Intelligent marine fish breeding device and method based on vision and eddy current

Abstract

This invention provides a vision- and eddy current-based intelligent aquaculture device and method for marine fish, relating to the field of aquaculture technology. The device includes: a layered monitoring module for collecting environmental data, fish body condition data, and residual feed data, wherein the environmental data includes water temperature and dissolved oxygen data, and the fish body condition data includes fish behavior data, cyst distribution data at the bottom of the container, and fish body surface data; an edge computing unit for calculating instructions based on thresholds corresponding to the environmental data, fish body condition data, and residual feed data to obtain eddy current control instructions and feeding instructions; an intelligent eddy current module for switching corresponding operating modes according to the eddy current control instructions, including a normal aquaculture mode and a cyst removal mode; and an intelligent feeder for feeding the aquaculture area according to the feeding instructions. This invention solves the problem of low efficiency in marine fish disease monitoring and prevention in existing technologies.

Description

Technical Field

[0001] This invention relates to the field of aquaculture technology, and in particular to a vision- and eddy current-based intelligent aquaculture device and method for marine fish. Background Technology

[0002] Cryptocryon irritans, a major parasitic ciliate of marine bony fish, has a life cycle consisting of four stages: trophozoite, cyst precursor, cyst, and larva. The cyst stage is the core link in the spread of disease. When the water temperature is between 25-27℃, a single cyst can hatch in 3-7 days, releasing hundreds of ciliate larvae. These larvae invade the fish through wounds on the body surface or thin mucus layers, leading to large-scale mortality of fish fry and causing billions of dollars in losses to the global marine aquaculture industry annually.

[0003] The main problems in pest and disease monitoring and control are as follows:

[0004] Outdated monitoring methods: Traditional manual observation relies on the experience of farmers, resulting in a missed detection rate of over 30% for cysts, and it is impossible to track the dynamic distribution of cysts in real time;

[0005] Limitations of control methods: Currently, the main control methods for cryptocaryonosis are chemical control, physical control, and immunotherapy. Chemical control uses drugs that are harmful to fish, users, and the environment, and are prone to drug residues; physical control is time-consuming and labor-intensive and can easily damage fish; immunotherapy has not yet been put into use; the newly proposed biological control relies on the natural environment and lacks a dynamic regulation mechanism.

[0006] In summary, there are still technological gaps in the efficient physical removal and intelligent control of the encapsulation stage. Summary of the Invention

[0007] In order to overcome the shortcomings of the prior art, the purpose of this invention is to provide a vision-based and eddy current-based intelligent aquaculture device and method for marine fish. This invention solves the problem of low efficiency in monitoring and protecting marine fish diseases in the prior art.

[0008] To achieve the above objectives, the present invention provides the following solution:

[0009] A vision- and eddy current-based intelligent aquaculture device for marine fish includes:

[0010] The system comprises an edge computing unit, a breeding tank body, and a layered monitoring module, an intelligent vortex module, and an intelligent feeder, all positioned at preset locations on the breeding tank body. The bottom of the breeding tank body is a conical surface with an inclination angle of 30°, and the interior of the breeding tank body includes a breeding zone and a vortex zone.

[0011] The edge computing unit is connected to the hierarchical monitoring module, the intelligent eddy current module, and the intelligent feeder, respectively.

[0012] The stratified monitoring module is used to collect environmental data, fish condition data, and residual feed data. The environmental data includes water temperature data, dissolved oxygen data, and cyst removal data at the bottom of the tank. The fish condition data includes fish behavior data and fish surface data. The edge computing unit is used to calculate instructions based on the thresholds corresponding to the environmental data, fish condition data, and residual feed data to obtain eddy current control instructions and feeding instructions. The intelligent eddy current module is used to switch the corresponding working mode according to the eddy current control instructions. The working modes include normal breeding mode and cyst removal mode. The intelligent feeder is used to feed the breeding area according to the feeding instructions.

[0013] Preferably, the tiered monitoring module includes:

[0014] Top underwater camera, side wall underwater camera, and water quality sensor;

[0015] The top underwater camera is used to collect fish behavior data and fish body surface data, the side underwater camera is used to collect data on the distribution of pods at the bottom of the bucket and data on remaining food, and the water quality sensor is used to collect water temperature data and dissolved oxygen data.

[0016] Preferably, the main body of the aquaculture tank includes:

[0017] Top support, outer shell of the breeding tank, mesh isolation layer, four corner supports of the outer tank, conical surface;

[0018] The top bracket is fixedly installed at the upper port of the aquaculture tank shell to support the top underwater camera;

[0019] The grid isolation layer is horizontally positioned above the junction of the inner wall of the breeding tank shell and the conical surface, with its height corresponding to the top of the central drain valve, and is used to separate the breeding area from the vortex area.

[0020] The four corner supports of the outer barrel are located around the bottom perimeter of the outer shell of the breeding barrel and are used to support the main body of the breeding barrel and adjust its level.

[0021] The conical surface is used to create a spiral sinking flow field in the bladder removal mode, which collects the bladders at the bottom of the barrel to the central drain valve.

[0022] Preferably, the intelligent eddy current module includes:

[0023] Edge tangential water inlet pipe, circulating water drain outlet and central drain valve;

[0024] The edge tangential water inlet pipes are evenly distributed circumferentially along the inner wall of the aquaculture tank shell, which is used to form a slow circumferential circulation at a flow rate of 0.2-0.3 m / s in normal aquaculture mode, and to increase it to 0.5-0.8 m / s in cyst removal mode to enhance the spiral vortex.

[0025] The circulating water drain outlet is located on the side wall of the aquaculture tank shell and is higher than the mesh isolation layer. It is used to maintain the water level and discharge surface residual feed and floating impurities during normal aquaculture mode.

[0026] The central drain valve is located at the lowest point of the conical surface and is used to fully open during the cyst removal mode to discharge the cysts and wastewater accumulated at the bottom of the tank, and to close during normal aquaculture mode to maintain water stability.

[0027] Preferably, the central drain valve is a solenoid valve with a diameter ranging from 10 to 15 cm.

[0028] Preferably, the time period of the capsule clearance mode is 30s-60s.

[0029] Preferably, the mesh diameter of the mesh isolation layer is in the range of 2-5 mm.

[0030] Preferably, the conical surface is made of smooth polyethylene material.

[0031] A vision- and eddy current-based intelligent aquaculture method for marine fish includes:

[0032] Image and environmental parameter acquisition: Top camera and inner wall camera collect fish behavior, bucket bottom sac distribution and fish body surface feature data in real time at preset frequency; Water quality sensor collects water temperature data and dissolved oxygen data simultaneously.

[0033] Image preprocessing and feature analysis: Noise removal and contrast enhancement are performed on the acquired images. The YOLO11 deep learning algorithm is applied to identify small targets in the capsule and fish behavior. Combined with morphological filtering and spatial localization, the capsule density and abnormal fish behavior indicators are output.

[0034] Dynamic decision-making and execution: When the cyst density exceeds the threshold, the inlet flow rate is automatically increased and the drain valve is opened to form a spiral vortex at the bottom of the tank, which concentrates and discharges the cysts. After each cleaning, the density is checked again at a set interval. If it is below the threshold, the process stops; if it is above the threshold, the process is repeated. The cumulative number of cleanings does not exceed 3.

[0035] Feeding amount control: Feed 3 times a day at set times, with each feeding amount being 4% of the initial weight of the feeder. The feeding strategy is controlled by the residual feed data collected by the side wall camera. If the camera identifies a residual feed coverage area >5% 1 hour after feeding, the next feeding amount will be reduced by 15%; if the camera identifies a residual feed coverage area <2% 2% 2 times in a row, the feeding amount will be increased by 10%.

[0036] Anomaly warning and remote management: Automatically issues real-time warnings for excessive cysts, abnormal water temperature data, and abnormal dissolved oxygen data, and pushes alarms to the administrator terminal, supporting remote intervention and manual operation.

[0037] The present invention discloses the following technical effects:

[0038] This invention provides a vision- and vortex-based intelligent aquaculture device for marine fish, comprising: an edge computing unit, a culture tank body, and a layered monitoring module, an intelligent vortex module, and an intelligent feeder disposed at preset positions on the culture tank body; wherein, the bottom of the culture tank body is a conical surface with an inclination angle of 30°, and the interior of the culture tank body includes a culture area and a vortex area; the edge computing unit is connected to the layered monitoring module, the intelligent vortex module, and the intelligent feeder respectively; the layered monitoring module is used to collect environmental data, fish body status data, and residual feed data, wherein... The environmental data includes water temperature and dissolved oxygen data. The fish status data includes fish behavior data, cyst distribution data at the bottom of the container, and fish surface data. The edge computing unit is used to calculate instructions based on thresholds corresponding to the environmental data, fish status data, and residual feed data, resulting in eddy current control instructions and feeding instructions. The intelligent eddy current module is used to switch the corresponding working mode according to the eddy current control instructions. The working modes include normal aquaculture mode and cyst removal mode. The intelligent feeder is used to feed the aquaculture area according to the feeding instructions. This invention uses multiple cameras to monitor cyst distribution and fish behavior in real time, utilizes the YOLO11 algorithm to accurately identify targets, and combines a 30° tilt angle at the bottom of the container with tangential water inflow at the edge to form a natural spiral eddy current, achieving efficient cyst removal and intelligent feeding at the bottom of the container. A closed-loop system of "monitoring-eddy current-feeding" is constructed, significantly improving the efficiency of Cryptocaryon stimulans cyst removal, reducing energy consumption and labor costs, and is suitable for industrial indoor aquaculture scenarios. It cuts off key links in disease transmission and ensures the healthy growth of grouper. Attached Figure Description

[0039] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0040] Figure 1 A three-dimensional schematic diagram of a vision- and eddy current-based intelligent aquaculture device for marine fish, provided for an embodiment of the present invention;

[0041] Figure 2 A schematic diagram of the internal structure of a vision- and eddy current-based intelligent aquaculture device for marine fish, provided for an embodiment of the present invention;

[0042] Figure 3 A schematic diagram of the bottom valve of a vision- and eddy current-based intelligent aquaculture device for marine fish, provided for an embodiment of the present invention;

[0043] Figure 4 This is a schematic diagram of the control logic provided in an embodiment of the present invention.

[0044] Explanation of reference numerals in the attached figures:

[0045] 1-Intelligent feeder; 2-Top bracket; 3-Water quality sensor; 4-Top underwater camera; 5-Outer shell of aquaculture tank; 6-Mesh isolation layer; 7-Four corner brackets of outer tank; 8-Edge tangential water inlet pipe; 9-Side wall underwater camera; 10-Conical surface; 11-Circulating water drain outlet; 12-Central drain valve. Detailed Implementation

[0046] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0047] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0048] like Figure 1-3 As shown, the present invention provides a vision- and eddy current-based intelligent aquaculture device for marine fish, comprising:

[0049] The system comprises an edge computing unit, a breeding tank body, and a layered monitoring module, an intelligent vortex module, and an intelligent feeder 1, all positioned at a preset location on the breeding tank body. The bottom of the breeding tank body is a conical surface 10 with an inclination angle of 30°, and the interior of the breeding tank body includes a breeding area and a vortex area.

[0050] The edge computing unit is connected to the hierarchical monitoring module, the intelligent eddy current module, and the intelligent feeder 1, respectively.

[0051] The stratified monitoring module is used to collect environmental data, fish body status data, and residual feed data. The environmental data includes water temperature data and dissolved oxygen data. The fish body status data includes fish behavior data, cyst distribution data at the bottom of the container, and fish body surface data. The edge computing unit is used to calculate instructions based on the thresholds corresponding to the environmental data, fish body status data, and residual feed data to obtain eddy current control instructions and feeding instructions. The intelligent eddy current module is used to switch the corresponding working mode according to the eddy current control instructions. The working modes include normal breeding mode and cyst removal mode. The intelligent feeder 1 is used to feed the breeding area according to the feeding instructions.

[0052] Specifically, the camera equipment consists of a top-mounted underwater camera (resolution ≥1080P, waterproof rating IP68) and an inner wall camera (180° wide-angle lens, waterproof rating IP68), fixed to the tank body via a corrosion-resistant bracket. This allows for 360° image acquisition within the tank, covering the water surface, middle layer, and bottom. The top camera monitors the grouper's swimming behavior and body surface condition, while the inner wall camera focuses on the distribution data of cysts and residual feed at the bottom of the tank. The sensor unit integrates a water temperature sensor (accuracy ±0.5℃) and a dissolved oxygen sensor (measurement range 0-20 mg / L, accuracy ±0.2 mg / L), collecting real-time environmental parameters of the tank to provide data support for dynamic adjustment of cyst detection thresholds and eddy current control strategies.

[0053] Bucket bottom structure design:

[0054] The bottom of the bucket is a conical surface 10 with a 30° inclination angle, and a smart drain valve (solenoid valve) with a diameter of 10-15cm is installed in the center to connect to the external circulating water system; the water inlet pipe is evenly distributed on the edge of the bottom of the bucket (50-80cm from the center), and adopts a tangential water inlet design (45° angle with the tangent of the bucket wall) to ensure that the water flows spirally towards the center along the inclination angle of the bottom of the bucket.

[0055] Work mode:

[0056] Normal farming method (without cysts):

[0057] The edge water inlet pipe continuously supplies water at a flow rate of 0.2 m / s, while the central valve at the bottom of the bucket is closed to maintain slow water circulation and prevent residual bait from settling.

[0058] Capsule clearance mode (capsule detected):

[0059] When the inner wall camera detects a water temperature of 25℃ and a cyst density at the bottom of the tank of >5 cysts / ㎡, the flow rate of the edge inlet pipe is increased to 0.5m / s, and the central valve is fully opened (100% opening). Utilizing the 30° tilt angle of the tank bottom and the tangential water inlet, a clockwise spiral vortex (flow rate 0.6-0.8m / s) is formed, causing the cysts to converge towards the center with the water flow and be discharged. The single cleaning time is 30-60s, and the system automatically switches back to normal mode after completion.

[0060] Control logic: Based on the characteristics of the bottom tilt angle of the tank, the capsule detection threshold is lowered by 1 capsule / ㎡ for every 1℃ increase in water temperature, and the clearing is triggered in advance under high temperature environment;

[0061] After a single cleaning, re-inspect 30 minutes later. If the density is still >5 particles / m², repeat the cleaning process. The total number of cleanings should be ≤3 to avoid excessive drainage affecting water quality.

[0062] Adaptive feeding system:

[0063] Feeding equipment includes a fish monitoring camera (6-megapixel resolution) and a quantitative feeder (accuracy ±2%). The former monitors the grouper's body surface condition and swimming behavior in real time, while the latter feeds the grouper three times a day at set times, each time at 4% of the initial weight.

[0064] Feeding control strategy: The feeding strategy is controlled by the residual feed data collected by the side wall camera. If the residual feed coverage area is >5% after 1 hour of feeding, the next feeding amount will be reduced by 15%; if the residual feed coverage area is <2% for two consecutive times, the feeding amount will be increased by 10%.

[0065] Edge computing unit:

[0066] It incorporates the YOLO11 target detection algorithm, trained on capsule morphology features (diameter 50-300μm, surface ridges), achieving an accuracy of ≥95% and a false positive rate of <5%; it is also trained on residual bait morphology features, achieving an accuracy of ≥95% and a false positive rate of <5%.

[0067] It integrates multimodal data fusion algorithms, combines data on cyst density, abnormal fish behavior, water temperature, and dissolved oxygen, and outputs control commands such as valve opening and closing and feeding amount adjustment to form a closed loop of "monitoring-analysis-execution".

[0068] Furthermore, the hierarchical monitoring module includes:

[0069] Top underwater camera 4, side wall underwater camera 9 and water quality sensor 3;

[0070] The top underwater camera 4 is used to collect fish behavior data and fish body surface data, the side underwater camera 9 is used to collect data on the distribution of pods at the bottom of the bucket and data on remaining food, and the water quality sensor 3 is used to collect water temperature data and dissolved oxygen data.

[0071] Furthermore, the main body of the aquaculture tank includes:

[0072] 2. Top support; 5. Outer shell of the breeding tank; 6. Mesh isolation layer; 7. Four-corner support of the outer tank; 10. Conical surface;

[0073] The top bracket 2 is fixedly installed on the upper port of the aquaculture tank shell 5 to support the top underwater camera 4;

[0074] The grid isolation layer 6 is horizontally positioned above the junction of the inner wall of the outer shell 5 of the breeding tank and the conical surface 10, with its height corresponding to the top of the central drain valve 12, and is used to separate the breeding area from the vortex area.

[0075] The four corner supports 7 of the outer barrel are located around the bottom periphery of the outer shell 5 of the breeding barrel, and are used to support the main body of the breeding barrel and adjust its level.

[0076] The conical surface 10 is used to form a spiral sinking flow field in the bladder removal mode, which collects the bladder at the bottom of the barrel to the central drain valve 12.

[0077] Furthermore, the intelligent eddy current module includes:

[0078] 8 edge tangential water inlet pipe, 11 circulating water outlet and 12 center drain valve;

[0079] The edge tangential water inlet pipe 8 is evenly distributed circumferentially along the inner wall of the outer shell 5 of the breeding tank, and is used to form a slow circumferential circulation at a flow rate of 0.2-0.3 m / s in the normal breeding mode, and to increase it to 0.5-0.8 m / s in the cyst removal mode to enhance the spiral vortex.

[0080] The circulating water drain outlet 11 is located on the side wall of the outer shell 5 of the breeding tank and is higher than the mesh isolation layer 6. It is used to maintain the water level and discharge surface residual feed and floating impurities in normal breeding mode.

[0081] The central drain valve 12 is located at the lowest point of the conical surface 10. It is used to fully open during the cyst removal mode to discharge the cysts and wastewater accumulated at the bottom of the tank, and to close during normal aquaculture mode to maintain water stability.

[0082] Furthermore, the central drain valve 12 is a solenoid valve with a diameter ranging from 10 to 15 cm.

[0083] Furthermore, the time period of the capsule clearance mode ranges from 30s to 60s.

[0084] Furthermore, the mesh diameter of the aforementioned mesh isolation layer 6 ranges from 2 to 5 mm.

[0085] Furthermore, the conical surface 10 is made of smooth polyethylene material.

[0086] like Figure 4 As shown, corresponding to the above-mentioned device, this embodiment also discloses specific control steps:

[0087] Feature extraction and analysis:

[0088] This step utilizes a multimodal data fusion algorithm and a hierarchical feature extraction framework to accurately quantify abnormal behavior and the number of cysts in grouper. The specific process is as follows:

[0089] Image acquisition and preprocessing:

[0090] The top camera (1080P, 5 frames / second) captures real-time images of the fish from above and on its surface. After removing water ripple noise through Gaussian filtering, the images are input into the YOLO11 model for fish detection and tracking.

[0091] Core indicator calculation:

[0092] Average swimming speed monitoring: A single grouper is tracked via video stream, and its total distance traveled within 300 frames (1 minute) is calculated to obtain its average swimming speed V (unit: cm / s). If V ≥ 10 cm / s is detected for 10 consecutive minutes (normal state V < 5 cm / s), it is judged as "abnormal behavior".

[0093] Exception triggering rules:

[0094] When ≥30% of the fish in a single frame exhibit the above-mentioned abnormality, and this is triggered ≥3 times within 10 consecutive minutes, the edge computing unit generates an abnormal behavior alarm and triggers the capsule detection module to start an expedited scan.

[0095] Detection of the number of capsules at the bottom of the bucket (precise identification of small targets):

[0096] Image acquisition and target recognition:

[0097] The camera on the inner wall of the grid platform (waterproof rating IP68) acquires images of the bottom of 10 barrels on the conical surface at a frequency of 2 frames / second. After contrast enhancement preprocessing, the images are input into the YOLO11 model to detect cysts (training data labeled cyst diameter 50-300μm, surface ridge features).

[0098] Detection of residual bait at the bottom of the bucket (precise identification of small targets):

[0099] Image acquisition and target recognition:

[0100] The camera on the inner wall of the grid platform (waterproof rating IP68) captures images of the bottom of 10 buckets on the conical surface at a frequency of 2 frames per second. After contrast enhancement preprocessing, the images are input into the YOLO11 model to detect residual bait.

[0101] Dynamic threshold adjustment:

[0102] The edge computing unit automatically corrects the detection threshold based on water temperature sensor data (accuracy ±0.5℃).

[0103] Benchmark threshold: 5 per square meter (at a water temperature of 25°C);

[0104] For every 1°C increase in temperature, the threshold is lowered by 1 per square meter (e.g., the threshold is adjusted to 3 per square meter at 27°C) to improve the sensitivity of capsule recognition under high temperature conditions.

[0105] Density quantization methods:

[0106] The pixel coordinates of the detected capsules and residual bait are converted into physical positions using camera calibration parameters. A 1m×1m grid is divided with the center of the bottom of the bucket as the origin. The number of capsules in each grid is calculated, and the density per unit area (capsules / m²) is output.

[0107] Multimodal data fusion and decision-making:

[0108] Edge computing units integrate two types of data using timestamp alignment technology (error < 100ms) and execute hierarchical control logic: Abnormal behavior linkage:

[0109] If the abnormal behavior alarm and the capsule density > 5 capsules / ㎡ are triggered simultaneously, immediately increase the water inflow by 50% and start the drain valve at the bottom of the tank (open in 10s / close in 30s). The initial duration is 2 minutes, and it will be dynamically extended according to the capsule density thereafter.

[0110] Environmental parameter warning:

[0111] If the water temperature is greater than 27℃ and the capsule density is greater than 3 capsules / m², a text message alert will be sent to the administrator via the 4G module to prompt manual intervention.

[0112] Capsule-assisted intelligent control:

[0113] Triggering conditions:

[0114] When the underwater camera under the grid platform detects that the density of cysts at the bottom of the tank at 25℃ is greater than 5 per square meter, it immediately triggers a 50% increase in water intake and activates the drain valve at the bottom of the tank (opens in 10 seconds and closes in 30 seconds), with an initial duration of 2 minutes.

[0115] Dynamic control rules:

[0116] If the capsule density is >5 capsules / m² at 25℃, the system will automatically increase the water inflow and extend the drain valve opening time to 4 minutes. After one flushing cycle, the system will automatically recheck the capsule density. If it is still >5 capsules / m², a second flush will be started after a 10-minute interval, until the density is <5 capsules / m².

[0117] Environmental parameter linkage:

[0118] When the water temperature is above 27℃, the capsule detection threshold is automatically lowered to 3 capsules / ㎡, and the flushing process start time is extended to 3 minutes to improve the disinfection efficiency during high-temperature periods.

[0119] Multi-dimensional abnormal data alerts:

[0120] Abnormal behavior in grouper lasts for ≥10 minutes (e.g., average swimming speed ≥10cm / s).

[0121] Water temperature >30℃ or <20℃;

[0122] Dissolved oxygen <5 mg / L or >12 mg / L, and sustained for ≥5 minutes.

[0123] Warning execution logic:

[0124] Abnormal data is sent to the administrator terminal in real time via the 4G communication module. The content includes:

[0125] Anomaly types (such as "excessive capsule density", "abnormal water temperature data", "abnormal dissolved oxygen data");

[0126] Specific parameter values ​​(such as "capsule density: 6 capsules / m²" and "water temperature: 31℃");

[0127] Suggested actions (e.g., "Please check if the water quality sensor is working properly").

[0128] This embodiment also provides a vision- and eddy current-based intelligent aquaculture method for marine fish, the method including:

[0129] Image and environmental parameter acquisition: Top camera and inner wall camera collect fish behavior data, bucket bottom bag distribution data, bucket bottom residual food distribution data, and fish body surface feature data in real time at preset frequency; Water quality sensor collects water temperature data and dissolved oxygen data simultaneously.

[0130] Image preprocessing and feature analysis: Noise removal and contrast enhancement are performed on the acquired images. The YOLO11 deep learning algorithm is applied to identify small targets in sacs and small targets with residual food, as well as fish behavior. Combined with morphological filtering and spatial localization, the density of sacs, density of residual food, and indicators of abnormal fish behavior are output.

[0131] Dynamic decision-making and execution: When the cyst density exceeds the threshold, the inlet flow rate is automatically increased and the drain valve is opened to form a spiral vortex at the bottom of the tank, which concentrates and discharges the cysts. After each cleaning, the density is checked again at a set interval. If it is below the threshold, the process stops; if it is above the threshold, the process is repeated. The cumulative number of cleanings does not exceed 3.

[0132] Feeding amount control: Feed 3 times a day at set times, with each feeding amount being 4% of the initial weight of the feeder. The feeding strategy is controlled by the residual feed data collected by the side wall camera. If the camera identifies a residual feed coverage area >5% 1 hour after feeding, the next feeding amount will be reduced by 15%; if the camera identifies a residual feed coverage area <2% 2% 2 times in a row, the feeding amount will be increased by 10%.

[0133] Anomaly warning and remote management: Automatically issues real-time warnings for excessive cysts, abnormal water temperature data, and abnormal dissolved oxygen data, and pushes alarms to the administrator terminal, supporting remote intervention and manual operation.

[0134] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0135] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. Furthermore, those skilled in the art will recognize that, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A vision- and eddy current-based intelligent aquaculture method for marine fish, applied to a vision- and eddy current-based intelligent aquaculture device for marine fish, characterized in that, The device includes: an edge computing unit, a culture tank body, and a layered monitoring module, an intelligent vortex module, and an intelligent feeder disposed at preset positions on the culture tank body; wherein, the bottom of the culture tank body is a conical surface with an inclination angle of 30°, and the interior of the culture tank body includes a culture area and a vortex area; the edge computing unit is connected to the layered monitoring module, the intelligent vortex module, and the intelligent feeder respectively; the layered monitoring module is used to collect environmental data, fish body status data, and residual feed data, wherein the environmental data includes: water temperature data and dissolved oxygen data, and the fish body status data includes: fish behavior data, cyst distribution data at the bottom of the tank, and fish body surface data; the edge computing unit is used to perform instruction calculations based on the thresholds corresponding to the environmental data, fish body status data, and residual feed data to obtain vortex control instructions and feeding instructions; the intelligent vortex module is used to switch the corresponding working mode according to the vortex control instructions, the working modes including: normal culture mode and cyst removal mode. The intelligent feeder is used to feed the aquaculture area according to the feeding instructions. The aquaculture tank body includes: a top support, an outer shell, a mesh isolation layer, a four-corner support for the outer tank, and a conical surface. The top support is fixedly installed at the upper port of the outer shell of the aquaculture tank to support the top underwater camera. The mesh isolation layer is horizontally positioned above the junction of the inner wall of the outer shell and the conical surface, with its height corresponding to the top of the central drain valve, to separate the aquaculture area from the vortex area. The four-corner support for the outer tank is located around the bottom periphery of the outer shell of the aquaculture tank to support and level the aquaculture tank body. The conical surface is used to form a spiral sinking flow field in the blister removal mode, collecting the blister at the bottom of the tank to the central drain valve. The intelligent vortex module includes: an edge tangential water inlet pipe, a circulating water outlet, and a central drain valve. The edge tangential water inlet pipe is evenly distributed circumferentially along the inner wall of the outer shell of the aquaculture tank, used to form a slow circumferential circulation at a flow rate of 0.2-0.3 m / s in normal aquaculture mode, and increased to 0.5-0.3 m / s in the blister removal mode.8 m / s to enhance the spiral vortex; the circulating water drain outlet is located on the side wall of the outer shell of the breeding tank and is higher than the grid isolation layer, used to maintain the water level and discharge surface residual feed and floating impurities in normal breeding mode; the central drain valve is located at the lowest point of the conical surface, used to fully open during the cyst removal mode to discharge the cysts and sewage accumulated at the bottom of the tank, and to close in normal breeding mode to maintain water stability; the method includes: image and environmental parameter acquisition: the top camera and the inner side wall camera collect fish behavior data, cyst distribution data at the bottom of the tank, residual feed distribution data at the bottom of the tank, and fish surface feature data in real time at a preset frequency, and the water quality sensor simultaneously collects water temperature data and dissolved oxygen data; image preprocessing and feature analysis: noise removal and contrast enhancement processing are performed on the acquired images, the YOLO11 deep learning algorithm is applied to identify small targets of cysts and fish behavior, and the cyst density and abnormal fish behavior indicators are output by combining morphological filtering and spatial positioning; dynamic decision-making and execution: when the cyst density exceeds the threshold, the system automatically... The system automatically increases the inlet water flow rate and opens the drain valve, forming a spiral vortex at the bottom of the tank to concentrate and discharge cysts. After each clearing, the density is checked at set intervals. If it falls below a threshold, the process stops; if it exceeds the threshold, the process repeats, with a cumulative clearing count not exceeding 3 times. Feeding amount control: Feeding is done 3 times daily at set times, with each feeding amount being 4% of the initial weight of the feeder. The feeding strategy is controlled by residual feed data collected by a side-wall camera. If, one hour after feeding, the camera identifies a residual feed coverage area >5%, the next feeding amount is reduced by 15%; if the residual feed coverage area is <2% for two consecutive times, the feeding amount is increased by 10%. Abnormal warning and remote management: Automatic real-time warnings are issued for excessive cysts, abnormal water temperature data, and abnormal dissolved oxygen data, and alarms are pushed to the administrator terminal, supporting remote intervention and manual operation. The central drain valve is a solenoid valve with a diameter range of 10-15cm; the time cycle range of the cyst clearing mode is 30s-60s; the mesh diameter of the grid isolation layer ranges from 2-5mm. The edge computing unit detects the morphological features of the capsules, detects the number of capsules at the bottom of the bucket, and converts the pixel coordinates of the detected capsules and residual bait into physical positions through camera calibration parameters. The unit divides the bucket into grids with the center of the bottom as the origin, calculates the number of capsules in each grid, and outputs the density per unit area. By tracking a single fish through video streams and calculating its total distance traveled and average swimming speed, we can use these as indicators to judge abnormal fish behavior. The edge computing unit automatically corrects the capsule detection threshold based on water temperature sensor data.

2. The intelligent marine fish farming method based on vision and eddies according to claim 1, characterized in that, The stratified monitoring module includes: a top underwater camera, a side underwater camera, and a water quality sensor; the top underwater camera is used to collect fish behavior data and fish body surface data, the side underwater camera is used to collect data on the distribution of pods at the bottom of the bucket and data on residual food, and the water quality sensor is used to collect water temperature data and dissolved oxygen data.

3. The intelligent marine fish farming method based on vision and eddies according to claim 1, characterized in that, The conical surface is made of smooth polyethylene material.

Images