Factory-like intelligent recirculating aquaculture system and method for salmon and trout

Abstract

The invention relates to a salmon and trout factory-like intelligent circulating water culture system and method, and belongs to the technical field of salmon and trout culture. The salmon and trout factory-like intelligent circulating water culture system comprises a culture pond, a sedimentation pond, a water return pond, a solid and suspended matter separation device, a water purification system, a circulating water pump and a biological ammonia nitrogen removal device; a clear water outlet of the suspended matter separation device is communicated with the water return pond after passing through the water purification system, a muddy water outlet of the suspended matter separation device is communicated with the sedimentation pond, a clear water outlet of the sedimentation pond is communicated with the water return pond, the muddy water outlet of the sedimentation pond discharges sediments, the water return pond is provided with a circulating water pump, and the circulating water pump is communicated with the culture pond through the biological ammonia nitrogen removal device. The research and development aim of the invention is to solve the problems that the intelligence degree is low, the management and control refinement is insufficient and multiple links need to be manually controlled and operated in the land-based industrial aquaculture process of the salmon and trout, so that the intelligence degree is higher in the land-based industrial aquaculture process of the salmon and trout, and the aquaculture water quality meets the physiological needs of normal growth of the salmon and trout.

Description

Technical Field

[0001] The present invention relates to a salmon and trout industrial intelligent recirculating aquaculture system and method, belonging to the technical field of salmon and trout aquaculture. Background Art

[0002] In the process of circulating water aquaculture in fish facilities, harmful solid substances, suspended substances, soluble substances and gases such as excrement and residual bait of fish metabolism in the aquaculture water body need to be discharged from the water body or converted into harmless substances during the circulation of the water body, and necessary elements and substances are added to make the aquaculture water quality meet the physiological needs of the normal growth of fish. In order to ensure the reliability of recirculating aquaculture, the design of facilities and equipment, the application of water treatment technology and the control of water quality all require a very high technical level to make the whole system operate in a supporting manner during the operation process. However, in the process of land-based industrial aquaculture of salmon and trout, problems such as low intelligence level, insufficient refined management and control, and the need for manual control and operation in multiple links directly lead to the need to improve the aquaculture quality of salmon and trout.

[0003] Therefore, there is an urgent need to propose a new salmon and trout industrial intelligent recirculating aquaculture system and method to solve the above technical problems. Summary of the Invention

[0004] The research and development purpose of the present invention is to solve the problems of low intelligence level, insufficient refined management and control, and the need for manual control and operation in multiple links in the process of land-based industrial aquaculture of salmon and trout. A brief overview of the present invention is given below to provide a basic understanding of certain aspects of the present invention. It should be understood that this overview is not an exhaustive overview of the present invention. It is not intended to identify the key or important parts of the present invention, nor is it intended to limit the scope of the present invention.

[0005] Technical Solution of the Present Invention

[0006] Solution 1: A salmon and trout industrial intelligent recirculating aquaculture system includes a culture pond, a sedimentation pond, a return pond, a solid and suspended matter separation device, a water purification system, a circulation water pump and a biological ammonia and nitrogen removal device. The culture pond is communicated with the solid and suspended matter separation device. The clear water outlet of the suspended matter separation device is communicated with the return pond through the water purification system. The turbid water outlet of the suspended matter separation device is communicated with the sedimentation pond. The clear water outlet of the sedimentation pond is communicated with the return pond. The turbid water outlet of the sedimentation pond discharges sediment. A circulation water pump is arranged in the return pond. The water outlet of the circulation water pump is communicated with the culture pond through the biological ammonia and nitrogen removal device.

[0007] Preferably: The return pond is connected with a makeup water pond.

[0008] Preferably: The culture pond is connected with an automatic feeding device.

[0009] Preferably, the water purification system includes a temperature adjustment device, an oxygenation device, and a packing device. The clear water outlet of the solid and suspended matter separation device is connected to the return water tank after passing through the temperature adjustment device, the oxygenation device, and the packing device.

[0010] Preferably, the water purification system further includes a temperature sensor, an oxygen content detector, a chemical substance detector, and a pH sensor. The measuring ends of the temperature sensor, the oxygen content detector, the chemical substance detector, and the pH sensor measure the circulating water in the water purification system.

[0011] Solution Two: A method for intelligent recirculating aquaculture of salmon and trout in industrial scale is realized relying on the salmon and trout industrial intelligent recirculating aquaculture system described in Solution One, and includes the following steps:

[0012] Step 1: Start the circulating water pump in the return water tank, and the aquaculture water in the aquaculture pond automatically enters the solid and suspended matter separation device for solid and suspended matter separation treatment.

[0013] Step 2: The solids separated by the solid and suspended matter separation device enter the sedimentation tank through the turbid water outlet of the suspended matter separation device. The sediment separated in the sedimentation tank is discharged through the turbid water outlet of the sedimentation tank, and the circulating water separated in the sedimentation tank enters the return water tank through the clear water outlet.

[0014] Step 3: The circulating water separated by the solid and suspended matter separation device is subjected to circulating water treatment through the water purification system and then enters the return water tank.

[0015] Step 4: The circulating water pumped by the circulating water pump in the return water tank flows back into the aquaculture pond after harmful gas removal and disinfection treatment through the biological ammonia nitrogen removal device.

[0016] Preferably, the circulating water treatment includes temperature control treatment by the temperature adjustment device, oxygenation treatment by the oxygenation device, and chemical substance and pH treatment by the packing device.

[0017] The temperature control treatment includes: setting the temperature range of the aquaculture water, using the temperature sensor to measure the circulating water passing through the water purification system, and if the measured temperature is not within the temperature range of the aquaculture water, controlling the temperature adjustment device to exchange heat for the circulating water.

[0018] The oxygenation treatment includes: setting the oxygen content range of the aquaculture water, using the oxygen content detector to measure the circulating water passing through the water purification system, and if the measured oxygen content is lower than the oxygen content range of the aquaculture water, controlling the oxygenation device to oxygenate the circulating water.

[0019] The chemical substances and pH treatment include: setting the standard ranges of chemical substances and pH in the aquaculture water, measuring the circulating water passing through the water purification system using a chemical substance detector and a pH sensor. If the measured chemical substances and pH in the aquaculture water are not within the standard ranges of chemical substances and pH in the aquaculture water, the packing device is controlled to pack the circulating water until the chemical substances and pH reach the standard ranges of chemical substances and pH in the aquaculture water, and then it stops.

[0020] Furthermore, the automatic feeding device makes real-time predictions on the images of salmon and trout in the aquaculture pond collected by the automatic feeding device based on the salmon and trout target prediction module set inside, predicts the length and weight of the salmon and trout, and then based on the predicted length and weight of the salmon and trout, the automatic feeding device calculates the feeding amount through the feeding calculation module set inside to feed the salmon and trout.

[0021] Furthermore, the specific implementation method for the automatic feeding device to feed salmon and trout includes the following steps:

[0022] S1. The automatic feeding device collects images of salmon and trout in the aquaculture pond;

[0023] S2. Preprocess the images of salmon and trout collected in step S1 to obtain the preprocessed images of salmon and trout;

[0024] S2.1. Set the pixel value at the pixel coordinate (x, y) position of the collected salmon and trout image as P(x, y), where x and y are the pixel coordinates in the horizontal direction and the vertical direction respectively, and calculate the pixel gradient The difference P in the horizontal direction in the image x and the difference P in the vertical direction y , and the expressions are:

[0025] P x = P(x + 1, y) - P(x, y)

[0026] P y = P(x, y + 1) - P(x, y)

[0027] where P(x + 1, y) is the pixel value at the coordinate (x + 1, y) position of the image, and P(x, y + 1) is the pixel value at the coordinate (x, y + 1) position of the image;

[0028] Then calculate the pixel gradient The calculation formula is:

[0029]

[0030] The pixel gradient is used to determine the edge intensity at the image pixel points;

[0031] S2.2. Set the adjustment coefficient b(x,y) to determine the update degree of pixels at different positions (x,y) in the image. The calculation formula of the adjustment coefficient b(x,y) is as follows:

[0032]

[0033] Where d is a constant for controlling sensitivity, and g is a constant for controlling gradient components. Both constants are determined by experience;

[0034] S2.3. Perform pixel update of the image to obtain the pixel value P e (x,y) at the position of the updated image, and the calculation formula is:

[0035] P e (x,y) = P(x,y) + ∑ f b(x,y)(Pf - P(x,y))

[0036] Where f are the four pixel coordinates adjacent to the pixel coordinate (x,y) in the up, down, left, and right directions, and Pf refers to the pixel values of the four pixel coordinates adjacent in the up, down, left, and right directions;

[0037] S3. Input the pre-processed salmon and trout image into the salmon and trout target prediction module, perform feature extraction based on the trained neural network model, and obtain the length L of the i-th salmon and trout, i and then calculate the weight of the salmon and trout in the breeding pond based on the length of the salmon and trout. The calculation formula is:

[0038] W i = a × L i b

[0039] Where W i is the weight of the i-th salmon and trout, a is the first coefficient, and b is the second coefficient;

[0040] Then calculate the average value of the weights of all the salmon and trout obtained from the salmon and trout image to obtain the average weight of the salmon and trout The calculation formula is:

[0041]

[0042] Where m is the number of all the salmon and trout in the salmon and trout image;

[0043] Then calculate the total weight W total of the salmon and trout in the breeding pond, and the calculation formula is:

[0044]

[0045] Among them, n is the total number of salmon and trout in the aquaculture pond;

[0046] S4. Based on the estimated weight of the salmon and trout obtained, the automatic feeding device calculates the feeding amount through the internally set feeding calculation module. The feeding amount M total The calculation formula is:

[0047] M total = 2% × W total .

[0048] The present invention has the following beneficial effects:

[0049] 1. The salmon and trout industrialized intelligent circulating water aquaculture system of the present invention makes the salmon and trout more intelligent during the land-based industrialized aquaculture process, reduces the manual control link, reduces the risk of errors, saves manpower, improves efficiency, and makes the aquaculture water control more refined, so that the aquaculture water quality meets the physiological needs of the normal growth of fish;

[0050] 2. The circulating water aquaculture mode is adopted in the salmon and trout industrialized intelligent circulating water aquaculture system of the present invention, and adjustments are made in terms of energy-saving optimization of the system equipment, simplification of facilities and operation costs, which is more water-saving, land-saving, energy-saving and emission-reducing, so as to reduce the aquaculture cost and investment scale;

[0051] 3. The salmon and trout industrialized intelligent circulating water aquaculture system of the present invention can realize the whole cycle of the entire system only by relying on the water power of the circulating water pump. The design is ingenious, the structure is simple, the equipment cost and maintenance cost are reduced, and the practicability is stronger;

[0052] 4. The salmon and trout industrialized intelligent circulating water aquaculture system of the present invention uses an automatic feeding device for accurate feeding;

[0053] 5. The present invention not only organizes and establishes a salmon and trout industrialized intelligent circulating water aquaculture system, but also provides technical support for the freshwater fish aquaculture system. Description of the Drawings

[0054] Figure 1 is a flow chart of a salmon and trout industrialized intelligent circulating water aquaculture system;

[0055] Figure 2 is a control block diagram of the water purification system of the present invention;

[0056] In the figure, 1 - aquaculture pond, 2 - sedimentation pond, 3 - return water pond, 4 - solid and suspended matter separation device, 5 - water purification system, 6 - circulating water pump, 7 - biological ammonia and nitrogen removal device, 8 - makeup water pond, 9 - automatic feeding device, 10 - temperature regulating device, 11 - aeration device, 12 - packing device, 13 - temperature sensor, 14 - oxygen content detector, 15 - chemical substance detector, 16 - PH sensor. Detailed Implementation Modes

[0057] To make the objectives, technical solutions, and advantages of the present invention clearer and more understandable, the present invention will be described below through specific embodiments shown in the accompanying drawings. However, it should be understood that these descriptions are merely exemplary and are not intended to limit the scope of the present invention. In addition, in the following description, descriptions of well-known structures and technologies are omitted to avoid unnecessarily confusing the concepts of the present invention.

[0058] The connections mentioned in the present invention are divided into fixed connections and detachable connections. The fixed connections are non-detachable connections, including but not limited to conventional fixed connection methods such as hemming connections, rivet connections, bonding connections, and welding connections. The detachable connections include but are not limited to conventional disassembly methods such as threaded connections, snap connections, pin connections, and hinge connections. When the specific connection method is not clearly defined, it is defaulted that at least one connection method can always be found among the existing connection methods to achieve this function, and those skilled in the art can select it according to their needs. For example: welding connection is selected for the fixed connection, and hinge connection is selected for the detachable connection.

[0059] Detailed Implementation Mode 1: In combination with Figure 1 - Figure 2 To illustrate this implementation mode, an intelligent circulating water aquaculture system for salmon and trout in industrial scale of this implementation mode includes a culture pond 1, a sedimentation pond 2, a return pond 3, a solid and suspended matter separation device 4, a water purification system 5, a circulating water pump 6, and a biological ammonia and nitrogen removal device 7. The culture pond 1 is communicated with the solid and suspended matter separation device 4. The clear water outlet of the suspended matter separation device 4 is communicated with the return pond 3 through the water purification system 5. The turbid water outlet of the suspended matter separation device 4 is communicated with the sedimentation pond 2. The clear water outlet of the sedimentation pond 2 is communicated with the return pond 3. The turbid water outlet of the sedimentation pond 2 discharges sediment. A circulating water pump 6 is provided in the return pond 3. The water outlet of the circulating water pump 6 is communicated with the culture pond 1 through the biological ammonia and nitrogen removal device 7.

[0060] The return pond 3 is connected with a makeup water pond 8. When the circulating water in the system is lost and the water level drops, the pipeline connecting the makeup water pond 8 and the return pond 3 is opened to supplement circulating water into the return pond 3.

[0061] The culture pond 1 is connected with an automatic feeding device 9, and the automatic feeding device 9 regularly and quantitatively feeds salmon and trout aquaculture bait into the culture pond 1.

[0062] The water purification system 5 includes a temperature regulation device 10, an oxygenation device 11, and a packing device 12. The clear water outlet of the solid and suspended matter separation device 4 is communicated with the return pond 3 after passing through the temperature regulation device 10, the oxygenation device 11, and the packing device 12.

[0063] The temperature adjustment device 10 is an existing heat exchange device with temperature adjustment function, which can heat or cool the circulating water flowing through. The oxygenation device 11 is an oxygenation pump or other equipment that can increase the oxygen content in the water body, aiming to increase the oxygen content in the water.

[0064] The filler device 12 can automatically put at least two kinds of materials into the circulating water. These materials can adjust the chemical substances and pH value in the circulating water. The chemical substances include indicators such as ammonia nitrogen, nitrate nitrogen, calcium, and magnesium that need to be monitored for cultivating salmon and trout. For example, if the filler device 12 puts commonly used EM bacteria, Bacillus subtilis, Lactobacillus, etc. in aquaculture, it can adsorb and soften harmful substances such as ammonia nitrogen and nitrite in the pond and improve the water quality.

[0065] Furthermore, the inlet of the filler device 12 is connected to an ozone generator, an ultraviolet lamp sterilization device is installed in the filler device, and one or several of quicklime, potassium permanganate, and activated carbon are put in the filler device to neutralize the pH value and eliminate geosmin.

[0066] Furthermore, the mass ratio of quicklime, potassium permanganate, and activated carbon put in the filler device 12 is 0.1:0.1:1 - 10.

[0067] The water purification system 5 further includes a temperature sensor 13, an oxygen content detector 14, a chemical substance detector 15, and a pH sensor 16. The measuring ends of the temperature sensor 13, the oxygen content detector 14, the chemical substance detector 15, and the pH sensor 16 measure the circulating water in the water purification system 5. The temperature sensor 13, the oxygen content detector 14, the chemical substance detector 15, and the pH sensor 16 are all electrically connected to the control terminal machine. The control terminal machine can display the values detected by the temperature sensor 13, the oxygen content detector 14, the chemical substance detector 15, and the pH sensor 16. In the control terminal machine, the temperature range of the breeding water, the oxygen content range of the breeding water, the standard range of chemical substances in the breeding water, and the pH range can also be set. The temperature adjustment device 10, the oxygenation device 11, and the filler device 12 are all electrically connected to the control terminal machine. The control terminal machine can control the operating states of the temperature adjustment device 10, the oxygenation device 11, and the filler device 12. The control terminal machine can be a terminal device such as a computer or a single-chip microcomputer controller with a control system and a display.

[0068] The temperature adjustment device 10 is controlled by the temperature sensor 13 to select whether to heat or cool the circulating water. The oxygenation device 11 is controlled by the oxygen content detector 14 to be turned on or off. The filler device 12 is controlled by the chemical substance detector 15 and the pH sensor 16 to make the circulating water reach the standard range of chemical substances and the pH range in the breeding water.

[0069] Specific Embodiment 2: Combine Figure 1 - Figure 2Describing this embodiment, based on the salmon and trout industrial intelligent recirculating aquaculture system described in the specific embodiment 1, an aquaculture method for salmon and trout in this embodiment includes the following steps:

[0070] Step 1, start the circulating water pump 6 in the return water tank 3, and the aquaculture water in the aquaculture pond 1 automatically enters the solid and suspended matter separation device 4 for solid and suspended matter separation treatment;

[0071] Step 2, the solids separated by the solid and suspended matter separation device 4 enter the sedimentation tank 2 through the turbid water outlet of the suspended matter separation device 4. The sediment separated in the sedimentation tank 2 is discharged through the turbid water outlet of the sedimentation tank 2, and the circulating water separated in the sedimentation tank 2 enters the return water tank 3 through the clear water outlet;

[0072] Step 3, the circulating water separated by the solid and suspended matter separation device 4 is subjected to circulating water treatment by the water purification system 5 and then enters the return water tank 3;

[0073] Step 4, the circulating water pumped by the circulating water pump 6 in the return water tank 3 flows back into the aquaculture pond 1 after harmful gas removal and disinfection treatment by the biological ammonia nitrogen removal device 7.

[0074] The circulating water treatment includes temperature control treatment by the temperature regulation device 10, oxygenation treatment by the oxygenation device 11, and chemical substance and PH treatment by the packing device 12;

[0075] The temperature control treatment includes: setting the aquaculture water temperature range, measuring the circulating water passing through the water purification system 5 using the temperature sensor 13. If the measured temperature is not within the aquaculture water temperature range, then control the temperature regulation device 10 to exchange heat for the circulating water;

[0076] The oxygenation treatment includes: setting the dissolved oxygen content range of the aquaculture water, measuring the circulating water passing through the water purification system 5 using the dissolved oxygen detector 14. If the measured dissolved oxygen content is lower than the aquaculture water dissolved oxygen content range, then control the oxygenation device 11 to oxygenate the circulating water;

[0077] The chemical substance and PH treatment includes: setting the standard range of chemical substances and the PH range in the aquaculture water, measuring the circulating water passing through the water purification system 5 using the chemical substance detector 15 and the PH sensor 16. If the measured chemical substances and PH in the aquaculture water are not within the standard range of chemical substances and the PH range in the aquaculture water, then control the packing device 12 to pack the circulating water until the chemical substances and PH reach the standard range of chemical substances and the PH range in the aquaculture water and then stop.

[0078] Furthermore, the automatic feeding device 9 makes real-time predictions on the images of salmon and trout in the aquaculture pond 1 collected by the automatic feeding device 9 based on the salmon and trout target prediction module set inside, predicts the length and weight of the salmon and trout, and then based on the predicted length and weight of the salmon and trout, the automatic feeding device 9 calculates the feeding amount through the feeding calculation module set inside to feed the salmon and trout.

[0079] Furthermore, the specific implementation method for the automatic feeding device 9 to feed salmon and trout includes the following steps:

[0080] S1. The automatic feeding device 9 collects images of salmon and trout in the aquaculture pond 1;

[0081] S2. Preprocess the images of salmon and trout collected in step S1 to obtain the preprocessed images of salmon and trout;

[0082] S2.1. Set the pixel value at the pixel coordinate (x, y) position of the collected salmon and trout image as P(x, y), where x and y are the pixel coordinates in the horizontal direction and the vertical direction respectively. Calculate the horizontal difference P x and the vertical difference P y , and the expressions are:

[0083] P x = P(x + 1, y) - P(x, y)

[0084] P y = P(x, y + 1) - P(x, y)

[0085] where P(x + 1, y) is the pixel value at the coordinate (x + 1, y) position of the image, and P(x, y + 1) is the pixel value at the coordinate (x, y + 1) position of the image;

[0086] Then calculate the pixel gradient ▽P(x, y), and the calculation formula is:

[0087]

[0088] The pixel gradient is used to determine the edge intensity at the image pixel points;

[0089] S2.2. Set the adjustment coefficient b(x, y) to determine the update degree of pixels at different positions (x, y) in the image. The calculation formula of the adjustment coefficient b(x, y) is:

[0090]

[0091] where d is a constant controlling sensitivity, and g is a constant controlling the gradient component. Both constants are determined by experience;

[0092] S2.3. Update the pixels of the image to obtain the pixel value P at the position (x, y) of the updated image e (x, y), and the calculation formula is:

[0093] P e (x, y) = P(x, y) + ∑ f b(x, y)(Pf - P(x, y))

[0094] where f are the four pixel coordinates adjacent to the pixel coordinate (x, y) in the up, down, left, and right directions, and Pf refers to the pixel values of the four pixel coordinates adjacent to the pixel coordinate (x, y) in the up, down, left, and right directions;

[0095] S3. Input the pre - processed salmon and trout image into the salmon and trout target prediction module, extract features based on the trained neural network model, and obtain the length L of the i - th salmon and trout i , and then calculate the weight of the salmon and trout in the breeding pond based on the length of the salmon and trout. The calculation formula is:

[0096] W i = a × L i b

[0097] where W i is the weight of the i - th salmon and trout, a is the first coefficient, and b is the second coefficient;

[0098] Then calculate the average value of the weights of all the salmon and trout obtained from the salmon and trout image to get the average weight of the salmon and trout The calculation formula is:

[0099]

[0100] where m is the total number of all the salmon and trout in the salmon and trout image;

[0101] Then calculate the total weight W total of the salmon and trout in the breeding pond. The calculation formula is:

[0102]

[0103] where n is the total number of the salmon and trout in the breeding pond;

[0104] S4. Based on the obtained weight estimation of the salmon and trout, the automatic feeding device 9 calculates the feeding amount through the feeding calculation module set inside. The feeding amount M total The calculation formula is:

[0105] M total = 2% × W total .

[0106] Further, according to the obtained feeding amount, bait is added to the feed bin. A weighing sensor is provided at the bottom of the feed bin. When the system time of the automatic feeding device is equal to the feeding time set by the user, the feeding process starts. In order to avoid fish swarming to grab food, an intermittent feeding method is adopted. The feeding time and the interval time are set by the user. The spiral discharging motor works intermittently to batch-feed the bait in the feed bin into the breeding pond. During the feeding process, the weighing sensor continuously detects whether there is still bait in the feed bin. If there is bait, the above process is repeated until all the bait in the feed bin is fed out.

[0107] Further, the neural network model described in step S3 is composed of two parts, an encoder part and a decoder part;

[0108] S3.1. First, the pre-processed salmon and trout fish image obtained in step S3 is subjected to a channel number dimension transformation to be transformed into a d-dimensional feature map H and W respectively represent the height and width of the feature map, and then the spatial dimension of the d-dimensional feature map is compressed into a one-dimensional feature map Then it is input into the encoder part;

[0109] S3.2. The encoder part is composed of a multi-head self-attention module, a feed-forward network FFN, and a fixed position encoding pos; in the multi-head self-attention module, Q, K, and V are respectively set as the query matrix, key matrix, and value matrix of the input feature embedding, W q is the query learning matrix, W k is the key learning matrix, W v is the value learning matrix, and V is the value matrix, and the calculation expression is obtained as:

[0110] Q = W q × (z t + pos)

[0111] K = W k × (z t + pos)

[0112] V = W v × z t ;

[0113] Calculate the self-attention weight, and the calculation expression is obtained as:

[0114]

[0115] where a ji represents the self-attention weight of the i-th position in the feature to the j-th position, and Softmax is the normalized exponential function;

[0116] After obtaining the self-attention weights, multiply the self-attention weights by the value matrix to obtain the feature vector v of the multi-head self-attention module j , and the calculation formula is as follows:

[0117]

[0118] Add the feature vector v of the multi-head self-attention module j to the one-dimensional feature map z t , add and normalize them, then send them into the feed-forward network FFN, and then perform the addition and normalization operations again. Execute the above operations N times to complete the encoder part and obtain the feature vector output by the encoder part;

[0119] S3.3. Input the feature vector output by the encoder part into the decoder part. The decoder part consists of a multi-head self-attention module and a feed-forward network FFN. Define the number of detection targets to be output finally as m, which is called the object queries, and convert the m object queries into the final feature vector g through the decoder part t ; finally, input the feature vector g t into the feed-forward network FFN, and the calculation expression is:

[0120] b t = FFN 1 (g t )

[0121] c t = FFN 2 (g t )

[0122] Among them, FFN is composed of a three-layer perceptron with a ReLU activation function and a hidden dimension d and a linear mapping layer. b t represents the bounding box information of the target object in the detected image, and c t represents the length information of the target object in the detected image.

[0123] Furthermore, use the annotation tool for the pre-processed salmon and trout images obtained in step S2 to annotate the position of the salmon and trout and the aggregation degree of the salmon and trout in each image. The aggregation degree of the salmon and trout is divided into 1-10 levels to obtain the salmon and trout image annotation file, where level 1 is the sparsest aggregation and level 10 is the densest aggregation. The specific setting method is as follows;

[0124] The fish body spacing > 1 times the body length is set to level 1;

[0125] 0.95 times the body length < fish body spacing ≤ 1 times the body length is set to level 1;

[0126] When the fish body spacing is greater than 0.9 times the body length and less than or equal to 0.95 times the body length, it is set as level 2;

[0127] When the fish body spacing is greater than 0.85 times the body length and less than or equal to 0.9 times the body length, it is set as level 3;

[0128] When the fish body spacing is greater than 0.8 times the body length and less than or equal to 0.85 times the body length, it is set as level 4;

[0129] When the fish body spacing is greater than 0.75 times the body length and less than or equal to 0.8 times the body length, it is set as level 5;

[0130] When the fish body spacing is greater than 0.7 times the body length and less than or equal to 0.75 times the body length, it is set as level 6;

[0131] When the fish body spacing is greater than 0.65 times the body length and less than or equal to 0.7 times the body length, it is set as level 7;

[0132] When the fish body spacing is greater than 0.6 times the body length and less than or equal to 0.65 times the body length, it is set as level 8;

[0133] When the fish body spacing is greater than 0.5 times the body length and less than or equal to 0.6 times the body length, it is set as level 9;

[0134] When the fish body spacing is less than 0.5 times the body length, it is set as level 10;

[0135] Construct a data set, convert the obtained salmon and trout fish image annotation files into txt format, and divide the data set into a training set, a validation set, and a test set according to the ratio of 70%, 15%, and 15% for the images and the corresponding image annotation files;

[0136] Use the obtained training set to train the YOLOv10 model. First, configure the environment suitable for the operation of the YOLOv10 model, and set the learning rate, batch size, training cycle, and hyperparameters. Monitor the training progress using the loss during the training process and the performance on the validation set, and adjust the training parameters;

[0137] Use the precision and recall rate parameters to evaluate the effect of salmon and trout fish recognition and the degree of salmon and trout fish aggregation. When the precision and recall rate reach the set values, stop the training. At this time, the corresponding model is the salmon and trout fish aggregation degree recognition model.

[0138] Then, during the salmon and trout fish feeding process, collect salmon and trout fish images in real time, and input them into the salmon and trout fish aggregation degree recognition model for salmon and trout fish aggregation degree recognition; when it is continuously recognized as a salmon and trout fish aggregation degree of level 1 - 3 within 1 minute during the feeding process, control the automatic feeding device to stop feeding; when it is continuously recognized as a salmon and trout fish aggregation degree of level 8 - 10 within 1 minute after the feeding ends, control the automatic feeding device to continue feeding until the salmon and trout fish aggregation degree is recognized as a salmon and trout fish aggregation degree below level 6.

[0139] It should be noted that in the above embodiments, as long as the technical solutions do not conflict, they can be permutated and combined. Those skilled in the art can exhaust all possibilities based on the mathematical knowledge of permutation and combination. Therefore, the present invention will no longer explain each of the technical solutions after permutation and combination one by one, but it should be understood that the technical solutions after permutation and combination have been disclosed by the present invention.

[0140] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention can have various modifications and changes. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A salmon and trout factory-scale intelligent circulating aquaculture system, characterized by: The invention comprises a culture pond (1), a sedimentation pond (2), a return water pond (3), a solid and suspended matter separation device (4), a water purification system (5), a circulating water pump (6) and a biological ammonia nitrogen removal device (7). The culture pond (1) is connected to the solid and suspended matter separation device (4), the clean water outlet of the suspended matter separation device (4) is connected to the return water pond (3) after passing through the water purification system (5), the turbid water outlet of the suspended matter separation device (4) is connected to the sedimentation pond (2), the clean water outlet of the sedimentation pond (2) is connected to the return water pond (3), the turbid water outlet of the sedimentation pond (2) discharges sediment, and a circulating water pump (6) is arranged in the return water pond (3), and the water outlet of the circulating water pump (6) is connected to the culture pond (1) through the biological ammonia nitrogen removal device (7).

2. The intelligent circulating aquaculture system for salmon and trout according to claim 1 is characterized by: The return water tank (3) is connected to a water replenishment tank (8).

3. The intelligent circulating aquaculture system for salmon and trout according to claim 2 is characterized by: The breeding pond (1) is connected to an automatic feeding device (9).

4. The intelligent circulating aquaculture system for salmon and trout according to claim 3 is characterized by: The water purification system (5) comprises a temperature regulating device (10), an oxygenating device (11) and a filling device (12); the clean water outlet of the solid and suspended matter separation device (4) is connected to the return water tank (3) after passing through the temperature regulating device (10), the oxygenating device (11) and the filling device (12).

5. The intelligent circulating aquaculture system for salmon and trout according to claim 4 is characterized by: The water purification system (5) further comprises a temperature sensor (13), an oxygen content detector (14), a chemical substance detector (15) and a pH sensor (16), wherein the measuring ends of the temperature sensor (13), the oxygen content detector (14), the chemical substance detector (15) and the pH sensor (16) measure the circulating water in the water purification system (5).

6. A method for intelligent circulating aquaculture of salmon and trout in factories, which is realized by relying on the intelligent circulating aquaculture system for salmon and trout in factories according to claim 5, characterized in that: The following steps are involved: Step 1, starting the circulating water pump (6) in the return water tank (3), and the aquaculture water in the aquaculture tank (1) automatically enters the solid and suspended matter separation device (4) for solid and suspended matter separation treatment; Step 2, the solid separated by the solid and suspended matter separation device (4) enters the sedimentation tank (2) through the turbid water outlet of the suspended matter separation device (4), the sediment separated in the sedimentation tank (2) is discharged through the turbid water outlet of the sedimentation tank (2) for treatment, and the circulating water separated in the sedimentation tank (2) enters the recycle tank (3) through the clean water outlet; Step 3, the circulating water separated by the solid and suspended matter separation device (4) is processed by the water purification system (5) and then enters the return water tank (3); Step 4, the circulating water drawn by the circulating water pump (6) in the return pool (3) passes through the biological ammonia nitrogen removal device (7) to remove harmful gases and disinfect and sterilize, and then flows back into the breeding pond (1).

7. The method for industrialized intelligent circulating aquaculture of salmon and trout according to claim 6, characterized in that: The circulating water treatment includes temperature control treatment by a temperature regulating device (10), oxygenation treatment by an oxygenation device (11), and chemical substance and pH treatment by a filler device (12); The temperature control process comprises: setting a temperature range of the aquaculture water, using a temperature sensor (13) to measure the circulating water passing through the water purification system (5), and if the measured temperature is not within the aquaculture water temperature range, controlling the temperature regulating device (10) to perform heat exchange on the circulating water; The oxygenation treatment comprises: setting the oxygen content range of the aquaculture water, using an oxygen content detector (14) to measure the circulating water passing through the water purification system (5), and if the measured oxygen content is lower than the oxygen content range of the aquaculture water, controlling the oxygenation device (11) to oxygenate the circulating water; The chemical substance and pH treatment comprises: setting a standard range of chemical substances and a pH range in the aquaculture water, using a chemical substance detector (15) and a pH sensor (16) to measure the circulating water passing through the water purification system (5); if the measured chemical substances and pH in the aquaculture water are not within the standard range of chemical substances and the pH range in the aquaculture water, then controlling the filling device (12) to fill the circulating water until the chemical substances and the pH reach the standard range of chemical substances and the pH range in the aquaculture water and then stopping.

8. The method for intelligent circulating aquaculture of salmon and trout according to claim 7, characterized in that: The automatic feeding device (9) performs real-time prediction of the salmon and trout images in the breeding pond (1) collected by the automatic feeding device (9) based on the salmon and trout target prediction module arranged inside, and predicts the length and weight of the salmon and trout. Then, based on the predicted length and weight of the salmon and trout, the automatic feeding device (9) calculates the feeding amount through the feeding calculation module arranged inside, and feeds the salmon and trout.

9. The method for industrialized intelligent circulating aquaculture of salmon and trout according to claim 8, characterized in that: The specific implementation method of the automatic feeding device (9) for feeding salmon and trout comprises the following steps: S1. An automatic feeding device (9) collects images of salmon and trout in a breeding pond (1); S2. Pre-processing the salmon and trout image collected in step S1 to obtain a pre-processed salmon and trout image; S2.

1. Set the pixel value of the collected salmon image at the pixel coordinate (x, y) position to P(x, y), where x and y are the horizontal pixel coordinates and the vertical pixel coordinates, and calculate the horizontal difference P of the pixel gradient ▽P(x, y) in the image. x and the vertical difference P y , the expression is: P x =P(x+1,y)-P(x,y) P y =P(x,y+1)-P(x,y) Where P(x+1,y) is the pixel value of the image at the coordinate position (x+1,y), and P(x,y+1) is the pixel value of the image at the coordinate position (x,y+1); Then calculate the pixel gradient ▽P(x,y), the calculation formula is: Pixel gradient is used to determine the edge strength at image pixels; S2.

2. Set the adjustment coefficient b(x,y) to determine the update degree of pixels at different positions (x,y) in the image. The calculation formula of the adjustment coefficient b(x,y) is: Among them, d is a constant for controlling sensitivity, and g is a constant for controlling the gradient component, both constants are determined empirically; S2.

3. Update the pixel of the image and obtain the pixel value P of the updated image at the coordinate position (x, y) e (x,y), the calculation formula is: P e (x,y)=P(x,y)+∑ f b(x,y)(Pf-P(x,y)) Wherein, f is the four pixel coordinates adjacent to the pixel coordinate (x, y) in the four directions of up, down, left, and right, and Pf refers to the pixel value of the four pixel coordinates adjacent to the pixel coordinate (x, y) in the four directions of up, down, left, and right; S3. Input the pre-processed salmon and trout image into the salmon and trout target prediction module, perform feature extraction based on the trained neural network model, and obtain the length L of the i-th salmon and trout i , and then calculate the weight of the salmon and trout in the breeding pond based on the length of the salmon and trout, the calculation formula is: IN i =a×L i b Among them, W i is the weight of the ith salmon, a is the first coefficient, and b is the second coefficient; Then the weights of all the salmon trout obtained in the salmon trout image are averaged to get the average weight of the salmon trout The calculation formula is: Where m is the number of all salmon and trout in the salmon and trout image; Then calculate the total weight W of salmon and trout in the breeding pond total The calculation formula is: Where n is the total number of salmon and trout in the breeding pond; S4. Based on the obtained weight estimate of the salmon and trout, the automatic feeding device (9) calculates the feeding amount through the internal feeding calculation module, and the feeding amount M total The calculation formula is: M total =2%×W total 。

Images