An aquaculture water body regulation method, system, device and medium based on the Internet of Things

Abstract

This application discloses a method, system, equipment, and medium for aquaculture water body control based on the Internet of Things (IoT), relating to the field of aquaculture management technology. The method includes the following steps: real-time acquisition of time-series data of water quality parameters for each monitoring area; construction of a grey prediction model to predict the predicted values ​​of water quality parameters for each monitoring area within a preset time window; inputting the predicted water quality parameters into a pre-trained fuzzy neural network model to obtain the comprehensive ecological pressure index for each monitoring area; acquisition of the spatial adjacency relationship of each monitoring area and construction of a spatial adjacency matrix; identification of high-risk areas and their potential pollution propagation paths based on the comprehensive ecological pressure index and the spatial adjacency matrix; and generation of a control instruction set based on the identification results of high-risk areas and pollution propagation paths to control the corresponding aquaculture control equipment to perform corresponding actions. This application has the advantages of achieving accurate prediction of water quality change trends and reducing aquaculture risks and management costs.

Description

Technical Field

[0001] This application relates to the field of aquaculture management and control technology, and in particular to a method, system, equipment and medium for aquaculture water body control based on the Internet of Things. Background Technology

[0002] In aquaculture, water quality parameters (such as dissolved oxygen, pH, ammonia nitrogen concentration, nitrite concentration, etc.) directly affect the survival, growth and quality of farmed organisms. If abnormal water quality is not detected and controlled in time, it can easily lead to disease and death of farmed organisms, causing significant economic losses.

[0003] With the development of IoT technology, IoT sensors have been gradually applied to the monitoring of aquaculture water bodies. Existing technologies (such as patent document CN118228908A - an IoT-based aquaculture environment monitoring system) rely on monitoring the current state of exceeding standards, which is a post-event control. At this time, the deterioration of water quality may have already caused irreversible damage to the farmed organisms. The lack of accurate prediction of water quality change trends, accurate identification of risk areas, and analysis of pollution transmission paths leads to blind control measures, making it difficult to achieve dynamic, accurate, and efficient control of aquaculture water bodies, resulting in high aquaculture risks and management costs. Summary of the Invention

[0004] The main purpose of this application is to provide a method, system, equipment and medium for aquaculture water body control based on the Internet of Things, which aims to solve the technical problems of existing aquaculture water body control methods lacking accurate prediction of water quality change trends and having blind control measures.

[0005] To achieve the above objectives, this application provides a method for regulating aquaculture water bodies based on the Internet of Things, comprising the following steps:

[0006] The target aquaculture area is divided into multiple monitoring zones, and the time series data of water quality parameters for each monitoring zone are acquired in real time.

[0007] Based on time series data of water quality parameters, a grey prediction model is constructed to predict the predicted values ​​of water quality parameters for each monitoring area within a future preset time window.

[0008] The predicted water quality parameters are input into a pre-trained fuzzy neural network model to obtain the comprehensive ecological pressure index of each monitoring area.

[0009] Obtain the spatial adjacency relationships of each monitoring area and construct a spatial adjacency matrix;

[0010] Based on the comprehensive ecological pressure index and spatial adjacency matrix, high-risk areas and their potential pollution transmission routes are identified.

[0011] Based on the identification results of high-risk areas and pollution transmission paths, a set of control instructions is generated to control the corresponding aquaculture control equipment to perform corresponding actions; the set of control instructions includes control equipment identification, control parameters, and control priority.

[0012] Optionally, the step of constructing a grey prediction model based on water quality parameter time series data to predict the predicted values ​​of water quality parameters for each monitoring area within a future preset time window includes:

[0013] Collect single water quality parameters from each monitoring area at multiple consecutive sampling times to form a raw data sequence;

[0014] Perform an accumulation generation operation on the original data sequence to obtain the accumulated generation sequence;

[0015] Establish a first-order univariate differential equation for the cumulative generation sequence;

[0016] The development coefficient and grey action quantity of the first-order single-variable differential equation are solved by the least squares method.

[0017] By using the development coefficient and the grey action quantity, the time response function of the first-order univariate differential equation is solved to obtain the predicted values ​​of water quality parameters.

[0018] Optionally, the fuzzy neural network model includes an input layer, a fuzzification layer, a fuzzy rule layer, a normalization layer, and an output layer; wherein,

[0019] The nodes in the input layer correspond to the predicted values ​​of water quality parameters;

[0020] The fuzzification layer uses a Gaussian membership function to fuzzify the input parameters;

[0021] The fuzzy rule layer is used to perform fuzzy inference. Its fuzzy rule base encodes the nonlinear coupling relationship between different water quality parameters and their comprehensive impact weights on aquaculture ecology.

[0022] The normalization layer is used to normalize the output of the fuzzy rule layer;

[0023] The nodes in the output layer represent the comprehensive ecological pressure index.

[0024] Optionally, the input layer also includes macro-environmental correction parameters, which include atmospheric temperature, light radiation intensity, and chlorophyll a concentration retrieved from the target aquaculture water area by a meteorological satellite.

[0025] Optionally, the identification of high-risk areas and their potential pollution transmission pathways based on a comprehensive ecological pressure index and spatial adjacency matrix includes:

[0026] Monitoring areas whose comprehensive ecological pressure index exceeds a preset threshold are marked as initial high-risk areas;

[0027] Based on the spatial adjacency matrix and the preset spatial correlation model, all associated areas adjacent to the initial high-risk area are identified.

[0028] Associated areas with a spatial correlation greater than 0 with the initial high-risk areas are marked as secondary potential risk areas;

[0029] The direction from the initial high-risk area to the secondary potential risk area is determined as the pollution transmission path.

[0030] Optionally, the step of generating a control instruction set based on the identification results of high-risk areas and pollution transmission paths includes:

[0031] Based on the spatial correlation between the comprehensive ecological pressure index and pollution transmission pathways, the control priorities of each control area are obtained.

[0032] Based on the deviation between the predicted water quality parameters and the water quality standard values, the control parameters of the control equipment are obtained;

[0033] Integrate control equipment identification, control parameters, and control priorities to generate a control instruction set.

[0034] Optionally, the generation method of the control instruction set includes:

[0035] The optimal action combination determined by the current environmental state through a deep reinforcement learning model deployed in the cloud serves as the control instruction set. The deep reinforcement learning model uses the comprehensive ecological pressure index value, pollution propagation path and the availability of aquaculture control equipment as the state space, the combination of operating parameters of each aquaculture control equipment as the action space, and the weighted sum of the improvement of the ecological pressure index value and energy consumption cost as the reward function.

[0036] To achieve the above objectives, this application also provides an Internet of Things-based aquaculture water body control system, comprising:

[0037] The data acquisition module is used to divide the target aquaculture area into multiple monitoring zones and acquire the time series data of water quality parameters corresponding to each monitoring zone in real time.

[0038] The parameter prediction module is used to build a grey prediction model based on water quality parameter time series data to predict the predicted values ​​of water quality parameters for each monitoring area within a preset time window in the future.

[0039] The index acquisition module is used to input the predicted values ​​of water quality parameters into a pre-trained fuzzy neural network model to obtain the comprehensive ecological pressure index of each monitoring area.

[0040] The matrix construction module is used to obtain the spatial adjacency relationships of each monitoring area and construct a spatial adjacency matrix.

[0041] The path identification module is used to identify high-risk areas and their potential pollution transmission paths based on the comprehensive ecological pressure index and spatial adjacency matrix;

[0042] The control module is used to generate a control instruction set based on the identification results of high-risk areas and pollution transmission paths, so as to control the corresponding aquaculture control equipment to perform corresponding actions; the control instruction set includes control equipment identification, control parameters and start priority.

[0043] To achieve the above objectives, this application also provides a computer device, which includes a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the above-described method.

[0044] To achieve the above objectives, this application also provides a computer-readable storage medium storing a computer program, wherein a processor executes the computer program to implement the above-described method.

[0045] The beneficial effects that this application can achieve are as follows:

[0046] This application reconstructs the traditional passive "monitoring-feedback" control model into a proactive "prediction-intervention" control model. First, by introducing a grey prediction model, it solves the problem of untimely control caused by the lag in water quality parameter changes, achieving a leap from "treating existing problems" to "preventing future problems." Second, by using a fuzzy neural network model instead of a linear weighted model, it profoundly reveals the nonlinear coupling and synergistic toxicological effects among various water quality parameters, making the assessment results of the comprehensive ecological stress index closer to the actual physiological stress state of aquaculture organisms and avoiding the limitations of single-parameter threshold alarms. Third, by introducing a spatial adjacency matrix to analyze pollution propagation paths, it extends the assessment of isolated areas to spatial correlation analysis, enabling the prediction of pollution diffusion trends and generating spatially prioritized control instructions, effectively curbing the expansion of pollution range. In summary, this application significantly improves the intelligent level of aquaculture water body management from four dimensions: prediction accuracy, decision-making scientificity, control foresight, and system robustness, reducing aquaculture risks and management costs. Attached Figure Description

[0047] To more clearly illustrate the technical solutions in the specific embodiments of this application or the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn to scale.

[0048] Figure 1 This is a schematic flowchart of an IoT-based aquaculture water body control method in an embodiment of this application;

[0049] Figure 2 This is a schematic diagram of the framework structure of an Internet of Things-based aquaculture water body control system in an embodiment of this application.

[0050] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0051] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0052] It should be noted that if the embodiments of this application involve descriptions such as "first" or "second," these descriptions are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, features defined with "first" or "second" may explicitly or implicitly include at least one of those features. Furthermore, the technical solutions of the various embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. When the combination of technical solutions is contradictory or impossible to implement, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed in this application.

[0053] Example 1

[0054] Reference Figure 1 This embodiment provides a method for regulating aquaculture water bodies based on the Internet of Things, including the following steps:

[0055] Step S10: Divide the target aquaculture area into multiple monitoring zones and acquire the time series data of water quality parameters for each monitoring zone in real time;

[0056] Here, monitoring data can be collected by deploying IoT sensor nodes in each monitoring area, including dissolved oxygen sensors, pH sensors, ammonia nitrogen sensors, nitrite sensors, and water temperature sensors. At least one sensor node should be deployed in each monitoring area to ensure data coverage. The collected water quality parameter time series data is denoted as X(t)=[x1(t),x2(t),...,x...]. n (t)] T Where t is the sampling time, n is the type of water quality parameter (such as dissolved oxygen, pH value, etc.), and xn (t) represents the collected value of the nth water quality parameter at time t, where T represents matrix transpose.

[0057] Step S20: Based on the time series data of water quality parameters, construct a grey prediction model to predict the predicted values ​​of water quality parameters for each monitoring area within a preset time window in the future, specifically including:

[0058] Step S21: Collect single water quality parameters for each monitoring area at multiple consecutive sampling times to form a raw data sequence; let the raw data sequence of a certain type of water quality parameter in a certain monitoring area be... ,in To collect the number of samples, ( );

[0059] Step S22: Perform an accumulation generation operation on the original data sequence to obtain the accumulated generation sequence; in order to reduce the randomness of the original data and reveal its inherent pattern, the original data is... The formula for summation is:

[0060] ;

[0061] A new sequence is obtained The formula is:

[0062] ;

[0063] Step S23: Establish a first-order univariate differential equation for the cumulative generation sequence. The expression of the differential equation is as follows:

[0064] ;

[0065] in, The development coefficient reflects the development trend of the cumulatively generated sequence. If a < 0, the sequence exhibits exponential growth; if a > 0, the sequence exhibits exponential decay; the larger |a| is, the faster the rate of change. The gray action quantity reflects external disturbances or system driving terms, and is equivalent to a constant input.

[0066] Step S24: Solve the development coefficient and grey action of the first-order single-variable differential equation using the least squares method, assuming the parameters to be solved. The formula is:

[0067] ;

[0068] In the formula, B is the coefficient matrix and Y is the data vector. The formulas are as follows:

[0069] ;

[0070] ;

[0071] Step S25: Using the development coefficient and the grey action quantity, solve the time response function of the first-order single-variable differential equation to obtain the predicted values ​​of water quality parameters;

[0072] This step involves substituting the parameters of the development coefficient and the grey action quantity obtained from the solution into the differential equation to obtain the prediction function of the cumulative sequence, as shown in the following formula:

[0073] ;

[0074] When k=0, Ensure consistent initial conditions; when k≥1, obtain the cumulative predicted value for future time steps. ;

[0075] Then, the accumulated prediction sequence is subtracted and restored to obtain the final predicted values ​​of the original water quality parameters. The formula is:

[0076] ;

[0077] Using the above formula, the predicted values ​​of various water quality parameters for each monitoring area within a future preset time window (e.g., 1 hour, 2 hours) can be obtained, denoted as ,in For the future moment ( ).

[0078] In summary, this step, based on the construction of a grey prediction model, overcomes the limitations of existing technologies that can only achieve real-time monitoring but cannot predict water quality changes. It enables accurate prediction of water quality parameters within a preset time window, allowing for early detection of abnormal water quality trends and providing support for the early deployment of subsequent control measures, avoiding the passive situation of control only after water quality deteriorates. Furthermore, considering the characteristics of time series water quality parameters in aquaculture water bodies—"small sample size, nonlinearity, and strong randomness"—the grey prediction model eliminates the randomness of the original data by accumulating and generating sequences, improving prediction accuracy. Compared to traditional linear prediction models, it has stronger adaptability and smaller prediction errors. Secondly, the least squares method is used to solve the model parameters, ensuring the scientific validity and accuracy of the parameter solution, further improving the reliability of the prediction results. The prediction results cover various water quality parameters in each monitoring area, achieving accurate prediction by "region and parameter," providing targeted data support for differentiated control in each area and avoiding blind control. The prediction process can be completed automatically without human intervention, adapting to the intelligent needs of large-scale aquaculture and significantly reducing labor costs.

[0079] Step S30: Input the predicted water quality parameters into the pre-trained fuzzy neural network model to obtain the comprehensive ecological pressure index of each monitoring area;

[0080] In this step, a fuzzy neural network (FNN) is used, combining the fuzzy reasoning ability of fuzzy logic with the self-learning ability of neural networks to achieve the mapping of water quality parameters to the comprehensive ecological pressure index. The fuzzy neural network model includes an input layer, a fuzzification layer, a fuzzy rule layer, a normalization layer, and an output layer; among which,

[0081] The nodes in the input layer correspond to the predicted water quality parameters, meaning the input is the predicted water quality parameters obtained in step S20. ,common There are 1 input node, and each input node corresponds to a water quality parameter;

[0082] The fuzzification layer uses a Gaussian membership function to fuzzify the input parameters. The expression for the Gaussian membership function is:

[0083] ;

[0084] in, For the first The input parameter belongs to the first Membership degree of a fuzzy subset; For the first The input parameter of the first The center of a fuzzy subset; For the first The input parameter of the first The width of a fuzzy subset; , , For the first The number of fuzzy subsets of each input parameter;

[0085] The fuzzy rule layer is used to perform fuzzy inference. Its fuzzy rule base encodes the nonlinear coupling relationships between different water quality parameters and their comprehensive impact weights on aquaculture ecology. That is, each node corresponds to a fuzzy rule, and the rule output is the product of the membership degrees of each input, as shown in the formula:

[0086] ;

[0087] in, For the first Output of the fuzzy rules; K represents the total number of fuzzy rules, which is obtained by multiplying the number of fuzzy subsets of all water quality parameters, i.e. ; For the first The input parameter in the first... The corresponding fuzzy subset index in the rule;

[0088] The normalization layer is used to normalize the output of the fuzzy rule layer. The formula is:

[0089] ;

[0090] in, For the first The normalized value output by each rule;

[0091] The output layer nodes output the comprehensive ecological pressure index, specifically the comprehensive ecological pressure index P, using a linear output function, with the following formula:

[0092] ;

[0093] in, For the first The weights corresponding to the fuzzy rules are obtained through model pre-training; , The closer the value is to 1, the greater the ecological pressure and the higher the water quality risk in the monitored area.

[0094] In summary, this embodiment addresses the limitations of assessing water quality risk using a single water quality parameter. By employing a fuzzy neural network model, it integrates the influence of multiple water quality parameters (dissolved oxygen, pH, ammonia nitrogen, etc.) to generate a comprehensive ecological pressure index. This enables a comprehensive and objective assessment of water quality risk in each monitoring area, avoiding missed risk assessments due to a single parameter being normal while others are abnormal. Combining the fuzzy reasoning capabilities of fuzzy logic with the self-learning capabilities of neural networks, it can both align with practical aquaculture experience through fuzzy rules (such as the fuzzy classification of water quality parameters as "high, medium, and low") and adapt to the water quality characteristics of different aquaculture scenarios through self-learning capabilities, thus improving the accuracy and adaptability of risk assessment. The use of Gaussian membership functions for fuzzification of water quality parameters accurately characterizes their gradual changes, avoiding assessment biases caused by traditional "black-and-white" threshold judgments. The comprehensive ecological pressure index adopts... Interval quantification intuitively reflects the water quality risk level of each monitoring area, making it easier for staff to quickly identify high-risk areas and providing a quantitative basis for setting subsequent control priorities. After model pre-training, it can achieve automated calculation with high computational efficiency, adapt to real-time control needs, and complete the risk assessment of each area without manual intervention.

[0095] As an optional implementation, the input layer also includes macro-environmental correction parameters, which include atmospheric temperature, light radiation intensity, and chlorophyll a concentration retrieved from the target aquaculture water area by air meteorological satellites.

[0096] In this embodiment, the current scheme only monitors local water parameters. This embodiment accesses publicly available meteorological satellite (such as the Fengyun series) and remote sensing satellite (such as Landsat) data. Therefore, this embodiment extracts macro-environmental factors such as atmospheric temperature, light radiation intensity, and chlorophyll a concentration (indicating the risk of algal blooms) above the aquaculture area. These macro-environmental factors are used as additional input nodes (i.e., adding 3 inputs) to the fuzzy neural network model to correct the prediction results of local monitoring data. For example, predicted high temperatures will further amplify the weight of rising water temperature on ecological pressure. Therefore, this embodiment combines macro-meteorological disaster early warning with local micro-environment monitoring, solving the shortcomings of existing technologies that can only perceive local areas and cannot predict sudden environmental risks driven by macro-climate. It increases the prevention rate of aquaculture disasters under extreme weather conditions from less than 60% to over 95%, further improving the accuracy of preventive control of aquaculture water bodies.

[0097] Step S40: Obtain the spatial adjacency relationship of each monitoring area and construct a spatial adjacency matrix;

[0098] The target aquaculture area is divided into: There are several monitoring areas, numbered as follows: , build Spatial adjacency matrix Matrix elements The definition is as follows:

[0099] ;

[0100] in, Adjacent matrix It is a symmetric matrix, that is .

[0101] This step quantifies the spatial adjacency relationships of each monitoring area into a spatial adjacency matrix, achieving a digital and standardized expression of spatial relationships and providing a standardized computational foundation for subsequent pollution propagation path analysis. The matrix elements adopt a "0-1" binary definition, which is concise and clear, facilitating subsequent calculations of spatial correlation and accurately reflecting the adjacency relationships between areas (whether there are common boundaries), avoiding ambiguity in the description of spatial relationships. The adjacency matrix is ​​a symmetric matrix, conforming to the symmetry of spatial adjacency relationships (if area i is adjacent to area j, then area j is adjacent to area i), ensuring the rationality and scientific nature of the data. The construction process can be combined with the actual spatial distribution of aquaculture waters, flexibly adapting to aquaculture waters of different shapes and sizes, and has strong versatility. Through the spatial adjacency matrix, the abstract relationship of "spatial adjacency" is transformed into a computable matrix form, providing core support for subsequent analysis of pollution propagation paths in conjunction with the comprehensive ecological pressure index, and solving the problem that existing technologies cannot quantify spatial correlations and are difficult to analyze pollution diffusion trends.

[0102] Step S50: Based on the comprehensive ecological pressure index and spatial adjacency matrix, identify high-risk areas and their potential pollution transmission pathways, including:

[0103] Step S51: Mark the monitoring areas where the comprehensive ecological pressure index exceeds the preset threshold as initial high-risk areas;

[0104] Step S52: Based on the spatial adjacency matrix and the preset spatial correlation model, identify all associated areas adjacent to the initial high-risk area;

[0105] Step S53: Mark the associated areas with a spatial correlation greater than 0 with the initial high-risk areas as secondary potential risk areas;

[0106] Step S54: The direction from the initial high-risk area to the secondary potential risk area is determined as the pollution transmission path.

[0107] In this step, a preset threshold is set as follows: ( (This can be adjusted according to the actual situation such as the species being farmed and the farming environment). When the comprehensive ecological pressure index of a certain monitoring area... When this happens, the area can be determined as a high-risk area, and the set of high-risk areas is denoted as [missing information]. , This represents the comprehensive ecological pressure index of the i-th high-risk area, with a value range of [0,1], where the closer to 1, the higher the risk; then, based on the spatial adjacency matrix... Based on the comprehensive ecological pressure index, a spatial correlation model is used to calculate the spatial correlation of each area to identify pollution transmission paths. The expression for the spatial correlation model is as follows:

[0108] ;

[0109] in, Indicates spatial correlation, reflecting high-risk areas The degree of correlation with the ecological pressure of adjacent area j; P j This represents the comprehensive ecological pressure index of the j-th monitoring area adjacent to the i-th high-risk area. This represents the average of the comprehensive ecological pressure index for all areas. The standard deviation of the comprehensive ecological stress index;

[0110] when At that time, it indicates a high-risk area. The ecological pressure of adjacent area j is positively correlated, meaning that adjacent areas of a high-risk area are easily polluted, forming a pollution transmission path. In this case, the corresponding adjacent area j is marked as a secondary potential risk area. The direction of the pollution transmission path is from the high-risk area to the secondary potential risk area. and Then there is the possibility of coming from high-risk areas. To secondary potential risk areas Potential pollution transmission routes.

[0111] This step enables the accurate identification of high-risk areas by setting an adjustable comprehensive ecological pressure index threshold. This system adapts to the risk assessment needs of different aquaculture species (such as freshwater fish and shrimp) and different aquaculture environments, avoiding a one-size-fits-all approach to risk assessment and improving the targeting of identification. It overcomes the limitations of existing technologies that can only identify high-risk areas but cannot analyze pollution transmission paths. Through spatial correlation calculations, it accurately identifies potential pollution transmission paths in high-risk areas, clarifying the direction and scope of pollution spread, providing support for subsequent "source control + diffusion interception" control strategies. The spatial correlation model, combined with the average and standard deviation of the comprehensive ecological pressure index, can accurately quantify the degree of ecological pressure correlation between areas, avoiding the one-sidedness of judging transmission paths solely based on adjacency relationships. It clarifies the direction of pollution transmission paths (high ecological pressure → low ecological pressure), providing a basis for subsequent control priority division, allowing for priority control of high-risk source areas and key areas along transmission paths, improving control efficiency. The identification process is automated, quickly outputting sets of high-risk areas and pollution transmission paths, facilitating staff to quickly grasp the distribution and diffusion trends of water quality risks, and taking timely targeted measures to prevent the expansion of pollution.

[0112] Step S60: Based on the identification results of high-risk areas and pollution transmission paths, generate a set of control instructions to control the corresponding aquaculture control equipment to perform corresponding actions; wherein, the set of control instructions includes control equipment identification, control parameters and control priority.

[0113] As an optional implementation, the step of generating a control instruction set based on the identification results of high-risk areas and pollution transmission paths includes:

[0114] Based on the spatial correlation between the comprehensive ecological pressure index and pollution transmission pathways, the control priority of each control area is obtained; the expression for the control priority is:

[0115] ;

[0116] in, Indicates the priority of regulation. , These are the first weighting coefficient and the second weighting coefficient, respectively. Adjustments can be made based on actual regulatory needs (e.g., prioritizing high-risk areas). ); , The larger the value, the higher the priority of regulation.

[0117] Based on the predicted values ​​of water quality parameters (i.e.) ) and water quality standard value (denoted as The deviation of the control parameters is used to obtain the control parameters of the control equipment; taking dissolved oxygen control as an example, the control parameters (such as the speed of the aerator) are... The formula is:

[0118] ;

[0119] in, , These are the maximum and minimum operating parameters of the control equipment, respectively. This is the minimum allowable value for dissolved oxygen; the calculation methods for other water quality parameters (such as pH and ammonia nitrogen) are similar, and the formulas can be adjusted according to the characteristics of the corresponding control equipment.

[0120] Integrate and control equipment identifiers (such as aerator ID, water exchanger ID), control parameters (such as speed, water exchange rate), and control priority. Generate a set of control instructions, denoted as ,in In order to control the number of equipment, , For the first The identification of each control device Adjust its parameters, Set its startup priority.

[0121] Then, the control command set is received through the IoT gateway. According to the control equipment label The corresponding control instructions (including control parameters) will be sent. The instructions are sent to various aquaculture control equipment (such as aerators, water exchange equipment, disinfection equipment, etc.); after receiving the instructions, the equipment starts operation according to the control parameters to achieve precise control of the aquaculture water; at the same time, the IoT sensor nodes collect the water quality parameters after control in real time to form a closed-loop feedback. If the water quality parameters do not meet the standards, steps S20-S60 are repeated until the water quality meets the standards.

[0122] In this implementation, a quantitative division of control priorities is achieved. Through a control priority formula, the ecological pressure and pollution propagation correlation of high-risk areas are comprehensively considered to avoid disordered control and ensure that high-priority areas (high-risk sources and key nodes in the transmission path) receive priority control, thereby improving the timeliness of control. Weighting coefficients are used. , It can be flexibly adjusted to adapt to different control needs (such as prioritizing pollution control or high-risk area treatment), and has strong versatility. The control parameters are calculated based on the deviation between the predicted water quality value and the standard value, realizing "on-demand control" and avoiding resource waste (such as excessive oxygenation) or insufficient control caused by traditional fixed parameter control, thus improving the accuracy and energy efficiency of control. The control instruction set integrates three core pieces of information: equipment identification, control parameters, and start-up priority, clarifying "who controls, how to control, and when to control," providing a clear and standardized basis for subsequent instruction issuance and equipment execution, and avoiding control chaos. The instruction set generation process is automated. Combined with the data analysis results of the previous steps, it can generate targeted control instructions without manual intervention, adapting to the control needs of intelligent and large-scale aquaculture and greatly improving control efficiency.

[0123] As an optional implementation method, the generation of the control instruction set includes:

[0124] The optimal action combination determined by the current environmental state through a deep reinforcement learning model deployed in the cloud serves as the control instruction set. The deep reinforcement learning model uses the comprehensive ecological pressure index value, pollution propagation path and the availability of aquaculture control equipment as the state space, the combination of operating parameters of each aquaculture control equipment as the action space, and the weighted sum of the improvement of the ecological pressure index value and energy consumption cost as the reward function.

[0125] In this implementation, a deep reinforcement learning model (DQN) is introduced. The environmental state is defined as "the current comprehensive ecological pressure index, pollution propagation path, and equipment status of each area," the action is "selecting a set of control parameters (power and start / stop of each device)," and the reward is "the decrease in the ecological pressure index and the energy cost of control over a period of time after control." The DQN model is trained using historical interaction data, enabling it to autonomously learn the optimal and most energy-efficient control strategies for different scenarios and continuously evolve. For example, in a simulated environment, the state is defined as: pressure index [0.6, 0.9], propagation path [none, upstream, downstream], action as: aerator power [0, 30%, 60%, 100%], pump flow rate [0, 10, 20 m³ / h], and reward function R = αΔpressure index - βΔenergy consumption. After 10,000 iterations of training, the DQN learns a non-intuitive strategy: when there is a slight warning but the pollution source is clearly downstream, instead of immediately starting upstream aeration, it first starts a low-flow pump downstream to form a "water curtain isolation." This strategy is 37% more energy-efficient than the traditional "treat the symptom, not the cause" approach, and improves control effectiveness by 22%. Compared to existing technologies that use fixed rule bases and cannot cope with complex and dynamic scenarios, this embodiment solves the bottleneck of existing technologies' rigid rules and inability to adapt to complex and ever-changing aquaculture environments. It achieves self-evolution and continuous optimization of the control strategy, minimizing the overall energy consumption of the system while ensuring ecological safety, and truly realizing intelligent and economical precise control.

[0126] In summary, this application has the following advantages over the prior art:

[0127] First, it achieves a fundamental shift from passive response to proactive prediction. Existing technologies rely on monitoring the current state of exceeding standards, which is a reactive measure. By this time, water quality deterioration may have already caused irreversible damage to aquaculture organisms. Step S20 of this method constructs a grey prediction model and uses historical data to extrapolate the trend of water quality parameter changes in the near future. This allows the control process to be initiated when water quality parameters are about to exceed standards but have not yet, providing proactive protection for aquaculture organisms and significantly reducing stress response and mortality.

[0128] Second, a scientific and comprehensive decision-making and assessment mechanism has been established. Existing technologies use a linear weighting method to calculate the pollution index, which ignores the complex interactions between various physicochemical factors in the aquatic environment. For example, at the same ammonia nitrogen concentration, high water temperature or low dissolved oxygen can exponentially increase the toxicity of ammonia nitrogen. In step S30 of this method, a pre-trained fuzzy neural network model is used. This network learns from a large amount of historical data or expert knowledge, enabling it to automatically encode and simulate this nonlinear coupling relationship, thereby outputting a more scientific "comprehensive ecological pressure index" that is closer to the actual ecological risk, effectively avoiding misjudgment or omission.

[0129] Third, this method introduces dynamic control priorities based on spatial dimensions. Existing technologies assess each area in isolation, lacking a holistic understanding of the pollution diffusion trend. Steps S40 and S50 of this method, by constructing a spatial adjacency matrix and analyzing pollution propagation paths, can identify "pollution source" areas and "high-risk receptor" areas. The resulting control instruction set (step S60) is no longer indiscriminate global control, but rather dynamically allocates control resources based on the direction of pollution propagation (e.g., prioritizing the activation of oxygenation or water exchange equipment downstream to form an isolation zone), achieving precise and efficient targeted intervention and significantly saving energy and water resources.

[0130] Example 2

[0131] Reference Figure 2 Based on the same inventive concept as the foregoing embodiments, this embodiment also provides an Internet of Things-based aquaculture water body control system, including:

[0132] The data acquisition module is used to divide the target aquaculture area into multiple monitoring zones and acquire the time series data of water quality parameters corresponding to each monitoring zone in real time.

[0133] The parameter prediction module is used to build a grey prediction model based on water quality parameter time series data to predict the predicted values ​​of water quality parameters for each monitoring area within a preset time window in the future.

[0134] The index acquisition module is used to input the predicted values ​​of water quality parameters into a pre-trained fuzzy neural network model to obtain the comprehensive ecological pressure index of each monitoring area.

[0135] The matrix construction module is used to obtain the spatial adjacency relationships of each monitoring area and construct a spatial adjacency matrix.

[0136] The path identification module is used to identify high-risk areas and their potential pollution transmission paths based on the comprehensive ecological pressure index and spatial adjacency matrix;

[0137] The control module is used to generate a control instruction set based on the identification results of high-risk areas and the pollution transmission path, so as to control the corresponding aquaculture control equipment to perform corresponding actions; the control instruction set includes the control equipment identifier, control parameters and start priority.

[0138] The explanations and examples of the modules in this embodiment can be found in the methods of the foregoing embodiments, and will not be repeated here.

[0139] Example 3

[0140] Based on the same inventive concept as the foregoing embodiments, this embodiment provides a computer device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the above-described method.

[0141] Example 4

[0142] Based on the same inventive concept as the foregoing embodiments, this embodiment provides a computer-readable storage medium storing a computer program, and a processor executes the computer program to implement the above-described method.

[0143] In some embodiments, the computer-readable storage medium may be a memory such as FRAM, ROM, PROM, EPROM, EEPROM, flash memory, magnetic surface memory, optical disk, or CD-ROM; or it may be a variety of devices including one or any combination of the above-mentioned memories.

[0144] The above are merely preferred embodiments of this application and do not limit the patent scope of this application. Any equivalent structural or procedural transformations made using the content of this application's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.

Claims

1. A method for regulating aquaculture water bodies based on the Internet of Things, characterized in that, Includes the following steps: The target aquaculture area is divided into multiple monitoring zones, and the time series data of water quality parameters for each monitoring zone are acquired in real time. Based on time series data of water quality parameters, a grey prediction model is constructed to predict the predicted values ​​of water quality parameters for each monitoring area within a future preset time window. The predicted water quality parameters are input into a pre-trained fuzzy neural network model to obtain the comprehensive ecological pressure index of each monitoring area. Obtain the spatial adjacency relationships of each monitoring area and construct a spatial adjacency matrix; Based on the comprehensive ecological pressure index and spatial adjacency matrix, high-risk areas and their potential pollution transmission routes are identified. Based on the identification results of high-risk areas and pollution transmission paths, a set of control instructions is generated to control the corresponding aquaculture control equipment to perform corresponding actions; the set of control instructions includes control equipment identification, control parameters, and control priority.

2. The method for regulating aquaculture water bodies based on the Internet of Things as described in claim 1, characterized in that, The grey prediction model, constructed based on water quality parameter time series data, is used to predict the water quality parameter values ​​for each monitoring area within a preset time window in the future, including: Collect single water quality parameters from each monitoring area at multiple consecutive sampling times to form a raw data sequence; Perform an accumulation generation operation on the original data sequence to obtain the accumulated generation sequence; Establish a first-order univariate differential equation for the cumulative generation sequence; The development coefficient and grey action quantity of the first-order single-variable differential equation are solved by the least squares method. By using the development coefficient and the grey action quantity, the time response function of the first-order univariate differential equation is solved to obtain the predicted values ​​of water quality parameters.

3. The method for regulating aquaculture water bodies based on the Internet of Things as described in claim 1, characterized in that, The fuzzy neural network model includes an input layer, a fuzzification layer, a fuzzy rule layer, a normalization layer, and an output layer; among them, The nodes in the input layer correspond to the predicted values ​​of water quality parameters; The fuzzification layer uses a Gaussian membership function to fuzzify the input parameters; The fuzzy rule layer is used to perform fuzzy inference. Its fuzzy rule base encodes the nonlinear coupling relationship between different water quality parameters and their comprehensive impact weights on aquaculture ecology. The normalization layer is used to normalize the output of the fuzzy rule layer; The nodes in the output layer represent the comprehensive ecological pressure index.

4. The method for regulating aquaculture water bodies based on the Internet of Things as described in claim 3, characterized in that, The input layer also includes macro-environmental correction parameters, which include atmospheric temperature, light radiation intensity, and chlorophyll a concentration retrieved from the target aquaculture water area by atmospheric meteorological satellites.

5. The method for regulating aquaculture water bodies based on the Internet of Things as described in claim 1, characterized in that, The identification of high-risk areas and their potential pollution transmission pathways based on a comprehensive ecological pressure index and spatial adjacency matrix includes: Monitoring areas whose comprehensive ecological pressure index exceeds a preset threshold are marked as initial high-risk areas; Based on the spatial adjacency matrix and the preset spatial correlation model, all associated areas adjacent to the initial high-risk area are identified. Associated areas with a spatial correlation greater than 0 with the initial high-risk areas are marked as secondary potential risk areas; The direction from the initial high-risk area to the secondary potential risk area is determined as the pollution transmission path.

6. The method for regulating aquaculture water bodies based on the Internet of Things as described in claim 5, characterized in that, The process of generating a control instruction set based on the identification results of high-risk areas and pollution transmission paths includes: Based on the spatial correlation between the comprehensive ecological pressure index and pollution transmission pathways, the control priorities of each control area are obtained. Based on the deviation between the predicted water quality parameters and the water quality standard values, the control parameters of the control equipment are obtained; Integrate control equipment identification, control parameters, and control priorities to generate a control instruction set.

7. The method for regulating aquaculture water bodies based on the Internet of Things as described in claim 6, characterized in that, The generation methods of the control instruction set include: The optimal action combination determined by the current environmental state through a deep reinforcement learning model deployed in the cloud serves as the control instruction set. The deep reinforcement learning model uses the comprehensive ecological pressure index value, pollution propagation path and the availability of aquaculture control equipment as the state space, the combination of operating parameters of each aquaculture control equipment as the action space, and the weighted sum of the improvement of the ecological pressure index value and energy consumption cost as the reward function.

8. An Internet of Things-based aquaculture water body control system, characterized in that, include: The data acquisition module is used to divide the target aquaculture area into multiple monitoring zones and acquire the time series data of water quality parameters corresponding to each monitoring zone in real time. The parameter prediction module is used to build a grey prediction model based on water quality parameter time series data to predict the predicted values ​​of water quality parameters for each monitoring area within a preset time window in the future. The index acquisition module is used to input the predicted values ​​of water quality parameters into a pre-trained fuzzy neural network model to obtain the comprehensive ecological pressure index of each monitoring area. The matrix construction module is used to obtain the spatial adjacency relationships of each monitoring area and construct a spatial adjacency matrix. The path identification module is used to identify high-risk areas and their potential pollution transmission paths based on the comprehensive ecological pressure index and spatial adjacency matrix; The control module is used to generate a control instruction set based on the identification results of high-risk areas and pollution transmission paths, so as to control the corresponding aquaculture control equipment to perform corresponding actions; the control instruction set includes control equipment identification, control parameters and start priority.

9. A computer device, characterized in that, The computer device includes a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement an Internet of Things-based aquaculture water body control method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, and the processor executes the computer program to implement the Internet of Things-based aquaculture water body control method as described in any one of claims 1-7.

Images