Fish precision feeding method and system based on digital twin

Abstract

The present invention discloses a method and system for accurate feeding of fish based on digital twins, which belongs to the technical field of fish farming. The method comprises the following steps: S1: collecting and processing the real-time data of fish growth based on the Internet of Things; S2: constructing a digital twin model of fish growth based on machine learning based on machine learning technology according to the historical data of fish growth, and analyzing the real-time data of fish growth based on the Internet of Things, predicting the growth trend of fish, and determining the prediction result of the fish growth trend; S3: automatically adjusting the feeding amount according to the prediction result of the fish growth trend, and then accurately feeding the fish to ensure the nutritional balance required for fish growth. The present invention solves the problem that the existing accurate feeding cannot be performed according to the growth of fish, which increases the cost of fish farming and leads to poor fish farming effect. The present invention can accurately feed the fish according to the growth of fish, reduce the cost of fish farming, and improve the fish farming effect.

Description

Technical Field

[0001] The present invention relates to the technical field of fish farming, and specifically to a fish precision feeding method and system based on digital twins. Background Art

[0002] In aquaculture, it has always been a problem in the industry that the amount of feed to be fed is difficult to judge. The water temperature, dissolved oxygen and other environmental conditions in the breeding environment fluctuate greatly, which directly affects the feeding of fish. In addition, most fishermen feed based on experience. If the feeding amount is small, the fish grow slowly. If the feeding amount is too much, it will not only cause waste of feed, but also lead to deterioration of water quality, which has a negative impact on the survival status of fish.

[0003] Chinese patent publication number CN119054639A discloses a cage system for fish farms, comprising a movable or transportable cage and a fixed docking frame, the fixed docking frame is arranged below a predetermined number of meters below the sea surface, the docking frame is configured to receive and accommodate the cage therein, and the cage is configured for breeding fish therein; however, the patent has the following defects: Existing technology cannot accurately feed fish according to their growth conditions, which increases the cost of fish farming and leads to poor fish farming results. Summary of the invention

[0004] The purpose of the present invention is to provide a fish precision feeding method and system based on digital twins, which can accurately feed fish according to the growth conditions of fish, reduce fish farming costs, improve fish farming effects, and solve the problems raised in the above-mentioned background technology.

[0005] To achieve the above object, the present invention provides the following technical solutions: The fish precision feeding method based on digital twins includes the following steps: S1. Data collection and processing: collecting and processing the real-time data of fish growth based on the Internet of Things, and processing the real-time data of fish growth based on the Internet of Things, so that the real-time data of fish growth based on the Internet of Things is safely stored in the database; S2. Model prediction: Based on the historical data of fish growth and machine learning technology, a digital twin model of fish growth based on machine learning is constructed, and the real-time data of fish growth based on the Internet of Things is analyzed, and the fish growth trend is predicted to determine the prediction results of the fish growth trend; S3. Feeding control: Automatically adjust the feeding amount according to the predicted results of fish growth trend, and then feed the fish accurately to ensure the nutritional balance required for fish growth.

[0006] Preferably, in S1, collecting real-time fish growth data based on the Internet of Things includes: Install intelligent data collection equipment at the bottom and around the fish pond; Based on intelligent data collection equipment, the water temperature, dissolved oxygen, pH value, ammonia nitrogen concentration, nitrite concentration and suspended solids concentration of the fish pond are monitored and collected in real time to obtain the environmental parameters of the fish pond; Based on intelligent data collection equipment, fish weight, body length, growth rate, feed conversion rate and feeding activity are monitored and collected in real time to obtain fish physiological indicators; Among them, the real-time data of fish growth based on the Internet of Things is determined based on the fish pond environmental parameters and fish physiological indicators.

[0007] Preferably, in S1, the concentration of suspended matter is monitored and collected in real time, including: Extract the regional structure map of the fish pond; Dividing the regional structure of the fish pond into grids according to the regional structural characteristics of the fish pond; wherein the regional structural characteristics include the regional area of ​​the fish pond, the maximum water depth of the fish pond, and the water depth change rate; Obtaining the side length of the grid using the regional structural characteristics of the fish pond; The grid side length is obtained by the following formula: ; Where L represents the grid side length; H max Indicates the maximum water depth of the fish pond; L max represents the length of the pond bottom corresponding to the maximum water depth of the fish pond; A represents the area of ​​the fish pond; P represents the rate of change of water depth; A h represents the area of ​​the fish pond corresponding to the maximum water depth of the fish pond; n represents the number of water depth dimensions corresponding to the fish pond except the maximum water depth; L i represents the pond bottom extension length corresponding to the i-th water depth dimension; Divide the regional structure diagram of the fish pond into grids according to the side length of the grids to obtain multiple grid structures; The suspended matter concentration is collected according to the grid structure.

[0008] Preferably, collecting the concentration of suspended matter according to the grid structure includes: Collecting suspended matter concentration at the edge grid corresponding to the regional structure diagram of the fish pond to obtain edge suspended matter concentration values; Extracting the central grid corresponding to the regional structure diagram to collect suspended matter concentration and obtain the central suspended matter concentration value; The suspended matter concentration span coefficient is obtained by combining the suspended matter concentration value corresponding to each edge grid with the central suspended matter concentration value corresponding to the central grid; The suspended matter concentration span coefficient is obtained by the following formula: ; Where K represents the suspended matter concentration span coefficient; m represents the number of edge grids; C i represents the edge suspended matter concentration value corresponding to the i-th edge grid; Z p represents the average value of the central suspended matter concentration of the central grid; Z represents the central suspended matter concentration value of the central grid; C b represents the standard deviation of the edge suspended matter concentration of m edge grids; Z b Indicates the standard deviation of the central suspended matter concentration corresponding to multiple central grids; comparing the suspended matter concentration span coefficient with a preset coefficient threshold; When the suspended matter concentration span coefficient is lower than the preset coefficient threshold, the average value of the edge suspended matter concentration value and the center suspended matter concentration value is taken as the suspended matter concentration of the current fish pond; When the suspended matter concentration span coefficient is not lower than a preset coefficient threshold, the suspended matter concentration value of the target grid area between the edge grid and the center grid is selected to determine the suspended matter concentration of the current fish pond.

[0009] Preferably, when the suspended matter concentration span coefficient is not lower than a preset coefficient threshold, the suspended matter concentration value of the target grid area between the edge grid and the center grid is selected to determine the suspended matter concentration of the current fish pond, including: When the suspended matter concentration span coefficient is not lower than a preset coefficient threshold, the suspended matter concentration value corresponding to each edge grid and the suspended matter concentration value corresponding to the center grid are extracted; Obtaining a distance adjustment coefficient according to a concentration difference between a suspended matter concentration value corresponding to each edge grid and a suspended matter concentration value corresponding to the central grid; The distance adjustment coefficient is obtained by the following formula: ; Where S represents the distance adjustment coefficient; m represents the number of edge grids; C i represents the edge suspended matter concentration value corresponding to the i-th edge grid; Z p Indicates the average value of the central suspended matter concentration of the central grid; C b represents the standard deviation of the edge suspended matter concentration values ​​of m edge grids; Z represents the central suspended matter concentration value of the central grid; K represents the suspended matter concentration span coefficient; K y Indicates the preset coefficient threshold; P z Indicates the change rate of the suspended matter concentration corresponding to multiple central grids; The edge grid corresponding to the maximum value of the suspended matter concentration corresponding to the edge grid is used as the calibration grid; Extracting the straight-line distance between the calibration grid and the central grid; Acquire a calibration distance according to the straight-line distance between the center grids and a distance adjustment coefficient; The calibration distance is obtained by the following formula: ; Where D represents the calibration distance; D max represents the straight-line distance between the calibration grid and the central grid; S represents the distance adjustment coefficient; S c represents the preset reference value of the distance adjustment coefficient; x represents the preset adjustment coefficient, which is used to control the slope of the sigmoid function, thereby affecting the sensitivity of the calibration distance to the distance adjustment coefficient; Calibrate the distance along each edge grid toward the center grid, select a grid whose straight-line distance to the edge grid is the calibrated distance as a target grid, and group the target grids into a target grid area; Collecting the suspended matter concentration value in each grid contained in the target grid area; The overall average value of the suspended matter concentration value in each grid contained in the target grid area and the suspended matter concentration value corresponding to the edge grid is taken as the suspended matter concentration of the current fish pond.

[0010] Preferably, in S1, processing the real-time data of fish growth based on the Internet of Things includes: Clean the real-time fish growth data based on IoT based on data cleaning tools; Among them, the real-time data of fish growth based on the Internet of Things is checked, duplicate values, missing values ​​and abnormal values ​​in the real-time data of fish growth based on the Internet of Things are identified, and the duplicate values, missing values ​​and abnormal values ​​in the real-time data of fish growth based on the Internet of Things are processed; For repeated values ​​in the real-time data of fish growth based on the Internet of Things, the repeated values ​​are deleted; for missing values ​​in the real-time data of fish growth based on the Internet of Things, the missing values ​​are deleted or filled to make the missing values ​​complete; for outliers in the real-time data of fish growth based on the Internet of Things, the outliers are deleted or replaced to make them normal.

[0011] Preferably, in S1, processing the real-time data of fish growth based on the Internet of Things also includes: The real-time fish growth data based on the Internet of Things is converted, the dimensional differences between the real-time fish growth data based on the Internet of Things are reduced, the real-time fish growth data based on the Internet of Things are unified, and standardized real-time fish growth data based on the Internet of Things are determined; The standardized real-time fish growth data based on the Internet of Things is integrated, and the standardized real-time fish growth data based on the Internet of Things from different sources are integrated into a unified view. The integrated standardized real-time fish growth data based on the Internet of Things is verified to determine whether the integrated standardized real-time fish growth data based on the Internet of Things is missing. After the verification is qualified, the integrated standardized real-time fish growth data based on the Internet of Things is securely stored.

[0012] Preferably, in S2, constructing a fish growth digital twin model based on machine learning includes: According to the precise feeding requirements of fish based on digital twins, fish growth history data is collected and divided to determine the training set and test set; Based on machine learning technology, a training set is used to train the machine learning model, so that the machine learning model can autonomously learn the fish growth prediction process and predict the fish growth trend, and determine the fish growth digital twin model based on machine learning; Based on the test set, a performance test is conducted on the fish growth digital twin model based on machine learning to determine whether the fish growth digital twin model based on machine learning can achieve the effect of predicting the growth trend of fish. When the fish growth digital twin model based on machine learning cannot achieve the effect of predicting the growth trend of fish, the parameters and structure of the fish growth digital twin model based on machine learning are adjusted. After continuous iterative optimization, the optimal fish growth digital twin model is determined.

[0013] Preferably, in S2, predicting the growth trend of fish includes: Obtain the optimal fish growth digital twin model and deploy it in the actual fish growth trend prediction environment; The real-time fish growth data based on the Internet of Things is input into the optimal fish growth digital twin model. Based on the optimal fish growth digital twin model, the real-time fish growth data based on the Internet of Things is analyzed, and the fish growth trend is predicted to determine the fish growth trend prediction result.

[0014] Preferably, in S3, accurately feeding the fish comprises: Obtain the predicted results of fish growth trends, automatically adjust the feeding amount according to the predicted results, and then feed the fish accurately. At the same time, monitor and track the growth of fish in real time, and adjust the feeding plan in time to ensure the nutritional balance required for fish growth.

[0015] According to another aspect of the present invention, a fish precision feeding system based on digital twin is provided, which is used to implement the above-mentioned fish precision feeding method based on digital twin, including: Data acquisition module, used to collect fish pond environmental parameters and fish physiological indicators, and determine the real-time data of fish growth based on the Internet of Things; A data processing module is used to clean, convert and integrate the real-time data of fish growth based on the Internet of Things, and to store and back up the real-time data of fish growth based on the Internet of Things; The model prediction module is used to analyze the real-time data of fish growth based on the Internet of Things using the fish growth digital twin model, predict the fish growth trend, and determine the prediction result of the fish growth trend; The feeding control module is used to automatically adjust the feeding amount according to the predicted results of fish growth trend to ensure the nutritional balance required for fish growth.

[0016] Compared with the prior art, the present invention has the following beneficial effects: The present invention determines the real-time fish growth data based on the Internet of Things by collecting fish pond environmental parameters and fish physiological indicators, cleans, converts and integrates the real-time fish growth data based on the Internet of Things, and stores the real-time fish growth data based on the Internet of Things. According to the fish growth historical data, based on machine learning technology, a fish growth digital twin model based on machine learning is constructed, and the real-time fish growth data based on the Internet of Things is analyzed based on the optimal fish growth digital twin model, and the fish growth trend is predicted, and the fish growth trend prediction result is determined. The feeding amount is automatically adjusted according to the fish growth trend prediction result, and then the fish is accurately fed. At the same time, the growth of the fish is monitored and tracked in real time, and the feeding plan is adjusted in time to ensure the nutritional balance required for fish growth. Accurate feeding can be performed according to the growth of the fish, thereby reducing the cost of fish farming and improving the effect of fish farming. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 This is a flow chart of the digital twin-based fish precision feeding method of the present invention; Figure 2 This is a module diagram of the digital twin-based fish precision feeding system of the present invention. DETAILED DESCRIPTION

[0018] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0019] In order to solve the existing problem that the fish cannot be fed accurately according to the growth of fish, which increases the cost of fish farming and leads to poor fish farming results, please refer to Figure 1-Figure 2 , this embodiment provides the following technical solutions: The fish precision feeding method based on digital twins includes the following steps: S1. Data collection and processing: collecting and processing the real-time data of fish growth based on the Internet of Things, and processing the real-time data of fish growth based on the Internet of Things, so that the real-time data of fish growth based on the Internet of Things is safely stored in the database; In this embodiment, real-time fish growth data based on the Internet of Things is collected, including: Install intelligent data collection equipment at the bottom and around the fish pond; Based on intelligent data collection equipment, the water temperature, dissolved oxygen, pH value, ammonia nitrogen concentration, nitrite concentration and suspended solids concentration of the fish pond are monitored and collected in real time to obtain the environmental parameters of the fish pond; It should be noted that by using sensors and cameras to monitor and collect the water temperature, dissolved oxygen, pH value, ammonia nitrogen concentration, nitrite concentration and suspended solids concentration of the fish pond in real time, it is possible to achieve intelligent regulation of the fish growth environment; Among them, water temperature has an important impact on the growth, reproduction and metabolism of fish, and different fish have different adaptability to water temperature; dissolved oxygen is necessary for fish respiration, and its concentration has a direct impact on the growth and health of fish. Too low dissolved oxygen concentration will cause fish hypoxia and even death; pH value affects the physiological function and growth of fish, and different fish have different adaptability to pH value; ammonia nitrogen is a waste product produced by fish metabolism, and its excessive ammonia nitrogen concentration will be toxic to fish; nitrite is an intermediate product of ammonia nitrogen, and its excessive nitrite concentration will also be toxic to fish; the concentration of suspended matter affects the transparency and light of the water body, which in turn affects the growth and health of fish; Based on intelligent data collection equipment, fish weight, body length, growth rate, feed conversion rate and feeding activity are monitored and collected in real time to obtain fish physiological indicators; Among them, body weight is one of the important indicators for measuring fish growth; body length is also one of the important indicators for measuring fish growth; growth rate reflects the growth and health status of fish; feed conversion rate is an important indicator for measuring the efficiency of fish in utilizing feed; feeding activity is used to reflect the feeding and activity status of fish; Among them, the real-time data of fish growth based on the Internet of Things is determined based on the fish pond environmental parameters and fish physiological indicators.

[0020] Specifically, in S1, the concentration of suspended matter is monitored and collected in real time, including: Extract the regional structure map of the fish pond; Dividing the regional structure of the fish pond into grids according to the regional structural characteristics of the fish pond; wherein the regional structural characteristics include the regional area of ​​the fish pond, the maximum water depth of the fish pond, and the water depth change rate; Obtaining the side length of the grid using the regional structural characteristics of the fish pond; The grid side length is obtained by the following formula: ; Where L represents the grid side length; H max Indicates the maximum water depth of the fish pond; L max represents the length of the pond bottom corresponding to the maximum water depth of the fish pond; A represents the area of ​​the fish pond; P represents the rate of change of water depth; A h represents the area of ​​the fish pond corresponding to the maximum water depth of the fish pond; n represents the number of water depth dimensions corresponding to the fish pond except the maximum water depth; L i represents the pond bottom extension length corresponding to the i-th water depth dimension; Divide the regional structure diagram of the fish pond into grids according to the side length of the grids to obtain multiple grid structures; The suspended matter concentration is collected according to the grid structure.

[0021] The technical effect of the above technical solution is: by extracting the regional structure diagram of the fish pond, accurate basic data is first provided for the subsequent grid division. This ensures that various features of the fish pond (such as shape, boundary, etc.) are accurately captured, laying a solid foundation for further analysis and processing. The grid division is based on the regional structural characteristics of the fish pond, which include the regional area, maximum water depth and water depth change rate of the fish pond. This feature-based division method can more accurately reflect the actual topography and hydrological conditions of the fish pond, thereby improving the accuracy and reliability of subsequent analysis. The grid side length size is calculated by a given formula, which comprehensively considers multiple factors such as the maximum water depth of the fish pond, the extension length of the pond bottom, the regional area, the water depth change rate, and the extension length of the pond bottom corresponding to different water depth sizes. This dynamic calculation method makes the grid division more flexible and adaptable, and can be adjusted according to the specific conditions of different fish ponds, thereby improving the precision and practicality of the division. After the regional structure diagram of the fish pond is gridded according to the calculated grid side length size, multiple grid structures can be obtained. These grid structures provide clear sampling points and sampling areas for the collection of suspended matter concentration, thereby improving the collection efficiency and accuracy. At the same time, since the grid division is based on the actual characteristics of the fish pond, the collected suspended matter concentration data can better reflect the actual situation of the fish pond. This technical solution is not only suitable for the collection of suspended matter concentration in fish ponds, but can also be extended to other similar water environments or scenarios that require grid division and data analysis based on regional structural characteristics. Its feature-based grid division method and dynamic grid side length calculation method are highly practical and scalable, providing new ideas and methods for research and application in related fields.

[0022] In summary, this technical solution achieves accurate understanding and expression of the regional structure of fish ponds through precise regional structure extraction, feature-based grid division, dynamic grid edge length calculation, and efficient suspended matter concentration collection, improves the accuracy and efficiency of suspended matter concentration collection, and provides strong support for research and application in related fields.

[0023] Specifically, collecting the suspended matter concentration according to the grid structure includes: Collecting suspended matter concentration at the edge grid corresponding to the regional structure diagram of the fish pond to obtain edge suspended matter concentration values; Extracting the central grid corresponding to the regional structure diagram to collect suspended matter concentration and obtain the central suspended matter concentration value; The suspended matter concentration span coefficient is obtained by combining the suspended matter concentration value corresponding to each edge grid with the central suspended matter concentration value corresponding to the central grid; The suspended matter concentration span coefficient is obtained by the following formula: ; Where K represents the suspended matter concentration span coefficient; m represents the number of edge grids; C i represents the edge suspended matter concentration value corresponding to the i-th edge grid; Z p represents the average value of the central suspended matter concentration of the central grid; Z represents the central suspended matter concentration value of the central grid; C b represents the standard deviation of the edge suspended matter concentration of m edge grids; Z b Indicates the standard deviation of the central suspended matter concentration corresponding to multiple central grids; comparing the suspended matter concentration span coefficient with a preset coefficient threshold; When the suspended matter concentration span coefficient is lower than the preset coefficient threshold, the average value of the edge suspended matter concentration value and the center suspended matter concentration value is taken as the suspended matter concentration of the current fish pond; When the suspended matter concentration span coefficient is not lower than a preset coefficient threshold, the suspended matter concentration value of the target grid area between the edge grid and the center grid is selected to determine the suspended matter concentration of the current fish pond.

[0024] The technical effect of the above technical solution is: by collecting suspended matter concentration in the edge grid and the center grid corresponding to the regional structure diagram of the fish pond, more refined suspended matter concentration data can be obtained. This collection method not only takes into account the overall suspended matter concentration of the fish pond, but also pays special attention to the concentration difference between the edge and the center area, thereby improving the accuracy and representativeness of the data. The suspended matter concentration span coefficient is calculated by combining the suspended matter concentration value corresponding to each edge grid with the central suspended matter concentration value corresponding to the center grid. This coefficient can quantify the degree of difference between the suspended matter concentrations in the edge and center areas, and provides an important reference for subsequent concentration determination. Different concentration determination strategies are adopted according to the comparison results of the suspended matter concentration span coefficient and the preset coefficient threshold. When the span coefficient is lower than the threshold, it means that the concentration difference between the edge and the center area is small. At this time, the average value of the suspended matter concentration of the edge and the center can be directly used as the suspended matter concentration of the fish pond, which simplifies the calculation process. When the span coefficient is not lower than the threshold, it means that the concentration difference is large. At this time, it is necessary to consider the suspended matter concentration of the target grid area more carefully to ensure the accuracy of the concentration. Through the above technical solution, the suspended matter concentration can be flexibly determined according to the actual conditions of different fish ponds. Whether the concentration difference is small or large, a relatively accurate concentration value can be obtained. At the same time, the solution is also adaptable to meet the needs of suspended matter concentration collection and determination in different fish ponds and different environmental conditions. Through refined collection and flexible concentration determination strategies, this technical solution can optimize resource utilization and cost-effectiveness while ensuring accuracy. It avoids unnecessary repeated collection and calculation, and improves work efficiency and economic benefits.

[0025] In summary, the above technical solution achieves accurate, efficient and reliable determination of suspended matter concentration in fish ponds by refining suspended matter concentration collection, introducing suspended matter concentration span coefficient, flexible concentration determination strategy, and improving accuracy and adaptability, providing strong technical support for fish pond management, water quality monitoring and other fields.

[0026] Specifically, when the suspended matter concentration span coefficient is not lower than a preset coefficient threshold, the suspended matter concentration value of the target grid area between the edge grid and the center grid is selected to determine the suspended matter concentration of the current fish pond, including: When the suspended matter concentration span coefficient is not lower than a preset coefficient threshold, the suspended matter concentration value corresponding to each edge grid and the suspended matter concentration value corresponding to the center grid are extracted; Obtaining a distance adjustment coefficient according to a concentration difference between a suspended matter concentration value corresponding to each edge grid and a suspended matter concentration value corresponding to the central grid; The distance adjustment coefficient is obtained by the following formula: ; Where S represents the distance adjustment coefficient; m represents the number of edge grids; C i represents the edge suspended matter concentration value corresponding to the i-th edge grid; Z p Indicates the average value of the central suspended matter concentration of the central grid; C b represents the standard deviation of the edge suspended matter concentration values ​​of m edge grids; Z represents the central suspended matter concentration value of the central grid; K represents the suspended matter concentration span coefficient; K y Indicates the preset coefficient threshold; P z Indicates the change rate of the suspended matter concentration corresponding to multiple central grids; The edge grid corresponding to the maximum value of the suspended matter concentration corresponding to the edge grid is used as the calibration grid; Extracting the straight-line distance between the calibration grid and the central grid; Acquire a calibration distance according to the straight-line distance between the center grids and a distance adjustment coefficient; The calibration distance is obtained by the following formula: ; Where D represents the calibration distance; D max represents the straight-line distance between the calibration grid and the central grid; S represents the distance adjustment coefficient; S c represents the preset reference value of the distance adjustment coefficient; x represents the preset adjustment coefficient, which is used to control the slope of the sigmoid function, thereby affecting the sensitivity of the calibration distance to the distance adjustment coefficient; Calibrate the distance along each edge grid toward the center grid, select a grid whose straight-line distance to the edge grid is the calibrated distance as a target grid, and group the target grids into a target grid area; Collecting the suspended matter concentration value in each grid contained in the target grid area; The overall average value of the suspended matter concentration value in each grid contained in the target grid area and the suspended matter concentration value corresponding to the edge grid is taken as the suspended matter concentration of the current fish pond.

[0027] The technical effect of the above technical solution is: when the suspended matter concentration span coefficient is not lower than the preset coefficient threshold, the solution determines the target grid area by more carefully considering the concentration difference between the edge grid and the central grid, as well as the spatial relationship between them. This helps to capture subtle changes in the suspended matter concentration in the fish pond, thereby improving the accuracy of the overall concentration determination. By calculating the distance adjustment coefficient, the solution can consider the degree of concentration difference between the edge grid and the central grid, as well as the relative position of these grids in space. This helps to more reasonably select the target grid area, so that the selected area can more accurately reflect the distribution of suspended matter concentration in the fish pond. The calibration distance is adjusted by the sigmoid function, so that the sensitivity of the calibration distance to the distance adjustment coefficient is controllable. This helps to avoid the reduction of detection sensitivity due to too small concentration differences while maintaining a certain accuracy, thereby enhancing the stability and robustness of the solution. The solution not only considers the suspended matter concentration in the target grid area, but also combines the suspended matter concentration values ​​of the edge grids to determine the suspended matter concentration of the fish pond by calculating the overall concentration average. This helps to more comprehensively reflect the distribution of suspended matter in the fish pond and improve the accuracy of concentration determination. The scheme introduces multiple parameters (such as distance adjustment coefficient, calibration distance, etc.) and adjustable functions (such as sigmoid function) to adapt to the needs of suspended matter concentration collection and determination in different fish ponds and different environmental conditions. At the same time, each step and parameter in the scheme can be adjusted and optimized according to actual conditions, thereby improving the flexibility and adaptability of the scheme. By more accurately selecting the target grid area and calculating the suspended matter concentration, the scheme can reduce unnecessary collection and calculation work while ensuring accuracy, thereby optimizing resource utilization and reducing cost-effectiveness.

[0028] In summary, the above technical solution achieves a more accurate, reliable and efficient determination of suspended matter concentration in fish ponds by introducing distance adjustment coefficients, using sigmoid functions to adjust calibration distances, integrating suspended matter concentrations of multiple grids, and improving adaptability and flexibility. This provides strong technical support and decision-making basis for fish pond management, water quality monitoring and other fields.

[0029] In this embodiment, the real-time data of fish growth based on the Internet of Things is processed, including: Clean the real-time fish growth data based on IoT based on data cleaning tools; Among them, the real-time data of fish growth based on the Internet of Things is checked, duplicate values, missing values ​​and abnormal values ​​in the real-time data of fish growth based on the Internet of Things are identified, and the duplicate values, missing values ​​and abnormal values ​​in the real-time data of fish growth based on the Internet of Things are processed; For repeated values ​​in the real-time data of fish growth based on the Internet of Things, the repeated values ​​are deleted; for missing values ​​in the real-time data of fish growth based on the Internet of Things, the missing values ​​are deleted or filled to make the missing values ​​complete; for outliers in the real-time data of fish growth based on the Internet of Things, the outliers are deleted or replaced to make them normal.

[0030] It should be noted that by cleaning the real-time fish growth data based on the Internet of Things, duplicate values, missing values ​​and outliers in the real-time fish growth data based on the Internet of Things can be identified, and by processing the duplicate values, missing values ​​and outliers in the real-time fish growth data based on the Internet of Things, the subsequent processing accuracy and processing speed of the real-time fish growth data based on the Internet of Things can be improved.

[0031] In this embodiment, the real-time data of fish growth based on the Internet of Things is processed, and further includes: The real-time fish growth data based on the Internet of Things is converted, the dimensional differences between the real-time fish growth data based on the Internet of Things are reduced, the real-time fish growth data based on the Internet of Things are unified, and standardized real-time fish growth data based on the Internet of Things are determined; The standardized real-time fish growth data based on the Internet of Things is integrated, and the standardized real-time fish growth data based on the Internet of Things from different sources are integrated into a unified view. The integrated standardized real-time fish growth data based on the Internet of Things is verified to determine whether the integrated standardized real-time fish growth data based on the Internet of Things is missing. After the verification is qualified, the integrated standardized real-time fish growth data based on the Internet of Things is securely stored.

[0032] S2. Model prediction: Based on the historical data of fish growth and machine learning technology, a digital twin model of fish growth based on machine learning is constructed. The real-time data of fish growth based on the Internet of Things is analyzed, and the growth trend of fish is predicted. The prediction results of fish growth trend are determined to provide a basis for precise feeding. In this embodiment, a fish growth digital twin model based on machine learning is constructed, including: According to the precise feeding requirements of fish based on digital twins, fish growth history data is collected and divided to determine the training set and test set; It should be noted that by collecting a large amount of historical fish growth data, data support is provided for building a fish growth digital twin model based on machine learning.

[0033] Based on machine learning technology, a training set is used to train the machine learning model, so that the machine learning model can autonomously learn the fish growth prediction process and predict the fish growth trend, and determine the fish growth digital twin model based on machine learning; Based on the test set, a performance test is conducted on the fish growth digital twin model based on machine learning to determine whether the fish growth digital twin model based on machine learning can achieve the effect of predicting the growth trend of fish. When the fish growth digital twin model based on machine learning cannot achieve the effect of predicting the growth trend of fish, the parameters and structure of the fish growth digital twin model based on machine learning are adjusted. After continuous iterative optimization, the optimal fish growth digital twin model is determined.

[0034] In this embodiment, predicting the growth trend of fish includes: Obtain the optimal fish growth digital twin model and deploy it in the actual fish growth trend prediction environment; The real-time fish growth data based on the Internet of Things is input into the optimal fish growth digital twin model. Based on the optimal fish growth digital twin model, the real-time fish growth data based on the Internet of Things is analyzed, and the fish growth trend is predicted to determine the fish growth trend prediction result.

[0035] S3. Feeding control: Automatically adjust the feeding amount according to the predicted results of fish growth trend, and then feed the fish accurately to ensure the nutritional balance required for fish growth.

[0036] In this embodiment, accurate feeding of fishes includes: Obtain the predicted results of fish growth trends, automatically adjust the feeding amount according to the predicted results, and then feed the fish accurately. At the same time, monitor and track the growth of fish in real time, and adjust the feeding plan in time to ensure the nutritional balance required for fish growth.

[0037] In order to better demonstrate the principle of precise feeding of fish based on digital twins, this embodiment now provides a precise feeding system for fish based on digital twins, which is used to implement the precise feeding method for fish based on digital twins as described above, including: Data acquisition module, used to collect fish pond environmental parameters and fish physiological indicators, and determine the real-time data of fish growth based on the Internet of Things; A data processing module is used to clean, convert and integrate the real-time data of fish growth based on the Internet of Things, and to store and back up the real-time data of fish growth based on the Internet of Things; The model prediction module is used to analyze the real-time data of fish growth based on the Internet of Things using the fish growth digital twin model, predict the fish growth trend, and determine the prediction result of the fish growth trend; The feeding control module is used to automatically adjust the feeding amount according to the predicted results of fish growth trend to ensure the nutritional balance required for fish growth.

[0038] In summary, by collecting fish pond environmental parameters and fish physiological indicators, the real-time fish growth data based on the Internet of Things is determined, the real-time fish growth data based on the Internet of Things is cleaned, converted and integrated, and the real-time fish growth data based on the Internet of Things is stored. According to the historical data of fish growth, based on machine learning technology, a machine learning-based digital twin model of fish growth is constructed, and the real-time fish growth data based on the Internet of Things is analyzed based on the optimal fish growth digital twin model, and the fish growth trend is predicted, and the fish growth trend prediction result is determined. The feeding amount is automatically adjusted according to the fish growth trend prediction result, and the fish is accurately fed. At the same time, the growth of the fish is monitored and tracked in real time, and the feeding plan is adjusted in time to ensure the nutritional balance required for fish growth. Accurate feeding can be carried out according to the growth of the fish, which can reduce the cost of fish farming and improve the effect of fish farming.

[0039] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0040] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A fish precision feeding method based on digital twins, characterized in that: The steps include: S1. Data collection and processing: collecting and processing the real-time data of fish growth based on the Internet of Things, and processing the real-time data of fish growth based on the Internet of Things, so that the real-time data of fish growth based on the Internet of Things is safely stored in the database; S2. Model prediction: Based on the historical data of fish growth and machine learning technology, a digital twin model of fish growth based on machine learning is constructed, and the real-time data of fish growth based on the Internet of Things is analyzed, and the fish growth trend is predicted to determine the prediction results of the fish growth trend; S3. Feeding control: Automatically adjust the feeding amount according to the predicted results of fish growth trend, and then feed the fish accurately to ensure the nutritional balance required for fish growth.

2. The method for accurate fish feeding based on digital twins according to claim 1, characterized in that: In S1, real-time fish growth data based on the Internet of Things is collected, including: Install intelligent data collection equipment at the bottom and around the fish pond; Based on intelligent data collection equipment, the water temperature, dissolved oxygen, pH value, ammonia nitrogen concentration, nitrite concentration and suspended solids concentration of the fish pond are monitored and collected in real time to obtain the environmental parameters of the fish pond; Based on intelligent data collection equipment, fish weight, body length, growth rate, feed conversion rate and feeding activity are monitored and collected in real time to obtain fish physiological indicators; Among them, the real-time data of fish growth based on the Internet of Things is determined based on the fish pond environmental parameters and fish physiological indicators.

3. The method for accurate fish feeding based on digital twins according to claim 2, characterized in that: In S1, the concentration of suspended matter is monitored and collected in real time, including: Extract the regional structure map of the fish pond; Dividing the regional structure of the fish pond into grids according to the regional structural characteristics of the fish pond; wherein the regional structural characteristics include the regional area of ​​the fish pond, the maximum water depth of the fish pond, and the water depth change rate; Obtaining the side length of the grid using the regional structural characteristics of the fish pond; The grid side length is obtained by the following formula: ； Where L represents the grid side length; H max Indicates the maximum water depth of the fish pond; L max represents the length of the pond bottom corresponding to the maximum water depth of the fish pond; A represents the area of ​​the fish pond; P represents the rate of change of water depth; A h represents the area of ​​the fish pond corresponding to the maximum water depth of the fish pond; n represents the number of water depth dimensions corresponding to the fish pond except the maximum water depth; L i represents the pond bottom extension length corresponding to the i-th water depth dimension; Divide the regional structure diagram of the fish pond into grids according to the side length of the grids to obtain multiple grid structures; The suspended matter concentration is collected according to the grid structure.

4. The method for accurate fish feeding based on digital twins according to claim 3, characterized in that: The suspended matter concentration is collected according to the grid structure, including: Collecting suspended matter concentration at the edge grid corresponding to the regional structure diagram of the fish pond to obtain edge suspended matter concentration values; Extracting the central grid corresponding to the regional structure diagram to collect suspended matter concentration and obtain the central suspended matter concentration value; The suspended matter concentration span coefficient is obtained by combining the suspended matter concentration value corresponding to each edge grid with the central suspended matter concentration value corresponding to the central grid; The suspended matter concentration span coefficient is obtained by the following formula: ； Where K represents the suspended matter concentration span coefficient; m represents the number of edge grids; C i represents the edge suspended matter concentration value corresponding to the i-th edge grid; Z p represents the average value of the central suspended matter concentration of the central grid; Z represents the central suspended matter concentration value of the central grid; C b represents the standard deviation of the edge suspended matter concentration of m edge grids; Z b Indicates the standard deviation of the central suspended matter concentration corresponding to multiple central grids; comparing the suspended matter concentration span coefficient with a preset coefficient threshold; When the suspended matter concentration span coefficient is lower than the preset coefficient threshold, the average value of the edge suspended matter concentration value and the center suspended matter concentration value is taken as the suspended matter concentration of the current fish pond; When the suspended matter concentration span coefficient is not lower than a preset coefficient threshold, the suspended matter concentration value of the target grid area between the edge grid and the center grid is selected to determine the suspended matter concentration of the current fish pond.

5. The method for accurate fish feeding based on digital twins according to claim 4, characterized in that: When the suspended matter concentration span coefficient is not lower than the preset coefficient threshold, the suspended matter concentration value of the target grid area between the edge grid and the center grid is selected to determine the suspended matter concentration of the current fish pond, including: When the suspended matter concentration span coefficient is not lower than a preset coefficient threshold, the suspended matter concentration value corresponding to each edge grid and the suspended matter concentration value corresponding to the center grid are extracted; Obtaining a distance adjustment coefficient according to a concentration difference between a suspended matter concentration value corresponding to each edge grid and a suspended matter concentration value corresponding to the central grid; The distance adjustment coefficient is obtained by the following formula: ； Where S represents the distance adjustment coefficient; m represents the number of edge grids; C i represents the edge suspended matter concentration value corresponding to the i-th edge grid; Z p Indicates the average value of the central suspended matter concentration of the central grid; C b represents the standard deviation of the edge suspended matter concentration values ​​of m edge grids; Z represents the central suspended matter concentration value of the central grid; K represents the suspended matter concentration span coefficient; K y Indicates the preset coefficient threshold; P z Indicates the change rate of the suspended matter concentration corresponding to multiple central grids; The edge grid corresponding to the maximum value of the suspended matter concentration corresponding to the edge grid is used as the calibration grid; Extracting the straight-line distance between the calibration grid and the central grid; Acquire a calibration distance according to the straight-line distance between the center grids and a distance adjustment coefficient; The calibration distance is obtained by the following formula: ； Where D represents the calibration distance; D max represents the straight-line distance between the calibration grid and the central grid; S represents the distance adjustment coefficient; S c represents the preset reference value of the distance adjustment coefficient; x represents the preset adjustment coefficient, which is used to control the slope of the sigmoid function, thereby affecting the sensitivity of the calibration distance to the distance adjustment coefficient; Calibrate the distance along each edge grid toward the center grid, select a grid whose straight-line distance to the edge grid is the calibrated distance as a target grid, and group the target grids into a target grid area; Collecting the suspended matter concentration value in each grid contained in the target grid area; The overall average value of the suspended matter concentration value in each grid contained in the target grid area and the suspended matter concentration value corresponding to the edge grid is taken as the suspended matter concentration of the current fish pond.

6. The method for accurate fish feeding based on digital twins according to claim 2, characterized in that: In S1, the real-time data of fish growth based on the Internet of Things is processed, including: Clean the real-time fish growth data based on IoT based on data cleaning tools; Among them, the real-time data of fish growth based on the Internet of Things is checked, duplicate values, missing values ​​and abnormal values ​​in the real-time data of fish growth based on the Internet of Things are identified, and the duplicate values, missing values ​​and abnormal values ​​in the real-time data of fish growth based on the Internet of Things are processed; For repeated values ​​in the real-time data of fish growth based on the Internet of Things, the repeated values ​​are deleted; for missing values ​​in the real-time data of fish growth based on the Internet of Things, the missing values ​​are deleted or filled to make the missing values ​​complete; for outliers in the real-time data of fish growth based on the Internet of Things, the outliers are deleted or replaced to make them normal.

7. The method for accurate fish feeding based on digital twins according to claim 6, characterized in that: In S1, the real-time data of fish growth based on the Internet of Things is processed, and further includes: The real-time fish growth data based on the Internet of Things is converted, the dimensional differences between the real-time fish growth data based on the Internet of Things are reduced, the real-time fish growth data based on the Internet of Things are unified, and standardized real-time fish growth data based on the Internet of Things are determined; The standardized real-time fish growth data based on the Internet of Things is integrated, and the standardized real-time fish growth data based on the Internet of Things from different sources are integrated into a unified view. The integrated standardized real-time fish growth data based on the Internet of Things is verified to determine whether the integrated standardized real-time fish growth data based on the Internet of Things is missing. After the verification is qualified, the integrated standardized real-time fish growth data based on the Internet of Things is securely stored.

8. The method for accurate fish feeding based on digital twins according to claim 7, characterized in that: In S2, a fish growth digital twin model based on machine learning is constructed, including: According to the precise feeding requirements of fish based on digital twins, fish growth history data is collected and divided to determine the training set and test set; Based on machine learning technology, a training set is used to train the machine learning model, so that the machine learning model can autonomously learn the fish growth prediction process and predict the fish growth trend, and determine the fish growth digital twin model based on machine learning; Based on the test set, a performance test is conducted on the fish growth digital twin model based on machine learning to determine whether the fish growth digital twin model based on machine learning can achieve the effect of predicting the growth trend of fish. When the fish growth digital twin model based on machine learning cannot achieve the effect of predicting the growth trend of fish, the parameters and structure of the fish growth digital twin model based on machine learning are adjusted. After continuous iterative optimization, the optimal fish growth digital twin model is determined.

9. The method for accurate fish feeding based on digital twins according to claim 8, characterized in that: In S2, the growth trend of fish is predicted, including: Obtain the optimal fish growth digital twin model and deploy it in the actual fish growth trend prediction environment; Input the real-time fish growth data based on the Internet of Things into the optimal fish growth digital twin model, analyze the real-time fish growth data based on the Internet of Things based on the optimal fish growth digital twin model, predict the fish growth trend, and determine the fish growth trend prediction result; In S3, the fish are accurately fed, including: Obtain the prediction results of fish growth trends, automatically adjust the feeding amount according to the prediction results, and then feed the fish accurately. At the same time, monitor and track the growth of fish in real time, and adjust the feeding plan in time to ensure the nutritional balance required for fish growth.

10. A fish precision feeding system based on digital twins, used to implement the fish precision feeding method based on digital twins as described in any one of claims 1 to 9, characterized in that: include: Data acquisition module, used to collect fish pond environmental parameters and fish physiological indicators, and determine the real-time data of fish growth based on the Internet of Things; A data processing module is used to clean, convert and integrate the real-time data of fish growth based on the Internet of Things, and to store and back up the real-time data of fish growth based on the Internet of Things; The model prediction module is used to analyze the real-time data of fish growth based on the Internet of Things using the fish growth digital twin model, predict the fish growth trend, and determine the prediction result of the fish growth trend; The feeding control module is used to automatically adjust the feeding amount according to the predicted results of fish growth trend to ensure the nutritional balance required for fish growth.

Images