Cooperative regulation and control system for water environment of land-sea relay culture matched temporary culture pond

Abstract

The invention discloses a land-sea relay culture supporting temporary culture pond water environment cooperative regulation and control system, and relates to the technical field of culture, and the system comprises a remote data interaction module which is configured to obtain environment data in a culture cabin and a temporary culture pond, and carries out the mutual transmission and interaction of the data; the eddy current simulation module is used for performing eddy current simulation in the temporary rearing pond based on the eddy current data in the rearing cabin and synchronizing the eddy current data of the rearing cabin; the transition evaluation module monitors behavior data of the cultured organisms in the vortex, judges whether the current cultured organisms adapt to the current vortex environment of the culture cabin or not, and reduces the size of the vortex if the current cultured organisms do not adapt to the current vortex environment; and an iterative optimization module. Progressive transition and stress response automation are achieved, the iterative optimization module constructs a data set and performs iterative learning with the help of a reinforcement learning algorithm, a closed-loop management system is formed, data-driven decision making is achieved, an eco-friendly breeding mode is created, cross-scene adaptation capacity is achieved, and the aquaculture industry is promoted to be transformed to intelligence, intensification and sustainability.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of aquaculture technology, in particular to a water environment coordinated control system for a land-sea relay aquaculture supporting temporary breeding pond. BACKGROUND

[0002] With the continuous development of aquaculture industry, the land-sea relay aquaculture mode gradually emerges, which fully combines the advantages of land-based aquaculture and sea-based aquaculture. In the land-sea relay aquaculture, when the fish in the land-based aquaculture is transported to the aquaculture ship, due to the difference between the environment of the land-based temporary breeding pond and the environment of the aquaculture cabin, the fish directly transported is easy to have stress reaction due to environmental mutation, which affects the survival rate and growth state. Therefore, a period of time is needed for temporary breeding in the temporary breeding pond before transportation, and a vortex environment is simulated during the temporary breeding. At present, the existing temporary breeding pond mostly only has the basic temporary breeding function, cannot obtain and simulate the environmental data of the aquaculture cabin in real time, and cannot realize the effective coordination of the temporary breeding pond and the aquaculture cabin environment. At the same time, the temporary breeding pond lacks remote data interaction capability, cannot realize real-time data sharing and linkage control with the aquaculture ship, and the control parameters of the temporary breeding pond cannot be adjusted in time according to the environmental change of the aquaculture cabin, which cannot provide stable environmental transition support for the fish, and seriously restricts the development of the land-sea relay aquaculture mode. For this reason, we propose a water environment coordinated control system for a land-sea relay aquaculture supporting temporary breeding pond. SUMMARY

[0003] To solve the above technical problems, the water environment coordinated control system for a land-sea relay aquaculture supporting temporary breeding pond is provided, which solves the above problems.

[0004] To achieve the above purpose, the technical scheme adopted by the present application is as follows: the water environment coordinated control system for a land-sea relay aquaculture supporting temporary breeding pond comprises: a remote data interaction module configured to obtain the environmental data in the aquaculture cabin and the temporary breeding pond and to transmit and interact the data with each other; a vortex simulation module based on the vortex data in the aquaculture cabin to simulate the vortex in the temporary breeding pond and to synchronize the vortex data of the aquaculture cabin; a transition evaluation module to monitor the behavior data of the aquaculture organisms in the vortex and to judge whether the current aquaculture organisms are adapted to the current vortex environment of the aquaculture cabin, and to reduce the vortex size if not adapted; an iterative optimization module to record the coordinated control process to form a data set and to iteratively learn based on a reinforcement learning algorithm to form a closed-loop management system.

[0005] Preferably, the remote data interaction module obtains the environmental data in the aquaculture cabin and the temporary breeding pond through sensor network collection and video monitoring collection; and the transmission and interaction mode is based on 5G communication technology for transmission.

[0006] Preferably, the vortex simulation step is as follows: Preprocess the environmental data of the cultivation tank; Mine the data to extract vortex-related features, determine the vortex type and intensity; Based on the determined vortex type and intensity, select a fluid dynamics model, input the integrated data into the model, and construct the vortex simulation effect; Based on the fluid visualization software, convert the vortex simulation effect obtained by simulation calculation into a visual image.

[0007] Preferably, the step of extracting vortex-related features is: The cultivation tank is spatially gridded, and the finite difference method is used to calculate the vortex velocity gradient value; Obtain the morphological change characteristics of the cultivation tank by morphological algorithm, and obtain the water flow velocity change trend value; Combine the obtained vortex velocity gradient value and water flow velocity change trend value into a feature vector, use the random forest learning algorithm, train the model based on historical data, input the feature vector into the model, and predict the vortex type and intensity level.

[0008] Preferably, the step of converting the vortex simulation effect obtained by simulation calculation into a visual image is: Import the data into the visualization software and adapt it to the software format, determine the visualization type and parameters, and select the streamline diagram based on the visualization requirements; Adjust the time step and sampling rate parameters, construct a three-dimensional scene and adjust the viewing angle, and add background elements; Render and optimize the effect, set the light, material, handle the anti-aliasing, shadow and color calibration parameters; Output the visual image in the required format, and convert the vortex simulation data into image display.

[0009] Preferably, the specific operation steps in the transition evaluation module are: Based on the sensor, collect the behavior data of the cultured organisms in the vortex, preprocess and extract the collected data, extract the behavior characteristics of the cultured organisms, including swimming speed characteristics and breathing frequency characteristics; Based on the deep learning model, train and optimize the historical data to learn the mapping relationship between the behavior patterns of the cultured organisms in different vortex environments and the adaptation state; Based on the deep learning model, analyze the real-time behavior data and environmental data, and determine in real time whether the cultured organisms are adapted to the current vortex environment; If the adaptation is not suitable, an early warning signal is issued, and based on the preset strategy, the vortex size is adjusted and the environmental parameters are optimized.

[0010] Preferably, the swimming speed feature extraction is performed by tracking the trajectory of the cultured organism, obtaining motion trajectory data, calculating the displacement distance between adjacent time points, and calculating the swimming speed v based on v = d / t, where d is the displacement distance and t is the time interval. The respiratory frequency feature extraction is based on video to obtain the activity image of the cultured organism, uses the Canny algorithm to locate the gill cover contour, tracks the pixel motion through the optical flow method, quantifies the displacement and speed of the gill cover opening and closing, converts the features into a time series signal, removes noise through low-pass filtering, calculates the respiratory period and frequency through peak detection, and obtains the respiratory frequency feature.

[0011] Preferably, the mapping relationship between the behavior pattern and the adaptation state of the cultured organism in different vortex environments is learned through SHAP value analysis, which quantifies the contribution of each input feature to the model prediction result, and draws a heat map for display. The SHAP value calculation is performed by determining the reference value, arranging each sample feature and calculating the contribution value, calculating the contribution value of each feature under different arrangements through multiple random arrangements, and performing average processing.

[0012] Preferably, the reinforcement learning algorithm in the iterative optimization module is trained and optimized by defining the state space, action space and reward function.

[0013] Preferably, the optimized model in the iterative optimization module is integrated into the coordination and control system for monitoring, intelligent decision-making and automatic control, generating an analysis report, and forming a closed loop of "data acquisition - model learning - strategy execution - effect feedback - optimization iteration".

[0014] Compared with the prior art, the beneficial effects of the present application are as follows: The remote data interaction module of the present application can obtain and bidirectionally transmit the environment data of the culture tank in real time, improve the data accuracy and real-time performance, the vortex simulation module generates a dynamic vortex field based on real-time data, strengthens the simulation of the scene reality, the transition evaluation module evaluates the adaptability by monitoring the biological behavior data, automatically adjusts the vortex size, realizes the gradual transition and automatic stress response, the iterative optimization module constructs a data set and learns through reinforcement learning algorithm iteration, forms a closed-loop management system, realizes data-driven decision-making, creates an ecological friendly breeding mode, and has cross-scene adaptation capability, and promotes the transformation of the aquaculture industry to intelligence, intensification and sustainability. BRIEF DESCRIPTION OF DRAWINGS

[0015] Figure 1 The figure is a schematic diagram of the environment coordination and control system framework of the present application. DETAILED DESCRIPTION

[0016] The following description is intended to disclose the invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious variations will occur to those skilled in the art.

[0017] Reference Figure 1 As shown, the integrated water environment control system for temporary holding ponds in land-sea relay aquaculture includes: The remote data interaction module is configured to acquire environmental data in the breeding chamber and the temporary holding tank, and to transmit and interact with each other's data. The eddy current simulation module performs eddy current simulation in the temporary holding tank based on the eddy current data in the aquaculture chamber, and synchronizes the eddy current data in the aquaculture chamber. The transition assessment module monitors the behavioral data of cultured organisms within the vortex to determine whether the cultured organisms are adapted to the current vortex environment of the culture chamber. If they are not adapted, the size of the vortex is reduced. The iterative optimization module records the coordinated control process, forms a dataset, and iteratively learns based on reinforcement learning algorithms to form a closed-loop management system.

[0018] The remote data interaction module of this application can acquire and transmit aquaculture chamber environmental data in real time and bidirectionally, improving data accuracy and real-time performance. The eddy simulation module generates a dynamic eddy field based on real-time data, enhancing the realism of the simulation scenario. The transition assessment module evaluates adaptability by monitoring biological behavior data, automatically adjusts the eddy size, and realizes gradual transition and automated stress response. The iterative optimization module constructs a dataset and uses reinforcement learning algorithms for iterative learning to form a closed-loop management system, realizing data-driven decision-making. In addition, the system can optimize energy consumption and costs, create an eco-friendly aquaculture model, and has cross-scenario adaptability, promoting the transformation of aquaculture towards intelligence, intensification, and sustainability, and transforming the experience-driven aquaculture model into a data and algorithm-driven model.

[0019] The remote data interaction module acquires environmental data in the breeding chamber and temporary holding tank through sensor network collection and video monitoring; the transmission and interaction method is based on 5G communication technology.

[0020] This application uses a sensor network to sense environmental indicators such as temperature, humidity, and dissolved oxygen in real time, and video monitoring to intuitively capture images of the aquaculture area. The combination of these two data acquisition methods provides multi-dimensional and high-precision data support for aquaculture environment analysis.

[0021] The steps for eddy current simulation are as follows: Preprocess the environmental data of the aquaculture chamber; Data mining is performed to extract features related to eddies and determine the type and intensity of eddies; Based on the determined eddy type and intensity, a fluid dynamics model is selected, and the integrated data is input into the model to construct the eddy simulation effect; Based on fluid visualization software, the eddy current simulation results obtained from simulation calculations are transformed into visual images.

[0022] This application focuses on eddy current simulation and application, forming a closed loop from data to decision-making to provide technical support for aquaculture. Specifically, it preprocesses environmental data from the aquaculture chamber to clarify the type and intensity of eddies, making model construction more targeted. Based on the analysis results, it selects an appropriate fluid dynamics model, inputs multidimensional data, and precisely reproduces the physical processes of eddies, ensuring the accuracy of the simulation. Fluid visualization software is used to transform the simulation results into intuitive images, presenting the spatial distribution and evolution of eddies, helping aquaculture personnel quickly identify risks, predict trends, and drive scientific decision-making.

[0023] The steps for extracting features related to eddies are as follows: The aquaculture chamber was spatially gridded, and the finite difference method was used to calculate the eddy velocity gradient. Morphological algorithms were used to obtain the morphological change characteristics of the aquaculture tank and to obtain the trend value of water flow velocity changes. The obtained eddy velocity gradient values ​​and water flow velocity change trend values ​​are combined into a feature vector. The random forest learning algorithm is used to train the model based on historical data. The feature vector is then input into the model to predict the eddy type and intensity level.

[0024] The specific calculation formula is as follows: Spatial gridding processing discretizes the horizontal two-dimensional space of the aquaculture tank into... There are 1 grid, and the coordinates of each grid node are 1. The corresponding eddy current velocity component is (x direction) and (y direction), temperature is ; Calculate the velocity gradients in the x and y directions. The formula for calculating the velocity gradient in the x direction is: ; The formula for calculating the velocity gradient in the y-direction is: ; in , This refers to the grid spacing; The eddy current velocity gradient value is then expressed as: ; in This represents the eddy current velocity gradient value.

[0025] The steps to convert the eddy current simulation results obtained from the simulation calculations into a visual image are as follows: Import the data into the visualization software and adapt it to the software format; determine the visualization type and parameters; and select a streamline diagram based on the visualization requirements. Adjust the time step and sampling rate parameters, construct a 3D scene and adjust the viewpoint, and add background elements; Perform rendering and effects optimization, set lighting and materials, and handle anti-aliasing, shadows, and calibrate color parameters; Output visualizations in the required format and transform eddy current simulation data into visual displays.

[0026] Transforming eddy current simulations into visual images can effectively assist in aquaculture decision-making and management. Its advantages are significant: it transforms complex numerical data into intuitive graphs such as streamline diagrams, and displays the evolution of eddies through dynamic 3D scenes, greatly reducing the understanding threshold and enabling aquaculture personnel to quickly obtain key information and determine the relationship between eddies and aquaculture areas.

[0027] The specific operational steps within the transition assessment module are as follows: Data is collected on the behavior of cultured organisms within the vortex using sensors. The collected data is preprocessed and features are extracted to extract the behavioral characteristics of the cultured organisms, including swimming speed and respiratory frequency. Based on a deep learning model, historical data is trained and optimized to learn the mapping relationship between the behavioral patterns and adaptive states of farmed organisms under different eddy current environments. Based on a deep learning model, real-time behavioral and environmental data are analyzed to determine whether farmed organisms are adapted to the current eddy environment. If an unsuitable condition is detected, an early warning signal is issued, and environmental parameters are optimized by adjusting the eddy current size based on a preset strategy.

[0028] This application collects behavioral data of cultured organisms in a vortex using sensors. After preprocessing and feature extraction, it focuses on key indicators such as swimming speed and respiratory rate to accurately capture changes in the organism's state. The deep learning model is trained and optimized on historical data to uncover the potential mapping relationship between different vortex environments and the organism's behavioral patterns and adaptive states, making the analysis of the organism's adaptation more scientific and predictive.

[0029] Swimming speed feature extraction involves tracking the trajectory of farmed organisms to obtain motion trajectory data, calculating the displacement distance between adjacent time points, and calculating the swimming speed v based on v=d / t, where d is the displacement distance and t is the time interval. Respiratory frequency feature extraction is based on video images of aquaculture organisms. The Canny algorithm is used to locate the gill cover outline, and the pixel motion is tracked by optical flow method to quantify the displacement and velocity of gill cover opening and closing. After the features are converted into time series signals, low-pass filtering is used to remove noise, and peak detection is used to calculate the respiratory cycle and frequency to obtain respiratory frequency features.

[0030] The mapping relationship between the behavioral patterns and adaptive states of farmed organisms under different eddy environments is studied by analyzing SHAP values, quantifying the contribution of each input feature to the model prediction results, and displaying the results by drawing a heat map. The specific steps for calculating the SHAP value are as follows: The baseline value represents the model's predicted output when there are no specific sample inputs. It is determined by calculating the global mean, random sampling, or setting according to the business scenario. Subsequent SHAP value calculations will revolve around the baseline value, reflecting the contribution of features to the change of the prediction from the baseline value to the actual value. The features of the sample are arranged and combined in a random manner. Since the computational cost of full permutation is too large, Monte Carlo sampling is used to arrange them randomly. Each time the permutation is performed, the features are added to the model input in sequence. The conditional expectation of the model prediction value is calculated after each feature is added. The marginal contribution of each feature is equal to the change in the conditional expectation of the model prediction value after adding the feature. Thus, the contribution value of each feature under a single permutation is obtained. However, the results of a single permutation are random and biased. To ensure the stability and reliability of the results, multiple random permutations are required. The process of permutation and calculating contribution values ​​is repeated to obtain multiple contribution values ​​for each feature. These values ​​are then averaged, and the final average value is the SHAP value of that feature. In this way, the predicted value of the sample can be expressed as the sum of the baseline value and the SHAP values ​​of all features. This clearly shows the direction and degree of influence of each feature on the model's prediction results, providing a scientific and intuitive decision-making basis for practical applications such as the analysis of the behavior of aquatic organisms.

[0031] The reinforcement learning algorithm iteration within the iterative optimization module trains and optimizes the algorithm by clearly defining the state space, action space, and reward function.

[0032] The optimized model within the iterative optimization module is integrated into the coordination and control system for monitoring, intelligent decision-making, and automatic control, generating analysis reports and forming a closed loop of "data acquisition - model learning - strategy execution - effect feedback - optimization iteration".

[0033] After the optimized model of this application is integrated into the coordination and control system, it realizes the integration of monitoring, decision-making and control. The system can perceive changes in the environment and biological state in real time, make intelligent decisions quickly based on reinforcement learning models, automatically adjust the parameters of the aquaculture environment, and generate detailed analysis reports to summarize the control effects and problems. The closed-loop mechanism of "data collection-model learning-strategy execution-effect feedback-optimization iteration" enables the system to continuously accumulate experience from actual operation, dynamically improve the model and strategy based on feedback, continuously improve the scientificity and accuracy of aquaculture management, reduce the cost of manual intervention, enhance the adaptability of the aquaculture system to complex environments, and ultimately achieve the dual optimization of aquaculture benefits and resource utilization efficiency, promoting the development of aquaculture towards intelligence and sustainability.

[0034] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention.

Claims

1. A coordinated water environment control system for temporary holding ponds in land-sea relay aquaculture, characterized in that: include: The remote data interaction module is configured to acquire environmental data in the breeding chamber and the temporary holding tank, and to transmit and interact with each other's data. The eddy current simulation module performs eddy current simulation in the temporary holding tank based on the eddy current data in the aquaculture chamber, and synchronizes the eddy current data in the aquaculture chamber. The transition assessment module monitors the behavioral data of cultured organisms within the vortex to determine whether the cultured organisms are adapted to the current vortex environment of the culture chamber. If they are not adapted, the size of the vortex is reduced. The iterative optimization module records the coordinated control process, forms a dataset, and iteratively learns based on reinforcement learning algorithms to form a closed-loop management system.

2. The land-sea relay aquaculture supporting temporary holding pond water environment coordinated control system according to claim 1, characterized in that, The remote data interaction module acquires environmental data in the breeding chamber and temporary holding tank through sensor network collection and video monitoring; the transmission and interaction method is based on 5G communication technology.

3. The water environment coordinated control system for temporary holding ponds supporting land-sea relay aquaculture according to claim 1, characterized in that, The steps for eddy current simulation are as follows: Preprocess the environmental data of the aquaculture chamber; Data mining is performed to extract features related to eddies and determine the type and intensity of eddies; Based on the determined eddy type and intensity, a fluid dynamics model is selected, and the integrated data is input into the model to construct the eddy simulation effect; Based on fluid visualization software, the eddy current simulation results obtained from simulation calculations are transformed into visual images.

4. The land-sea relay aquaculture supporting temporary holding pond water environment coordinated control system according to claim 3, characterized in that, The steps for extracting features related to eddies are as follows: The aquaculture chamber was spatially gridded, and the finite difference method was used to calculate the eddy velocity gradient. Morphological algorithms were used to obtain the morphological change characteristics of the aquaculture tank and to obtain the trend value of water flow velocity changes. The obtained eddy velocity gradient values ​​and water flow velocity change trend values ​​are combined into a feature vector. The random forest learning algorithm is used to train the model based on historical data. The feature vector is then input into the model to predict the eddy type and intensity level.

5. The land-sea relay aquaculture supporting temporary holding pond water environment coordinated control system according to claim 3, characterized in that, The steps to convert the eddy current simulation results obtained from the simulation calculations into a visual image are as follows: Import the data into the visualization software and adapt it to the software format; determine the visualization type and parameters; and select a streamline diagram based on the visualization requirements. Adjust the time step and sampling rate parameters, construct a 3D scene and adjust the viewpoint, and add background elements; Perform rendering and effects optimization, set lighting and materials, and handle anti-aliasing, shadows, and calibrate color parameters; Output visualizations in the required format and transform eddy current simulation data into visual displays.

6. The land-sea relay aquaculture supporting temporary holding pond water environment coordinated control system according to claim 1, characterized in that, The specific operational steps within the transition assessment module are as follows: Data is collected on the behavior of cultured organisms within the vortex using sensors. The collected data is preprocessed and features are extracted to extract the behavioral characteristics of the cultured organisms, including swimming speed and respiratory frequency. Based on a deep learning model, historical data is trained and optimized to learn the mapping relationship between the behavioral patterns and adaptive states of farmed organisms under different eddy current environments. Based on a deep learning model, real-time behavioral and environmental data are analyzed to determine whether farmed organisms are adapted to the current eddy environment. If an unsuitable condition is detected, an early warning signal is issued, and environmental parameters are optimized by adjusting the eddy current size based on a preset strategy.

7. The land-sea relay aquaculture supporting temporary holding pond water environment coordinated control system according to claim 6, characterized in that, Swimming speed feature extraction involves tracking the trajectory of farmed organisms to obtain motion trajectory data, calculating the displacement distance between adjacent time points, and calculating the swimming speed v based on v=d / t, where d is the displacement distance and t is the time interval. Respiratory frequency feature extraction is based on video images of aquaculture organisms. The Canny algorithm is used to locate the gill cover outline, and the pixel motion is tracked by optical flow method to quantify the displacement and velocity of gill cover opening and closing. After the features are converted into time series signals, low-pass filtering is used to remove noise, and peak detection is used to calculate the respiratory cycle and frequency to obtain respiratory frequency features.

8. The land-sea relay aquaculture supporting temporary holding pond water environment coordinated control system according to claim 6, characterized in that, The mapping relationship between the behavioral patterns and adaptive states of farmed organisms under different eddy environments is studied by analyzing SHAP values, quantifying the contribution of each input feature to the model prediction results, and displaying the results by drawing a heat map. The SHAP value is calculated by determining a baseline value, arranging the features of each sample and calculating their contribution value, and then averaging the contribution value of each feature under different arrangements through multiple random arrangements.

9. The land-sea relay aquaculture supporting temporary holding pond water environment coordinated control system according to claim 1, characterized in that, The reinforcement learning algorithm iteration within the iterative optimization module trains and optimizes the algorithm by clearly defining the state space, action space, and reward function.

10. The land-sea relay aquaculture supporting temporary holding pond water environment coordinated control system according to claim 1, characterized in that, The optimized model within the iterative optimization module is integrated into the coordination and control system for monitoring, intelligent decision-making, and automatic control, generating analysis reports and forming a closed loop of "data acquisition - model learning - strategy execution - effect feedback - optimization iteration".

Images