (Convolution Neural Network), one of the AI deep learning functions, it is possible to determine the state of a fish even without direct observation of the fish, and as a countermeasure to this, a biofloc can be supplied. It has the effect of providing a method of providing aquaculture facilities using deep learning with artificial intelligence.

In addition, it is a kind of image recognition technology using a deep learning system using image pattern information based on a convolutional neural network (CNN) that can improve the quality of image learning information that is vulnerable to various environmental problems (shake, illuminance, noise, lower recognition rate, etc.) It has the effect of providing a method of providing aquaculture facilities using artificial intelligence deep learning that shows accuracy that exceeds SIFT (Scale Invariant Figure trensform) or HOG (Histogram of Oriented Gradient).

In addition, in the operation and management of these farms, it is possible to observe the right place at each site to prevent or deal with negative factors in the shortest time in terms of climatic conditions, the occurrence of artificial or natural physical situations, and security factors. It has the effect of providing a method of providing aquaculture facilities using artificial intelligence deep learning.

In addition, in performing these observations, it is possible to provide a more stable and reliable structural means in response to various factors, so it is possible to respond to the events of the entire farm based on observation and to prevent cost loss and damage through this. It has the effect of providing a method of providing aquaculture facilities using deep learning.

1 is a diagram illustrating the structure of a general deep learning system based on a convolutional neural network (CNN).
2 is a view showing a farm management FEMS system according to an embodiment of the present invention.
3 is a view showing the configuration of a linear pump according to an embodiment of the present invention.
4 is an exemplary view showing a linear motor according to an embodiment of the present invention.
5 is a photograph in which photographic images are exemplarily classified for the case of having a disease with respect to various fish.
can do.
6 is a diagram schematically showing the configuration of a convolutional neural network (CNN) according to an embodiment of the present invention.
7 is a diagram schematically showing the structure of a fish drone according to an embodiment of the present invention.
8 to 9 are schematic diagrams illustrating main components among the components according to FIG. 2 .

Hereinafter, a detailed description of a preferred embodiment of the present invention will be described with reference to the accompanying drawings. In the following description of the present invention, if it is determined that a detailed description of a related well-known function or configuration may unnecessarily obscure the gist of the present invention, the detailed description thereof will be omitted.

Let's take a look at the smart farm management FEMS (Factory Energy Management System) constituting the present invention. 2 is a view showing such a farm management FEMS system in the present invention according to an embodiment of the present invention. Referring to FIG. 2, the farm management FEMS system includes a reverse osmosis membrane 430 that pumps water and a chamber 410 for storing water, and uses reverse osmosis to separate water into fresh water and concentrated water, The fresh water separated by the reverse osmosis membrane 430 moves to the water chamber 460 , the concentrated concentrated water is discharged to the sewer, water is supplied to the reverse osmosis membrane 430 , and the fresh water moves to the water chamber 360 . A pump 420 may be included.

A configuration that enables the operation of the reverse osmosis device in the farm management FEMS system will be described. The chamber 410 stores water capable of supplying fresh water, and reverse osmosis occurs by the reverse osmosis membrane 430 to separate fresh water and concentrated water. The chamber 410 draws in water and is pumped into the reverse osmosis membrane 430 . The reverse osmosis membrane 430 pumps water and separates the water into fresh water and concentrated water through reverse osmosis. By pumping water to the reverse osmosis membrane 430 , fresh water is produced through reverse osmosis, and the concentrated water is discharged to a sewer. The fresh water separated by the reverse osmosis membrane 430 moves to the water chamber 460, and the concentrated water is discharged into a river. The linear pump 420 is configured such that fresh water is made through reverse osmosis by pumping water stored in the farm, and the concentrated concentrated water is discharged into a river.

The linear pump 420 may perform a pumping operation through a reciprocating motion. The linear pump 420 periodically repeats reverse osmosis of the reverse osmosis membrane 430 by applying a reciprocating motion for alternately performing a meme and a pulling action to the reverse osmosis membrane 430 . The linear pump 420 periodically repeats reverse osmosis to draw in water discharged from the farm, and discharges fresh water to repeatedly perform reverse osmosis. A controller and a sensor are configured in the linear pump 420 to measure the discharge amount of fresh water, and to control the driving force of the linear pump 420 to match the required discharge amount to the discharge amount of the fresh water. The controller may adjust the required amount of fresh water by controlling the driving force of the linear pump 420 according to the amount of fresh water. The controller may increase the driving force of the linear pump 420 when the required discharge of fresh water is high.

3 is a view showing the configuration of a linear pump according to an embodiment of the present invention. Referring to FIG. 3 , the linear pump 420 uses the driving force of the linear motor 510 , and the linear motor 510 forms a pair of an electromagnet and a magnet, and flows a current to the electromagnet so that the magnet and the repulsive or attractive force act. to increase or decrease the distance

You can make a linear motion. The configuration of the linear motor 510 is as follows. 4 is an exemplary view showing a linear motor according to an embodiment of the present invention. Referring to FIG. 4 , a wire 620 for supplying current to the coil 610 is configured in a case 630 . The wire 620 is vertically configured on the inside of the case 630 and is configured to contact the coil 610 contact. The coil 610 moves up and down in a state where the contact point of the coil 610 is in contact with the electric wire 620 . The wire 620 may be connected to the driving circuit. The current supplied to the wire 620 is the coil

It is transmitted to the coil 610 through the 610 contact. The coil 610 is wound around a plastic block, and a contact at the end of the coil 610 is in contact with the electric wire 620 . go to the plastic block

There is an id so that it slides up and down along the case 630 . The coil 610 uses a non-conductive plastic block so that it does not stick to the magnet when no current is applied. The case 630 protects the plurality of coils 610 and the permanent magnet 640 . A plurality of coils 610 and permanent magnets 640 move within the case 630 .

Wires 620 for supplying current to the first coil and the second coil are each configured inside the case 630 . The wire 620 has an anode and a cathode. When current is supplied to the first coil, the permanent magnet 640 is pushed down, and when current is supplied to the second coil, the permanent magnet 640 is pushed up. The position of the permanent magnet 640 may be changed by controlling the current supplied to the first coil and the second coil. A bar is configured in the permanent magnet 640 so that the bar moves up and down according to the movement of the permanent magnet 640 . The movement of the rod transmits the driving force to the outside. The driving circuit includes a distribution circuit that divides a DC voltage into a first voltage and a second voltage, which are driving voltages, according to a target voltage, and divides the driving current into a first current and a second current according to the driving voltage. and a differential circuit for driving the coil and the second coil.

The driving circuit supplies the first current and the second current to the first coil and the second coil, respectively, to place the permanent magnet 640 at a designated position. The first coil and the second coil are a plurality of coils composed of a coil block. The coil blocks are wound in the same direction or in different directions to output a repulsive force to each other. The repulsive force serves to narrow or increase the distance so that the permanent magnet 640 is placed in a designated position by configuring the coil blocks to repel each other. As the distance between the coil blocks increases or becomes narrower, the overall length of the first coil and the second coil changes, so that the position of the permanent magnet 640 is changed. The magnitude of the repulsive force varies according to the magnitude of the current supplied to the coil 610 . The driving circuit is to differently control the total length of the plurality of coil blocks constituting the first coil and the plurality of coil blocks constituting the second coil by differently controlling the magnitude of the repulsive force.

The control unit controls the driving circuit. The control unit adjusts the input values of the first coil current and the second coil current input to the driving circuit so that the driving circuit can control the first coil current and the second coil current according to the input values. By controlling the operation of the linear motor 510 in the same manner as described above, it may be possible to control the water supplied to the farm FEMS. As cold-blooded animals, fish have some degree of adaptability to the environment, but if they exceed that limit, they cause physiological disturbances. In fish, disease refers to a state in which health can no longer be maintained with respect to internal and external environments, such as those of terrestrial animals.

A disease is a phenomenon that appears as a result of the correlation between host factors, pathogenic factors, and the environment. does not In other words, disease does not occur when the balance between the pathogenic factor, the host factor, and the environment is well established, and occurs when the balance is broken.

Pathways for pathogens to enter the body of fish can be seen through the skin, gills, nasal passages, and digestive tract. These organs are protected by the secretion and action of mucus, enzymes, antibacterial and bactericidal substances, antibodies, phagocytes, etc., but these protective substances are damaged by mechanical damage during sorting, wounds caused by parasites, poor breeding management, etc. When it is worn, the pathogen enters the body through it, and then through the bloodstream it reaches an organ or tissue to form a lesion.

Factors exacerbating these diseases include overcrowding and contamination of the farm environment by feed residues and excrement, and the pathogens have long-term viability and are always present, and have the potential to parasitize fish. Once a fish disease occurs, it often spreads rapidly to all farms. When seedlings carrying pathogens are supplied to aquaculture farms in other areas, or when fish (fish) are moved for one year for two-year breeding, it can cause the disease-causing area to expand.

The skin mucous membrane is weakened by the administration of low-grade feed, and the intrusion of pathogens from the outside cannot be prevented. In addition, when the density in the cage and tank is high, the secretion of mucus is small and the defense function is lowered. In addition, wounds caused by parasitic infections, etc., easily allow the intrusion of pathogens from the outside.

Convolution (CNN) is one of the methods for determining the condition of fish using deep running in the present invention, because an expert on diseases cannot visit each farm to diagnose the disease. We will use a method using a neural network.

The essence of Convoluted Neural Network (CNN) is that it learns photos. The purpose of this is to give an image file with a label as an input and learn a lot of images to attach a label exactly when a new image is input later. For example, the purpose of CNN is to show images of various animals such as dogs, cats, and birds, and to make the learned computer judge the image as a dog when a new dog image is input. CNNs use the same way living things process visual information. Therefore, in operating a fish farm, it can be said that a database file storing an image of a disease information about a fish is necessary when determining whether a fish has a disease or not. 5 can be said to be a photograph in which photographic images are exemplarily classified for the case of having a disease with respect to various kinds of fish. Referring to FIG. 5 (a), it is a photograph showing the symptoms of galligus, which can be called a parasitic disease. In the case of parasites, parasites are often parasitic on the surface of fish, so as in the present invention, important data is extracted using photographic data, and images are compared using the CNN method with the existing database on fish diseases. Immediate action can be taken only when a database for this has been established.

In particular, information (images) captured by a mobile camera in a convolutional neural network and indexing data stored in a database containing information on diseases of fish are compared with images extracted from the results of photographing the fish This is because not only can accurate diagnosis of disease be performed through the process of comparing with

5 (b) and (c) are pictures showing the aspect of niche, a viral disease of pufferfish, respectively, and it can be said that it is a photograph showing anorexia, a trophic disease of red sea bream. Such data may be stored in a disease database (database). Such a disease database can be a measure of whether an image obtained and extracted from a photographic image taken by a mobile camera using a convolutional neural network (CNN) algorithm has a disease.

6 is a diagram schematically showing the configuration of a convolutional neural network (CNN) according to an embodiment of the present invention.

Referring to FIG. 6 , the CNN may be divided into a convolution layer, a pooling layer, a ReLu layer imparting nonlinearity, and a fully connected layer. First, it can be expressed in the structure of an image that is input data to the CNN 100 (in the present invention, an image transmitted to a computer through a mobile camera). The structure of the image can be expressed by three numbers, and it is a three-dimensional array and has the meaning of width * height * color. (x, y, 0) contains R information, (x, y, 1) contains G information, and (x, y, 2) contains B information.

The output value is presented as the probability of which label the input image will have. In the all-connection layer (FC layer), it is decided which label the corresponding image will have by using softmax classification. First, the convolutional layer undergoes a process of extracting characteristic values of an image captured by a mobile camera using a filter (kernel, neuron). The array of data extracted through the filter in this way is called a feature map (activation map). The number of feature maps has the same characteristics as the number of filters used. If convolution is frequently performed, the size of the feature map is inevitably smaller than that of the original image. A padding value is given here to prevent the size from being reduced. Data that has undergone the convolution process consists of addition and multiplication. Since such linear data has a vulnerability to complex data classification, a non-Linearity property can be assigned using ReLU. The pooling layer can be viewed as an operation that performs a subsampling function. It can be seen that the sample is sampled once more from the convolutional data.

At this time, as shown in FIG. 6 , the pooling that selects the largest value from the matrix is called max pooling. If you go through the pooling process like this, you can extract only the data you really need, and you get the double effect of reducing the size of the data you understand. Even if the original image contains noise that does not fit the context, if it goes through the pooling process, some of the noise can be removed and data can be trained. Finally, the all-connection layer (FC layer) performs Softmax classification by collecting all previous information. With the final data classified in this way, it is determined whether or not the image taken from the mobile camera taken through the comparison with the disease server data already acquired through the CNN process has disease.

If it is determined that the image captured by the mobile camera has a disease, the disease treatment unit is started. In the present invention, a technology called biofloc is used to treat these diseases. Before that, let's take a look at a fish drone that can transmit data (image) as described above. A fish drone can be said to be a technology applied to underwater exploration or disease detection or disease treatment as in the present invention, which is a type of aerial vehicle. 7 is a diagram schematically showing the structure of a fish drone according to an embodiment of the present invention. Referring to FIG. 7 , the fish drone 300 is formed in a streamlined shape, and the fish drone body 310 is provided with a gyro control device 330 to enable yaw, pitch, and roll movements, and is a fish-shaped wearable camera. By having a mobile camera (not shown), it may be possible to photograph the state of the fish. In addition, the fish drone 300 may also serve to drop a vaccine or supply food. In addition, since the pyramid gyro 320 having a gyro sensor is provided, posture control and the like may be possible.

Here, yaw refers to the rotational motion of the fish drone 300 in the z-axis direction, and the pitch refers to rotation in the y-axis direction. Also, roll means rotation in the x-axis direction as shown. In the water, as in the sky, it must be able to move up, down, left and right to transmit images without colliding with the fish.

It can be said to be a configuration for performing a function. Biofloc usually refers to a small mass of microorganisms, algae, protozoa, and microparticles aggregated together. It forms plaque and develops to a size that can be identified with the naked eye (50-200micron) and is easily precipitated. The biofloc formed in this way has high nutritional value, and in general, dried biofloc consists of 30-45% protein, 1-5% lipid, and the rest of various vitamins and minerals, and also acts as a probiotic.

Biofloc technology (BFT) using such biofloc can raise a large amount of fish in a space where water is scarce and the site is small. More specifically, the microorganisms in the breeding water act as decomposers, microalgae play the role of primary producers, and zooplankton using particulate materials such as uneaten feed, excrement of cultured organisms, microorganisms, and microalgae that are generated during the breeding process of aquaculture organisms. The role of this primary consumer, the microbes in the breeding water play the role of decomposers, and the microalgae play the role of primary producers by using particulate matter such as uneaten feed, excrement of cultured organisms, microorganisms, and microalgae generated during the breeding process of aquaculture organisms. , As one of the ecological aquaculture methods that enables a small aquaculture ecosystem to be constituted by zooplankton acting as the primary consumer and the aquaculture organism as the final predator, microorganisms decompose ammonia, nitrite, nitrate, and organic matter in the breeding water to produce amino acids or small molecules. It is an eco-friendly aquaculture technology that can be converted into protein or converted into new organic components through biosynthesis to purify water quality and use it as food for aquaculture organisms. By using such a virofloc, disease treatment can be performed on fish, and a treatment process for sick fish can be performed through a vaccine or a food supply device mounted on the fish drone 300 .

Hereinafter, the operation method of the smart farm management system using artificial intelligence deep learning will be described in detail. First of all, as an indoor aquaculture farm, the water supply is a water purification method. High-pressure 200,000V (20kV) and AC are flowed between the electrode plates, causing the bacteria in the water to be electrocuted and killed. After connecting the rubber hose, the water quality is improved in real time by electrolysis so that it falls from top to bottom like a waterfall. In addition, in the case of a farm, if wastes accumulate, it may have an adverse effect on the fish in the farm, so it may be necessary to periodically check the temperature, pH, and dissolved oxygen (Dissolved Oxygen). [0110] This can be said to be a basic measure to prevent the death of fish due to internal circumstances of the farm. In order to perform this function, it is attached as an accessory for a fixed camera arranged in each room or in each passage to enable continuous reporting (s10, s20). Based on the above-mentioned contents, referring to FIGS. , A method of providing aquaculture facilities using artificial intelligence deep learning, the method comprising: reviewing and selecting a location of aquaculture facilities; installing the aquaculture facility at a selected location; and operating the installed aquaculture facility to perform aquaculture, wherein the aquaculture facility includes a root-shaped aquaculture farm formed so that the fish can be scattered into each room when the fish is released.

Here, in the farm, the facilities and equipment are managed through a fixed camera with a sensor that is installed in each room and passage installed in the farm to measure and transmit water temperature, pH, and DO (Dissolved Oxygen), and wind, solar power Energy is stored in wjsrfmf in an energy storage system and used in cogeneration to provide reserve power when power is insufficient.

In addition, it includes an observation unit for monitoring a part of the farm to prevent unauthorized personnel from entering the farm, and for observing the state of the farmed fish from the outside. The observation unit, the base body 1110 at the bottom, the charging module 1120 installed on the upper part of the base body 1110, and a first temporary wall panel 1130 provided on one side of the charging module 1120 And, a second temporary wall panel 1140 provided on the other side of the charging module 1120, both ends are interlocked with the first temporary wall panel 1130 and the second temporary wall panel 1140, and a plurality of first A flight photographing body (D1) is provided and includes a supply holding module (1170) for supplying power supplied from the charging module (1120) to the first flight photographing body (D1).

On the other hand, the supply mounting module 1170 flows outwardly from the first temporary wall panel 1130 to the first bottom connection body 1150 on one side, and the second temporary wall panel to the second bottom connection body 1160 on the other side. The first flight photographing body D1 in a state in which it retracts at 1140 and is elevated to a certain height based on the retractable flow of the first temporary wall panel 1130 and the second temporary wall panel 1140 ) start to fly, and the charging module 1120 performs a plurality of step-by-step high-speed charging for the supply holding module 1170 .

A plurality of first cooling jets 1131 are provided on the inside of the first temporary wall panel 1130 in a height direction, and a plurality of second cooling jets 1141 are provided inside the second temporary wall panel 1140 in height. direction, and the first cooling jets 1131 to the second cooling jets 1141 are operated in a progressively increased number as the stage of the high-speed charging increases, so that for the charging module 1120 It is operable in the first mode to suppress overheating of the charging module 1120 by spraying cold air, and is located inside the first temporary wall panel 1130, below the first cooling jets 1131 . Each of the plurality of first temperature measurement modules 1132 is provided, and the temperature of the opposite surface of the charging module 1120 of each of the first temperature measurement modules 1132 is measured.

Here, the first cooling jets 1131 each spray cold air toward the temperature measurement part when a preset temperature value is exceeded based on the temperature measurement, and inside the second temporary wall panel 1140, the A plurality of second temperature measurement modules 1142 are respectively provided to be positioned under the second cooling jets 1141 , and each of the second temperature measurement modules 1142 measures the opposite surface temperature of the charging module 1120 . , and the second cooling jets 1141 are operable in the second mode for individually injecting cold air toward the temperature measurement site when a preset temperature value is exceeded based on the temperature measurement.

Meanwhile, the observation unit is driven upward from the first sub temporary wall panel 2101 provided on the outside of the first temporary wall panel 1130 of the base body 1110 and the first sub temporary wall panel 2101 A first connecting pillar 2102 to be formed, a second sub temporary wall panel 2201 provided on the outside of the second temporary wall panel 1140 of the base body 1110, and the second sub temporary wall panel 2201 A second connecting post 2202 driven upward in the , and an upper mounting module 2300 between the first connecting post 2102 and the second connecting post 2202 are further included.

Here, the upper mounting module 2300 is provided with a plurality of recessed spaces opened toward the upper portion, and the second flying photographing bodies D2 for photographing are accommodated and mounted in the recessed space. The first flight photographing body (D1) and the second flight photographing body (D2) perform a flight for photographing, but in the process of performing the flight, the supply holding module 1170 and the mutually for power supply and flight The position is moved alternately between the upper mounting modules 2300 by rotation.

In addition, on the first sub temporary wall panel 2101 and the second sub temporary wall panel 2201, the first connecting post 2102 and the second connecting post 2202 move up and down, and the upper mounting module (2300) is a variable height. The upper mounting module 2300 is axially rotated at a predetermined angle on the first connecting pole 2102 and the second connecting pole 2202 to adjust the initial angle at which the second flight photographing body D2 is scrambled, A plurality of interlocking modules are provided between the supply holding module 1170 and the upper holding module 2300 .

The interlocking module is installed on the supply holding module 1170 and has a pillar part 3101 that is inserted and mounted on the upper holding module 2300 upward, and an outer body which is installed on the pillar part 3101 to the outside. 3101 and a pair of insertion protectors 3103 inserted from the outer body 3101 into the upper mounting module 2300 are included.

The insertion protection body 3103 is inserted and mounted on the insertion protection body 3103 so as to be located outside the pillar part 3101, and the room holding module 2300 and the pillar part 3101 based on the left and right movement. The gap between them is filled by reinforcing, and the pillar part 3101 is in conduction with the supply holding module 1170, and supplies the power supplied from the charging module 1120 to the upper holding module 2300, and the second 2 The flight camera D2 receives power through the upper mounting module 2300 , and the supply holding module 1170 supplies power at a higher speed than the upper mounting module 2300 .

A first outer provisional wall panel 4100 provided on the outside of the first sub provisional wall panel 2101 and a second outer provisional wall panel 4200 provided on the outside of the second sub provisional wall panel 2201 are further added. include Here, a first connecting structure provided in plurality in the height direction of the first outer temporary wall panel 4100 and a second connecting structure provided in plurality in the height direction of the second outer temporary wall panel 4200 are further included.

The first connection structure is mounted to the first temporary wall panel 1130 through the first sub temporary wall panel 2101, and the second connection structure passes through the first sub temporary wall panel 2101 to be mounted as the second temporary wall panel 1140, the first linkage structure having a bar-shaped first body portion 4110, and a first end portion that is a region of an end rotated from the first body portion 4110 An entire 4111 and a first intermediate rotating body adjacent to the first end rotating body 4111 and rotating from the first body 4110 are provided.

The second linkage structure includes a bar-shaped second body portion 4210, a second end rotating body 4211 that is an end region rotated from the second body 4210, and the second end rotating body ( A second intermediate rotating body 4212 which is adjacent to 4211 and rotates from the second body 4210 is provided.

The first end rotating body 4111 is a polygonal cross-sectional shape and is inserted on the first temporary wall panel 1130 to fix the first temporary wall panel 1130 in a rotational manner, and the first intermediate rotating body is As a polygonal cross-sectional shape, the first sub temporary wall panel 2101 is fixed in a rotation manner while being inserted on the first sub wall panel 2101, and the second end rotating body 4211 is a polygonal cross-sectional shape. As an upper body, the second temporary wall panel 1140 is fixed in a rotational manner while being inserted on the second temporary wall panel 1140, and the second intermediate rotating body 4212 is a polygonal cross-sectional shape, and the second sub While being inserted on the wall panel 2201, the second sub is fixed to the wall panel 2201 in a rotational manner.

In a state in which the first end rotating body 4111, the second end rotating body 4211, the first intermediate rotating body and the second intermediate rotating body 4212 are respectively inserted and fixed in rotation, the first outer The wall panel 4100 applies a pressing and fixing force by completely or selectively reversing the first interlocking structure to the inside, and the second outer temporary wall panel 4200 is pressing the second connecting structure as a whole or selectively backwards to the inside. apply fixing force.

The driving method of the physical components described above is based on a motor, an actuator, etc., and the forward and backward movement and rotational movement are made, and the shape and size of each component can be variously selected and provided according to the installation site and materials to be implemented. . Although the present invention has been described with reference to the embodiment shown in the drawings, which is merely exemplary, it will be understood by those skilled in the art that various modifications and equivalent other embodiments are possible therefrom. Accordingly, the true technical protection scope of the present invention should be determined by the technical spirit of the appended claims.

100: convolutional neural network
110: video input unit
112: Learning Department CNN
114: learning information
120: Execution unit CNN

Claims (3)

As a method of providing aquaculture facilities using artificial intelligence deep learning,
reviewing and selecting the location of the aquaculture facility;
installing the aquaculture facility at a selected location; and
Including the step of operating the installed aquaculture facility to perform aquaculture,
The aquaculture facility includes a root-shaped aquaculture farm formed so that when the fish are released, they can be scattered into each room,
In the farm, management is performed on facilities and equipment through a fixed camera with a sensor that is installed in each room and passage installed in the farm to measure and transmit water temperature, pH, and DO (Dissolved Oxygen), and wind and solar energy is stored in an energy storage system (Energy Storage System) and used in cogeneration to provide reserve power when power is insufficient,
and an observation unit for monitoring a part of the farm to prevent unauthorized personnel from entering the farm, and for observing the state of the farmed fish from the outside,
The observation unit is
and the base body 1110 at the bottom,
A charging module 1120 installed on the upper portion of the base body 1110, and
A first temporary wall panel 1130 provided on one side of the charging module 1120 and,
a second temporary wall panel 1140 provided on the other side of the charging module 1120;
Both ends are interlocked with the first temporary wall panel 1130 and the second temporary wall panel 1140 , and a plurality of first flight photographing bodies D1 are provided on the upper part, and the power supplied from the charging module 1120 is supplied to the first temporary wall panel 1140 . 1 It includes a supply holding module 1170 to deliver and supply to the flight camera (D1),
The supply mounting module 1170 flows in and out of the first temporary wall panel 1130 to a first bottom connector 1150 on one side, and the second temporary wall panel ( 1140), and the first flight photographing body (D1) in a state in which the first temporary wall panel 1130 and the second temporary wall panel 1140 are elevated to a certain height based on the retractable flow of the first temporary wall panel 1140. they start flying
The charging module 1120 performs a plurality of step-by-step high-speed charging for the supply holding module 1170, a method of providing aquaculture facilities using artificial intelligence deep learning. delete According to claim 1,
A plurality of first cooling jets 1131 are provided on the inside of the first temporary wall panel 1130 in the height direction, and a plurality of second cooling jets 1141 are provided inside the second temporary wall panel 1140. It is provided in the height direction,
The first cooling jets 1131 to the second cooling jets 1141 are operated in a progressively increased number as the stage of the high-speed charging increases to inject cold air to the charging module 1120, It is operable in the first mode to suppress overheating of the charging module 1120,
Inside the first temporary wall panel 1130,
A plurality of first temperature measuring modules 1132 are respectively provided to be positioned under the first cooling jets 1131 , and the temperature of the opposite surface of the charging module 1120 of each of the first temperature measuring modules 1132 . take measurements,
The first cooling jets 1131 are,
When a preset temperature value is exceeded based on the temperature measurement, cold air is sprayed toward each temperature measurement part,
Inside the second temporary wall panel 1140,
A plurality of second temperature measuring modules 1142 are respectively provided to be positioned under the second cooling jets 1141 , and the temperature of the opposite surface of the charging module 1120 of each of the second temperature measuring modules 1142 . take measurements,
The second cooling jets 1141 are,
When a preset temperature value is exceeded based on the temperature measurement, it is possible to operate in the second mode of individually spraying cold air toward the temperature measurement site,
The observation unit is
A first sub temporary wall panel 2101 provided on the outside of the first temporary wall panel 1130 of the base body 1110, and a first connecting column driven upward from the first sub temporary wall panel 2101 ( 2102) and
A second sub temporary wall panel 2201 provided on the outside of the second temporary wall panel 1140 of the base body 1110, and a second connecting column driven upward from the second sub temporary wall panel 2201 ( 2202) and
An upper mounting module 2300 is further included between the first connecting post 2102 and the second connecting post 2202,
The upper mounting module 2300 is provided with a plurality of recessed spaces opened toward the upper portion, and the second flight photographing bodies D2 for photographing are accommodated and mounted in the recessed space,
The first flight photographing body (D1) and the second flight photographing body (D2) perform a flight for photographing, and the supply holding module 1170 and each other for power supply and flight in the process of performing the flight A method of providing aquaculture facilities using artificial intelligence deep learning, in which the position is shifted by rotation between the upper mounting modules 2300.

Images