CN113951196B - Intelligent feeding method and device based on machine vision and environment dynamic coupling - Google Patents

Intelligent feeding method and device based on machine vision and environment dynamic coupling Download PDF

Info

Publication number
CN113951196B
CN113951196B CN202111227095.7A CN202111227095A CN113951196B CN 113951196 B CN113951196 B CN 113951196B CN 202111227095 A CN202111227095 A CN 202111227095A CN 113951196 B CN113951196 B CN 113951196B
Authority
CN
China
Prior art keywords
feeding
time
dissolved oxygen
cultured organisms
machine vision
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN202111227095.7A
Other languages
Chinese (zh)
Other versions
CN113951196A (en
Inventor
叶章颖
倪伟强
赵建
朋泽群
魏丹
韩志英
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Zhejiang University ZJU
Original Assignee
Zhejiang University ZJU
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Zhejiang University ZJU filed Critical Zhejiang University ZJU
Priority to CN202111227095.7A priority Critical patent/CN113951196B/en
Publication of CN113951196A publication Critical patent/CN113951196A/en
Application granted granted Critical
Publication of CN113951196B publication Critical patent/CN113951196B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • A—HUMAN NECESSITIES
    • A01—AGRICULTURE; FORESTRY; ANIMAL HUSBANDRY; HUNTING; TRAPPING; FISHING
    • A01K—ANIMAL HUSBANDRY; CARE OF BIRDS, FISHES, INSECTS; FISHING; REARING OR BREEDING ANIMALS, NOT OTHERWISE PROVIDED FOR; NEW BREEDS OF ANIMALS
    • A01K61/00—Culture of aquatic animals
    • A01K61/80—Feeding devices
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
    • Y02A40/00—Adaptation technologies in agriculture, forestry, livestock or agroalimentary production
    • Y02A40/80—Adaptation technologies in agriculture, forestry, livestock or agroalimentary production in fisheries management
    • Y02A40/81—Aquaculture, e.g. of fish

Abstract

The invention discloses an intelligent feeding method and device based on machine vision and environment dynamic coupling, the method determines a feeding mode according to a weather monitoring result, when the monitoring result is no rain, the machine vision is adopted to control feeding, namely, an image processing technology is adopted to analyze and obtain the real-time ingestion desire of a fish school, and the strength of the ingestion desire is judged so as to determine the feeding time and the feeding amount. And when the monitoring shows that the rainfall is rainfall, fuzzy control feeding is adopted, namely according to a pre-established fuzzy control module, the type and the weight of the fish school, the culture water temperature and the dissolved oxygen are used as input, the feeding rate and the feeding rate coefficient are used as output, and the feeding time and the feeding amount are determined. The method can automatically switch feeding control modes according to weather change so as to realize intelligent and accurate feeding.

Description

Intelligent feeding method and device based on machine vision and environment dynamic coupling
Technical Field
The invention belongs to the technical field of aquaculture, and relates to an intelligent feeding method and device based on machine vision and environment dynamic coupling.
Background
With the rapid development of economic society and the remarkable improvement of the living standard of people, the demand of people on fish protein is continuously increased, the global fish price is in a rising channel, the increase of the income of residents in aquatic product consumption markets and the increase of the demand of people on protein and other aquatic products are the main reasons for driving the price to rise due to the factors of population increase and farmland limitation. In addition, in order to ensure the sustainability of oceans and environments, the fishery policy in China gradually reduces the yield of the wild fishing fishery, and simultaneously slows down the yield increase of the breeding fishery under the condition that the cost (manpower, feed, energy and the like) continuously rises.
Currently, aquaculture modes in China mainly include industrial aquaculture, deep water cage culture, pond culture and the like. Wherein the influence of pond culture because environmental factor, degree of mechanization is on the low side, and common pond is thrown and is raised the mode and have two kinds: manual feeding and pond feeding machines.
The manual feeding mainly comprises the steps of manually throwing the feed by using manual tools such as a spoon and a shovel, and judging the amount of the feed required by a breeding object by eyes according to experience. In general, it is difficult to grasp the most suitable demand level of a culture subject. The manual feeding can not ensure the feeding uniformity, and has the disadvantages of time and labor consumption, high labor intensity, low efficiency, fish feed waste and pond water quality pollution. Influences the growth and development of the fish and increases the culture cost. The feeder integrates fixed point, timing and quantification, and has the advantages of wide feeding area, uniform feeding and the like. Not only reduces the labor intensity of fishermen, but also increases the yield of fish. However, the existing feeder mostly adopts a simple control system with mechanical timing, and the feeding amount cannot be automatically adjusted according to the actual feeding requirement of the fish. When the feeding amount is less than the actual feeding demand of the fish, the severe food snatching phenomenon occurs, so that the fish collide with each other and even the surface of the fish body is damaged, and the damaged fish and the weak and small fish on the surface are easy to infect certain fish diseases, so that the aquaculture water environment bears larger pressure and has adverse effect on the growth of the fish; when the feeding amount is larger than the actual feeding demand of the fish, not only the breeding cost is increased, but also the breeding environment is seriously polluted by redundant feed. The feed is therefore fed in an amount as consistent as possible with the actual food intake requirements of the fish.
The computer vision technology is a technology which can judge the ingestion demand of fishes in real time and is convenient to match with a feeder to carry out feeding operation, but for pond feeding, the interference of environmental factors of rainy weather often exists, so that the feeding judgment carried out by a machine vision method is not very accurate. The fuzzy control feeding based on water quality monitoring can determine the feeding rate and the feeding rate coefficient of the fish according to the collected water temperature, dissolved oxygen, fish body weight, relevant literature experimental data and expert empirical data, and obtain the feeding amount of the fish for feeding.
Based on the problems, the invention provides a pond intelligent feeding method and device based on machine vision and environment dynamic coupling, which comprises the following steps: the machine vision is combined with the water quality monitoring technology, the control mode is automatically switched according to weather prediction so as to achieve accurate feeding operation, food and nutrition required by growth are provided for fish, labor cost is reduced, and pollution to water quality is reduced.
Disclosure of Invention
The invention provides an intelligent feeding method and device based on machine vision and environment dynamic coupling.
An intelligent feeding method based on machine vision and environment dynamic coupling is characterized in that a feeding mode is determined according to weather monitoring results, machine vision control feeding is adopted when no rain exists, and fuzzy control feeding is adopted when rain falls are monitored.
The machine vision control feeding is to adopt an image processing technology to analyze and obtain the real-time ingestion desire of the cultured organisms and judge the strength of the ingestion desire so as to determine the feeding time and the feeding amount. The method comprises the following steps:
firstly, performing conventional feeding twice, analyzing the feeding activity degree of the cultured organisms fed twice before, and then determining the subsequent feeding state, specifically: after feeding begins, shooting a feeding real-time video picture of a cultured organism, converting an image from an RGB color space into an HSV color space, respectively extracting color components of the image under a saturation S component and a brightness V component, and segmenting and extracting a light reflection area:
Figure BDA0003314551290000021
wherein Is(x, y) and Iv(x, y) denotes saturation and brightness of the image at (x, y), respectively, TsAnd TvRespectively representing a saturation threshold value and a brightness threshold value, wherein f (x, y) represents the value of the pixel point (x, y) after binarization processing;
extracting the change characteristics of the water surface light reflecting area generated by the movement of the cultured organisms by using an optical flow method, wherein the optical flow between the continuous 2 frames of images is set as F, and the change amplitude of the target area is expressed as
Figure BDA0003314551290000022
(x, y) represents the coordinates of the reflection area of the current frame, and N is the total number of non-zero motion vectors in the current frame;
classifying the motion vectors in 2 continuous frames of target images into different intervals according to the velocity change amplitude v for statistics, dividing the velocity range into m intervals, calculating the number and probability of the motion vectors appearing at different velocities,
Figure BDA0003314551290000031
wherein j is more than or equal to 0 and less than or equal to m, and k (j) is the number of motion vectors falling into the speed interval; p (j) is the probability of a motion vector falling within the velocity interval;
measuring the irregularity degree of the change characteristic distribution probability of the water body reflecting region by using the information entropy so as to realize the analysis of the irregularity degree of the movement of the cultured organisms;
Figure BDA0003314551290000032
and establishing a modified kinetic energy model for analyzing the feeding activity intensity of the cultured organisms, Ek＝CEv2，EkKinetic energy for the cultured organisms; cEThe degree of irregularity of the target (reflective) area variation;
according to the calculated time period t before feeding is started1The average value E1 of the kinetic energy of the ingestion area and the feeding interval time t2Feeding area kinetic energy mean E2 is compared:
Figure BDA0003314551290000033
wherein
Figure BDA0003314551290000034
Is the kinetic energy value of the alpha second,
Figure BDA0003314551290000035
the kinetic energy value of theta second; if E2 is more than 1.3E1, performing the next feeding round; predicting the feed amount required to be fed by the current feeding node according to the correlation of the feeding activity degrees of the two adjacent feeding nodes in the single-round feeding:
Figure BDA0003314551290000036
and n belongs to Z; q (n) and E (n) are respectively the predicted feeding amount and the movement kinetic energy of the cultured organisms of the current feeding node, and n is the feeding times; the feeding time is as follows:
Figure BDA0003314551290000037
wherein l is the feeding rate of the feeder, T1The preheating time is the preheating time when the feeder is started; if E2 is less than 1.3E1, stopping feeding and waiting for the start of the next feeding work.
The fuzzy control feeding specifically comprises the following steps:
firstly, establishing a fuzzy control module which takes the species and the weight of cultured organisms, the culture water temperature and the dissolved oxygen as input and takes the feeding rate and the feeding rate coefficient as output according to relevant literature experimental data and expert experience data;
and (3) measuring the height A and the width B of the cultivated organisms by an estimation device to calculate the quality of the cultivated organisms: m = aAb+cBd+e(A·B)f+ g; wherein M is the quality of the cultured organisms, and a, b, c, d, e, f and g are constants which are obtained by measuring under the experimental condition in advance and fitting by a least square method, wherein a, b, c, d and g>0,e, f, g have no range requirement; calculating the average weight of the cultured organisms:
Figure BDA0003314551290000038
p is 5 hours before feedingThe number of the cultured organisms is measured in the middle period;
fuzzifying the real-time data of the average quality of the cultured organisms, the culture water temperature and the dissolved oxygen, inputting the data into a fuzzy control module, fuzzifying and deducing the culture water temperature and the average quality output feeding rate K of the cultured organisms according to an established fuzzy rule table, and fuzzifying the obtained feeding rate and feeding rate coefficient by the culture water temperature and the dissolved oxygen output feeding rate coefficient R to calculate the feeding amount required by the pond: v = KjRjMpu; wherein the content of the first and second substances,jand RjThe feed rate and the feed rate coefficient obtained after the ambiguity resolution are obtained, u is the number of the cultured organisms, and the feed time is
Figure BDA0003314551290000041
Wherein l is the feeding rate of the feeder, T1The preheating time is the preheating time when the feeder is started.
The water quality is monitored in real time in the whole breeding process, the oxygenation is closed 1 hour before feeding and during feeding, when the dissolved oxygen in a detected water body is lower than 3mg/L, the feeding is stopped, the aeration is carried out, when the dissolved oxygen is higher than 5mg/L, the aeration is stopped, the feeding is recovered, in other time, when the dissolved oxygen in the detected water body is lower than 3mg/L, the oxygenation is started, and when the dissolved oxygen is higher than 5mg/L, the oxygenation is closed.
The invention has the beneficial effects that:
the intelligent feeding method and device based on machine vision and environment dynamic coupling, disclosed by the invention, have the advantages of simple structure and simple and convenient control mode, not only can judge the actual appetite of the cultured organisms for feeding by utilizing a machine vision technology, but also can make a prejudgment on weather in advance by utilizing a meteorological monitoring device under the condition that the machine vision judgment is greatly influenced in rainy days, automatically switch the fuzzy control feeding mode in advance, effectively combine the water quality condition to finish feeding work, avoid the problems of feed waste, water quality pollution, yield influence and the like, and further achieve the purpose of intelligent and accurate feeding.
Drawings
The invention is described in detail below with reference to the drawings and the detailed description;
fig. 1 is a schematic structural diagram of an example of an intelligent feeding device based on machine vision and environment dynamic coupling according to the present invention.
FIG. 2 is a graph of membership function for water temperature in accordance with one embodiment of the present invention.
FIG. 3 is a graph of membership functions for weights in accordance with one embodiment of the present invention.
FIG. 4 is a graph of membership function for dissolved oxygen in one embodiment of the invention.
FIG. 5 is a graph of membership functions for feed rate in one embodiment of the invention.
FIG. 6 is a graph of membership functions for feed rate coefficients in accordance with one embodiment of the invention.
FIG. 7 is a graph of fuzzy variable input-output for feed rate in accordance with an embodiment of the present invention.
FIG. 8 is a graph of the feed rate coefficient fuzzy variable input-output relationship in accordance with one embodiment of the present invention.
In the figure: 1-high-definition waterproof camera 2-meteorological monitoring device 3-PLC 4-aerator 5-water quality monitoring device 6-grating biomass estimation device 7-aeration plate 8-feeder 9-steel wire hose 10-blower 11-frame 12-digital signal processor.
Detailed Description
The technical scheme of the invention is further explained by using pond culture fish in combination with the attached drawings.
As shown in fig. 1, the device of the invention comprises a high-definition waterproof camera 1, a meteorological monitoring device 2, a PLC3, an aerator 4, a water quality monitoring device 5, a grating biomass estimation device 6, an aeration disc 7, a feeder 8, a steel wire hose 9, a blower 10, a frame 11 and a digital signal processor 12;
the high-definition waterproof camera 1 is installed at the bank of the pond and fixed at the upper end of the rack 11, and the high-definition waterproof camera 1 is connected with the input end of the digital signal processor 12; the installation position of the camera can ensure that the camera can shoot the whole feeding area;
the weather monitoring device 2 is arranged on one side of the rack 11, and the output end of the weather monitoring device 2 is connected with the input end of the PLC 3; the weather monitoring device monitors weather environment conditions through a weather monitoring system and predicts whether rainfall occurs or not, a prediction result is input into a PLC, and the PLC switches and selects fuzzy control feeding under the rainfall condition and machine vision control feeding under the no-rain condition;
the water quality monitoring device 5 is arranged in a feeding area of the pond and floats on the water surface of the pond by virtue of a hollow sphere at the upper end, the water quality monitoring device comprises a water temperature sensor and a dissolved oxygen sensor, and the output end of the water quality monitoring device 5 is connected with the input end of the PLC 3;
the digital signal processor 12 and the PLC3 are both arranged on one side of the rack 11, and the output end of the digital signal processor 12 is connected with the input end of the PLC 3; under the condition of no rain, the digital signal processor receives image information input by the camera and sound information input by the hydrophone and performs corresponding processing, firstly, the real-time eating desire of the fish is analyzed through an image processing technology, whether the feeder performs feeding operation or not is determined, and if the digital signal processor determines that the eating desire is strong, the processing result is transmitted to the PLC to control the feeder to work, wherein the feeding time and the feeding amount are included; under the rainfall condition, the PLC receives water temperature and dissolved oxygen data output by the water quality monitoring device and average fish weight data output by the grating fish biomass estimation device, fuzzifies the data, outputs feeding rate and feeding rate coefficient through a fuzzy feeding model, defzifies the feeding rate and feeding rate coefficient, calculates feeding amount and feeding time, and controls the working time of the feeder;
the feeder 8 is arranged at the bank of the pond, and the feeder 8 is connected with the output end of the PLC 3;
the aerator 4 is arranged in the center of the pond, and the aerator 4 is connected with the output end of the PLC 3;
the blower 10 is arranged at the bank side of the pond, and the blower 10 is connected with the output end of the PLC 3;
one end of the steel wire hose 9 is connected with an air outlet of the blower 10, and the other end is connected with an air inlet of the aeration disc 7;
the aeration disc 7 is arranged at the position of half the water depth of the pond feeding area, and four long rods at the lower end of the aeration disc 7 are inserted into the bottom of the pond for fixing;
the grating biomass estimation device 6 is arranged on the upper side of the beam of the aeration disc 7, and the output end of the grating biomass estimation device 6 is connected with the input end of the PLC 3; the grating biomass estimating device estimates the weight of the fish body by measuring the length and the width of the fish body by the gratings on the left side and the right side.
The intelligent fish feeding method by using the device comprises the following steps:
1) Predicting whether the feeding period is rained or not according to the weather monitoring device 2, if no rain exists, automatically closing fuzzy control feeding by the PLC3, and starting machine vision control feeding; if the rainfall occurs, the PLC automatically closes the machine vision control feeding and starts the fuzzy control feeding;
2) The blower aeration device, the aerator 4 and the water quality monitoring device 5 are combined through the PLC 3; in order to prevent water turbidity and fish stress during feeding, the automatic aerator is in a closed state 1 hour before feeding and during feeding, when the water quality monitoring device detects that the dissolved oxygen is lower than 3mg/L, the feeder 8 stops working, the PLC controls the aerator 10 to work, when the dissolved oxygen is higher than 5mg/L, the aerator stops working, the feeder recovers to a working state, and when the water quality monitoring device detects that the dissolved oxygen is lower than 3mg/L, the PLC controls to start the automatic aerator, and when the dissolved oxygen is higher than 5mg/L, the automatic aerator is closed;
3) If no rain exists, the PLC3 controls the feeder 8 to work at regular time, the feeding time is 10s, the feeding interval time is 50s, and the feeding work is finished by controlling the feeder at regular time by the PLC in the first two times of feeding; after 120s from the beginning of feeding, the digital signal processor 12 analyzes the feeding activity degrees of the previous two times to determine the subsequent working state of the feeder. After feeding begins, the high-definition waterproof camera 1 transmits shot real-time video pictures to the digital signal processor in real time; the digital signal processor converts the image from RGB color space to HSV color space, and respectively extracts the color components of the image under S component (saturation) and V component (brightness), and divides and extracts the light reflecting region:
Figure BDA0003314551290000061
wherein Is(x, y) and Iv(x, y) denotes saturation and brightness of the image at (x, y), respectively, TsAnd TvRespectively representing a saturation threshold value and a brightness threshold value, wherein f (x, y) represents the value of the pixel point (x, y) after binarization processing;
4) Extracting the change characteristics of the water surface light reflecting area generated by the movement of the cultured organisms by using an optical flow method, wherein the optical flow between the continuous 2 frames of images is set as F, and the change amplitude of the target area is expressed as
Figure BDA0003314551290000062
(x, y) represents the coordinates of the reflection area of the current frame, and N is the total number of non-zero motion vectors in the current frame;
5) Classifying the motion vectors in 2 continuous frames of target images into different intervals according to the velocity change amplitude v for statistics, dividing the velocity range into m intervals, calculating the number and probability of the motion vectors appearing at different velocities,
Figure BDA0003314551290000063
wherein j is more than or equal to 0 and less than or equal to m, and k (j) is the number of motion vectors falling into the speed interval; p (j) is the probability of a motion vector falling within the velocity interval;
6) Measuring the irregularity degree of the change characteristic distribution probability of the water reflection area by using the information entropy so as to analyze the irregularity degree of the fish school movement;
Figure BDA0003314551290000071
7) Establishing a modified kinetic energy model for analyzing the feeding activity intensity of the cultured organisms, Ek＝CEv2，EkKinetic energy for the cultured organisms; cEDegree of irregularity of the target (light reflection) region variation;
8) According to the calculated time period t before feeding is started1The average value E1 of the kinetic energy of the ingestion area and the feeding interval time t2The average kinetic energy E2 of the feeding area is compared:
Figure BDA0003314551290000072
wherein
Figure BDA0003314551290000073
Is the kinetic energy value of the alpha second,
Figure BDA0003314551290000074
the kinetic energy value of theta second; if E2 is more than 1.3E1, performing the next feeding round; 9) Predicting the feed amount required to be fed by the current feeding node according to the correlation of the feeding activity degrees of the two adjacent feeding nodes in the single-round feeding:
Figure BDA0003314551290000075
and n belongs to Z; q (n) and E (n) are respectively the predicted feeding amount and the movement kinetic energy of the cultured organisms of the current feeding node, and n is the feeding times; the feeding time is as follows:
Figure BDA0003314551290000076
wherein l is the feeding rate of the feeder, T1The preheating time is the preheating time when the feeder is started; if E2 is less than 1.3E1, stopping feeding and waiting for the start of the next feeding work.
10 Whether the feeding period is rained or not is predicted according to the meteorological monitoring device 2, if the feeding period is rained, the PLC3 automatically turns off the machine vision control feeding system, and turns on the fuzzy control system for feeding;
11 According to relevant literature experimental data and expert experience data, establishing a fuzzy control module which takes the weight of a fish school, the water temperature of a pond and dissolved oxygen as input and takes a feeding rate and a feeding rate coefficient as output;
12 The fish body mass is calculated by measuring the height a and width B of the fish body by the rastered fish biomass estimation means 6: m = aAb+cBd+e(A·B)f+ g; wherein M is the mass of the fish, and a, b, c, d, e, f and g are constants which are obtained by measuring under the experimental condition in advance and fitting by a least square method, wherein a, b, c, d>0,e, f, g have no range requirement;
13 Fish mass data obtained by the raster fish biomass estimation device 6 are input into the PLC3 to calculate the average weight of fish:
Figure BDA0003314551290000077
p is the number of the fish to be measured in 5 hours before feeding, and the average fish body mass is fuzzified and then input into a fuzzy control module;
14 The water quality monitoring device 5 transmits culture water temperature and dissolved oxygen real-time data to be fuzzified and then inputs the data into a fuzzy control module in the PLC3, the culture water temperature and the average culture biological quality output feeding rate K are fuzzified and deduced according to an established fuzzy rule table, the obtained feeding rate and the feeding rate coefficient R are defuzzified according to the culture water temperature and the dissolved oxygen output feeding rate coefficient, and the required feeding amount of the pond is calculated: v = Kj*Rj*Mp* u; wherein, KjAnd RjThe feed rate and the feed rate coefficient obtained after the ambiguity resolution are obtained, u is the number of the cultured organisms, and the feed time is
Figure BDA0003314551290000081
Wherein l is the feeding rate of the feeder, T1The preheating time is the preheating time when the feeder is started.
15 Calculates the working time of the feeder 8 according to the feeding amount,
Figure BDA0003314551290000082
wherein l is the feeding rate of the feeder, T1Is the preheating time when the feeder is started.
In clear weather, intelligent feeding based on a machine vision technology has universal applicability, and a fuzzy control feeding mode in rainy days needs to determine fuzzy feeding according to relevant literature experimental data and expert breeding data, taking tilapia as an example;
according to the water temperature monitoring and known discourse range [15,40] of the tilapia breeding data, dividing temperature variables into 4 fuzzy subsets according to grades, wherein the fuzzy subsets are S1, S2, S3 and S4 respectively. Wherein, the S1 and the S2 use triangle membership functions, and the S3 and the S4 use trapezoid membership functions. (S1-trimf, S2-trimf, S3-trapmf, S4-trapmf), and the membership function curve of the water temperature is shown in FIG. 2.
The range of the weight determination universe of tilapia [0,1000], is also classified into four fuzzy subsets, Z1, Z2, Z3, Z4, according to grades. Wherein Z1, Z2 and Z3 are triangle membership functions, and Z4 is trapezoid membership functions. (Z1-trimf, Z2-trimf, Z3-trimf, Z4-trapmf) and the weight membership function curve are shown in FIG. 3.
The discourse domain of dissolved oxygen is [1.5,12], which is also classified into four fuzzy subsets according to grades, namely O1, O2, O3 and O4. Wherein O1, O2, O3, O4 are all represented by triangular membership functions (O1-trimf, O2-trimf, O3-trimf, O4-trimf), and the curve of the membership function of dissolved oxygen is shown in FIG. 4.
The argumentation of the feeding rate is [0,0.06]. And four fuzzy subsets are divided according to grades, namely K1, K2, K3 and K4. Wherein K1 and K4 are trapezoidal membership functions and K2 and K3 are triangular membership functions (K1-trapmf, K2-trimf, K3-trimf, K4-trapmf). The membership function curve of the feeding rate is shown in FIG. 5.
The feeding rate coefficient is the adjustment range of the basic feeding amount during the near feeding, the domain of discourse is [0,1.2], and the four fuzzy subsets are divided into four fuzzy subsets according to the grades, wherein the four fuzzy subsets are respectively R1, R2, R3 and R4. Wherein R1, R2 and R3 use triangle membership functions, and R4 uses trapezoid membership functions (R1-trimf, R2-trimf, R3-trimf and R4-trapnf). The membership function curve of the feeding rate coefficient is shown in FIG. 6.
Based on the water temperature and dissolved oxygen measured by the water quality monitoring device and the fish weight obtained by the raster fish biomass estimation device, the water temperature, dissolved oxygen and weight values are firstly fuzzified, and the fuzzy states of the feeding rate and the feeding rate coefficient are determined according to the corresponding IF-THEN type conditional statement rule table.
The IF-THEN type conditional statement rule is determined by establishing a fuzzy rule table, the fuzzy rule table is formulated based on relevant literature experimental data and expert experience data, the input variables of water temperature, dissolved oxygen and weight are divided into 4 linguistic variables, and the output variables of feeding rate and feeding rate coefficient are also divided into 4 linguistic variables. Therefore, the fuzzy inference system has 32 rules, and is composed of a series of IF-THEN type condition statements, wherein the input variable water temperature, the input variable weight and the output variable feeding rate IF-THEN type condition statements are as follows:
if (water temperature is S1) and (weight is Z1) then (feed rate is K2)
If (water temperature is S2) and (weight is Z1) then (feed rate is K3)
If (water temperature is S3) and (weight is Z1) then (feed rate is K4)
If (water temperature is S4) and (weight is Z1) then (feed rate is K1)
If (water temperature is S1) and (weight is Z2) then (feed rate is K1)
If (water temperature is S2) and (weight is Z2) then (feed rate is K2)
……
The specific fuzzy rule table according to the rule sentence is shown in table 1, the input-output relationship of the feeding rate fuzzy variable is shown in fig. 7, and table 1 is an inference rule table of the feeding rate.
TABLE 1 inference rule table of feeding rate K
Figure BDA0003314551290000091
The input variables are water temperature, dissolved oxygen and output variable feeding rate coefficient IF-THEN type conditional statements are as follows:
if (water temperature is S1) and (dissolved oxygen is O1) then (feed rate coefficient is R1)
If (water temperature is S2) and (dissolved oxygen is O1) then (feed rate coefficient is R2)
If (water temperature is S3) and (dissolved oxygen is O1) then (feed rate coefficient is R2)
If (water temperature is S4) and (dissolved oxygen is O1) then (feed rate coefficient is R1)
If (water temperature is S1) and (dissolved oxygen is O2) then (feed rate coefficient is R1)
If (water temperature is S2) and (dissolved oxygen is O2) then (feed rate coefficient is R3)
……
The specific fuzzy rule table according to the rule sentence is shown in table 2, the input-output relationship of the fuzzy variable of the feeding rate coefficient is shown in fig. 8, and table 2 is an inference rule table of the feeding rate coefficient.
TABLE 1 inference rule table of feeding rate coefficient
Figure BDA0003314551290000101
Obtaining the feeding rate and the feeding rate coefficient according to a fuzzy rule table, and obtaining the feeding rate K by defuzzifying through a gravity center methodLAnd feeding rate coefficient RLActual feeding amount VL＝KL*RL*Mp* u. The feeding time is
Figure BDA0003314551290000102
Wherein l is the feeding rate of the feeder, T1The preheating time is the preheating time when the feeder is started.
The above disclosure is only for the specific embodiment of the present invention, but the present invention is not limited thereto, and it should be understood by those skilled in the art that the modifications made without departing from the present invention shall fall within the protection scope of the present invention.

Claims (2)

1. An intelligent feeding device based on machine vision and environment dynamic coupling is characterized by comprising a camera, a meteorological monitoring device, an aerator, a water quality monitoring device, a grating biomass estimation device, an aerator, a feeder and a processor; the camera is used for shooting real-time pictures of the aquaculture water body, and the camera, the meteorological monitoring device, the water quality monitoring device and the grating biomass estimation device are all used for outputting monitoring results to the processor; the processor processes the received data and then controls the aerator, the aeration device and the feeder; determining a feeding mode according to a meteorological monitoring result, adopting machine vision to control feeding when no rain exists, adopting the culture water temperature and the average mass of cultured organisms to output feeding rate when rain falls, and adopting the fuzzy control of the culture water temperature and the dissolved oxygen output feeding rate coefficient to control feeding; the machine vision control feeding is to adopt an image processing technology to analyze and obtain the real-time ingestion desire of the cultured organisms and judge the strength of the ingestion desire so as to determine the feeding time and the feeding amount; the machine vision control feeding comprises:
firstly, carrying out conventional feeding twice, analyzing the feeding activity degree of the cultured organisms fed twice before, and then determining the subsequent feeding state, specifically: after feeding begins, shooting a feeding real-time video picture of a cultured organism, converting an image from an RGB color space into an HSV color space, respectively extracting color components of the image under a saturation S component and a brightness V component, and segmenting and extracting a light reflecting area:
Figure FDA0003835481970000011
wherein Is(x, y) and Iv(x, y) denotes saturation and brightness of the image at (x, y), respectively, TsAnd TvRespectively representing a saturation threshold value and a brightness threshold value, wherein f (x, y) represents the value of the pixel point (x, y) after binarization processing;
extracting the change characteristics of the water surface light reflecting area generated by the movement of the cultured organisms by using an optical flow method, wherein the optical flow between the continuous 2 frames of images is set as F, and the change amplitude of the target area is expressed as
Figure FDA0003835481970000012
(x, y) represents the coordinates of the reflection area of the current frame, and N is the total number of non-zero motion vectors in the current frame;
classifying the motion vectors in 2 continuous frames of target images into different intervals according to the velocity change amplitude v for statistics, dividing the velocity range into m intervals, calculating the number and probability of the motion vectors appearing at different velocities,
Figure FDA0003835481970000013
wherein j is more than or equal to 0 and less than or equal to m, and k (j) is the number of motion vectors falling into the speed interval; p (j) is the probability of a motion vector falling within the velocity interval;
measuring the irregularity degree of the change characteristic distribution probability of the water reflection area by using the information entropy so as to realize the analysis of the irregularity degree of the movement of the cultured organisms;
Figure FDA0003835481970000014
and establishing cultivationModified kinetic energy model of biological feeding activity intensity analysis, Ek＝CEv2，EkKinetic energy for the cultured organisms; cEDegree of irregularity of the target (light reflection) region variation;
according to the calculated time period t before feeding is started1The average value E1 of the kinetic energy of the ingestion area and the feeding interval time t2Feeding area kinetic energy mean E2 is compared:
Figure FDA0003835481970000021
wherein
Figure FDA0003835481970000022
Is the kinetic energy value of the alpha second,
Figure FDA0003835481970000023
the kinetic energy value of theta second; if E2 is more than 1.3E1, carrying out the next round of feeding; predicting the feed amount required to be fed by the current feeding node according to the correlation of the feeding activity degrees of the two adjacent feeding nodes in the single-round feeding:
Figure FDA0003835481970000024
n is more than or equal to 3 and n belongs to Z; q (n) and E (n) are respectively the predicted feeding amount and the movement kinetic energy of the cultured organisms of the current feeding node, and n is the feeding times; the feeding time is as follows:
Figure FDA0003835481970000025
wherein l is the feeding rate of the feeder, T1The preheating time is the preheating time when the feeder is started; if E2 is less than 1.3E1, stopping feeding and waiting for the start of the next feeding work;
the fuzzy control feeding specifically comprises the following steps:
firstly, establishing a fuzzy control module which takes the species and the weight of cultured organisms, the culture water temperature and the dissolved oxygen as input and takes the feeding rate and the feeding rate coefficient as output according to relevant literature experimental data and expert experience data; the height A and the width B of the farmed organisms are measured by the estimation device to calculate the farmed organismsQuality: m = aAb+cBd+e(A·B)f+ g; wherein M is the quality of the cultured organisms, and a, b, c, d, e, f and g are constants which are obtained by measuring under the experimental condition in advance and fitting by a least square method, wherein a, b, c, d and g>0,e, f, g have no range requirement; calculating the average weight of the cultured organisms:
Figure FDA0003835481970000026
p is the number of the detected cultured organisms in 5 hours before feeding;
fuzzifying the real-time data of the average quality of the cultured organisms, the culture water temperature and the dissolved oxygen, inputting the data into a fuzzy control module, fuzzifying and deducing the culture water temperature and the average quality output feeding rate K of the cultured organisms according to an established fuzzy rule table, and fuzzifying the obtained feeding rate and feeding rate coefficient by the culture water temperature and the dissolved oxygen output feeding rate coefficient R to calculate the feeding amount required by the pond: v = KjRjMpu; wherein, KjAnd RjThe feed rate and the feed rate coefficient obtained after the ambiguity resolution are shown in the specification, u is the number of the cultured organisms, and the feed time is
Figure FDA0003835481970000027
Wherein l is the feeding rate of the feeder, T1The preheating time is the preheating time when the feeder is started.
2. The intelligent feeding device based on machine vision and environment dynamic coupling of claim 1, characterized in that, the whole course needs to monitor the water quality in real time, the oxygen increasing is closed 1 hour before feeding and during feeding, when the dissolved oxygen in the water body is detected to be lower than 3mg/L, the feeding is stopped, the aeration is performed, when the dissolved oxygen is higher than 5mg/L, the aeration is stopped, the feeding is resumed, and in other time, when the dissolved oxygen in the water body is detected to be lower than 3mg/L, the oxygen increasing is started, and when the dissolved oxygen is higher than 5mg/L, the oxygen increasing is stopped.
CN202111227095.7A 2021-10-21 2021-10-21 Intelligent feeding method and device based on machine vision and environment dynamic coupling Active CN113951196B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111227095.7A CN113951196B (en) 2021-10-21 2021-10-21 Intelligent feeding method and device based on machine vision and environment dynamic coupling

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111227095.7A CN113951196B (en) 2021-10-21 2021-10-21 Intelligent feeding method and device based on machine vision and environment dynamic coupling

Publications (2)

Publication Number Publication Date
CN113951196A CN113951196A (en) 2022-01-21
CN113951196B true CN113951196B (en) 2022-11-01

Family

ID=79465870

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111227095.7A Active CN113951196B (en) 2021-10-21 2021-10-21 Intelligent feeding method and device based on machine vision and environment dynamic coupling

Country Status (1)

Country Link
CN (1) CN113951196B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115211410B (en) * 2022-08-30 2023-06-02 四川农业大学 Underwater grading fishing device and method for small-sized aquaculture fish pond

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107114300A (en) * 2017-04-28 2017-09-01 中国水产科学研究院渔业机械仪器研究所 A kind of regulation and control modification method of leg Shrimp Litopenaeus vannamei Culture Raceway-type Ponds pool substrate
CN107535399A (en) * 2017-09-13 2018-01-05 寿县绿园特种水产生态养殖加工有限责任公司 A kind of method for being combined cultivation Macrobrachium rosenbergii with gas stone oxygenation using aeration oxygen replenishing

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
NO331769B1 (en) * 2010-05-18 2012-03-26 Uni I Stavanger System and method for controlled feeding of farmed fish
CN106773721B (en) * 2017-02-08 2019-07-05 南京渔管家物联网科技有限公司 A kind of aquaculture intelligence feeding system and its control method based on fuzzy control
CN108375906A (en) * 2018-03-20 2018-08-07 北京农业信息技术研究中心 A kind of aquatic fish feeding volume predictor method based on fuzzy algorithmic approach
CN112213962A (en) * 2020-08-21 2021-01-12 四川渔光物联技术有限公司 Intelligent feeding system and method based on growth model and sonar feedback

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107114300A (en) * 2017-04-28 2017-09-01 中国水产科学研究院渔业机械仪器研究所 A kind of regulation and control modification method of leg Shrimp Litopenaeus vannamei Culture Raceway-type Ponds pool substrate
CN107535399A (en) * 2017-09-13 2018-01-05 寿县绿园特种水产生态养殖加工有限责任公司 A kind of method for being combined cultivation Macrobrachium rosenbergii with gas stone oxygenation using aeration oxygen replenishing

Also Published As

Publication number Publication date
CN113951196A (en) 2022-01-21

Similar Documents

Publication Publication Date Title
CN107094683B (en) Automatic feeding and water quality monitoring control system for aquaculture
JP3101938B2 (en) Automatic feeding device and method for aquatic organisms
US11925173B2 (en) Data collection system and method for feeding aquatic animals
WO2021143783A1 (en) Swimming-type fish self-adaptive feeding device and method based on light-sound coupling technique
CN112213962A (en) Intelligent feeding system and method based on growth model and sonar feedback
CN110074030B (en) Feedback type pond circulating water intelligent feeding system integrating machine vision and infrared detection technology
CN103761565B (en) The juvenile crab quantity survey of fish and shrimp under water and behavior monitoring device and method based on computer vision
CN113951196B (en) Intelligent feeding method and device based on machine vision and environment dynamic coupling
AU2011255706A1 (en) System and method for controlling feeding of farmed fish
CN114467824B (en) Intelligent bait casting boat
CN112400773B (en) Greenhouse fry intelligent feeding device and method based on machine vision technology
CN111838027A (en) Pure oxygen supply aquaculture system and method
JP2000201568A (en) Automatic feeding system for aquatic organism
CN112506120A (en) Wisdom fishery management system based on thing networking
Huang et al. The prototype of a smart underwater surveillance system for shrimp farming
CN110089477A (en) A kind of fish welfare intelligent cultivation system and method for circulating water cultivation mode
CN106614243B (en) A kind of efficient dirt collection and intelligent feeding system for pond circulation flowing water culture
CN110476855A (en) A kind of far-reaching extra large aquafarm
TW202209965A (en) A smart shrimp and/or crab feeding management system and the method thereof
CN113749030B (en) Fish welfare self-adaptive feeding system suitable for circulating water aquaculture mode
CN107730495A (en) A kind of fish pond anoxic detection method based on background modeling
JPH0785080B2 (en) Fish condition monitor
JPH0668489B2 (en) Aquatic animal monitoring device
CN117011795B (en) River crab growth state nondestructive monitoring and evaluating platform and method based on Gaussian-like fuzzy support degree
CN116307893B (en) Fish proliferation releasing water area suitability evaluation system based on data analysis

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant