An intelligent monitoring and management system for aquaculture based on Internet of Things technology

Abstract

The present invention provides an intelligent aquaculture monitoring and management system based on Internet of Things technology, which relates to the field of aquaculture technology. The method includes: a deployment module for determining key locations in an aquaculture environment and deploying multiple sensor nodes at the key locations; a communication module for wirelessly connecting to the multiple sensor nodes to transmit aquaculture data monitored by each sensor node; a data analysis module connected to the communication module for receiving and analyzing the aquaculture data transmitted by the communication module and determining dynamic correction values; and using the dynamic correction values, adjusting the current sensor node positions to obtain the final position nodes. The present invention can improve the intelligence level of aquaculture.

Description

Technical Field

[0001] The present invention relates to the technical field of aquaculture, and in particular to an intelligent monitoring and management system for aquaculture based on the Internet of Things technology. Background Art

[0002] With the rapid development of aquaculture, the demand for refined and intelligent management of the aquaculture environment is growing. However, some traditional aquaculture monitoring and management solutions have some shortcomings. Specifically, traditional technical solutions have the following defects:

[0003] For example, traditional aquaculture typically relies on regular manual sampling and laboratory testing to obtain key parameters such as water quality and weather conditions. This method is not only time-consuming and labor-intensive, but also difficult to ensure the real-time and accurate data. Furthermore, there is a lack of effective means to monitor the growth status of farmed organisms, making it difficult to detect abnormalities in a timely manner.

[0004] Due to the lack of advanced early warning systems, traditional aquaculture solutions are somewhat slow to respond to emergencies such as water quality deterioration, resulting in unnecessary losses. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide an intelligent monitoring and management system for aquaculture based on Internet of Things technology, which can improve the intelligence level of aquaculture.

[0006] In order to solve the above technical problems, the technical solutions of the present invention are as follows:

[0007] In the first aspect, an intelligent monitoring and management system for aquaculture based on Internet of Things technology includes:

[0008] A deployment module is used to identify key locations in an aquaculture environment and deploy multiple sensor nodes at the key locations;

[0009] A communication module is used to wirelessly connect to multiple sensor nodes to send the farming data monitored by each sensor node;

[0010] a data analysis module connected to the communication module, configured to receive and analyze the farming data sent by the communication module and determine a dynamic correction value; and adjust the current sensor node position using the dynamic correction value to obtain a final position node;

[0011] A construction module is used to obtain aquaculture data monitored by each sensor node based on the final location node, where the aquaculture data includes monitored water quality parameters, meteorological parameters, and growth status parameters of aquaculture organisms; fuse the aquaculture data from each sensor node to form a comprehensive data set, and extract key features reflecting the status and change trends of the aquaculture environment from the comprehensive data set; and construct a prediction model based on the key features to simulate the water flow dynamics, water quality distribution, meteorological conditions, and growth status of aquaculture organisms in the aquaculture pond;

[0012] The early warning module is used to generate corresponding early warning information based on the prediction results and preset thresholds;

[0013] The user interaction interface is connected to the warning module and is used to display warning information to the user and receive user operation instructions.

[0014] Furthermore, key locations within the aquaculture environment were identified, including:

[0015] Evaluate the overall layout of the aquaculture environment, water flow dynamics, and the shape and size of the aquaculture ponds to obtain environmental assessment results;

[0016] Based on the environmental assessment results, determine the preliminary site selection, which may include the intersection of water flows, the edge of the aquaculture pond, or near the aeration equipment;

[0017] Construct a two-dimensional matrix to represent the planar layout of the aquaculture environment, where each element of the matrix corresponds to a specific location in the aquaculture environment;

[0018] Based on the preliminary selected points, calculate the weight values ​​corresponding to the corresponding elements in the position matrix;

[0019] Create a feature matrix that includes various factors that affect the selection of key locations, including the rate of change of water quality parameters, the stability of meteorological parameters, and the distribution density of aquaculture organisms. Each factor corresponds to a column in the feature matrix.

[0020] Perform matrix multiplication on the position matrix and the feature matrix to obtain a comprehensive value for each position;

[0021] According to the comprehensive value obtained by the matrix multiplication operation, all positions are sorted according to the size of the comprehensive value to obtain the corresponding key position.

[0022] Furthermore, wireless communication is performed with multiple sensor nodes to transmit the farming data monitored by each sensor node, including:

[0023] According to the key positions, the location, number and type of sensor nodes are determined, and the initial parameters of the genetic algorithm are set, including population size, crossover rate, mutation rate and number of iterations;

[0024] The wireless communication path is represented by binary coding, and a population is initialized, which contains multiple individuals, each of which represents a wireless communication scheme;

[0025] Define a fitness function to evaluate the pros and cons of each wireless communication scheme;

[0026] According to the fitness function, the corresponding individuals are selected to enter the next generation. Two individuals are randomly selected to exchange the corresponding part of their genes to generate new individuals. Through multiple iterations, the individuals in the population are continuously optimized, that is, the wireless communication scheme. After each iteration, the fitness function is used to evaluate the quality of the individuals in the current population, and selection and genetic operations are performed. When the preset number of iterations is reached, the iteration is stopped and the final wireless communication scheme, that is, the corresponding individual, is output.

[0027] Configure sensor nodes and wireless communication network according to the final wireless communication solution;

[0028] According to the configured sensor nodes and wireless communication network, the breeding data monitored by each sensor node is sent.

[0029] Furthermore, the calculation formula of the dynamic correction value is:

[0030] ;

[0031] in, Indicates dynamic correction value; represents the distance from the i-th sensor node to the target location; represents the real-time monitoring value of the i-th sensor node; represents the reference value of the i-th sensor node; represents the standard deviation of the Gaussian distribution; represents the number of sensor nodes; and Represent the signal strength of the i-th sensor and the j-th sensor respectively; i and j are indexes used to represent different sensors.

[0032] Furthermore, the dynamic correction value is used to adjust the current sensor node position to obtain the final position node, including:

[0033] Determine the original sensor node position. The position of each sensor node is represented by three-dimensional coordinates (x, y, z), where x and y represent the position on the horizontal plane and z represents the water depth.

[0034] Define the position adjustment step size based on the size and shape of the monitoring area and the current distribution of sensor nodes;

[0035] Decompose the dynamic correction value into three dynamic correction components, corresponding to the x-axis, y-axis, and z-axis directions respectively. Multiply the dynamic correction component in each direction by the adjustment step size to obtain a three-dimensional adjustment vector. The dynamic correction value is decomposed into (dx, dy, dz), and the adjustment vector is (dx × step size, dy × step size, dz × step size);

[0036] The calculated adjustment vector is added to the original three-dimensional coordinates of each sensor node to obtain the final position node, that is, the final x coordinate = original x coordinate + dx × step length; the final y coordinate = original y coordinate + dy × step length; the final z coordinate = original z coordinate + dz × step length.

[0037] Furthermore, the dynamic correction value is decomposed into three dynamic correction components, corresponding to the x-axis, y-axis and z-axis directions respectively, including:

[0038] Collect a dataset containing sensor node locations and readings;

[0039] Preprocess the data set and generate corresponding dx, dy, and dz components as training labels based on the position changes of sensor nodes in historical data;

[0040] Build a convolutional neural network model and define the input and output layers. The input layer contains sensor readings and position coordinates. The output layer has three nodes, corresponding to the predicted values ​​of the three components dx, dy, and dz. Set up the hidden layer and determine the corresponding ReLU function.

[0041] Divide the dataset into training, validation, and test sets;

[0042] Use the training set to train the convolutional neural network model, and optimize the parameters of the convolutional neural network model through the back propagation algorithm and gradient descent algorithm to obtain the final convolutional neural network model;

[0043] The reading and position coordinates of the current sensor node are input into the final convolutional neural network model, and the predicted dx, dy, and dz component values ​​are obtained through the forward propagation calculation of the final convolutional neural network model.

[0044] Furthermore, corresponding warning information is generated based on the prediction results and preset thresholds, including:

[0045] Obtain key indicators of simulation data from the prediction model in real time, including predicted values ​​of water flow dynamics, water quality distribution, meteorological conditions, and the growth status of aquaculture organisms;

[0046] Based on historical data analysis, set corresponding warning thresholds for each key indicator;

[0047] Compare key indicators with warning thresholds to determine whether each key indicator exceeds or approaches the warning threshold to obtain comparison results;

[0048] Based on the comparison results, determine whether to trigger an early warning.

[0049] Secondly, an intelligent monitoring and management method for aquaculture based on Internet of Things technology includes:

[0050] Identify key locations in the aquaculture environment and deploy multiple sensor nodes at key locations;

[0051] Wirelessly connect to multiple sensor nodes to send farming data monitored by each sensor node;

[0052] Receive and analyze the farming data sent by the communication module and determine a dynamic correction value; use the dynamic correction value to adjust the current sensor node position to obtain a final position node;

[0053] Based on the final location node, the aquaculture data monitored by each sensor node is obtained. The aquaculture data includes monitored water quality parameters, meteorological parameters, and growth status parameters of the aquaculture organisms. The aquaculture data from each sensor node is integrated to form a comprehensive data set. Key features reflecting the status and changing trends of the aquaculture environment are extracted from the comprehensive data set. Based on the key features, a prediction model is constructed to simulate the water flow dynamics, water quality distribution, meteorological conditions, and growth status of the aquaculture organisms in the aquaculture pond.

[0054] Generate corresponding warning information based on the prediction results and preset thresholds.

[0055] According to a third aspect, a computing device includes:

[0056] one or more processors;

[0057] The storage device is used to store one or more programs, and when the one or more programs are executed by the one or more processors, the one or more processors implement the method.

[0058] In a fourth aspect, a computer-readable storage medium stores a program, which implements the method when executed by a processor.

[0059] The above solution of the present invention includes at least the following beneficial effects:

[0060] By precisely identifying key locations within the aquaculture environment and deploying sensor nodes, comprehensive, real-time monitoring of water quality parameters, meteorological parameters, and the growth status of aquaculture organisms is achieved. This deployment strategy ensures comprehensive and accurate data, providing a solid foundation for subsequent data analysis and decision-making.

[0061] The communication module interacts with multiple sensor nodes through wireless communication connections, realizing real-time and efficient collection of aquaculture data. This not only reduces the cost and time of manual data collection, but also greatly improves the timeliness and reliability of the data, providing timely data support for aquaculture managers.

[0062] The data analysis module conducts in-depth analysis of the collected breeding data, determines dynamic correction values, and adjusts the positions of sensor nodes based on these values. This dynamic adjustment mechanism makes the monitoring system more intelligent and flexible, and can optimize the layout of monitoring points according to actual conditions, thereby improving the accuracy and effectiveness of monitoring.

[0063] The building module forms a comprehensive data set by fusing the data from each sensor node, and extracts key features from it to build a predictive model. This comprehensive analysis and modeling method not only helps managers fully understand the current status and changing trends of the breeding environment, but also provides a scientific basis for predicting future situations.

[0064] The early warning module generates early warning information based on the prediction model and preset thresholds, enabling timely discovery and early warning of potential problems. This early warning mechanism can effectively reduce breeding risks, help managers take timely measures to deal with abnormal situations, and ensure the smooth progress of breeding activities.

[0065] The user interface provides managers with an intuitive and convenient platform to view warning information and receive operational instructions. This not only improves the transparency and readability of information, but also enhances the interactivity and user-friendliness of the system, enabling managers to manage and control the breeding process more efficiently. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] Figure 1 The present invention provides a flow chart of an intelligent monitoring and management method for aquaculture based on Internet of Things technology.

[0067] Figure 2 This is a schematic diagram of an intelligent monitoring and management system for aquaculture based on Internet of Things technology provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0068] Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0069] like Figure 1As shown, the embodiment of the present invention provides an intelligent monitoring and management system for aquaculture based on Internet of Things technology, including:

[0070] A deployment module is used to identify key locations in an aquaculture environment and deploy multiple sensor nodes at the key locations;

[0071] A communication module is used to wirelessly connect to multiple sensor nodes to send the farming data monitored by each sensor node;

[0072] a data analysis module connected to the communication module, configured to receive and analyze the farming data sent by the communication module and determine a dynamic correction value; and adjust the current sensor node position using the dynamic correction value to obtain a final position node;

[0073] A construction module is used to obtain aquaculture data monitored by each sensor node based on the final location node, where the aquaculture data includes monitored water quality parameters, meteorological parameters, and growth status parameters of aquaculture organisms; fuse the aquaculture data from each sensor node to form a comprehensive data set, and extract key features reflecting the status and change trends of the aquaculture environment from the comprehensive data set; and construct a prediction model based on the key features to simulate the water flow dynamics, water quality distribution, meteorological conditions, and growth status of aquaculture organisms in the aquaculture pond;

[0074] The early warning module is used to generate corresponding early warning information based on the prediction results and preset thresholds;

[0075] The user interaction interface is connected to the warning module and is used to display warning information to the user and receive user operation instructions.

[0076] In an embodiment of the present invention, by deploying multiple sensor nodes at key locations, the system can comprehensively and in real time monitor various parameters within the aquaculture environment, such as water quality, weather conditions, and the growth status of aquaculture organisms. This distributed monitoring approach not only improves data accuracy and precision, but also significantly enhances monitoring efficiency and response speed. The data analysis module receives and analyzes sensor data in real time, adjusting the positions of sensor nodes using dynamic correction values. This dynamic adjustment mechanism ensures the flexibility and adaptability of the monitoring system, enabling it to optimize according to actual changes in the aquaculture environment, thereby more accurately reflecting the true state of the aquaculture environment. The construction module extracts key features from the fused comprehensive data set and constructs a predictive model that simulates and predicts water flow dynamics, water quality distribution, weather conditions, and the growth status of aquaculture organisms in the aquaculture pond. This predictive capability provides important decision-making support for aquaculture managers, helping them to take necessary measures in advance, optimize aquaculture conditions, and improve aquaculture efficiency. The early warning module generates corresponding warning information based on the prediction results and preset thresholds, which helps to promptly identify potential risks and problems, such as water quality deterioration and meteorological disasters. By promptly displaying these warning information to users, the system can help users take timely countermeasures, effectively preventing and controlling risks and reducing losses. The user interaction interface not only displays rich monitoring data and early warning information to users, but also receives user operation instructions, providing users with an intuitive and friendly operation platform. This interactive design greatly improves the user experience and convenience, allowing users to more easily manage and control the entire breeding process.

[0077] In a preferred embodiment of the present invention, determining a critical location in an aquaculture environment comprises:

[0078] The overall layout, water flow dynamics, shape and size of the aquaculture environment are assessed to obtain environmental assessment results. Specifically, the assessment includes: collecting relevant information on the aquaculture environment, including design drawings of the aquaculture ponds, historical data on water flow dynamics, meteorological records, etc.; conducting on-site inspections of the aquaculture ponds to observe water flow conditions, pond shape, size, and the layout of existing equipment; analyzing the collected data to understand the main flow paths, speed changes, and water quality differences in different areas of the ponds; and determining the overall characteristics and key areas of the aquaculture environment based on the analysis and survey results.

[0079] Based on the results of the environmental assessment, preliminary selection points are determined. The preliminary selection points include water flow intersections, edge areas of the aquaculture pond, or areas near the aeration equipment. Specifically, based on the environmental assessment report, areas where water flows converge, mix, or generate vortices are identified. These areas are key points for water quality changes; the edge areas of the aquaculture pond are marked, especially the boundaries where substances and energy are exchanged with the external environment; based on factors such as aeration equipment that may have a significant impact on local water quality, monitoring points are selected around them. These identified preliminary selection points are marked on the floor plan of the aquaculture pond, and their coordinates and characteristics are recorded.

[0080] Construct a two-dimensional matrix to represent the planar layout of the aquaculture environment, where each element of the matrix corresponds to a specific location in the aquaculture environment. Specifically, the planar layout of the aquaculture pond is divided into a fine grid, where each grid represents a specific location; coordinate values ​​are assigned to each grid, represented by a two-dimensional coordinate system (x, y); and a two-dimensional matrix is ​​constructed based on the coordinates of the grid, where each element in the matrix corresponds to a specific location in the aquaculture pond.

[0081] According to the preliminary selected points, calculate the weight values ​​corresponding to the corresponding elements in the position matrix, where the weight values The calculation formula is:

[0082] ;

[0083] in, 、 、 and They are the weight coefficients of water flow dynamics, water quality changes, the impact of aeration equipment and the impact of marginal areas; It's location Water flow velocity at It is the maximum velocity of water flow in the entire aquaculture pond; It's location The rate of change of water quality at the location; is the maximum water quality change rate at all locations in the aquaculture pond; It's location Distance to aeration equipment; It's location Distance to the edge of the culture pond.

[0084] Create a feature matrix that includes various factors that affect the selection of key locations, including the rate of change of water quality parameters, the stability of meteorological parameters, and the distribution density of aquaculture organisms. Each factor corresponds to a column of the feature matrix. Specifically, the main factors that affect the selection of key locations, such as the rate of change of water quality parameters, the stability of meteorological parameters, and the distribution density of aquaculture organisms, are selected. For each factor, relevant data at different locations in the aquaculture pond are collected or measured. From the collected data, numerical values ​​or indicators that can represent the characteristics of each factor are extracted. The extracted eigenvalues ​​are arranged according to the factors to form columns of the feature matrix, and each row represents the characteristic combination of a location.

[0085] Perform matrix multiplication on the location matrix and the feature matrix to obtain a composite value for each location. This involves ensuring that the location matrix and the feature matrix have the same number of rows to facilitate matrix multiplication. The weight values ​​in the location matrix are multiplied by the eigenvalues ​​of the corresponding location in the feature matrix, and the sum is calculated to obtain a composite value for each location. This process is equivalent to a weighted summation, where the weights reflect the importance of the initial selected location, and the eigenvalues ​​reflect the influence of each factor.

[0086] According to the comprehensive value obtained by the matrix multiplication operation, all positions are sorted according to the size of the comprehensive value to obtain the corresponding key positions, specifically including: sorting the comprehensive values ​​of all positions from high to low or from low to high, and according to the sorting results, selecting several positions with the highest comprehensive values ​​as key positions. These positions are representative or special in the aquaculture environment and are key areas for subsequent monitoring and management.

[0087] In an embodiment of the present invention, by comprehensively evaluating the overall layout of the aquaculture environment, water flow dynamics, and the shape and size of the aquaculture pond, the method can scientifically determine preliminary selected points and then accurately locate key locations using matrix operations. This layout and positioning method ensures the representativeness and effectiveness of the monitoring points, laying a solid foundation for subsequent data collection and analysis. When constructing the characteristic matrix, the method comprehensively considers multiple factors such as the rate of change of water quality parameters, the stability of meteorological parameters, and the distribution density of aquaculture organisms. This diversified consideration makes the determination of key locations more comprehensive and accurate, and can truly reflect the complexity and dynamic nature of the aquaculture environment. Through matrix multiplication operations, the method calculates a comprehensive value for each location, thereby achieving quantitative analysis of key locations. This quantitative method not only facilitates the comparison and ranking of different locations, but also provides aquaculture managers with intuitive decision-making basis, helping them optimize the layout of monitoring points and resource allocation. The accurate determination of key locations helps improve the efficiency and effectiveness of monitoring. By deploying sensor nodes at key locations, it is possible to maximize the capture of important information and changing trends in the aquaculture environment, reduce the collection of redundant data, and thus improve the response speed and accuracy of the entire monitoring system. By optimizing the layout of monitoring points, problems in the breeding environment can be discovered and solved in a timely manner, breeding risks can be reduced, and breeding benefits can be improved.

[0088] In a preferred embodiment of the present invention, wireless communication is connected with a plurality of sensor nodes to transmit the farming data monitored by each sensor node, including:

[0089] Based on the key positions, the position, number and type of sensor nodes are determined, and the initial parameters of the genetic algorithm are set, including population size, crossover rate, mutation rate and number of iterations. Specifically, the following are carried out: survey the breeding environment to determine the key areas that need to be monitored, such as locations where environmental factors such as temperature, humidity, and light vary greatly; select suitable sensor nodes based on the type of parameters that need to be monitored at each key position (such as temperature, humidity, pH value, etc.); comprehensively consider monitoring accuracy, cost budget and system complexity to determine the number and specific location of sensor nodes placed at each key position; set the appropriate population size, that is, the number of individuals in each generation, according to the complexity of the problem and computing resources, set the probability of crossover operation to control the speed of new individuals generation, set the probability of mutation operation to increase the diversity of the population, and set a reasonable number of iterations as the condition for algorithm termination based on the accuracy and time requirements of the problem solution.

[0090] Binary coding is used to represent wireless communication paths. A population is initialized, which contains multiple individuals. Each individual represents a wireless communication scheme. Specifically, the problem of selecting a wireless communication path is converted into a binary string representation using binary coding. According to the coding scheme, an initial population is randomly generated. Each individual (i.e., binary string) represents a possible wireless communication scheme.

[0091] A fitness function is defined to evaluate the pros and cons of each wireless communication scheme. The calculation formula of the fitness function is:

[0092] ;

[0093] in, represents the fitness function; Represents an individual (wireless communication scheme, including node locations and communication paths between nodes); 、 、 and Represents the weight factor, which adjusts the importance of each indicator; Indicates the effective coverage area; Indicates the total area of ​​the target area; represents the distance between the i-th and j-th nodes; represents the transmission delay of the i-th node (including processing time and link delay); represents the transmission failure probability of the i-th node; i and j represent two different nodes respectively; n is the total number of nodes.

[0094] Based on the fitness function, individuals are selected to advance to the next generation. Two individuals are randomly selected to exchange some of their genes to generate new individuals. Through multiple iterations, the individuals in the population, i.e., the wireless communication scheme, are continuously optimized. After each iteration, the fitness function is used to evaluate the quality of individuals in the current population, and selection and genetic operations are performed. When the preset number of iterations is reached, the iterations are terminated and the final wireless communication scheme, i.e., the corresponding individuals, is output. Specifically, the fitness function is used to evaluate the quality of individuals in the current population, and selection and genetic operations are performed. When the preset number of iterations is reached, the iterations are terminated and the final wireless communication scheme, i.e., the corresponding individuals, is output. Specifically, the fitness function is used to evaluate the quality of individuals in the current population. Common selection methods include roulette wheel selection and tournament selection. Two individuals are randomly selected to exchange some of their genes according to a set crossover rate and crossover method (such as single-point crossover or multi-point crossover) to generate new individuals. The newly generated individuals are randomly mutated according to a set mutation rate, i.e., the values ​​of certain genes are changed to increase the diversity of the population. The selection, crossover, and mutation operations are repeated until the preset number of iterations is reached. After each iteration, the fitness function is used to evaluate the quality of individuals in the current population.

[0095] According to the final wireless communication scheme, the sensor nodes and wireless communication network are configured. Specifically, when the preset number of iterations is reached, the iteration is stopped, the individual with the highest fitness is selected from the current population as the final wireless communication scheme, and the selected individual (binary string) is decoded into a specific wireless communication path and configuration parameters.

[0096] According to the configured sensor nodes and wireless communication network, the aquaculture data monitored by each sensor node is sent. Specifically, according to the decoded plan, sensor nodes and wireless communication equipment are deployed in the aquaculture environment, and the parameters of the wireless communication network, such as communication frequency and transmission rate, are configured to ensure that the sensor nodes can smoothly access the network and send data. Each sensor node collects monitoring data of the aquaculture environment at the set frequency, and sends the collected data to the data center or the designated receiving end in real time through the configured wireless communication network.

[0097] In an embodiment of the present invention, by analyzing key locations to determine the location, number, and type of sensor nodes, it is possible to ensure that key areas in the aquaculture environment are effectively monitored, thereby improving the comprehensiveness and accuracy of the data. The application of genetic algorithms can search for more optimal wireless communication path solutions, reduce signal interference and transmission delays, and improve the efficiency and stability of data transmission. Through algorithm optimization, unnecessary sensor nodes and communication equipment can be reduced while meeting monitoring needs, thereby reducing the hardware cost and maintenance cost of the entire system. The genetic algorithm used in this solution has good flexibility and scalability, and can adapt to aquaculture scenarios of different scales, as well as possible future increases in sensor nodes or adjustments to the network structure. The optimized wireless communication solution can ensure the real-time and accuracy of aquaculture data.

[0098] In a preferred embodiment of the present invention, the calculation formula of the dynamic correction value is:

[0099] ;

[0100] in, Indicates dynamic correction value; represents the distance from the i-th sensor node to the target location; represents the real-time monitoring value of the i-th sensor node; represents the reference value of the i-th sensor node; represents the standard deviation of the Gaussian distribution; represents the number of sensor nodes; and Represent the signal strength of the i-th sensor and the j-th sensor respectively; i and j are indexes used to represent different sensors.

[0101] In the embodiment of the present invention, the formula can make weighted corrections in real time based on the monitoring value of the sensor node and its distance to the target location by introducing a dynamic correction value. This correction method helps to reduce data errors caused by sensor location, environmental interference or other factors, thereby improving the data accuracy and reliability of the entire monitoring system. The Gaussian distribution function is used in the formula to consider the distance from the sensor node to the target location, which means that the sensor node closer to the target will be given a higher weight in the correction value calculation. At the same time, by introducing the signal strength and This formula also reflects the relative importance of different sensor nodes, further enhancing the accuracy of the correction. Because the formula is dynamic, it updates the correction value as the real-time monitoring values ​​of the sensor nodes change, thus ensuring the real-time performance of the system. By comparing the real-time monitoring values ​​with the reference values, the formula can identify and suppress the impact of outliers to a certain extent. When the monitoring value of a sensor node deviates from the normal range, its correction value is adjusted accordingly, thereby reducing the impact of abnormal data on the overall results. By comprehensively considering multiple factors (such as distance, signal strength, and real-time monitoring values), the formula can enhance the robustness of the entire monitoring system. Even if some sensor nodes fail or data anomalies occur, the remaining functioning nodes can still provide relatively accurate correction values.

[0102] In a preferred embodiment of the present invention, the current sensor node position is adjusted using the dynamic correction value to obtain the final position node, including:

[0103] Determine the original sensor node position. The position of each sensor node is represented by three-dimensional coordinates (x, y, z), where x and y represent the position on the horizontal plane and z represents the water depth. Specifically, it includes: obtaining the original position information of all sensor nodes, which is achieved through GPS and underwater positioning systems; recording the position of each sensor node as three-dimensional coordinates (x, y, z), where x and y represent the projected position of the sensor on the horizontal plane, expressed by latitude and longitude or relative coordinate system; z represents the depth of the sensor underwater.

[0104] Define the position adjustment step size based on the size and shape of the monitoring area and the current distribution of sensor nodes. Specifically, analyze the size and shape of the monitoring area, understand the scope of the entire monitoring environment, and evaluate the current distribution of sensor nodes, including node density and whether there are coverage blind spots. Based on the above analysis, set a suitable adjustment step size. This step size should be able to ensure the accuracy of the adjustment without causing excessive adjustment. The size of the step size may require multiple tests and optimizations based on actual conditions.

[0105] Decompose the dynamic correction value into three dynamic correction components, corresponding to the x-axis, y-axis, and z-axis directions respectively. Multiply the dynamic correction component in each direction by the adjustment step size to obtain a three-dimensional adjustment vector. The dynamic correction value is decomposed into (dx, dy, dz), and the adjustment vector is (dx × step size, dy × step size, dz × step size);

[0106] The calculated adjustment vector is added to the original 3D coordinates of each sensor node to obtain the final position node. That is, the final x-coordinate = original x-coordinate + dx × step length; the final y-coordinate = original y-coordinate + dy × step length; and the final z-coordinate = original z-coordinate + dz × step length. Specifically, the dynamic correction component in each direction is multiplied by the adjustment step length defined in step 2 to obtain the adjustment amount in each direction. These three values, dx × step length, dy × step length, and dz × step length, constitute a 3D adjustment vector. The calculated adjustment vector is added to the original 3D coordinates of each sensor node. Specifically, dx × step length is added to the original x-coordinate, dy × step length is added to the original y-coordinate, and dz × step length is added to the original z-coordinate to obtain the adjusted new coordinates.

[0107] In an embodiment of the present invention, by introducing a dynamic correction value and decomposing it into correction components in three directions, the position of each sensor node can be adjusted more accurately. This fine-grained adjustment method helps reduce errors in the original position coordinates, thereby improving the positioning accuracy of the entire sensor network. The method can flexibly define the position adjustment step size based on the size and shape of different monitoring areas and the distribution of sensor nodes. This adaptability makes the method widely applicable to various scenarios, enabling effective position adjustment for both small and large monitoring areas. By dynamically adjusting the positions of sensor nodes, the distribution of nodes within the monitoring area can be optimized, which helps ensure adequate coverage of key areas while avoiding over-deployment of nodes in non-key areas, thereby improving the efficiency and performance of the entire monitoring system. When certain sensor nodes experience data anomalies due to failures or environmental factors, the method can identify and adjust the positions of these nodes through dynamic correction values. This adjustment helps reduce the impact of abnormal nodes on the entire system, thereby enhancing the robustness and stability of the system. Since the dynamic correction value is calculated in real time, this method can adjust the position of the sensor node in real time. This real-time nature enables the system to respond quickly to changes in the environment and adjust the node position in time to maintain the best monitoring effect. At the same time, the real-time feedback mechanism also helps to promptly discover and correct problems in the system.

[0108] In a preferred embodiment of the present invention, the dynamic correction value is decomposed into three dynamic correction components corresponding to the x-axis, y-axis and z-axis directions respectively, including:

[0109] Collect a data set containing sensor node locations and readings. Specifically, this involves: determining the data source, determining the type and deployment location of the sensor nodes, and ensuring that data containing the sensor node locations and readings can be obtained; collecting data through the sensor network, recording the location coordinates (x, y, z) and corresponding readings of each sensor node, and organizing the collected data into a standard data set format for subsequent processing and analysis.

[0110] The dataset is preprocessed, and the corresponding dx, dy, and dz components are generated as training labels based on the position changes of the sensor nodes in the historical data. Specifically, this includes: removing outliers, missing values, or duplicate data in the dataset to ensure data quality, extracting features related to the position changes of the sensor nodes from the dataset, such as timestamps and reading change rates, etc. Based on the position changes of the sensor nodes in the historical data, the true values ​​of the three components dx, dy, and dz are calculated and used as training labels. This usually involves differential operations on the position coordinates of adjacent time points.

[0111] Construct a convolutional neural network model and define the input layer and output layer. The input layer contains sensor readings and position coordinates; the output layer has three nodes, corresponding to the predicted values ​​of the three components dx, dy, and dz, respectively. Set the hidden layer and determine the corresponding ReLU function. Specifically, it includes: designing the structure of the convolutional neural network, including the input layer, hidden layer, and output layer. The input layer should contain sensor readings and position coordinates, and the output layer should have three nodes corresponding to the predicted values ​​of dx, dy, and dz, respectively. Determine the number of hidden layers and the number of neurons in each layer, and select a suitable activation function (such as the ReLU function). Initialize the weights and bias of the model. Usually, random initialization or pre-trained models are used for parameter initialization.

[0112] Divide the dataset into training, validation, and test sets. Specifically, randomly divide the organized dataset into training, validation, and test sets, typically in a ratio of 70%:15%:15%. Prepare a data loader for model training to ensure efficient data loading in batches.

[0113] Use the training set to train the convolutional neural network model, and optimize the parameters of the convolutional neural network model through the back propagation algorithm and the gradient descent algorithm to obtain the final convolutional neural network model, including: selecting a suitable optimizer (such as Adam) and learning rate, defining the loss function of the model, usually using the mean square error (MSE) to measure the gap between the predicted value and the true value, using the training set to train the model for multiple iterations, and optimizing the parameters of the model through the back propagation algorithm and the gradient descent algorithm. After each training cycle, use the validation set to evaluate the performance of the model, and adjust the parameters or improve the model as needed. When the model achieves satisfactory performance on the validation set, save the parameters and structure of the model for subsequent use.

[0114] The reading and position coordinates of the current sensor node are input into the final convolutional neural network model, and the predicted dx, dy, and dz component values ​​are obtained through the forward propagation calculation of the final convolutional neural network model. Specifically, the method includes: loading the trained final convolutional neural network model, organizing the reading and position coordinates of the current sensor node into a format that matches the model input layer, passing the input data to the model, and obtaining the predicted dx, dy, and dz component values ​​through the forward propagation calculation of the model. The predicted results are output for subsequent position adjustment.

[0115] In an embodiment of the present invention, a convolutional neural network model is constructed and trained using historical data to learn the complex relationship between sensor readings and position changes. This method enables more accurate prediction of the three components dx, dy, and dz, thereby improving the accuracy of position adjustment. Using a CNN model for prediction automates and intelligentizes position adjustment. Once the model is trained, it automatically and rapidly calculates the required position adjustment based on the current sensor readings and position coordinates, without requiring manual intervention. Due to its powerful feature extraction capabilities, the CNN model is adaptable to varying environmental conditions and sensor configurations. Even with environmental changes or sensor replacements, retraining the model allows it to continue providing accurate position adjustment recommendations. The CNN model's fast forward propagation computation speed allows it to calculate position adjustment recommendations in real time based on current sensor data. By training with a large amount of historical data, the CNN model learns the patterns of position change under various conditions, resulting in robustness and the ability to provide relatively accurate position adjustment recommendations even in the presence of noisy data or abnormal conditions. This method can be easily scaled to larger sensor networks by simply collecting more historical data and adjusting the CNN model's structure and parameters accordingly.

[0116] In a preferred embodiment of the present invention, aquaculture data monitored by each sensor node is obtained based on the final location node, wherein the aquaculture data includes monitored water quality parameters, meteorological parameters, and growth status parameters of the aquacultured organisms; the aquaculture data of each sensor node are fused to form a comprehensive data set, and key features reflecting the status and change trends of the aquaculture environment are extracted from the comprehensive data set; and a prediction model is constructed based on the key features to simulate the water flow dynamics, water quality distribution, meteorological conditions, and growth status of the aquacultured organisms in the aquaculture pond, specifically including:

[0117] To determine the location node, first determine that the "final location node" refers to the exact location of the sensor node determined after dynamic correction; based on this location node information, obtain aquaculture data from each sensor node in real time or periodically through a preset data interface or communication protocol; analyze the obtained data to extract water quality parameters (such as dissolved oxygen, ammonia nitrogen, pH value, etc.), meteorological parameters (such as temperature, humidity, wind speed, light intensity, etc.) and growth status parameters of the aquaculture organisms (such as body length, weight, food intake, etc.).

[0118] Define a unified data format, typically a structured data model such as a database table, a CSV file format, or a data object model. This format should include all necessary fields, such as timestamp, sensor ID, parameter type (e.g., water quality, meteorology), and value. Store the parsed data in the defined format in a suitable data storage system, such as a relational database (e.g., MySQL, PostgreSQL), a NoSQL database (e.g., MongoDB, Cassandra), or a distributed file system (e.g., HDFS). Because different sensors may use different clock sources, it is necessary to ensure that all data timestamps are synchronized. This is typically achieved through NTP (Network Time Protocol), ensuring that all sensor nodes and the data center are synchronized. If some sensor nodes collect data less frequently than others, or if data arrives at different times due to network latency, data interpolation may be necessary to ensure that all data are time-aligned. Interpolation methods such as linear interpolation and nearest neighbor interpolation can be used. The Kalman filter is an effective data fusion algorithm that recursively estimates the state of dynamic systems in the presence of noise. For the same type of monitoring data (such as temperature and humidity), the Kalman filter can be used to fuse data from different sensor nodes. Implementing a Kalman filter involves initializing the state vector and covariance matrix, then updating them based on the measured and predicted values ​​at each time step. This process gradually reduces noise in the data and improves data accuracy. Kalman filter parameters, such as the covariance of process and measurement noise, are adjusted based on the application to achieve optimal fusion results. During the data fusion process, thresholds are set to detect outliers, which may be caused by sensor failure, transmission errors, or external interference. Once an outlier is detected, various strategies can be used to address it, such as ignoring it, replacing it with the previous valid value or average, or using interpolation. The specific strategy should be selected based on the actual situation and data characteristics. The fused and outlier-processed data are then integrated into a unified dataset, which contains the monitoring data from all sensor nodes and the fused data.

[0119] Basic statistical analysis is performed on various parameters in the comprehensive dataset (such as water quality and meteorological parameters), including calculation of mean, median, standard deviation, maximum, and minimum values, to understand the data distribution and fluctuation range. For time series data, methods such as Fourier transform (FFT) or wavelet transform can be used to convert the data from the time domain to the frequency domain to analyze the periodicity and frequency characteristics of the data. This helps to extract key features related to the status and changing trends of the aquaculture environment. Based on the results of statistical analysis and time-frequency transformation, key features that reflect the status and changing trends of the aquaculture environment are extracted. These features include statistical quantities of certain parameters, periodic components, and trend terms. Correlations between different features are analyzed using correlation coefficients (such as the Pearson correlation coefficient) or mutual information methods to identify those with significant impacts on the aquaculture environment. This helps understand the relationships between features and provides a basis for subsequent feature selection. Based on the results of the correlation analysis, the most representative feature subset is selected. Feature selection algorithms (such as tree-based feature importance assessment) can be used to assist in this process. Selected features are standardized (e.g., using z-score standardization) or normalized (e.g., using min-max normalization) to eliminate the impact of dimensionality differences on model training. This helps ensure that different features have equal weight and influence in the model. Based on the complexity of the problem and the characteristics of the data, a long short-term memory (LSTM) network was selected as the prediction model. LSTM is a deep learning model suitable for sequential data, capable of capturing long-term dependencies within sequences. Using the extracted key features as input and historical data on water flow dynamics, water quality distribution, meteorological conditions, and the growth status of aquaculture organisms as target outputs, an LSTM model was constructed and trained. During training, model parameters (e.g., number of layers, number of units, learning rate, etc.) and learning algorithms (e.g., optimizer selection, loss function selection, etc.) were adjusted to optimize model performance. The trained LSTM model was validated using a validation dataset to evaluate performance metrics such as prediction accuracy and generalization. Mean squared error (MSE) was used to measure the model's predictive performance. Based on the validation results, the model is adjusted and optimized as necessary. This includes adjusting the model structure (such as increasing or decreasing the number of layers or changing the number of units), adjusting hyperparameters (such as learning rate and batch size), or trying different learning algorithms and loss functions. Through continuous iteration and optimization, the model's predictive performance is improved. The trained and tuned LSTM model is deployed in the actual application environment to ensure that the model can predict and analyze the aquaculture environment in real time or periodically.

[0120] In a preferred embodiment of the present invention, generating corresponding warning information based on the prediction results and the preset threshold value includes:

[0121] Key indicators of simulation data are obtained from the prediction model in real time. Key indicators include water flow dynamics, water quality distribution, meteorological conditions and predicted values ​​of the growth status of farmed organisms. Specifically, the early warning system regularly (such as every second, every minute) pulls the latest simulation data from the prediction model through the data interface, and parses the obtained simulation data into specific key indicators, including water flow dynamics (such as flow velocity and flow direction), water quality distribution (such as dissolved oxygen and pH value), meteorological conditions (such as temperature, humidity and wind speed) and the growth status of farmed organisms (such as growth rate and health status prediction value).

[0122] Based on historical data analysis, corresponding warning thresholds are set for each key indicator. Specifically, this includes: collecting a large amount of historical data, including key indicator values ​​under normal and abnormal conditions; using statistical methods to analyze historical data, understanding the fluctuation range of each key indicator under normal conditions and its characteristics under abnormal conditions; and setting reasonable warning thresholds for each key indicator based on the analysis results. These thresholds can be single values ​​(such as upper or lower limits). Using a portion of historical data to verify the set thresholds to ensure their accuracy and effectiveness.

[0123] Compare key indicators with warning thresholds to determine whether each key indicator exceeds or approaches the warning threshold to obtain comparison results, specifically including: comparing the key indicator values ​​obtained in real time with the preset warning thresholds one by one, and based on the comparison results, determine whether each key indicator exceeds the warning threshold (indicating an abnormal situation) or approaches the warning threshold (indicating potential risks), and record the comparison results in the database, including key indicator values, warning thresholds, comparison time and status (normal, approaching warning, exceeding warning).

[0124] Based on the comparison results, determine whether to trigger an early warning, specifically including: defining the logical rules for early warning triggering, such as if one or more key indicators exceed the early warning threshold for multiple consecutive times, or if a key indicator suddenly deviates significantly from the normal value, etc. Based on the comparison results and early warning logical rules, determine whether an early warning needs to be triggered. If it is determined that an early warning is triggered, generate early warning information (including warning type, level, trigger time and recommended measures, etc.) and send it to relevant personnel via SMS, email or other means, and record the triggered early warning information in the database for subsequent query and analysis.

[0125] In an embodiment of the present invention, by real-time monitoring and comparing key indicators with warning thresholds, the system can issue timely warnings when key indicators approach or exceed safe ranges. This helps relevant personnel respond quickly and prevent potential problems from escalating. The warning system can proactively identify potential risks, such as abnormal water flow, deteriorating water quality, adverse weather conditions, or abnormal growth of aquaculture organisms. Through early intervention, the impact of these risks on aquaculture and other related activities can be significantly reduced. Warning information provides managers with important decision-making support. After receiving a warning, managers can formulate appropriate response measures based on the warning type and level, thereby optimizing resource allocation and improving management efficiency. Through the warning system, operators can more accurately grasp the operational status, adjust operational strategies in a timely manner, reduce unnecessary waste, and improve overall operational efficiency. As an integral part of the overall monitoring system, the warning mechanism enhances the system's robustness. Even in the face of emergencies or abnormal situations, the system can use the warning mechanism to promptly detect and respond to problems.

[0126] like Figure 2 As shown, an embodiment of the present invention further provides an intelligent monitoring and management method for aquaculture based on Internet of Things technology, comprising:

[0127] Identify key locations in the aquaculture environment and deploy multiple sensor nodes at key locations;

[0128] Wirelessly connect to multiple sensor nodes to send farming data monitored by each sensor node;

[0129] Receive and analyze the farming data sent by the communication module and determine a dynamic correction value; use the dynamic correction value to adjust the current sensor node position to obtain a final position node;

[0130] Based on the final location node, the aquaculture data monitored by each sensor node is obtained. The aquaculture data includes monitored water quality parameters, meteorological parameters, and growth status parameters of the aquaculture organisms. The aquaculture data from each sensor node is integrated to form a comprehensive data set. Key features reflecting the status and changing trends of the aquaculture environment are extracted from the comprehensive data set. Based on the key features, a prediction model is constructed to simulate the water flow dynamics, water quality distribution, meteorological conditions, and growth status of the aquaculture organisms in the aquaculture pond.

[0131] Generate corresponding warning information based on the prediction results and preset thresholds.

[0132] It should be noted that this system is a system corresponding to the above method, and all implementation methods in the above method embodiment are applicable to this embodiment and can achieve the same technical effects.

[0133] An embodiment of the present invention further provides a computing device comprising: a processor and a memory storing a computer program, wherein the computer program, when executed by the processor, performs the above-described method. All implementations in the above-described method embodiments are applicable to this embodiment and can achieve the same technical effects.

[0134] The embodiment of the present invention further provides a computer-readable storage medium storing instructions, which, when executed on a computer, causes the computer to execute the above-described method. All implementations in the above-described method embodiment are applicable to this embodiment and can achieve the same technical effects.

Claims

1. An intelligent monitoring and management system for aquaculture based on Internet of Things technology, characterized in that: include: The layout module is used to determine the key locations in the aquaculture environment, including: evaluating the overall layout of the aquaculture environment, water flow dynamics, and the shape and size of the aquaculture pond to obtain environmental assessment results; determining preliminary selection points based on the environmental assessment results, including water flow intersections, edge areas of the aquaculture pond, or near aeration equipment; constructing a two-dimensional matrix to represent the plane layout of the aquaculture environment, with each element of the matrix corresponding to a specific location in the aquaculture environment; calculating the weight values ​​corresponding to the corresponding elements in the position matrix based on the preliminary selection points, where the weight values ​​are obtained based on water flow dynamics, water quality changes, the impact of aeration equipment, and the impact of edge areas; creating a feature matrix that contains various factors that affect the selection of key locations, including the rate of change of water quality parameters, the stability of meteorological parameters, and the distribution density of aquaculture organisms, with each factor corresponding to a column of the feature matrix; performing matrix multiplication on the position matrix and the feature matrix to obtain a comprehensive value for each location; based on the comprehensive value obtained by the matrix multiplication operation, sorting all locations according to the size of the comprehensive value to obtain the corresponding key locations, and deploying multiple sensor nodes at the key locations; A communication module is used to wirelessly connect to multiple sensor nodes to send the farming data monitored by each sensor node; The data analysis module is connected to the communication module and is used to receive and analyze the breeding data sent by the communication module, and calculate the difference between the real-time monitoring value of the node and the reference value of the node according to the Gaussian distribution function, the signal strength and the difference between the real-time monitoring value of the node and the reference value of the node. and The dynamic correction value is determined by the ratio of ; the dynamic correction value is used to adjust the current sensor node position to obtain the final position node; A construction module is used to obtain aquaculture data monitored by each sensor node based on the final location node, where the aquaculture data includes monitored water quality parameters, meteorological parameters, and growth status parameters of aquaculture organisms; fuse the aquaculture data from each sensor node to form a comprehensive data set, and extract key features reflecting the status and change trends of the aquaculture environment from the comprehensive data set; construct a prediction model based on the key features to simulate the water flow dynamics, water quality distribution, meteorological conditions, and growth status of aquaculture organisms in the aquaculture pond; and perform real-time prediction of the future aquaculture environment based on the prediction model and real-time aquaculture data to obtain prediction results; The early warning module is used to generate corresponding early warning information based on the prediction results and preset thresholds; Wirelessly connect to multiple sensor nodes to send the farming data monitored by each sensor node, including: According to the key positions, the location, number and type of sensor nodes are determined, and the initial parameters of the genetic algorithm are set, including population size, crossover rate, mutation rate and number of iterations; The wireless communication path is represented by binary coding, and a population is initialized, which contains multiple individuals, each of which represents a wireless communication scheme; Define a fitness function to evaluate the pros and cons of each wireless communication scheme; According to the fitness function, the corresponding individuals are selected to enter the next generation. Two individuals are randomly selected to exchange the corresponding part of the genes to generate new individuals. Through multiple iterations, the individuals in the population are continuously optimized, that is, the wireless communication scheme. After each iteration, the fitness function is used to evaluate the quality of the individuals in the current population, and selection and genetic operations are performed. When the preset number of iterations is reached, the iteration is stopped and the final wireless communication scheme, that is, the corresponding individual, is output. The calculation formula of the fitness function is: ； in, represents the fitness function; Represents an individual, i.e., a wireless communication scheme, which includes node locations and communication paths between nodes; 、 、 and Represents the weight factor, which adjusts the importance of each indicator; Indicates the effective coverage area; Indicates the total area of ​​the target area; represents the distance between the i-th and j-th nodes; represents the transmission delay of the i-th node, including processing time and link delay; represents the transmission failure probability of the i-th node; i and j represent two different nodes; n is the total number of nodes; Configure sensor nodes and wireless communication network according to the final wireless communication solution; According to the configured sensor nodes and wireless communication network, the farming data monitored by each sensor node is sent; The calculation formula of the dynamic correction value is: ； in, Indicates dynamic correction value; represents the distance from the i-th sensor node to the target location; represents the real-time monitoring value of the i-th sensor node; represents the reference value of the i-th sensor node; represents the standard deviation of the Gaussian distribution; represents the number of sensor nodes; and Represents the signal strength of the i-th sensor and the j-th sensor respectively; i and j are indexes used to represent different sensors; Using the dynamic correction value, the current sensor node position is adjusted to obtain the final position node, including: Determine the original sensor node position. The position of each sensor node is represented by three-dimensional coordinates (x, y, z), where x and y represent the position on the horizontal plane and z represents the water depth. Define the position adjustment step size based on the size and shape of the monitoring area and the current distribution of sensor nodes; Decompose the dynamic correction value into three dynamic correction components, corresponding to the x-axis, y-axis, and z-axis directions respectively. Multiply the dynamic correction component in each direction by the adjustment step size to obtain a three-dimensional adjustment vector. The dynamic correction value is decomposed into (dx, dy, dz), and the adjustment vector is (dx × step size, dy × step size, dz × step size); The calculated adjustment vector is added to the original three-dimensional coordinates of each sensor node to obtain the final position node, that is, the final x coordinate = original x coordinate + dx × step length; the final y coordinate = original y coordinate + dy × step length; the final z coordinate = original z coordinate + dz × step length; The dynamic correction value is decomposed into three dynamic correction components, corresponding to the x-axis, y-axis and z-axis directions respectively, including: Collect a dataset containing sensor node locations and readings; Preprocess the data set and generate corresponding dx, dy, and dz components as training labels based on the position changes of sensor nodes in historical data; Build a convolutional neural network model and define the input and output layers. The input layer contains sensor readings and position coordinates. The output layer has three nodes, corresponding to the predicted values ​​of the three components dx, dy, and dz. Set up the hidden layer and determine the corresponding ReLU function. Divide the dataset into training, validation, and test sets; Use the training set to train the convolutional neural network model, and optimize the parameters of the convolutional neural network model through the back propagation algorithm and gradient descent algorithm to obtain the final convolutional neural network model; The reading and position coordinates of the current sensor node are input into the final convolutional neural network model, and the predicted dx, dy, and dz component values ​​are obtained through the forward propagation calculation of the final convolutional neural network model; Generate corresponding warning information based on the prediction results and preset thresholds, including: Obtain key indicators of simulation data from the prediction model in real time, including predicted values ​​of water flow dynamics, water quality distribution, meteorological conditions, and the growth status of aquaculture organisms; Based on historical data analysis, set corresponding warning thresholds for each key indicator; Compare key indicators with warning thresholds to determine whether each key indicator exceeds or approaches the warning threshold to obtain comparison results; Based on the comparison results, determine whether to trigger an early warning.

2. An intelligent monitoring and management method for aquaculture based on Internet of Things technology, characterized in that: The method is applied to the system according to claim 1, comprising: Identify key locations in the aquaculture environment and deploy multiple sensor nodes at key locations; Wirelessly connect to multiple sensor nodes to send farming data monitored by each sensor node; Receive and analyze the farming data sent by the communication module and determine a dynamic correction value; use the dynamic correction value to adjust the current sensor node position to obtain a final position node; Based on the final location node, the aquaculture data monitored by each sensor node is obtained. The aquaculture data includes monitored water quality parameters, meteorological parameters, and growth status parameters of the aquaculture organisms. The aquaculture data from each sensor node is integrated to form a comprehensive data set. Key features reflecting the status and changing trends of the aquaculture environment are extracted from the comprehensive data set. Based on the key features, a prediction model is constructed to simulate the water flow dynamics, water quality distribution, meteorological conditions, and growth status of the aquaculture organisms in the aquaculture pond. Generate corresponding warning information based on the prediction results and preset thresholds.

3. A computing device, characterized in that include: one or more processors; A storage device for storing one or more programs, which, when executed by the one or more processors, causes the one or more processors to implement the method according to claim 2.

4. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a program, which implements the method according to claim 2 when executed by a processor.

Images