A multi-state nature-imitating fishway system and a method for operating the same

Abstract

The application discloses a multi-state natural fishway system and a running method thereof, and the system comprises a multi-state flow field shaping module, a fish school tracking module, a fish passing evaluation module and a control module; wherein the multi-state flow field shaping module is used for creating a flow field meeting the upstream migration demands of various fishes; the fish school tracking module is used for shooting and tracking the motion track of the fish school in the upstream migration process; the fish passing evaluation module is used for evaluating the preferred upstream migration track and the rest area of the fishes; and the control module regulates the multi-state flow field shaping module to finally meet the common and special hydraulic demands of the upstream migration of different fishes.

Description

Technical Field

[0001] This invention belongs to the field of water ecological protection technology, specifically relating to a multi-modal, natural-inspired fishway system and its design and operation method. Background Technology

[0002] The ecological dilemmas in rivers caused by hydropower projects are drawing widespread attention. The resulting decline in river connectivity and habitat fragmentation pose significant challenges to the survival and reproduction of freshwater fish worldwide. Against the backdrop of ecological civilization construction, major river basins in my country have successively planned, constructed, and improved a series of fish conservation measures over the past few years.

[0003] Due to the diverse range of fish species and their varying morphological, physiological, and behavioral characteristics, swimming abilities, and upstream behaviors, the upstream performance of different fish species within the same fishway can vary significantly. This can result in insufficient upstream stimulation for larger fish while creating flow velocity barriers for smaller fish, leading to inconsistent upstream performance among individuals of the same species but different sizes. Consequently, the operational effectiveness of current fish passage facilities falls far short of expectations. Compared to traditional technical fishways, simulated natural fishways, by replicating natural water flow conditions, offer broader applicability to different fish species and higher fish passage efficiency. How to effectively utilize the advantages of simulated natural fishways to create conditions that accommodate the upstream movement of various fish species, including different species or different sizes and behavioral characteristics of the same species—in other words, multi-target fish passage—remains a recent hot topic and challenge in ecohydraulics research. Related inventions are expected to refine the hydraulic design standards for simulated natural fishways. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing technologies and provide a multi-modal, natural-like fishway system and its operation method. This system can precisely track the preferred upstream trajectories and resting areas of fish during their ascent, targeting different water flow patterns and different types of target fish populations. By adjusting the multi-modal flow field shaping module, it can create a flow field that satisfies the preferences of various fish species during their ascent, thereby improving the efficiency of fish passage within the fishway and better protecting flow pattern diversity and habitat diversity.

[0005] Technical solution: To solve the above-mentioned technical problems, this invention proposes a multi-mode natural fishway system, which includes a multi-mode flow field shaping module (1), a fish swarm tracking module (2), a fish passage evaluation module (3), and a control module (4).

[0006] The multi-state flow field shaping module (1) consists of multiple sub-flow field shaping modules (11). By laying and combining multiple sub-flow field shaping modules (11), the water flow state is restored and a flow field that meets the upstream needs of various fish species is constructed.

[0007] The fish tracking module (2) consists of multiple composite cameras (21), including a main camera (211) and an auxiliary camera (212). The optical centers of the main camera (211) and the auxiliary camera (212) are perpendicular to the plane of the multi-state flow field shaping module (1) and are used to capture the movement trajectory of the fish during the upstream process.

[0008] The fish evaluation module (3) is used to identify the movement trajectory and resting area of ​​the target fish, including a 3D trajectory generation unit (31), a fish kinematics calculation unit (32), a machine learning training unit (33), and a flow field monitoring unit (34).

[0009] The control module (4) adjusts the multi-state flow field shaping module (1) according to the results of the fish passage assessment module (3) to meet the commonalities and hydraulic requirements of different fish species to swim upstream.

[0010] Furthermore, the sub-flow field shaping module (11) is composed of a column structure (111), a ball bearing device (112), a base (113), and a toothed wall structure (114). The column structure (111) is connected to the base through the ball bearing device (112). The column structure (111) can move on the base through the ball bearing device (112) to meet the deployment requirements at different positions. The base (113) is fixed to the bottom of the riverbed through the toothed wall structure (114).

[0011] Furthermore, the method includes the following steps:

[0012] S1, based on the target fish group's body length, growth status, and lateral line sensing area, each sub-flow field shaping module is divided into multiple grid units. The division of the sub-flow field shaping module is based on a three-dimensional Cartesian coordinate system, and each grid unit after division is denoted as C. (i，j，k) Wherein, the longitudinal, transverse, and vertical lengths d of each grid cell x d y d z Specifically, it is calculated using the following formula:

[0013]

[0014]

[0015]

[0016] In the formula, δ1, δ2, and δ3 are the body length correction coefficients in three directions, and R x R y R z These represent the size of the lateral sensing area in three directions, B. h L represents the average body length of the target fish group. tThese represent the average growth cycle of the target fish population;

[0017] S2, during each upstream movement of the target fish school, the fish school tracking module (2) records the upstream video of the target fish school recorded by the main camera (211) and the auxiliary camera (212) in the composite camera (21), and marks the position of the centroid of the fish in the video frame by frame, and establishes the two-dimensional coordinate data arrays Dateset-1 and Dateset-2 of the fish; based on the two-dimensional coordinate data arrays Dateset-1 and Dateset-2, the 3D trajectory generation unit (31) establishes the three-dimensional coordinate data arrays Dateset-3 of the fish.

[0018] S3, using the fish three-dimensional coordinate data array Dateset-3, the fish kinematics calculation unit (32) calculates the swimming distance ΔS, relative movement distance D, instantaneous swimming velocity V, and instantaneous swimming acceleration a of the fish at each Δt time. The duration T of the fish in each grid cell is also calculated. L According to the duration T L With threshold time T v Comparative judgment;

[0019] If T L >T v If so, the grid cell is determined to be a rest area grid cell. R ;

[0020] If T L <T v If so, the mesh cell is determined to be a moving region mesh cell. S ;

[0021] S4, based on the rest area grid cell determined in S3. R and motion region grid cell S This is used to construct the fish movement trajectory and the range of the fish resting area within the entire multi-state flow field shaping module;

[0022] S5, a database is constructed based on each target fish, the target fish's movement trajectory, and the fish's resting area. The target fish's category is used as input, and the fish's movement trajectory and the fish's resting area are used as output. The machine learning training unit (33) trains the neural network to obtain a neural network model that can identify the fish's movement trajectory and the fish's resting area based on the fish's category.

[0023] S6, input the target fish into the neural network model trained in step S5 to obtain the fish movement trajectory and fish resting area of ​​the target fish. Based on the upward movement trajectory and resting area of ​​the target fish, the control module (4) and the flow field monitoring unit (34) adjust the ball bearing device to move the column to ensure that each flow field shaping module meets the flow velocity requirements of the target fish.

[0024] Furthermore, the requirements include that the velocity of the flow field at the fish's movement trajectory does not exceed the target fish's maximum swimming speed, and the velocity of the flow field in the resting area does not exceed the target fish's resting swimming speed.

[0025] Furthermore, the total number of mesh elements after the sub-flow field shaping module is divided is obtained by the following formula:

[0026]

[0027] In the formula, The total number of grid cells after dividing each subflow field is given, where i, j, and k are the number of grid cell sequences along the longitudinal, transverse, and vertical directions, respectively, and a, b, and c are the number of grid cells that can be divided along the longitudinal, transverse, and vertical directions, respectively. , , These are the three dimensions of the sub-flow field shaping module.

[0028] Furthermore, in step S2, the 3D trajectory calculation unit converts the two-dimensional coordinate data array into a three-dimensional coordinate data array of fish, including the following steps:

[0029] S21, perform camera calibration on the parameters of the main camera (211) and auxiliary camera (212) in the composite camera (21). The relevant parameters include the camera's focal length, principal point offset, distortion coefficient, rotation matrix, and translation vector.

[0030] S22, establish the feature association between the coordinates of the main and auxiliary cameras and the fish centroid coordinates of Dateset-1 and Dateset-2, and use the camera rotation matrix and translation vector parameters to convert the two-dimensional centroid coordinates of the fish in Dateset-1 and Dateset-2 into the corresponding centroid world coordinates;

[0031] S23, using the transformed centroid world coordinates and the coordinates of the main and auxiliary cameras to perform triangulation to obtain the three-dimensional coordinate data array Dateset-3 of the fish relative to the tank space.

[0032] Furthermore, in step S3, the swimming distance ΔS, relative movement distance D, instantaneous swimming velocity V, and instantaneous swimming acceleration a are calculated using the following formulas:

[0033]

[0034]

[0035]

[0036]

[0037] In the formula, Δx, Δy, and Δz are the three-dimensional distances in different grid cells of the sub-flow field shaping module, L is the body length of the target fish, and V T+Δt and V T The swimming speed of fish is divided into time T and time T+ΔT. For time intervals.

[0038] Beneficial effects: Compared with the prior art, the technical solution of the present invention has the following beneficial technical effects:

[0039] (1) By using the multi-state flow field shaping module, the natural water flow state is restored, which has a wider range of fish applicability and higher fish passage efficiency.

[0040] (2) From determining the grid unit to finally determining the fish's preferred trajectory and resting area, the relationship between fish kinematics and fish's preference for swimming upstream was further clarified, and the hydraulic design standards for imitation natural fishways were further refined. Attached Figure Description

[0041] Figure 1 This is a schematic diagram illustrating the operation of a polymorphic, natural-inspired fishway.

[0042] Figure 2 A schematic diagram of a polymorphic, natural-inspired fishway system;

[0043] Figure 3 A schematic diagram illustrating the multi-mode flow field arrangement that mimics a natural fishway;

[0044] Figure 4 A schematic diagram of the sub-flow field shaping module structure. Detailed Implementation

[0045] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments.

[0046] To achieve the above objectives, this invention provides a multi-modal mimicking fishway system and its operation method, which can precisely track and quantify the preferred upstream trajectories and resting areas of fish during their upstream migration, create a flow field that satisfies the preferences of various fish species, promote fish passage efficiency within the fishway, and better protect flow pattern diversity and habitat diversity.

[0047] The present invention proposes a multi-mode natural fishway system comprising a multi-mode flow field shaping module (1), a fish swarm tracking module (2), a fish passage evaluation module (3), and a control module (4);

[0048] The multi-state flow field shaping module (1) consists of multiple sub-flow field shaping modules (11). By laying and combining multiple sub-flow field shaping modules (11), the water flow state is restored and a flow field that meets the upstream needs of various fish species is constructed.

[0049] The fish tracking module (2) consists of multiple composite cameras (21), including a main camera (211) and an auxiliary camera (212). The optical centers of the main camera (211) and the auxiliary camera (212) are perpendicular to the plane of the multi-state flow field shaping module (1) and are used to capture the movement trajectory of the fish during the upstream process.

[0050] The fish evaluation module (3) is used to identify the movement trajectory and resting area of ​​the target fish, including a 3D trajectory generation unit (31), a fish kinematics calculation unit (32), a machine learning training unit (33), and a flow field monitoring unit (34).

[0051] The control module (4) adjusts the multi-state flow field shaping module (1) according to the results of the fish passage assessment module (3) to meet the commonalities and hydraulic requirements of different fish species to swim upstream.

[0052] Furthermore, the sub-flow field shaping module (11) is composed of a column structure (111), a ball bearing device (112), a base (113), and a toothed wall structure (114). The column structure (111) is connected to the base through the ball bearing device (112). The column structure (111) can move on the base through the ball bearing device (112) to meet the deployment requirements at different positions. The base (113) is fixed to the bottom of the riverbed through the toothed wall structure (114).

[0053] Furthermore, the method includes the following steps:

[0054] S1, based on the target fish group's body length, growth status, and lateral line sensing area, each sub-flow field shaping module is divided into multiple grid units. The division of the sub-flow field shaping module is based on a three-dimensional Cartesian coordinate system, and each grid unit after division is denoted as C. (i，j，k) Wherein, the longitudinal, transverse, and vertical lengths d of each grid cell x d y d z Specifically, it is calculated using the following formula:

[0055]

[0056]

[0057]

[0058] In the formula, δ1, δ2, and δ3 are the body length correction coefficients in three directions, and R x R y R z These represent the size of the lateral sensing area in three directions, B. h L represents the average body length of the target fish group. t These represent the average growth cycle of the target fish population;

[0059] S2, during each upstream movement of the target fish school, the fish school tracking module (2) records the upstream video of the target fish school recorded by the main camera (211) and the auxiliary camera (212) in the composite camera (21), and marks the position of the centroid of the fish in the video frame by frame, and establishes the two-dimensional coordinate data arrays Dateset-1 and Dateset-2 of the fish; based on the two-dimensional coordinate data arrays Dateset-1 and Dateset-2, the 3D trajectory generation unit (31) establishes the three-dimensional coordinate data arrays Dateset-3 of the fish.

[0060] S3, using the fish three-dimensional coordinate data array Dateset-3, the fish kinematics calculation unit (32) calculates the swimming distance ΔS, relative movement distance D, instantaneous swimming velocity V, and instantaneous swimming acceleration a of the fish at each Δt time. The duration T of the fish in each grid cell is also calculated. L According to the duration T L With threshold time T v Comparative judgment;

[0061] If T L >T v If so, the grid cell is determined to be a rest area grid cell. R ;

[0062] If T L <T v If so, the mesh cell is determined to be a moving region mesh cell. S ;

[0063] S4, based on the rest area grid cell determined in S3. R and motion region grid cell S This is used to construct the fish movement trajectory and the range of the fish resting area within the entire multi-state flow field shaping module;

[0064] S5, a database is constructed based on each target fish, the target fish's movement trajectory, and the fish's resting area. The target fish's category is used as input, and the fish's movement trajectory and the fish's resting area are used as output. The machine learning training unit (33) trains the neural network to obtain a neural network model that can identify the fish's movement trajectory and the fish's resting area based on the fish's category.

[0065] S6, input the target fish into the neural network model trained in step S5 to obtain the fish movement trajectory and fish resting area of ​​the target fish. Based on the upward movement trajectory and resting area of ​​the target fish, the control module (4) and the flow field monitoring unit (34) adjust the ball bearing device to move the column to ensure that each flow field shaping module meets the flow velocity requirements of the target fish.

[0066] Furthermore, the requirements include that the velocity of the flow field at the fish's movement trajectory does not exceed the target fish's maximum swimming speed, and the velocity of the flow field in the resting area does not exceed the target fish's resting swimming speed.

[0067] Furthermore, the total number of mesh elements after the sub-flow field shaping module is divided is obtained by the following formula:

[0068]

[0069] In the formula, The total number of grid cells after dividing each subflow field is given, where i, j, and k are the number of grid cell sequences along the longitudinal, transverse, and vertical directions, respectively, and a, b, and c are the number of grid cells that can be divided along the longitudinal, transverse, and vertical directions, respectively. , , These are the three dimensions of the sub-flow field shaping module.

[0070] Furthermore, in step S2, the 3D trajectory calculation unit converts the two-dimensional coordinate data array into a three-dimensional coordinate data array of fish, including the following steps:

[0071] S21, perform camera calibration on the parameters of the main camera (211) and auxiliary camera (212) in the composite camera (21). The relevant parameters include the camera's focal length, principal point offset, distortion coefficient, rotation matrix, and translation vector.

[0072] S22, establish the feature association between the coordinates of the main and auxiliary cameras and the fish centroid coordinates of Dateset-1 and Dateset-2, and use the camera rotation matrix and translation vector parameters to convert the two-dimensional centroid coordinates of the fish in Dateset-1 and Dateset-2 into the corresponding centroid world coordinates;

[0073] S23, using the transformed centroid world coordinates and the coordinates of the main and auxiliary cameras to perform triangulation to obtain the three-dimensional coordinate data array Dateset-3 of the fish relative to the tank space.

[0074] Furthermore, in step S3, the swimming distance ΔS, relative movement distance D, instantaneous swimming velocity V, and instantaneous swimming acceleration a are calculated using the following formulas:

[0075]

[0076]

[0077]

[0078]

[0079] In the formula, Δx, Δy, and Δz are the three-dimensional distances in different grid cells of the sub-flow field shaping module, L is the body length of the target fish, and V T+Δt and V T The swimming speed of fish is divided into time T and time T+ΔT. For time intervals.

[0080] Taking a 1:5 scale partial model test of a simulated natural fish passage at a hydropower station in the lower reaches of the Yangtze River as an example, a series of studies were conducted on the multi-modal simulated natural fish passage system of this invention. Based on field surveys, the cross-section of the pool chamber in the individual fish passage model was set to 2.5m, the height to 0.6m, the experimental water depth to 0.30m, the total length of the fish passage to 11m, and the core experimental section to 0.65m. The fish passage test subjects were *Schizothorax chevron*, with a body length ranging from 10cm to 2cm. Based on the total length of the fish passage and relevant parameters of the experimental fish (body length, growth state, lateral line sensing range), the two-dimensional range of each sub-flow field shaping module was determined to be 0.25m * 0.25m. Combined with the experimental water depth, the sub-flow field shaping modules were divided into x, y, and z directions. Specific parameters and division results are shown in [the table below]. Figure 1 .

[0081] Before each upstream test of the target fish species, the test fish were temporarily held in a rectangular pool of 1m×1m×1m for 2 weeks. The holding water was circulated deep well water, and the water temperature was maintained at about 17℃. The holding pool was aerated to maintain the dissolved oxygen concentration above 7mg / L. The lighting was natural indoor light, and feeding was stopped 2 days before the test.

[0082] A fish tracking module was installed above the core test section model. The main and auxiliary cameras consisted of two Lt-425 C model cameras. A coordinate system XOY was established based on the resolution of the captured video, with the origin O at the top left corner of the video, the positive X-axis pointing horizontally to the right, and the positive Y-axis pointing vertically downwards. Both the main and auxiliary cameras had a resolution of 1920×1080, therefore the maximum resolution coordinate point (Xmax, Ymax) corresponded to (1920, 1080). The target object's resolution coordinates (X, Y) were rotated, scaled, and corrected to generate position coordinates (x, y) in CAD. The fish tracking module processed the captured video frame by frame and marked the centroid position of the fish at each moment and in each frame according to the aforementioned coordinate system. Based on the two-dimensional fish coordinate data arrays Dateset-1 and Dateset-2, an appropriate translation was determined, and the data was further transformed to the actual three-dimensional coordinate system of the water tank, establishing the three-dimensional fish coordinate data array Dateset-3. The fish kinematics calculation unit calculates the swimming distance ΔS, relative movement distance D, instantaneous swimming velocity V, and instantaneous swimming acceleration of the fish at each time step Δt. Throughout the upstream process, the fish is either swimming or resting. Based on previous experimental results and relevant data from this model's experiment, the target fish is considered to be in a resting state if its time in a single grid cell exceeds 5 seconds, and in a swimming state if its time in a single grid cell is less than 5 seconds. Based on this threshold, T... v =5S, evaluate each grid cell in the 3D array of Dateset-3. All grid cells determined to be rest areas are... R This is a resting area for fish during their upstream migration; all grid cells identified as movement areas are considered as such. S The movement trajectory of the fish during its upstream migration was determined. Multiple sets of upstream migration experiments were repeated. For each upstream migration, the fish's movement trajectory and resting area were constructed. A machine learning training unit was used to build a neural network database. The ResNet network underwent 100,000 iterations. After the cross-entropy value stabilized within the iteration count range, the preferred movement trajectory and resting area of ​​the target fish were determined. The control module, in conjunction with the flow field monitoring unit, adjusted the multi-state flow field shaping module to ensure that each flow field shaping module met the flow velocity requirements of the target fish. During the movement trajectory, the flow velocity was lower than the target fish's continuous swimming speed. At the boundary between the resting area and the movement trajectory, the flow velocity was lower than the target fish's burst swimming speed. Within the resting area, the flow velocity was sufficient to allow the fish to rest undisturbed.

Claims

1. A method for operating a polymorphic simulated natural fishway based on a polymorphic simulated natural fishway system, characterized in that, The method includes the following steps: S1, based on the target fish group's body length, growth status, and lateral line sensing area, each sub-flow field shaping module is divided into multiple grid units. The division of the sub-flow field shaping module is based on a three-dimensional Cartesian coordinate system, and each grid unit after division is denoted as C. (i，j，k) Wherein, the longitudinal, transverse, and vertical lengths d of each grid cell x d y d z Specifically, it is calculated using the following formula: ； ； ； In the formula, δ1, δ2, and δ3 are the body length correction coefficients in three directions, and R x R y R z These represent the size of the lateral sensing area in three directions, B. h L represents the average body length of the target fish group. t These represent the average growth cycle of the target fish population; S2, during each upstream movement of the target fish school, the fish school tracking module (2) records the upstream video of the target fish school recorded by the main camera (211) and the auxiliary camera (212) in the composite camera (21), and marks the position of the centroid of the fish in the video frame by frame, and establishes the two-dimensional coordinate data arrays Dateset-1 and Dateset-2 of the fish; based on the two-dimensional coordinate data arrays Dateset-1 and Dateset-2, the 3D trajectory generation unit (31) establishes the three-dimensional coordinate data arrays Dateset-3 of the fish. S3, using the fish three-dimensional coordinate data array Dateset-3, the fish kinematics calculation unit (32) calculates the swimming distance ΔS, relative movement distance D, instantaneous swimming velocity V, and instantaneous swimming acceleration a of the fish at each Δt time. The duration T of the fish in each grid cell is also calculated. L According to the duration T L With threshold time T v Comparative judgment; If T L >T v If so, the grid cell is determined to be a rest area grid cell. R ; If T L <T v If so, the mesh cell is determined to be a moving region mesh cell. S ; S4, based on the rest area grid cell determined in S3. R and motion region grid cell S This is used to construct the fish movement trajectory and the range of the fish resting area within the entire multi-state flow field shaping module; S5, a database is constructed based on each target fish, the target fish's movement trajectory, and the fish's resting area. The target fish's category is used as input, and the fish's movement trajectory and the fish's resting area are used as output. The machine learning training unit (33) trains the neural network to obtain a neural network model that can identify the fish's movement trajectory and the fish's resting area based on the fish's category. S6, input the target fish into the neural network model trained in step S5 to obtain the fish movement trajectory and fish resting area of ​​the target fish. Based on the upward movement trajectory and resting area of ​​the target fish, the control module (4) and the flow field monitoring unit (34) adjust the ball bearing device to move the column to ensure that each flow field shaping module meets the flow velocity requirements of the target fish. The multi-mode natural fishway system includes a multi-mode flow field shaping module (1), a fish swarm tracking module (2), a fish passage evaluation module (3), and a control module (4). The multi-state flow field shaping module (1) consists of multiple sub-flow field shaping modules (11). By laying and combining multiple sub-flow field shaping modules (11), the water flow state is restored and a flow field that meets the upstream needs of various fish species is constructed. The fish tracking module (2) consists of multiple composite cameras (21), including a main camera (211) and an auxiliary camera (212). The optical centers of the main camera (211) and the auxiliary camera (212) are perpendicular to the plane of the multi-state flow field shaping module (1) and are used to capture the movement trajectory of the fish during the upstream process. The fish evaluation module (3) is used to identify the movement trajectory and resting area of ​​the target fish, including a 3D trajectory generation unit (31), a fish kinematics calculation unit (32), a machine learning training unit (33), and a flow field monitoring unit (34). The control module (4) adjusts the multi-state flow field shaping module (1) according to the results of the fish passage assessment module (3) to meet the commonalities and hydraulic requirements of different fish species to migrate upstream.

2. The method according to claim 1, characterized in that, The sub-flow field shaping module (11) consists of a column structure (111), a ball bearing device (112), a base (113), and a toothed wall structure (114). The column structure (111) is connected to the base through the ball bearing device (112). The column structure (111) can move on the base through the ball bearing device (112) to meet the deployment requirements at different locations. The base (113) is fixed to the bottom of the riverbed through the toothed wall structure (114).

3. The method according to claim 1, characterized in that, The requirements include: the velocity of the flow field at the fish's movement trajectory does not exceed the target fish's maximum swimming speed, and the velocity of the flow field in the resting area does not exceed the target fish's resting swimming speed.

4. The method according to claim 1, characterized in that, The total number of mesh elements after the sub-flow field shaping module is divided is obtained by the following formula: ； In the formula, The total number of grid cells after dividing each subflow field is given, where i, j, and k are the number of grid cell sequences along the longitudinal, transverse, and vertical directions, respectively, and a, b, and c are the number of grid cells that can be divided along the longitudinal, transverse, and vertical directions, respectively. , , These are the three dimensions of the sub-flow field shaping module.

5. The method according to claim 1 or 3, characterized in that, In step S2, the 3D trajectory calculation unit converts the two-dimensional coordinate data array into a three-dimensional coordinate data array for fish, including the following steps: S21, perform camera calibration on the parameters of the main camera (211) and auxiliary camera (212) in the composite camera (21). The relevant parameters include the camera's focal length, principal point offset, distortion coefficient, rotation matrix, and translation vector. S22, establish the feature association between the coordinates of the main and auxiliary cameras and the fish centroid coordinates of Dateset-1 and Dateset-2, and use the camera rotation matrix and translation vector parameters to convert the two-dimensional centroid coordinates of the fish in Dateset-1 and Dateset-2 into the corresponding centroid world coordinates; S23, using the transformed centroid world coordinates and the coordinates of the main and auxiliary cameras to perform triangulation to obtain the three-dimensional coordinate data array Dateset-3 of the fish relative to the tank space.

6. The method according to claim 1 or 3, characterized in that, In step S3, the swimming distance ΔS, relative movement distance D, instantaneous swimming velocity V, and instantaneous swimming acceleration a are calculated using the following formulas: ； ； ； ； In the formula, Δx, Δy, and Δz are the three-dimensional distances in different grid cells of the sub-flow field shaping module, L is the body length of the target fish, and V T+Δt and V T Fish swimming speeds are divided into time T and time T+ΔT. For time intervals.

Images