Automatic feeding support device and automatic feeding support method

Abstract

An automatic feeding support device and an automatic feeding support method are provided that are capable of optimizing feeding by a feeding device. [Solution] An acquisition control unit (101) uses multiple ultrasonic transmitters / receivers to emit ultrasonic waves over a wide range from the water surface to the depths of a farmed fish pen to acquire echograms that show changes in the reflected intensity of the ultrasonic waves over time with respect to the depth from the water surface of the pen. A calculation control unit (102) analyzes the acquired echograms to calculate fish behavior information that indicates the distribution of the farmed fish at the depth from the water surface, the movement direction of the farmed fish, the movement speed of the farmed fish, and the feeding status of the farmed fish, or a combination of these. An estimation control unit (103) uses the calculated fish behavior information to estimate the feeding activity of all fish F in the pen. An execution control unit (104) starts or stops feeding of a feeding device that supplies food to the pen depending on the estimated feeding activity.

Description

[Technical Field]

[0001] The present invention relates to an automatic feeding assistance device and an automatic feeding assistance method. [Background technology]

[0002] Conventionally, there are technologies for automatically feeding fish in a fish pen. For example, Japanese Patent Laid-Open Publication No. 04-075544 (Patent Document 1) discloses an automatic fish pen feeder comprising a feed hopper, a fish measuring means, and a control means. The feed hopper supplies feed, the fish measuring means measures whether fish are densely packed in a predetermined area inside the fish pen, and the control means controls the feed supply based on the measurement results from the fish measuring means. If fish are not densely packed in the predetermined area while feed is being supplied, the control means sends a stop signal to the feed hopper to stop the feed supply, and if a preset time has elapsed since the feed supply was stopped, sends a start signal to the feed hopper to start the feed supply again. This allows feed to be released only when fish are eating.

[0003] Furthermore, Japanese Patent Laid-Open Publication No. 05-276849 (Patent Document 2) discloses an automatic feeding device comprising a pecking target, an approaching fish detection device, and a start signal generation unit. The pecking target is the object that the farmed fish peck when they are hungry, and the approaching fish detection device uses ultrasound to detect farmed fish approaching the pecking target. When the detection signal sent out by the approaching fish detection device is a fish approach signal, the start signal generation unit generates a feeding start signal that instructs the start of feeding based on the received fish approach signal. This allows feeding to be carried out in accordance with the condition of the fish in the fish tank.

[0004] Furthermore, Japanese Patent Application Laid-Open Publication No. 2020-018312 (Patent Document 3) discloses an automatic feeding support device including a remaining amount information acquisition unit, a judgment unit, a transmission unit, and a statistical processing unit. The remaining amount information acquisition unit acquires remaining amount information regarding the amount of food remaining in an automatic feeder that provides food to fish, and the judgment unit determines whether the remaining amount of food is low enough that the remaining amount information satisfies a predetermined condition. The transmission unit transmits order information regarding the order of food when the judgment unit determines that the remaining amount of food is low enough to satisfy the predetermined condition. The statistical processing unit receives remaining amount information from two or more automatic feeders, statistically processes the two or more remaining amount information received from the two or more automatic feeders, and obtains the statistical processing results. The transmission unit changes the timing of sending the order information or the order amount in the order information based on the statistical processing results. This allows food to be ordered for fish using the automatic feeder at an optimal time, or supports user ordering.

[0005] Furthermore, Japanese Patent Laid-Open Publication No. 2020-145959 (Patent Document 4) discloses a feeding controller including a feeding continuation learning data storage unit, a feeding cessation learning data storage unit, a knowledge cluster generation unit, a feeding continuation knowledge cluster storage unit, a feeding cessation knowledge cluster storage unit, and a determination unit. The feeding continuation learning data storage unit stores at least the sensor data during feeding continuation as feeding continuation learning data in chronological order, among the sensor data consisting of sets of measurement values ​​obtained by arithmetic processing of various measurement data for each element acquired from various sensors at predetermined time intervals. The feeding cessation learning data storage unit stores at least the sensor data when an operator performs a feeding cessation operation as feeding cessation learning data in chronological order. The knowledge cluster generation unit determines the similarity between the multiple feeding continuation learning data stored in the feeding continuation learning data storage unit or the multiple feeding cessation learning data stored in the feeding cessation learning data storage unit, and generates a predetermined plurality of feeding continuation knowledge clusters or a predetermined plurality of feeding cessation knowledge clusters. The continuation feeding knowledge cluster storage unit and the cessation feeding knowledge cluster storage unit store the continuation feeding knowledge cluster and the cessation feeding knowledge cluster generated by the knowledge cluster generation unit, respectively. The determination unit uses sensor data obtained a predetermined time after the start of feeding as data to be determined, calculates the degree of match with the multiple continuation feeding knowledge clusters and the multiple cessation feeding knowledge clusters, and determines whether to continue feeding. This allows the system to learn the operator's decision-making status on whether to continue or cessation feeding based on environmental conditions, and to control the stopping of the feeding device at a timing close to that expected by the operator.

[0006] Furthermore, Japanese Patent Application Laid-Open Publication No. 2021-164444 (Patent Document 5) discloses an aquatic animal detection device that includes an image acquisition unit, a detection unit, a notification determination unit, a notification generation unit, and a notification output unit. The image acquisition unit acquires images captured by a camera, and the detection unit detects aquatic animals that appear in the images. The notification determination unit determines whether notification conditions related to the detection status of an aquatic animal by the detection unit are met, and the notification generation unit generates notification information when the notification determination unit determines that the notification conditions are met, and the notification output unit outputs the generated notification information. This makes it possible to easily perform tasks that involve detecting aquatic animals in water.

[0007] Furthermore, Japanese Patent Laid-Open Publication No. 2023-028328 (Patent Document 6) discloses a cultivation support device including a raw information storage unit, a specification reception unit, a raw material acquisition unit, and a cultivation raw material output unit. The raw information storage unit stores raw information that is the basis for acquiring one or more pieces of cultivation raw material information related to the cultivation raw materials used in cultivating the marine resources, using one or more pieces of specification information that specify the required specifications of the quality, growth cost, or production volume of the marine resources. The specification reception unit accepts one or more pieces of specification information for the marine resources, the raw material acquisition unit acquires one or more pieces of cultivation raw material information using the one or more pieces of specification information and raw information accepted by the specification reception unit, and the cultivation raw material output unit outputs the one or more pieces of cultivation raw material information acquired by the raw material acquisition unit. This makes it possible to easily acquire cultivation raw material information for cultivating marine resources that meet the required specifications.

[0008] Furthermore, Japanese Patent Laid-Open Publication No. 2023-073461 (Patent Document 7) discloses an asset value output device including an aquatic organism information receiving unit, an asset value acquisition unit, and an asset value output unit. The aquatic organism information receiving unit receives aquatic organism information including one or more of the following: food information regarding the food given to two or more aquatic organisms as breeding targets; and size information regarding the size of the targets. The asset value acquisition unit acquires the asset value of the aquatic organisms using the aquatic organism information, and the asset value output unit outputs the asset value. The aquatic organism information has a feed cost ratio, which is the proportion of feed cost to breeding cost, and the feed cost ratio is managed according to each of two or more types of information indicating the types of the targets. The device is equipped with a type information acquisition means for acquiring the type information. The asset value output unit acquires the feed cost ratio according to the type information acquired by the type information acquisition means, acquires the breeding cost using the feed cost ratio, and acquires the asset value using the breeding cost. This makes it possible to easily acquire the asset value of two or more aquatic organisms as breeding targets.

[0009] Furthermore, Japanese Patent Laid-Open Publication No. 2024-120003 (Patent Document 8) discloses an information processing device comprising an acquisition unit, an estimation unit, and a feeding control unit. The acquisition unit acquires sound information related to sounds from a microphone that collects sounds in the fish farming environment, and the estimation unit determines whether the feeding activity of the fish is high based on the sound information. If the estimation unit determines that the feeding activity of the fish is low, the feeding control unit generates control information indicating the timing to stop the feeding operation of the feeding device. This is said to enable the optimization of feeding by the feeding device. [Prior art documents] [Patent documents]

[0010] [Patent Document 1] Japanese Patent Application Publication No. 04-075544 [Patent Document 2] Japanese Patent Application Publication No. 05-276849 [Patent Document 3] Japanese Patent Publication No. 2020-018312 [Patent Document 4] Japanese Patent Application Publication No. 2020-145959 [Patent Document 5] Patent Publication No. 2021-164444 [Patent Document 6] Japanese Patent Publication No. 2023-028328 [Patent Document 7] Japanese Patent Publication No. 2023-073461 [Patent Document 8] Japanese Patent Application Laid-Open No. 2024-120003 Summary of the Invention [Problem to be solved by the invention]

[0011] For fish farmers, the method of feeding the farmed fish in the fish pens (when to start and stop feeding) is important from the perspective of preventing wasteful consumption of feed. The amount of feed given to the farmed fish in the fish pens basically depends on the type, number, and body length of the farmed fish in the pens, but it also needs to be changed depending on factors such as the water temperature in the pens.

[0012] Here, farmed fish in a fish tank usually sink to the depths of the tank and swim slowly. However, when they are hungry, depending on the type of farmed fish, they may drift slowly in the depths of the tank as usual, swim slightly shallower than the normal depth, move around at a faster speed than normal, or gather close to the surface of the tank, exhibiting a variety of behaviors. Therefore, when a fish farmer who raises farmed fish in a fish tank feeds them using a specified feeding device, the farmed fish move around quickly and gather close to the surface of the tank to feed. After the farmed fish continue to feed for a certain period of time, they become full, slow down, swim slowly to the depths of the tank again, and sink. Farmed fish repeat these behaviors within the tank.

[0013] On the other hand, although aquaculture farmers have some knowledge of the behavior of farmed fish described above, they are unable to observe the behavior of all the farmed fish in the pens, so they feed them a predetermined amount of food using feeding devices based on their own intuition and customs.As a result, in the aquaculture industry, farmers continue to feed their farmed fish even when they are full.In fact, farmers' feed expenditures account for approximately 80% of their total expenditures, and with the recent rise in prices, feed costs have also risen, causing a headache for farmers.Therefore, there has been a need in the aquaculture industry for a method to properly feed farmed fish in pens.

[0014] There are various technologies for evaluating the behavior of farmed fish and automatically feeding them. For example, technologies using cameras installed in the air or underwater monitor the behavior of farmed fish in fish pens, evaluate their behavior near the water surface or underwater, and adjust the timing of feeding. Furthermore, technologies using ultrasonic transducers installed underwater measure echoes from farmed fish in fish pens with ultrasound, evaluate the depth distribution of the farmed fish from the intensity distribution of the echoes, and adjust the timing of feeding.

[0015] However, when evaluating the behavior of farmed fish using an aerial camera, only those near the water surface are evaluated. Similarly, when evaluating the behavior of farmed fish using an underwater camera, only those within the camera's field of view are evaluated. Therefore, camera-based techniques have the drawback of being unable to evaluate the behavior of farmed fish throughout the entire fish pen. Furthermore, while ultrasonic transducer-based techniques can evaluate the depth distribution of farmed fish, they have the drawback of being unable to specifically evaluate changes in the behavior of farmed fish.

[0016] The technology described in Patent Document 1 controls feeding by using ultrasound to measure whether fish are crowded below a feed hopper. The technology described in Patent Document 2 controls feeding by using a pecked object that fish peck hungrily at and an approaching fish detection device that uses ultrasound to detect fish approaching the pecked object. The technologies described in Patent Documents 3-4 and 6-7 relate to information processing of feeding data and feed management. The technology described in Patent Document 5 detects fish using a camera. And the technology described in Patent Document 8 collects sounds of fish in a fish pen using a microphone. However, these technologies have the problem of not being able to specifically evaluate changes in fish behavior because they do not measure information such as the speed and direction of fish movement.

[0017] Therefore, the present invention has been made to solve the above-mentioned problems, and aims to provide an automatic feeding support device and an automatic feeding support method that can optimize feeding by a feeding device by specifically estimating the behavior of all farmed fish in a fish tank. [Means for solving the problem]

[0018] The automatic feeding support device according to the present invention includes an acquisition control unit, a calculation control unit, an estimation control unit, and an execution control unit. The acquisition control unit uses multiple ultrasonic transmitters / receivers that emit ultrasonic waves over a wide range from the water surface to the depths of a fish pen for farmed fish to acquire echograms that show changes in the reflected intensity of ultrasonic waves over time with respect to the depth from the water surface of the fish pen. The calculation control unit analyzes the echograms to calculate fish behavior information that shows the distribution of the farmed fish at the depth from the water surface, the movement direction of the farmed fish, the movement speed of the farmed fish, the feeding status of the farmed fish, or a combination of these. The estimation control unit uses the calculated fish behavior information to estimate the feeding activity of all the farmed fish in the pen. The execution control unit starts or stops feeding by a feeding device that feeds the fish pen according to the estimated feeding activity.

[0019] The automatic feeding assistance method according to the present invention also includes an acquisition control step, a calculation control step, an estimation control step, and an execution control step. Each control step of the automatic feeding assistance method according to the present invention corresponds to each control unit of the automatic feeding assistance device according to the present invention. [Effects of the Invention]

[0020] According to the present invention, by specifically estimating the behavior of all the cultured fish in the fish pen, it is possible to optimize feeding by the feeding device. [Brief explanation of the drawings]

[0021] [Figure 1] Figure 1A shows an example of farmed fish in a fish tank when they are full, Figure 1B shows an example of farmed fish in a fish tank when they are hungry, Figure 1C shows an example of farmed fish when they are being fed, and Figure 1D shows an example of farmed fish when they are full after being fed. [Figure 2] 2A is a conceptual diagram showing an example of an automatic feeding assistance device according to an embodiment of the present invention, and FIG. 2B is a conceptual diagram showing an example of an arrangement of a plurality of ultrasonic transmitter / receivers. [Figure 3] 1 is a functional block diagram showing an example of an automatic feeding support device according to an embodiment of the present invention. [Figure 4] 1 is a flowchart illustrating an example of an automatic feeding assistance method according to an embodiment of the present invention. [Figure 5] This is a conceptual diagram (Figure 5A) showing an example of an automatic feeding support device installed in a fish tank, a diagram (Figure 5B) showing an example of an echogram from an ultrasonic transmitter / receiver, and an image (Figure 5C) showing an example of a linked echogram created from the echogram. [Figure 6] 6A is an enlarged view showing an example of various types of linked echoes in a linked echogram, FIG. 6B is a view showing an example of a linked echogram divided into a predetermined number of sections, and FIG. 6C is a view showing an example of a tail echogram. [Figure 7] 7A is a diagram showing an example of a distance difference average value gram and a distance difference variance gram, and FIG. 7B is a diagram showing an example of a slope average value gram and a slope variance gram. [Figure 8] 8A is a diagram showing an example of a TS average value gram and a TS variance gram, and FIG. 8B is a diagram showing an example of an FL average value gram and an FL variance gram. [Figure 9] 9A shows an example of an echogram in a normal state and an echogram in a fasting state, and FIG. 9B shows an example of an average linked ping count gram and a variance linked ping count gram. [Figure 10] Figure 10A shows the experimental equipment for acquiring bait echoes and an example of two resulting echograms; Figure 10B shows an example of an echogram including bait echoes when hungry and full; and Figure 10C shows an example of an echogram including bait echoes, fish pen net echoes, and wild fish echoes when hungry and full. [Figure 11] FIG. 11A shows an example of a case where feeding activity is low and feeding has not started, and FIG. 11B shows an example of a case where feeding activity is high and feeding has started. [Figure 12] 12A is a conceptual diagram showing an example after feeding has started, FIG. 12B is a diagram showing an example of an echogram after feeding has started, and FIG. 12C is a diagram showing an example of a few grams of tail. [Figure 13] 13A is a diagram showing an example of a distance difference average value gram and a distance difference variance gram, and FIG. 13B is a diagram showing an example of a gradient average value gram and a gradient variance gram. [Figure 14] 14A is a diagram showing an example of a TS average value gram and a TS variance gram, and FIG. 14B is a diagram showing an example of an FL average value gram and an FL variance gram. [Figure 15] 15A shows an example of an echogram when fed and an echogram when full (FIG. 15A), and FIG. 15B shows an example of an average linked ping count gram and a variance linked ping count gram. [Figure 16] FIG. 16A shows an example of a case where feeding activity is high and feeding continues, and FIG. 16B shows an example of a case where feeding activity is low and feeding has stopped. [Figure 17]FIG. 17A shows an example of an automatic feeding support device installed at the center of a fish pen, and FIG. 17B shows an example of an automatic feeding support device installed at one end of the fish pen. [Figure 18] FIG. 19A shows an example of a scanning type configuration in which ultrasonic transmitter / receiver units are arranged in a fan shape, and FIG. 19B shows an example of a scanning type configuration in which ultrasonic transmitter / receiver units are arranged in a straight line. [Figure 19] 19A is a diagram showing an example of a case where an ultrasonic transmitter / receiver is rotated 360 degrees, and FIG. 19B is a diagram showing an example of a case where the ultrasonic transmitter / receiver is moved back and forth. [Figure 20] FIG. 1 shows an example of an echogram obtained by five ultrasonic transducers for a fish of fish species A. [Figure 21] FIG. 10 is a diagram showing another example of an echogram obtained by five ultrasonic transmitters and receivers for a fish of fish species A. [Figure 22] FIG. 10 is a diagram showing an example of an echogram acquired by five ultrasonic transducers for a fish of fish species B. [Figure 23] FIG. 10 is a diagram showing another example of an echogram obtained by five ultrasonic transmitters and receivers for a fish of fish species B. [Figure 24] FIG. 10 is a diagram showing an example of an echogram acquired by five ultrasonic transducers for a fish of fish species C. [Figure 25] FIG. 10 is a diagram showing another example of an echogram obtained by five ultrasonic transmitters and receivers for a fish of fish species C. [Figure 26] 26A is a diagram showing an example of grams of tails of fish of fish species C, and FIG. 26B is a diagram showing an example of average distance difference grams and variance distance difference grams of fish of fish species C. [Figure 27] 27A is a diagram showing an example of a slope average value gram and a slope variance gram for fish of fish species C, and FIG. 27B is a diagram showing an example of a TS average value gram and a TS variance gram for fish of fish species C. [Figure 28]Figure 28A shows an example of an FL average value gram and an FL variance gram for fish of fish species C, and Figure 28B shows an example of an average value gram of connected ping counts and a variance gram of connected ping counts for fish of fish species C. DETAILED DESCRIPTION OF THE INVENTION

[0022] Hereinafter, an embodiment of the present invention will be described with reference to the accompanying drawings to help understand the present invention. Note that the following embodiment is an example of the present invention, and is not intended to limit the technical scope of the present invention.

[0023] First, we will explain the behavior of the fish in the fish pen. As shown in Figure 1A, when the farmed fish F are full, they drift slowly in various directions from the water surface W to the deep depths D of the fish pen. In this case, since the farmed fish F are full, they do not need to search for food, and the farmed fish F become passive overall. This state continues even if the farmed fish are not particularly hungry.

[0024] When the farmed fish F become hungry, they behave in a variety of ways depending on the type of farmed fish, but as shown in Figure 1B, they may drift slowly in the deep part D of the fish pen W as usual, swim slightly shallower than the usual depth D, move around at a faster speed than their usual speed, or gather close together near the water surface W of the fish pen. In other words, when the farmed fish F become hungry, they behave in a variety of ways compared to when they are full.

[0025] Therefore, when the fish farmer uses the feeding device 10 to feed the feed 11 onto the water surface W of the fish pen, the farmed fish F quickly move toward the feed 11 on the water surface W and ingest (feed) the feed 11, as shown in Figure 1C. In this case, the farmed fish F are still hungry and need to ingest the feed 11 before anyone else, so the farmed fish F become active overall.

[0026] Then, when the fish farmer continues to feed the feed 11 onto the water surface W of the fish pen using the feeding device 10, most of the cultured fish F continue to ingest the feed 11 and become full. Here, the cultured fish F that are full do not need to ingest the feed 11, so they move slowly and sink from the water surface W of the fish pen towards the deeper depths, as shown in Figure 1D. Also, because the cultured fish F are not interested in the feed 11, the feed 11 sinks to the bottom without being eaten by the cultured fish F. In this case, the cultured fish F become passive overall.

[0027] In other words, when the farmed fish F are hungry, their behavior will vary, and when feeding begins, the overall behavior of the farmed fish F will become active, and when the farmed fish F are full, their overall behavior will become calm. In the present invention, by understanding the behavior of the farmed fish F using echograms, feeding by the feeding device 10 can be optimized.

[0028] As shown in Fig. 2A, an automatic feeding assistance device 1 according to an embodiment of the present invention comprises feeding device 10, a plurality of ultrasonic transmitters / receivers 20, and control device 30. Feeding device 10 and each of the plurality of ultrasonic transmitters / receivers 20 are electrically connected to control device 30 and are controlled by control device 30.

[0029] The feeding device 10 is a device for feeding fish. The configuration of the feeding device 10 is not particularly limited, but may include, for example, a container 100 for storing bait 11, a hole 101 provided on the underside of the container 100 for discharging the bait 11, and a lid 102 for opening and closing the hole 101. In this feeding device 10, when the lid 102 is opened, the bait 11 in the container 100 falls through the hole 101 under its own weight and is dispersed within the fish pen. Dispersion of the bait 11 can be stopped by closing the lid 102. The lid 102 can be opened and closed automatically by the control device 30. Other examples of the feeding device 10 include a blower-type (air-blowing) feeding device, a screw-type feeding device, and a throw-in type feeding device. The type of bait 11 is not particularly limited, but may include, for example, dry solid bait, raw solid bait, semi-solid bait, or liquid bait.

[0030] The ultrasonic transmitter / receiver 20 is a device that transmits and receives ultrasonic waves. Ultrasonic waves are transmitted by applying a high-frequency voltage signal from the control device 30 to a piezoelectric element such as a piezo element, causing the piezoelectric element to vibrate, generating ultrasonic waves that are then emitted (irradiated) into the water. Ultrasonic waves are received by receiving ultrasonic waves reflected by fish or objects in the water (including bait, as described below) with the piezoelectric element, causing the piezoelectric element to vibrate, and detecting the weak voltage generated by the vibration of the piezoelectric element as the reflected ultrasonic wave intensity.

[0031] Furthermore, there is no particular limitation on the number of ultrasonic transmitters / receivers 20 (channels), but it can be in the range of 1 to 100, for example, and can be in the range of 1 to 16. The greater the number of ultrasonic transmitters / receivers 20, the greater the number of echograms, allowing for a more detailed evaluation of the overall behavior of the farmed fish F. There is no particular limitation on the arrangement of the ultrasonic transmitters / receivers 20, but for example, one ultrasonic transmitter / receiver 20 can be arranged facing directly below the fish pen. Alternatively, as shown in FIG. 2A, three ultrasonic transmitters / receivers 20a, 20b, and 20c can be arranged facing directly below the fish pen, diagonally downward to the left, and diagonally downward to the right, respectively. In this case, the three ultrasonic transmitters / receivers 20a to 20c radiate ultrasonic waves to an area Sa directly below the fish pen, an area Sb diagonally downward to the left of the fish pen, and an area Sc diagonally downward to the right of the fish pen. In this way, the arrangement and configuration of the ultrasonic transmitter / receiver 30 can be changed as appropriate depending on the size and shape of the fish pen and the area where it is desired to radiate ultrasonic waves and measure fish behavior.

[0032] As shown in FIG. 2B, the ultrasonic wave emission directions of the left and right ultrasonic transmitters 20b and 20c are arranged at left and right angles α1 and α2 relative to the ultrasonic wave emission direction of the central ultrasonic transmitter 20a. The left and right angles α1 and α2 may be the same or different, and can be set within a range of 1 to 90 degrees by the fish farmer depending on the shape and size of the fish pen. Alternatively, five ultrasonic transmitters 20a to 20e may be arranged so that they face directly below the fish pen, diagonally downward to the left of the fish pen, diagonally downward to the right of the fish pen, diagonally downward in front of the fish pen, and diagonally downward in back of the fish pen. Using five ultrasonic transmitters 20a to 20e, it is possible to measure fish behavior not only directly below the fish pen but also in all four directions around it. Furthermore, although the range in which the behavior of the cultured fish F can be measured may be narrowed due to the influence of the directional characteristics of the ultrasonic transmitter / receiver 20, by increasing the number of ultrasonic transmitter / receivers 20 and making them multi-channel, it is possible to evaluate the behavior of the cultured fish F over a wide spatial range, just like with a camera. Another method is to arrange nine ultrasonic transmitters / receivers 20 so that they face directly below the fish pen and in eight directions obtained by dividing the diagonal direction below the pen into eight equal parts. The arrangement of multiple ultrasonic transmitters / receivers 20 can be determined appropriately by the fish farmer depending on the shape and size of the fish pen.

[0033] Control device 30 is a computer, a microcontroller, etc. Control device 30 controls feeding by feeding device 10, controls the transmission and reception of ultrasonic waves by ultrasonic transmitter / receiver 20, and processes and analyzes the received ultrasonic waves.

[0034] The control device 30 also incorporates a CPU (GPU), ROM, RAM, etc. (not shown), and the CPU uses, for example, the RAM as a work area to execute programs stored in the ROM, etc. The CPU also executes programs to realize the functions of each control unit, which will be described later.

[0035] Next, the configuration and execution procedures according to an embodiment of the present invention will be described with reference to Figures 3 to 16. First, as shown in Figure 5A, an aquaculture farmer, aquaculture supporter, or the like installs a feeding device 10, multiple ultrasonic transmitter / receivers 20, and a control device 30 near the water surface W of the net cage to prepare for feeding. Here, the aquaculture farmer places a net cage N in the deep part D of the net cage to separate the inside and outside of the net cage, and raises farmed fish F inside the net cage N. Note that wild fish NF other than farmed fish F roam outside the net cage N.

[0036] Next, when the farmer turns on the power of the control device 30 and presses the start key, the feeding start program of the control device 30 is executed, and the control device 30 accepts from the farmer whether to use the evaluation feeding setting or the normal feeding setting (Figure 4: S101).

[0037] Here, there are no particular limitations on the method for accepting the evaluation feeding setting by the control device 30. As described above, since the cultured fish F exhibit a wide variety of behaviors when full, the present invention allows for various feeding start settings. For example, in the evaluation feeding setting, the ultrasonic transmitter / receiver 20 is used to analyze echograms, confirm behavioral information of the cultured fish F in the fish pen, and determine whether to start feeding. On the other hand, in the normal feeding setting, feeding is started based on a preset time or time set by the farmer, or by the farmer selecting a feeding start key. In the present invention, the evaluation feeding setting will be described first, followed by the normal feeding setting. For example, the control device 30 displays an evaluation feeding setting key and a normal feeding setting key selectable on the display unit and accepts key selection from the farmer. Here, the display unit can be, for example, a desktop display unit or a touch panel display unit. Furthermore, the key selection method can be, for example, using the operation unit or touch panel display unit of the control device 30.

[0038] Now, when the farmer selects the evaluation feeding setting key, the control device 30 accepts the selection of the evaluation feeding setting key (Figure 4: S101 YES), and then the acquisition control unit 101 of the control device 30 uses multiple ultrasonic transmitter / receivers 20 to irradiate ultrasonic waves over a wide range from the water surface W to the depth D of the fish F's fish pen (Figure 4: S102), as shown in Figure 5A.

[0039] Here, there is no particular limitation on the method of irradiating ultrasound by the acquisition control unit 101, but an example is a method of irradiating ultrasound at a time interval ranging from 10 to 30 times per second. Here, there is no particular limitation on the configuration of the ultrasound, but an example is an ultrasound beam that spreads in a conical or quadrangular pyramid shape. There is also no particular limitation on the frequency of the ultrasound, but for example, a range of 200 kHz to 300 kHz can be used. There is also no particular limitation on the directivity angle of the ultrasound, but for example, a range of 5 degrees to 10 degrees can be used.

[0040] The time interval between ultrasonic waves is called a ping. For example, if the acquisition control unit 101 emits ultrasonic waves 10 times per second, the pings will be at time intervals of 0.1 seconds. By reducing the time interval between pings, the acquisition control unit 101 can separate and receive ultrasonic waves (echoes) reflected from individual farmed fish F, even if the farmed fish F are moving fast. The emitted ultrasonic waves are reflected by the farmed fish F and farm nets N in the fish pen, the wild fish NF, and the bait 11 in the fish pen. The acquisition control unit 101 receives the reflected ultrasonic waves using the ultrasonic transmitter / receiver 20.

[0041] Then, based on the reflection intensity of the ultrasonic waves received by the ultrasonic transmitter / receiver 20, the acquisition control unit 101 acquires an echogram that shows the change in the reflection intensity of the ultrasonic waves over time with respect to the depth from the water surface W of the fish pen (Figure 4: S103).

[0042] Here, as shown in FIG. 5B , the echogram is data showing the reflection intensity of ultrasonic waves at each time and each depth, with the horizontal axis representing time and the vertical axis representing distance (depth). The time on the horizontal axis is calculated by multiplying the ultrasonic irradiation number (number of irradiations) by the time interval between pings. The ultrasonic irradiation number is a number assigned by the acquisition control unit 101 to identify ultrasonic waves when transmitting them from the ultrasonic transmitter / receiver 20. For example, if the ultrasonic irradiation number is 5000 and the time interval between pings is 0.1 seconds, the time for which the ultrasonic waves were irradiated is 5000 × 1 / 10 seconds = 500 seconds. The depth is the distance from the transmitting / receiving surface of the ultrasonic transmitter / receiver 20 and is calculated by multiplying the ultrasonic velocity (approximately 1500 m / s in water) by the time taken from transmitting to receiving the ultrasonic waves and dividing the result by 2. Based on the above, the acquisition control unit 101 creates and acquires an echogram based on the irradiation number of the transmitted ultrasonic waves and the reflection intensity of the received ultrasonic waves. By acquiring an echogram, it is possible to visualize the change over time in the reflected intensity of ultrasound at each depth as a two-dimensional distribution.

[0043] Now, once the acquisition control unit 101 acquires the echogram, the calculation control unit 102 of the control device 30 then analyzes the echogram to calculate fish behavior information indicating the distribution of the farmed fish F at the depth from the water surface W, the movement direction of the farmed fish F, the movement speed of the farmed fish F, the feeding status of the farmed fish F, or a combination of these (Figure 4: S104).

[0044] Here, the calculation control unit 102 may calculate fish behavior information in any manner. First, as shown in FIG. 5B, the acquired echogram is represented by dotted or line echoes (reflected ultrasonic waves) containing reflection intensities. The stronger the reflection intensity of an echo, the more emphasized the color of the echo is. Therefore, the calculation control unit 102 creates a connected echo by connecting the points (peaks) of the strongest reflection intensity of adjacent echoes within a predetermined depth range in the acquired echogram. By connecting the peaks of adjacent echoes, it is possible to determine that the connected echoes are from the same farmed fish F. Furthermore, the connected echo can also be called a connected ping, since it represents the cumulative number of reflections of multiple consecutive ultrasonic pings. By repeatedly creating connected echoes for the echogram, the calculation control unit 102 creates a connected echogram that shows the changes in the connected echoes over time at each depth, as shown in FIG. 5C.

[0045] Here, as shown in FIG. 6A , the linked echo E is represented by a line segment connecting multiple points. The number of linked echoes E corresponds to the number of farmed fish F within the ultrasonic irradiation range. The direction from the starting point ES to the ending point EE of the linked echo E indicates the direction of movement of the farmed fish F. For example, as shown in FIG. 6A , if the starting point ES of a first linked echo E1 is at a deep position at depth D and the ending point EE of the first linked echo E1 is at a shallow position at depth D, this means that the farmed fish F corresponding to the first linked echo E1 is moving upward from a deep position to a shallow position. If the starting point ES of a second linked echo E2 is at a deep position at depth D and the ending point EE of the second linked echo E2 is at the same deep position, this means that the farmed fish F corresponding to the second linked echo E2 is moving horizontally (left and right) at the same position. 6A, the second linked echo E2 corresponding to the farmed fish F moving horizontally is represented as a dogleg. If the starting point ES of the third linked echo E3 is at a shallow position of depth D and the ending point EE of the third linked echo E3 is at a deep position of depth D, this means that the farmed fish F corresponding to the third linked echo E3 is moving downward from a shallow position to a deep position. In other words, the behavior of the farmed fish F can be analyzed by analyzing the linked echo E.

[0046] Next, a specific method by which the calculation control unit 102 calculates fish behavior information will be described. For example, when focusing on the distribution of cultured fish F at depth from the water surface W (e.g., the depth distribution of cultured fish F and the directional distribution of cultured fish F), as shown in FIG. 6B, the calculation control unit 102 divides the linked echogram into a grid at predetermined depth intervals and predetermined time intervals to create multiple divided sections P, and calculates the number of linked echoes in each section P as the number of cultured fish F. The calculation control unit 102 then places the calculated number of cultured fish F in each section P to create a fish count gram that shows the change over time in the number of cultured fish F at each depth, as shown in FIG. 6C. Here, the greater the number of fish, the brighter the color of the fish count gram section P, and the fewer the number of fish, the darker the color of the section P.

[0047] For example, if the cultured fish F are non-migratory, as shown in Figure 6C, when the cultured fish F become hungry, the color of the section P corresponding to a few grams of tails in the shallow depth area becomes brighter. This means that the cultured fish F usually swim in deep water, but when they want to feed, they gather at the water surface W. As described above, when the cultured fish F drift slowly in the deep part D of the fish pen W as usual, the color of the section P corresponding to a few grams of tails in the deep depth area becomes brighter. Furthermore, when the cultured fish F swim in a position slightly shallower than the normal depth D, the color of the section P corresponding to a few grams of tails in the predetermined shallow depth area becomes brighter. Therefore, depending on the type of cultured fish F, it is possible to determine when the cultured fish F are hungry by utilizing the number of grams of tails when the cultured fish F are hungry. Note that in Figure 6C, feeding is performed within a range of approximately 1080 seconds to 1190 seconds.

[0048] Next, when focusing on the movement direction of the cultured fish F at depth from the water surface W (e.g., the swimming direction of the fish, upward, downward), the calculation control unit 102 divides the linked echogram into multiple sections P and calculates the depth subtraction value, which is the depth of the starting point ES of the linked echo E in each section P minus the depth of the ending point EE of the linked echo E, as the distance difference (depth difference). Here, a positive distance difference indicates that the cultured fish F is moving downward (toward deeper water), and a negative distance difference indicates that the cultured fish F is moving upward (toward shallower water). The calculation control unit 102 then calculates the distance difference of each linked echo E in each section P, calculates the average value and variance of the distance differences of all the linked echoes E in section P, and arranges the calculated average value and variance of the distance differences in each section P to create a distance difference average value gram showing the change over time in the average value of the distance differences at each depth and a distance difference variance gram showing the change over time in the variance of the distance differences at each depth, as shown in FIG. 7A. Here, the larger the positive average value, the brighter the color of section P of the distance difference average value gram, and the larger the negative average value, the darker the color of section P. Furthermore, the larger the variance, the brighter the color of section P of the distance difference variance gram, and the smaller the variance, the darker the color of section P. If section P is all white at a given time, this indicates a time when measurement by the ultrasonic transmitter / receiver 20 has been stopped, and if section P is colored, then white indicates that no farmed fish F are present.

[0049] For example, in a cultured fish F that swims near the water surface W when hungry, as shown in FIG. 7A, the color of section P of the distance difference average value gram becomes brighter in shallower depth regions, and the color of section P of the distance difference variance gram also becomes brighter. Furthermore, in a cultured fish F that swims in deep water D even when hungry, the color of section P of the distance difference average value gram becomes brighter in deeper depth regions, and the color of section P of the distance difference variance gram also becomes brighter. Furthermore, in a cultured fish F that swims slightly shallower than the usual depth D when hungry, the color of section P of the distance difference average value gram becomes brighter in a predetermined shallower depth region, and the color of section P of the distance difference variance gram also becomes brighter. Therefore, depending on the type of cultured fish F, it is possible to determine when the cultured fish F is hungry by utilizing the distance difference average value gram and distance difference variance gram when the cultured fish F is hungry.

[0050] The calculation control unit 102 also divides the connected echogram into multiple sections P, and calculates, for the entire connected echo E in section P, a depth subtraction value obtained by subtracting the depth of the starting point ES of the connected echo E from the depth of the ending point EE, and a time subtraction value obtained by subtracting the time of the ending point EE of the connected echo E from the time of the starting point ES, and calculates the slope division value obtained by dividing the time subtraction value by the depth subtraction value as the slope of the connected echo E. Here, a positive slope means that the farmed fish F are moving downward (towards deeper water), and a negative slope means that the farmed fish F are moving upward (towards shallower water). The calculation control unit 102 then calculates the slope of each linked echo E present in section P, calculates the average value and variance of the slopes of all linked echoes E in section P, and assigns the calculated average value and variance of the slopes to each section P. As shown in FIG. 7B , the calculation control unit 102 creates an average value slope gram showing the change in the average value of the slope over time at each depth and an average value slope variance gram showing the change in the variance of the slope over time at each depth. Here, the larger the average value (positive value), the brighter the color of section P in the average value slope gram; the larger the average value (negative value), the darker the color of section P. Furthermore, the larger the variance, the brighter the color of section P in the slope variance gram; the smaller the variance, the darker the color of section P. If all sections P are white at a given time, this indicates a time when measurement by the ultrasonic transmitter / receiver 20 was stopped. If a section P has a color, white section P indicates that no farmed fish F are present.

[0051] For example, for cultured fish F that swim near the water surface W when hungry, the color of section P of the slope average value gram will be lighter in shallower depth regions, and the color of section P of the slope variance gram will be lighter, as shown in Figure 7B. Furthermore, for cultured fish F that swim in deep water D even when hungry, the color of section P of the slope average value gram will be lighter in deeper depth regions, and the color of section P of the slope variance gram will be lighter. Furthermore, for cultured fish F that swim slightly shallower than the usual depth D when hungry, the color of section P of the slope average value gram will be lighter in a predetermined shallower depth region, and the color of section P of the slope variance gram will be lighter. Therefore, depending on the type of cultured fish F, it is possible to determine whether the cultured fish F is hungry by utilizing the slope average value gram and slope variance gram when the cultured fish F is hungry.

[0052] In the above description, the calculation control unit 102 calculates the gradient of the connected echo E based on the entire connected echo E, but this is not limiting, and for example, the gradient of the connected echo E may be calculated based on echoes within a predetermined range including the peak of the connected echo E. In other words, by focusing on the peak of the connected echo E, it is possible to calculate the gradient of the connected echo E focusing on one of the farmed fish F.

[0053] Furthermore, when focusing on the variation in the movement direction of the cultured fish F at a depth from the water surface W (for example, variation in the swimming direction of the cultured fish F), the calculation control unit 102 divides the linked echogram into multiple sections P and calculates TS (Target Strength), which is the ratio of the intensity I1 of the ultrasonic waves transmitted from the ultrasonic transmitter / receiver 20 to the intensity I2 of the reflected ultrasonic waves, for the linked echo E in each section P. Here, TS is expressed by the following equation (1). Note that TS may be subjected to a predetermined normalization process as appropriate, and may be normalized as the reflection intensity converted into a distance of 1 m, for example.

[0054] TS=10×log(I2 / I1) (1)

[0055] Here, when the TS value is negative and its absolute value is small (for example, around -20 dB), it means that the reflection intensity is high and there are many farmed fish F moving in the deeper direction. On the other hand, when the TS value is negative and its absolute value is large (for example, around -60 dB), it means that the reflection intensity is low and there are many farmed fish F moving in the shallower direction. Furthermore, an extremely small TS value means that there are no farmed fish F. The calculation control unit 102 then calculates the TS of the linked echoes E present in section P for each linked echo E, calculates the average value and variance of the TS of all linked echoes E in section P, and places the calculated TS average value and variance in each section P to create a TS average value gram showing the change in the average value of TS at each depth over time and a TS variance gram showing the change in the TS variance at each depth over time, as shown in FIG. 8A. Here, the smaller the average value is, the brighter the color of section P in the TS average value gram; and the larger the average value is, the darker the color of section P is. Furthermore, the greater the variance, the brighter the color of section P of the TS variance gram, and the smaller the variance, the darker the color of section P. If section P is all white at a given time, this indicates a time when measurement by the ultrasonic transmitter / receiver 20 has been stopped, and if section P is colored and white, this indicates that no farmed fish F are present.

[0056] For example, for cultured fish F that swim near the water surface W when hungry, the color of section P in the TS average value gram will be lighter, and the color of section P in the TS variance gram will be lighter, as shown in FIG. 8A. Furthermore, for cultured fish F that swim at deep depths D even when hungry, the color of section P in the TS average value gram will be lighter, and the color of section P in the TS variance gram will be lighter in deeper areas. Furthermore, for cultured fish F that swim slightly shallower than the normal depth D when hungry, the color of section P in the TS average value gram will be lighter, and the color of section P in the TS variance gram will be lighter in predetermined shallower areas. Therefore, depending on the type of cultured fish F, it is possible to determine when cultured fish F is hungry by utilizing the TS average value gram and TS variance gram when the cultured fish F is hungry.

[0057] The calculation control unit 102 also calculates the fish length FL using the above-mentioned TS. The relationship between TS and the fish length FL is expressed by the following equation (2), where A is a constant determined by the signal frequency and the fish species.

[0058] TS = 20 × log(FL) + A (2)

[0059] Here, a large FL value indicates a large body length of the cultured fish F, and a small FL value indicates a small body length of the cultured fish F. The vertical movement of the cultured fish F is determined based on the depth distribution of the TS. The calculation control unit 102 then calculates the FL of each linked echo E present in section P, calculates the average value and variance of FL for all linked echoes E in section P, and places the calculated FL average value and variance in each section P. As shown in FIG. 8B , the calculation control unit 102 creates an FL average value gram showing the change in the average value of FL over time at each depth, and an FL variance gram showing the change in the variance of FL over time at each depth. Here, the smaller the average value is, the brighter the color of section P in the FL average value gram; and the larger the average value is, the darker the color of section P in the FL variance gram. Furthermore, the smaller the variance, the brighter the color of section P in the FL variance gram, and the larger the variance, the darker the color of section P. Furthermore, if all sections P are white at a given time, this indicates that measurement by the ultrasonic transmitter / receiver 20 has been stopped, and if section P is colored and section P is white, this indicates that no farmed fish F are present.

[0060] For example, for cultured fish F that swim near the water surface W when hungry, the color of section P of the FL average value gram will be lighter, and the color of section P of the FL variance gram will be lighter, as shown in Figure 8B. Furthermore, for cultured fish F that swim at deep depth D even when hungry, the color of section P of the FL average value gram will be lighter, and the color of section P of the FL variance gram will be lighter in deep areas. Furthermore, for cultured fish F that swim slightly shallower than the normal depth D when hungry, the color of section P of the FL average value gram will be lighter, and the color of section P of the FL variance gram will be lighter in predetermined shallow areas. Therefore, depending on the type of cultured fish F, it is possible to determine when cultured fish F is hungry by utilizing the FL average value gram and FL variance gram when the cultured fish F is hungry.

[0061] Incidentally, with regard to the above-mentioned TS and FL, the reflection of ultrasound on the farmed fish F depends on the angle of incidence of the ultrasound with respect to the body axis of the farmed fish F, and when the orientation of the farmed fish F changes, the reflected strength of the ultrasound changes significantly. Therefore, when there are various movements of the farmed fish F, the posture angles of the farmed fish F also differ for each farmed fish F, which causes variation. Depending on the degree of this variation, it is possible to determine whether the farmed fish F are hungry or full.

[0062] Furthermore, when focusing on the movement speed of the cultured fish F at a depth below the water surface W, the calculation control unit 102 divides the linked echogram into multiple sections P and calculates the number of linked points (also called the number of pings) that make up the linked echo E in each section P as the number of linked pings. Here, as shown in FIG. 9A , in the echogram under normal conditions, the cultured fish F swim slowly, so the reflected ultrasonic intensity is continuous. In contrast, in the echogram under fasting conditions, the cultured fish F moves quickly, so the reflected ultrasonic intensity is intermittent. This characteristic is utilized. The calculation control unit 102 then calculates the number of linked pings for each linked echo E in each section P, calculates the average value and variance of the number of linked pings for all linked echoes E in section P, and places the calculated average value and variance of the number of linked pings in each section P. As shown in FIG. 9B , a linked ping average value gram showing the change over time in the average value of the number of linked pings at each depth and a linked ping variance gram showing the change over time in the variance of the number of linked pings at each depth are created. Here, the larger the average value, the brighter the color of section P of the linked ping count average value gram, and the smaller the average value, the darker the color of section P. Furthermore, the larger the variance, the brighter the color of section P of the linked ping count variance gram, and the smaller the variance, the darker the color of section P. Furthermore, if section P is all white at a given time, this indicates a time when measurement by the ultrasonic transmitter / receiver 20 has been stopped, and if section P is colored, then white indicates that no farmed fish F are present.

[0063] Here, for example, for cultured fish F that swim near the water surface W when hungry, as shown in Figure 9B, the color of section P of the linked ping count average value gram will be lighter in shallower depth areas, and the color of section P of the linked ping count variance gram will also be lighter. Also, for cultured fish F that swim in deep water D even when hungry, the color of section P of the linked ping count average value gram will be lighter in deeper depth areas, and the color of section P of the linked ping count variance gram will also be lighter. And for cultured fish F that swim slightly shallower than the usual deep water D when hungry, the color of section P of the linked ping count average value gram will be lighter in a predetermined shallower depth area, and the color of section P of the linked ping count variance gram will also be lighter. Therefore, depending on the type of cultured fish F, it is possible to determine when cultured fish F is hungry by utilizing the linked ping count average value gram and linked ping count variance gram when cultured fish F is hungry.

[0064] In the above description, the calculation control unit 102 calculated the number of linked pings of the linked echo E as the movement speed of the cultured fish F. However, this is not limited to this. For example, if the distance from the ultrasonic transmitter / receiver 20 to the target of the cultured fish F does not change significantly from ping to ping, it can be determined that the cultured fish F is moving perpendicular to the ultrasonic transmitter / receiver 20, that is, the cultured fish F is swimming left and right. Therefore, the calculation control unit 102 may calculate the distance corresponding to the directivity angle from the directivity angle of the ultrasonic transmitter / receiver 20 and the distance to the target of the cultured fish F in the linked echogram, and divide this distance by the number of pings to calculate the movement speed of the cultured fish F. Alternatively, if the distance from the ultrasonic transmitter / receiver 20 to the target of the cultured fish F changes significantly from ping to ping, it can be assumed that the cultured fish F is moving along the irradiation direction (irradiation axis) of the ultrasonic transmitter / receiver 20 or within the range of the directivity angle of the ultrasonic transmitter / receiver 20. Therefore, the calculation control unit 102 may calculate the movement speed of the farmed fish F by dividing the distance between connected echoes E between pings in the connected echogram by the number of pings.

[0065] Incidentally, an echogram can also detect bait as an echo. For example, as shown in FIG. 10A , the inventors installed an ultrasonic transmitter / receiver 20 at the top of a cylindrical aquarium 4. When bait was dropped from the top, ultrasonic waves were transmitted to the bait, and the change in the reflection intensity of the ultrasonic waves reflected from the bait was obtained as an echogram. As a result, as shown in FIG. 10A , a linear bait echo FE was detected, representing the bait gently falling from top to bottom. It was also understood that the bait echo FE has a relatively weak reflection intensity. Therefore, in this invention, when focusing on the feeding status of the cultured fish F at a depth below the water surface W, the calculation control unit 102 sets a range of bait reflection intensity that is weaker than the reflection intensity of the cultured fish F, and detects echoes within the set bait reflection intensity range as bait echo FE in the echogram. This allows for the bait echo FE to be separated from the echoes of the cultured fish F. The calculation control unit 102 then detects the distribution of the bait echo FE and calculates the number of bait echoes FE. If the farmed fish F is hungry, the feed echo FE is not detected, and if the farmed fish F is full, the feed echo FE is detected.

[0066] Here, for example, as shown in Figure 10B, when the farmed fish F becomes hungry, the feed echo FE is not detected, but when the farmed fish F becomes full, it loses interest in the feed and the feed echo FE is detected. Here, because the feed falls downward over time, the feed echo FE is detected in a linear pattern from above toward the lower right. In this way, by using the feed echo FE, it is possible to determine whether the farmed fish F is hungry or full.

[0067] Furthermore, when the cultured fish F in the cage are full, the bait passes through the cage net N. Since wild fish NF wander below the cage net N, the wild fish NF ingest the dropped bait. As shown in FIG. 10C , the inventor conducted repeated tests and found that a linear echo NE having a predetermined width corresponding to the cage net N can be detected in the echogram. When the cultured fish F are full, an echo NFE of the wild fish NF below the cage net N can be detected. Therefore, in this invention, when focusing on the feeding status of the cultured fish F at a depth below the water surface W, the calculation control unit 102 detects a linear echo having a predetermined width in the echogram as the echo NE of the cage net N, and detects the echo NFE of the wild fish NF directly below the echo NE of the cage net N. This makes it possible to separate the echoes of the cultured fish F from the echoes of the wild fish NF. The calculation control unit 102 then detects the distribution of the number and depth of the echoes NFE of the wild fish NF. If the farmed fish F are hungry, the wild fish NF will not gather below the net N of the fish preserve, and therefore the echoes NFE of the wild fish NF will not be detected. If the farmed fish F are full, the feed passes through the net N of the fish preserve, and the wild fish NF will gather below the net N of the fish preserve, and therefore the echoes NFE of the wild fish NF will be detected. Furthermore, the level of remaining feed can be evaluated based on the number of echoes NFE of the wild fish NF. In Figure 10C, feeding is carried out in the ranges of 0 to 240 seconds and 850 to 900 seconds.

[0068] Here, for example, as shown in Figure 10C, when the farmed fish F becomes hungry, the echo NFE of the wild fish NF is not detected, but when the farmed fish F becomes full, the echo NFE of the wild fish NF is detected. In this way, by using the echo NFE of the wild fish NF, it is possible to determine whether the farmed fish F is hungry or full.

[0069] As described above, in the present invention, the distribution of the farmed fish F at depth from the water surface W is defined as the number of linked echoes, the movement direction of the farmed fish F is defined as the distance difference between the linked echoes and the slope of the linked echo, the variation in the movement direction of the farmed fish F is defined as the TS of the linked echo and the fish body length FL of the linked echo, the movement speed of the farmed fish F is defined as the number of linked pings, and the feeding status of the farmed fish F is defined as the echo FE of the feed and the echo NFE of the wild fish NF. Note that the present invention employs any one or a combination of these pieces of fish behavior information, but is not limited to these, and behavior information of the farmed fish F calculated by analyzing the echogram may be added as appropriate.

[0070] Now, once the calculation control unit 102 has completed calculating the fish behavior information, the estimation control unit 103 of the control device 30 then uses the calculated fish behavior information to estimate the feeding activity of all the farmed fish F in the fish pen (Figure 4: S105).

[0071] Here, the estimation control unit 103 may estimate the feeding activity level using any method. However, the estimation control unit 103 may estimate the feeding activity level according to the type of calculated fish behavior information. For example, if the calculated fish behavior information is only one item, the distribution of the cultured fish F, the estimation control unit 103 estimates the feeding activity level by quantifying the value of one piece of fish behavior information. Furthermore, if the calculated fish behavior information is two or more items, such as the distribution of the cultured fish F and the movement direction of the cultured fish F, the estimation control unit 103 may quantify the two or more pieces of fish behavior information and estimate the combined value of the respective fish behavior information as the feeding activity level, or may assign a predetermined weight to each piece of fish behavior information and estimate the combined value of the weighted values ​​as the feeding activity level. Here, depending on the type of cultured fish F and the type of fish pen, one piece of fish behavior information may be a signal and the other piece of fish behavior information may be noise. Therefore, the fish farmer or the like may appropriately select and combine pieces of fish behavior information to estimate the feeding activity level of the entire cultured fish F.

[0072] Here, if the value of feeding activity, which is a quantification of fish behavior information, is large, it can be determined that the entire farmed fish F is active in terms of feeding, and on the other hand, if the value of feeding activity is small, it can be determined that the entire farmed fish F is passive in terms of feeding.

[0073] Furthermore, as a setting of feeding activity, for example, the estimation control unit 103 may regard one piece of fish behavior information in which numerical values ​​are assigned to each section in the linked echogram as one pattern, and estimate the pattern of one piece of fish behavior information as feeding activity, or may regard two or more pieces of fish behavior information as patterns, and estimate an integrated pattern obtained by integrating the patterns of each piece of fish behavior information as feeding activity.

[0074] Here, if the color (value) of the patterned feeding activity classification is bright (for example, if there are a predetermined number or more bright colored classifications), it is possible to determine that the entire farmed fish F is active in response to feeding, whereas if the color of the feeding activity classification is dark (for example, if there are a predetermined number or more dark colored classifications), it is possible to determine that the entire farmed fish F is reluctant to feed. When feeding activity is a pattern, it is possible to determine whether the entire farmed fish F is active by taking into account the position of the classification in addition to the color of the classification, allowing for more detailed responses.

[0075] Once the estimation control unit 103 has completed the estimation of the feeding activity, the executive control unit 104 of the control device 30 then controls the start of feeding by the feeding device 10, which supplies food to the fish pen, according to the estimated feeding activity. Specifically, the executive control unit 104 determines whether the feeding activity has changed (Fig. 4: S106).

[0076] Here, there is no particular limitation on the determination method of the executive control unit 104, but for example, when the feeding activity is a value, the executive control unit 104 detects that the feeding device 10 has not started feeding, and then obtains a first threshold for starting feeding from a predetermined memory. Here, the first threshold is appropriately designed depending on the type of farmed fish F and the type of fish pen, with reference to the selection of the fish behavior information and the number of feeding activity levels described above. Next, the executive control unit 104 determines whether the estimated feeding activity is equal to or greater than the first threshold.

[0077] Here, for example, as shown in FIG. 11A, when the cultured fish F are full, they swim slowly below the cage, and the value of the feeding activity decreases. As a result, the feeding activity is determined to be less than the first threshold, and the executive control unit 104 determines that the feeding activity does not change and that the fish F in the cage are full (FIG. 4: S106 NO). In this case, the process returns to S102, and the process from the acquisition control unit 101 irradiating ultrasound (FIG. 4: S102) to the estimation control unit 103 estimating the feeding activity (FIG. 4: S105) is repeated. This prevents the feeding device 10 from feeding the cultured fish F even when they are full, resulting in the waste of the food 11.

[0078] On the other hand, for example, in the case of farmed fish F that move quickly upward toward the water surface W when hungry, the value of the feeding activity level increases, as shown in Figure 11B. As a result of the determination, the feeding activity level becomes equal to or greater than the first threshold, and the executive control unit 104 determines that the feeding activity level has changed and that the fish F in the fish pen are hungry (Figure 4: S106 YES). Then, the executive control unit 104 executes the feeding device 10 to start feeding (Figure 4: S107).

[0079] Here, there are no particular limitations on the method by which the execution control unit 104 starts feeding, but for example, the execution control unit 104 sends a feeding start signal to the feeding device 10, opens the lid 102 of the feeding device 10, and scatters the feed 11 in the container 100 into the cage through the hole 101. This allows the farmed fish F to be fed at the right time when they are hungry, thereby optimizing feeding.

[0080] While the above description describes a determination of whether the feeding activity level is equal to or greater than the first threshold, the present invention is not limited to this. For example, it is also possible to determine whether the change (difference) in the feeding activity level estimated over time is equal to or greater than the first threshold, or whether the change (difference) in the feeding activity level over time is less than the first threshold. As described above, the behavior of farmed fish F when hungry is diverse, so it is necessary to determine the change in feeding activity level and use the above-mentioned determination method differently depending on the type of farmed fish F. Furthermore, while the above description describes a case where the feeding activity level is a value, the present invention is not limited to this. For example, if the feeding activity level is a pattern, the executive control unit 104 may determine whether the feeding activity level has changed by determining whether the pattern of feeding activity level corresponds to the first pattern for starting feeding. Here, when determining whether a feeding activity pattern corresponds to a first pattern, for example, it is sufficient to perform pattern matching between the feeding activity pattern and the first pattern to determine whether the shape of the feeding activity pattern matches (is close to) the shape of the first pattern. Alternatively, for example, it is sufficient to input the feeding activity pattern and the first pattern into a predetermined clustering means or classifier to determine the attributes of the feeding activity pattern and the first pattern, and then determine whether the attributes of the feeding activity pattern match (are close to) the attributes of the first pattern.

[0081] Furthermore, in the above description, the execution control unit 104 starts feeding (FIG. 4: S107) when the feeding activity level changes (FIG. 4: S106: YES). However, this is not limiting. For example, it may determine whether the feeding activity level has remained constant for a predetermined period of time (FIG. 4: S106), and if the feeding activity level is constant (FIG. 4: S106: YES), start feeding (FIG. 4: S107). That is, as described above, behaviors of farmed fish F when hungry vary widely depending on the type of fish. For example, when hungry, farmed fish F may swim slowly in a deep area as usual for a certain period of time, with their feeding activity level remaining almost unchanged. To address this situation, it is possible to start feeding in response to the above-described hunger state of the farmed fish F by determining whether the feeding activity level has remained constant for a predetermined period of time (FIG. 4: S106), and if the feeding activity level is constant (FIG. 4: S106: YES), start feeding. Here, the predetermined period is set appropriately depending on the type of farmed fish F.

[0082] In the above description, the execution control unit 104 is configured to automatically start feeding by the feeding device 10. However, this is not limiting. For example, the execution control unit 104 may display fish behavior information and feeding activity on the display unit in real time, and when the feeding activity changes, display a selectable feeding start key. Then, when a fish farmer or the like decides the timing to start feeding while looking at the display unit and selects the feeding start key, the execution control unit 104 may accept the selection of the feeding start key and start feeding by the feeding device 10. This allows the start of feeding to be controlled based on the user's intuition and judgment in addition to the mechanical criteria of the control device 30.

[0083] Furthermore, if the farmer selects the normal feeding setting key in S101, the following occurs: The control device 30 accepts the selection of the normal feeding setting key (FIG. 4: S101 NO), and the execution control unit 104 executes the feeding device 10 to start feeding (FIG. 4: S107).

[0084] Here, the method by which the execution control unit 104 starts feeding is not particularly limited. For example, the execution control unit 104 may accept a configurable time or hour for starting feeding, or may accept the selection of a feeding start key. Thus, when the farmer or the like sets a specific time (e.g., "1 minute"), the execution control unit 104 executes feeding at specific time intervals. Alternatively, when the farmer or the like sets a specific time (e.g., "9:00"), the execution control unit 104 executes feeding at the specific time. Alternatively, when the farmer or the like manually selects the feeding start key, the execution control unit 104 executes feeding from the point in time when the selection of the feeding start key is accepted. By allowing flexibility in the timing of starting feeding in this way, since the behavior of farmed fish F when hungry is diverse and difficult to predict, as described above, smooth feeding can be achieved by the farmer or the like arbitrarily deciding on feeding to a certain extent.

[0085] Furthermore, as described above, since it is difficult to determine the behavior of farmed fish F when hungry depending on the type of farmed fish F, for example, the feeding activity level at the start of feeding for a specific farmed fish F may be stored in a database, and the hunger time or appropriate feeding start time for the specific farmed fish F may be estimated based on the changes in feeding activity level and the range of feeding activity level in the database, and feeding may be started at the estimated hunger time or appropriate feeding start time for the specific farmed fish F. The frequency of feeding start may be increased or decreased based on the feeding activity level in the database. Typically, when there is a large change in feeding activity level, it can be assumed that the hunger level of the farmed fish F is high. As a result, since specific farmed fish F are fed repeatedly every day, the behavior of the specific farmed fish F can be understood based on the changes in feeding activity level of the specific farmed fish F at that time, and more appropriate feeding can be achieved.

[0086] When the feed 11 is scattered in the fish pen, the farmed fish F gather close together toward the water surface W of the fish pen and start eating the feed 11, as shown in Figure 12A. When feeding is started by the execution control unit 104, the feeding start program of the control device 30 ends, and then the feeding stop program of the control device 30 is executed. The feeding stop program may also be executed by the fish farmer or the like selecting a key.

[0087] Specifically, when the executive control unit 104 continues feeding (FIG. 4: S201), the acquisition control unit 101 then uses the multiple ultrasonic transmitter / receivers 20 to irradiate ultrasonic waves over a wide range from the water surface W to the depth D of the fish pen for the cultured fish F, as described above (FIG. 4: S202). Next, the acquisition control unit 101 acquires an echogram based on the reflection intensity of the ultrasonic waves received by the ultrasonic transmitter / receivers 20 (FIG. 4: S203).

[0088] Here, when feeding of the farmed fish F has begun, it can be seen that the echogram shows that many echoes of the farmed fish F are detected in shallow areas, as shown in Figure 12B. In Figure 12B, feeding is carried out in the ranges of 0 to 300 seconds and 600 to 1080 seconds.

[0089] Once the acquisition control unit 101 acquires the echogram, the calculation control unit 102 then analyzes the echogram to calculate fish behavior information (FIG. 4: S204).

[0090] Here, there are no particular limitations on the method by which the calculation control unit 102 calculates the fish behavior information, but for example, as described above, the fish behavior information is any one or a combination of the distribution of the farmed fish F at the depth from the water surface W, the movement direction of the farmed fish F, the movement speed of the farmed fish F, and the feeding status of the farmed fish F. The calculation method is the same as described above, so it will be omitted.

[0091] For example, when the calculation control unit 102 calculates the number of connected echoes as the distribution of the cultured fish F at depth from the water surface W, it creates a connected echogram from the echogram, and then creates a tail count gram from the connected echogram, which shows the change in the number of cultured fish F at each depth over time, as shown in Figure 12C. Here, as the cultured fish F become full, the color of the tail count gram division P becomes brighter over time in shallower areas. In other words, by using the number of cultured fish F, it is possible to determine that the cultured fish F are full.

[0092] Furthermore, when the calculation control unit 102 calculates the distance difference of the linked echoes as the movement direction of the cultured fish F at a depth from the water surface W, it creates a distance difference average value gram that shows the change over time in the average value of the distance differences at each depth, and a distance difference variance gram that shows the change over time in the variance of the distance differences at each depth, as shown in Figure 13A. Here, when the cultured fish F becomes full, for example, in shallow areas, the color of section P of the distance difference average value gram becomes brighter over time, and the color of section P of the distance difference variance gram becomes brighter. In other words, by using the distance difference of the linked echoes, it is possible to determine that the cultured fish F is full.

[0093] Furthermore, when the calculation control unit 102 calculates the gradient of the linked echo as the movement direction of the cultured fish F at a depth from the water surface W, it creates a gradient average value gram showing the change over time in the average value of the gradient at each depth, and a gradient variance gram showing the change over time in the variance of the gradient at each depth, as shown in Figure 13B. Here, when the cultured fish F become full, for example, in the deeper regions, the color of section P in the gradient average value gram becomes lighter over time, and the color of section P in the gradient variance gram becomes lighter. In other words, by using the gradient of the linked echo, it is possible to determine that the cultured fish F are full.

[0094] Furthermore, when the calculation control unit 102 calculates the TS of the linked echoes as the variation in the movement direction of the cultured fish F at different depths from the water surface W, it creates a TS average value gram showing the change over time in the average value of TS at each depth, and a TS variance gram showing the change over time in the variance of TS at each depth, as shown in Figure 14A. Here, when the cultured fish F become full, the color of section P in the TS average value gram and the color of section P in the TS variance gram become lighter over time in the deeper regions. In other words, by using the TS of the linked echoes, it is possible to determine that the cultured fish F are full.

[0095] Furthermore, when the calculation control unit 102 calculates the fish body length FL of the linked echo as the variation in the movement direction of the cultured fish F at the depth from the water surface W, it creates an FL average value gram showing the change over time in the average value of FL at each depth, and an FL variance gram showing the change over time in the variance of FL at each depth, as shown in Figure 14B. Here, when the cultured fish F becomes full, the color of section P of the FL average value gram becomes lighter over time in the deeper areas, and the color of section P of the FL variance gram becomes lighter. In other words, by using the FL of the linked echo, it is possible to determine that the cultured fish F is full.

[0096] 15A, in the echogram taken during feeding, the cultured fish F moves quickly and actively, resulting in intermittent ultrasonic reflection intensity, whereas in the echogram taken during full stomach, the cultured fish F swims slowly, resulting in continuous ultrasonic reflection intensity. Therefore, when the calculation control unit 102 calculates the number of linked pings as the movement speed of the cultured fish F at a depth from the water surface W, it creates a linked ping number average value gram showing the change over time in the average value of the number of linked pings at each depth, and a linked ping number variance gram showing the change over time in the variance of the number of linked pings at each depth, as shown in FIG. 15B. Here, when the cultured fish F becomes full, for example, in the deeper regions, the color of section P in the linked ping number average value gram and the color of section P in the linked ping number variance gram become lighter over time. In other words, by using the number of linked pings, it is possible to determine whether the cultured fish F is full.

[0097] Furthermore, when the calculation control unit 102 calculates the echo FE of the bait and the echo NFE of the wild fish NF as the feeding status of the cultured fish F at a depth below the water surface W, as shown in Figures 10B and 10C, it detects the echo FE of the bait from the echogram, detects the net N of the fish preserve, and then detects the echo NFE of the wild fish NF below the net N. Here, when the cultured fish F are full, the echo FE of the bait and the echo NFE of the wild fish NF are detected, and by utilizing these, it can be determined that the cultured fish F are full.

[0098] Once the calculation control unit 102 has completed the calculation of the fish behavior information, the estimation control unit 103 then uses the calculated fish behavior information to estimate the feeding activity of all the cultured fish F in the fish pen (FIG. 4: S205).

[0099] Here, the estimation control unit 103 may estimate the feeding activity level by any method, but for example, as described above, the feeding activity level may be estimated by quantifying a single piece of fish behavior information, or by adding up the quantified values ​​of multiple pieces of fish behavior information, or by weighting each piece of fish behavior information and estimating the sum of the weighted values ​​as the feeding activity level. Also, as described above, one piece of fish behavior information may be regarded as a pattern, and the pattern of one piece of fish behavior information may be estimated as the feeding activity level, or an integrated pattern obtained by integrating two or more patterns of fish behavior information may be estimated as the feeding activity level. Furthermore, the feeding activity level at the start of feeding and the feeding activity level at the stop of feeding may be the same or different.

[0100] After the estimation control unit 103 has completed the estimation of the feeding activity level, the executive control unit 104 controls the stopping of feeding according to the change in the estimated feeding activity level. Specifically, the executive control unit 104 determines whether the feeding activity level has changed.

[0101] Here, there are no particular limitations on the determination method of the executive control unit 104. For example, when the feeding activity level is a value, the executive control unit 104 detects that the feeding device 10 is continuing feeding and then obtains a second threshold for stopping feeding from a predetermined memory. Here, the second threshold is appropriately designed depending on the type of farmed fish F and the type of fish pen, taking into account the selection of the fish behavior information and the number of feeding activity levels described above. Furthermore, the second threshold may be the same as or different from the first threshold. Next, the executive control unit 104 determines whether the estimated feeding activity level has become less than the second threshold.

[0102] Here, as shown in FIG. 16A, when the cultured fish F is being fed, the cultured fish F moves quickly on the water surface W of the fish pen in search of food, and the value of the feeding activity level increases. As a result of the determination, the feeding activity level becomes equal to or greater than the second threshold, and the executive control unit 104 determines that the feeding activity level has not changed and that the fish F in the fish pen are hungry (FIG. 4: S206 NO). In this case, the process returns to S202, and the processes from the acquisition control unit 101 irradiating ultrasound (FIG. 4: S202) to the estimation control unit 103 estimating the feeding activity level (FIG. 4: S205) are repeated. This makes it possible to prevent feeding from being stopped even when the cultured fish F is hungry, and to prevent feeding from being stopped before the cultured fish F is fully fed.

[0103] On the other hand, as shown in Figure 16B, when the cultured fish F are full, they move slowly downward toward the depth D of the fish pen, and the value of the feeding activity decreases. As a result, the feeding activity becomes less than the second threshold, and the executive control unit 104 determines that the feeding activity has changed and that the cultured fish F in the fish pen are full (Figure 4: S206 YES). Then, the executive control unit 104 stops feeding by the feeding device 10 (Figure 4: S207).

[0104] Here, there are no particular limitations on the method by which the executive control unit 104 stops feeding, but for example, the executive control unit 104 sends a feeding stop signal to the feeding device 10, which closes the lid 102 of the feeding device 10 and stops the dispersal of the feed 11 in the container 100. This allows feeding to be stopped in a timely manner when the cultured fish F become full, thereby optimizing feeding. In this way, the present invention makes it possible to optimize feeding by the feeding device by specifically estimating the behavior of all the cultured fish in the fish pen.

[0105] In the above description, it is determined whether the feeding activity level has fallen below the second threshold value, but the present invention is not limited to this. For example, it may be determined whether the change (difference) in the feeding activity level estimated over time has fallen below the second threshold value. In addition, in the above description, the feeding activity level is a value, but the present invention is not limited to this. For example, if the feeding activity level is a pattern, the execution control unit 104 may determine whether the feeding activity level has changed by determining whether the pattern of the feeding activity level corresponds to the second pattern for stopping feeding. The method of determining the pattern is the same as described above.

[0106] In the above description, the execution control unit 104 is configured to automatically stop feeding by the feeding device 10. However, this is not limiting. For example, after the feeding device 10 starts feeding, the execution control unit 104 displays fish behavior information and feeding activity on the display unit in real time, and if the feeding activity changes, displays a feed stop key so that it can be selected. The user can then determine the timing to stop feeding while looking at the display unit and select the feed stop key. The execution control unit 104 can then accept the user's selection of the feed stop key and cause the feeding device 10 to stop feeding. This allows feeding stop control to be based on the user's intuition and judgment in addition to the mechanical criteria of the control device 30.

[0107] Furthermore, the execution control unit 104 may, apart from the judgment process, always display a feed start key or a feed stop key that can be selected, so that the user can select the feed start key or the feed stop key at any time while looking at the display unit.

[0108] In the above description, the calculation control unit 102 calculates preset fish behavior information, the estimation control unit 103 estimates preset feeding activity, and the execution control unit 104 uses preset first thresholds (first pattern) and second thresholds (second pattern). However, the present invention is not limited to these. For example, previous echogram data, fish behavior information, feeding activity, first threshold, and second threshold may be input to a predetermined machine learning unit, and the machine learning unit may output optimal fish behavior information, feeding activity, first threshold, and second threshold, and the calculation control unit 102, estimation control unit 103, and execution control unit 104 may use the output optimal values. Furthermore, the machine learning unit may be input with information about the fish tank from which echograms are obtained and information about the farmed fish F, and may output fish behavior information, feeding activity, first threshold, and second threshold taking into account the correlation between various parameters. This allows the machine learning unit to actively output data and values ​​that the user is not aware of, thereby achieving optimal feeding using the machine learning unit.

[0109] There is no particular limitation on the installation location of the automatic feeding support device 1, and for example, as shown in Figure 17A, the automatic feeding support device 1 may be installed at or near the center of the fish tank, or at or near one end of the fish tank, as shown in Figure 17B. Of course, the shape of the fish tank is not limited to a rectangular parallelepiped, and various shapes such as a cylinder can be used. Furthermore, depending on the installation location of the automatic feeding support device 1, the arrangement method of the ultrasonic transmitter / receiver 20 and the direction in which it transmits and receives ultrasonic waves can be changed as appropriate.

[0110] Furthermore, the configuration of the ultrasonic transmitter / receiver 20 is not particularly limited, and in addition to the fixed configuration shown in Fig. 2, a scanning configuration can be used. As a scanning configuration, for example, as shown in Fig. 18A, a configuration in which ultrasonic transmitter / receiver units 21 are arranged in a fan shape and rotated around a rotation axis 22a by a motor 22 to scan the inside of the fish pen can be used. Also, as shown in Fig. 18B, a configuration in which ultrasonic transmitter / receiver units 21 are arranged in a line can be used.

[0111] Here, there is no particular limitation on the method of using the scanning ultrasonic transmitter / receiver 20, but for example, as shown in Figure 19A, the ultrasonic transmitter / receiver 20 may be installed at or near the center of the fish tank and rotated 360 degrees, or as shown in Figure 19B, the ultrasonic transmitter / receiver 20 may be installed at or near one end of the fish tank and moved back and forth.

[0112] Next, an example of the present invention will be described. First, the inventors used five ultrasonic transmitters / receivers 20a to 20e to obtain echograms of farmed fish F of a predetermined fish species A (red sea bream). Here, the measurement time for one measurement was 300 seconds, and after one measurement, the measurement was stopped for several minutes, and then the next measurement was resumed. As shown in Figure 20, the echogram contains data for a total of 3000 seconds, including the time when the measurement was stopped. It can be seen that the feeding activity of farmed fish F of fish species A increases and changes at a predetermined time point.

[0113] Next, feeding was started and the situation thereafter was observed. As shown in Figure 21, at the initial stage, feeding had just started so feeding activity was high, but as farmed fish F became full, feeding activity gradually decreased, indicating a change. In this way, feeding activity can be intuitively grasped by checking only the echogram.

[0114] The inventor also used five ultrasonic transmitters / receivers 20a to 20e to obtain echograms of farmed fish F of another fish species B (chub mackerel). As shown in Figure 22, it can be seen that the farmed fish F of fish species B is moving back and forth to the water surface, and therefore the feeding activity level increases and changes at a predetermined cycle.

[0115] Next, feeding was started and the situation was observed afterwards. As shown in Figure 23, the feeding activity was high at the initial stage, but as the farmed fish F became full, the feeding activity gradually decreased, demonstrating a change. After a certain time had passed, echoes of the bait 11, the fish cage net N, and the wild fish NF were confirmed.

[0116] Note that the tail count grams in Figure 6C, the distance difference average value grams and distance difference variance grams in Figure 7A, the slope average value grams and slope variance grams in Figure 7B, the TS average value grams and TS variance grams in Figure 8A, the FL average value grams and FL variance grams in Figure 8B, the linked ping average value grams and linked ping number variance grams in Figure 9B, the tail count grams in Figure 12C, the distance difference average value grams and distance difference variance grams in Figure 13A, the slope average value grams and slope variance grams in Figure 13B, the TS average value grams and TS variance grams in Figure 14A, the FL average value grams and FL variance grams in Figure 14B, and the linked ping average value grams and linked ping number variance grams in Figure 15B use echograms of fish species A and B.

[0117] Furthermore, the inventors used five ultrasonic transmitters / receivers 20a to 20e to obtain echograms of farmed fish F of another fish species C (striped jack). As shown in Figure 24, it can be seen that the farmed fish F of fish species C are always crowded on the water surface, and therefore their feeding activity is always high and fluctuating.

[0118] Next, feeding was started and the subsequent behavior was observed. As shown in Figure 25, the feeding activity was high at the initial stage, but as farmed fish F became full, the feeding activity gradually decreased, indicating a change.

[0119] Here, we show the results of calculating various fish behavior information for fish species C. Fig. 26A shows the number of tails grams for fish species C, Fig. 26B shows the average distance difference grams and distance difference variance grams for fish species C, Fig. 27A shows the average tilt value grams and tilt variance grams for fish species C, Fig. 27B shows the average TS value grams and TS variance grams for fish species C, Fig. 28A shows the average FL value grams and FL variance grams for fish species C, and Fig. 28B shows the average number of connected pings grams and variance grams for fish species C. As shown in Figs. 26 to 28, although the behavior of fish species C differs significantly from that of fish species A and B, it can be seen that certain patterns are observed.

[0120] As such, although the behavior of farmed fish F varies depending on the type, certain patterns can be seen, and as described above, it is possible to utilize fish behavior information to optimize feeding.

[0121] In the embodiment of the present invention, the automatic feeding support system 1 is configured to include each control unit, but it may also be configured to store a program that realizes each control unit on a storage medium and provide the storage medium. In this configuration, the program is read by a device, and the device realizes each control unit. In this case, the program itself read from the recording medium achieves the effects of the present invention. Furthermore, it is also possible to provide a method for storing the steps executed by each control unit on a hard disk. [Industrial Applicability]

[0122] As described above, the automatic feeding support device and automatic feeding support method of the present invention are extremely useful in all fields, including fish farming, as well as the fisheries, marine, and livestock industries, and are effective as a fish number estimation system and method that can specifically estimate the behavior of all farmed fish in a fish pen, thereby optimizing feeding by the feeding device. [Explanation of symbols]

[0123] 1 Automatic feeding support device 10 Feeding Device 20 Ultrasonic Transmitter / Receiver 30 Control device 101 Acquisition control unit 102 Calculation control unit 103 Estimation control unit 104 Execution control section

Claims

1. an acquisition control unit that acquires an echogram showing a time series of changes in the reflection intensity of the ultrasonic waves with respect to the depth from the water surface of the fish pen by using a plurality of ultrasonic transmitter-receivers arranged so as to face directly below the fish pen containing the farmed fish, diagonally downward to the left of the fish pen, diagonally downward to the right of the fish pen, diagonally downward in front of the fish pen, and diagonally downward in back of the fish pen, and irradiates ultrasonic waves over a wide range from the water surface to the depth of the fish pen; a calculation control unit that creates a connected echo by connecting points of the strongest reflection intensity of adjacent echoes within a predetermined depth range in the echogram, and creates a connected echo that shows changes in the connected echo at each depth over time by repeating the creation of the connected echo for the echogram, and regards the connected echo of the connected echogram as the farmed fish, and calculates fish behavior information that shows the distribution of the farmed fish at a depth from the water surface, the movement direction of the farmed fish, the movement speed of the farmed fish, the feeding status of the farmed fish, or a combination of these; an estimation control unit that estimates the feeding activity of all the cultured fish in the fish cage using the calculated fish behavior information; an execution control unit that detects that a feeding device that supplies feed to the fish pen continues feeding, obtains a predetermined threshold value from a predetermined memory, determines whether a previously estimated feeding activity level has fallen below the threshold value, and stops feeding by the feeding device when the feeding activity level has fallen below the threshold value; An automatic feeding assistance device equipped with:

2. an acquisition control unit that acquires an echogram showing a time series of changes in the reflection intensity of the ultrasonic waves with respect to the depth from the water surface of the fish pen by using a plurality of ultrasonic transmitter-receivers arranged so as to face directly below the fish pen containing the farmed fish, diagonally downward to the left of the fish pen, diagonally downward to the right of the fish pen, diagonally downward in front of the fish pen, and diagonally downward in back of the fish pen, and irradiates ultrasonic waves over a wide range from the water surface to the depth of the fish pen; a calculation control unit that sets a range of reflection intensities of feed that are weaker than the reflection intensities of the farmed fish, detects echoes in the echogram that fall within the set range of reflection intensities of feed as feed echoes, detects the distribution of the feed echoes, and calculates the number of feed echoes as the feeding status of the farmed fish, or detects linear echoes with a predetermined width in the echogram as echoes of a net for a fish cage, detects echoes of wild fish directly below the echoes of the net for a fish cage, and calculates the number of echoes of the wild fish or their distribution, including depth, as the feeding status of the farmed fish; an estimation control unit that estimates the feeding activity of all the cultured fish in the fish cage using the calculated fish behavior information; an execution control unit that detects that a feeding device that supplies feed to the fish pen continues feeding, obtains a predetermined threshold value from a predetermined memory, determines whether a previously estimated feeding activity level has fallen below the threshold value, and stops feeding by the feeding device when the feeding activity level has fallen below the threshold value; An automatic feeding assistance device equipped with:

3. an acquisition control process for acquiring an echogram showing a time series of changes in the reflection intensity of the ultrasonic waves with respect to the depth from the water surface of the fish pen, using a plurality of ultrasonic transmitter-receivers arranged so as to face directly below the fish pen containing the farmed fish, diagonally downward to the left of the fish pen, diagonally downward to the right of the fish pen, diagonally downward in front of the fish pen, and diagonally downward in back of the fish pen, and irradiating ultrasonic waves over a wide range from the water surface to the depth of the fish pen; a calculation and control process for creating a connected echo by connecting points of the strongest reflection intensity of adjacent echoes within a predetermined depth range in the echogram, and creating a connected echo that shows the change in the connected echo over time at each depth by repeating the creation of the connected echo for the echogram, and regarding the connected echo of the connected echogram as the farmed fish, and calculating fish behavior information that shows the distribution of the farmed fish at a depth from the water surface, the movement direction of the farmed fish, the movement speed of the farmed fish, the feeding status of the farmed fish, or a combination of these; an estimation control step of estimating the feeding activity of all the cultured fish in the fish cage using the calculated fish behavior information; an execution control process for detecting a state in which a feeding device that supplies feed to the fish pen continues feeding, retrieving a predetermined threshold value from a predetermined memory, determining whether a previously estimated feeding activity level has fallen below the threshold value, and stopping feeding by the feeding device when the feeding activity level has fallen below the threshold value; An automatic feeding assistance method comprising:

4. an acquisition control process for acquiring an echogram showing a time series of changes in the reflection intensity of the ultrasonic waves with respect to the depth from the water surface of the fish pen, using a plurality of ultrasonic transmitter-receivers arranged so as to face directly below the fish pen containing the farmed fish, diagonally downward to the left of the fish pen, diagonally downward to the right of the fish pen, diagonally downward in front of the fish pen, and diagonally downward in back of the fish pen, and irradiating ultrasonic waves over a wide range from the water surface to the depth of the fish pen; a calculation control process for setting a range of reflection intensities of bait that are weaker than the reflection intensities of the farmed fish, detecting echoes in the echogram that fall within the set range of reflection intensities of the bait as bait echoes, detecting the distribution of the bait echoes, and calculating the number of bait echoes as the feeding status of the farmed fish, or detecting linear echoes having a predetermined width in the echogram as echoes of a net for a fish cage, detecting echoes of wild fish directly below the echoes of the net for a fish cage, and calculating the number of echoes of the wild fish or their distribution, including depth, as the feeding status of the farmed fish; an estimation control step of estimating the feeding activity of all the cultured fish in the fish cage using the calculated fish behavior information; an execution control process for detecting a state in which a feeding device that supplies feed to the fish pen continues feeding, retrieving a predetermined threshold value from a predetermined memory, determining whether a previously estimated feeding activity level has fallen below the threshold value, and stopping feeding by the feeding device when the feeding activity level has fallen below the threshold value; An automatic feeding assistance method comprising:

Images