Aquaculture water quality monitoring and self-adaptive regulation and control platform and method

Abstract

The invention relates to the technical field of water quality monitoring, in particular to an aquaculture water quality monitoring and self-adaptive regulation and control platform and method.The aquaculture water quality monitoring and self-adaptive regulation and control platform comprises a data processing module used for obtaining water quality parameter data of aquaculture at multiple monitoring points and multiple time points and conducting preprocessing and time sequence serialization fitting on the water quality parameter data to obtain a data processing module; obtaining a water quality parameter time sequence; the feature extraction module is used for carrying out space-time semantic recognition and feature extraction on the water quality parameter time sequence and constructing a feature sequence of each water quality parameter; and the risk early warning module is used for carrying out inter-parameter space-time correlation calculation and dynamic risk threshold value adaptation according to the feature sequence to obtain water quality risk early warning features. By combining the water quality parameters and the culture organism state data, the system can accurately evaluate the risk level of the water quality, timely warn potential risk factors, help a culturist to take effective precautionary measures, and avoid the death or health problem of the culture organism.

Description

Technical Field

[0001] This invention relates to the field of water quality monitoring technology, specifically to a platform and method for monitoring and adaptively controlling water quality in aquaculture. Background Technology

[0002] Aquaculture is an agricultural activity that artificially raises aquatic animals or plants to produce aquatic products for human consumption or other uses. It typically includes the cultivation of aquatic organisms such as fish, shellfish, shrimp, and algae. Unlike traditional fishing, aquaculture aims to achieve economic benefits by cultivating and reproducing aquatic organisms in a controlled environment to ensure their growth, reproduction, and health management.

[0003] Currently, many traditional systems rely on only a few monitoring points and have a low data collection frequency; this makes it impossible for the system to capture rapid changes in water quality or potential risks in real time; and traditional systems often rely on fixed control strategies, which cannot flexibly respond to changes in water quality or fluctuations in the aquaculture environment.

[0004] In addition, most traditional systems only provide data monitoring and lack sufficient risk warning mechanisms; they cannot issue timely alarms when abnormal changes occur in water quality and lack targeted preventive measures; and traditional systems usually ignore the ecological relationship between water quality and aquaculture organisms, resulting in resource waste or low aquaculture efficiency. Summary of the Invention

[0005] To achieve the above objectives, the present invention provides the following technical solution: an aquaculture water quality monitoring and adaptive control platform, comprising: The data processing module is used to acquire water quality parameter data of aquaculture at multiple monitoring points and multiple time points, and to preprocess and time-series sequence fitting the water quality parameter data to obtain a water quality parameter time-series sequence. The feature extraction module is used to perform spatiotemporal semantic recognition and feature extraction on the time series of water quality parameters, and to construct a feature sequence for each water quality parameter; The risk warning module is used to calculate the spatiotemporal correlation between parameters and adapt the dynamic risk threshold based on the feature sequence to obtain water quality risk warning features; The relationship parsing module is used to acquire aquaculture organism status data, combine the water quality risk warning features and the aquaculture organism status data to construct a water quality-organism-ecology coupling model, and parse the niche correlation relationship; The benefit assessment module is used to predict the time-series evolution based on the water quality-biology-ecology coupling model, obtain multi-scenario control schemes, evaluate the benefits of the multi-scenario control schemes, and obtain optimized control strategies. The strategy control module is used to drive the actuator to perform adaptive control according to the optimized control strategy, and feed back the controlled water quality data to the water quality parameter time series to form a closed-loop control.

[0006] Preferably, the water quality parameters include water temperature, dissolved oxygen, ammonia nitrogen concentration, nitrite concentration, pH value, transparency, and plankton community density; The water quality parameter data is preprocessed and time-series sequenced for fitting to obtain a water quality parameter time-series sequence, including: Obtain raw data packets of water quality parameters within the monitoring area, and clean and standardize the raw data packets of water quality parameters to obtain preprocessed data packets of water quality parameters. Dynamic and static features are extracted from the water quality parameter preprocessing data package. The dynamic features are water quality parameter features that change with monitoring time, and the static features are water quality parameter inherent attribute features that do not change with monitoring time. The dynamic features and the static features are combined to obtain a water quality parameter feature set. The time-series segmentation threshold is determined based on the time correlation of the water quality parameter feature set, and the water quality parameter feature set is divided into multiple water quality parameter time-series segments according to the time-series segmentation threshold. The trend features of each water quality parameter time series segment are extracted, and a time series fitting model is constructed based on the trend features. The time series fitting model is used to fit each water quality parameter time series segment to obtain the initial time series sequence of water quality parameters. The fitting error of the initial time series sequence of the water quality parameters is calculated, the parameters of the time series fitting model are adjusted according to the fitting error, and the initial time series sequence of the water quality parameters is corrected by the adjusted time series fitting model to obtain the optimized time series sequence of the water quality parameters. Determine the time alignment parameter of the optimized water quality parameter time series, and integrate all the optimized water quality parameter time series according to the time alignment parameter to obtain the water quality parameter time series.

[0007] Preferably, spatiotemporal semantic recognition and feature extraction are performed on the time-series sequence of the water quality parameters to construct a feature sequence for each water quality parameter, including: Spatiotemporal semantic recognition is performed on the time series of water quality parameters to identify the parameter distribution characteristics at different monitoring points and the change patterns over different time periods; The spatiotemporal semantic features of each water quality parameter are extracted in multiple dimensions to obtain multidimensional features; wherein, the spatiotemporal semantic features include temporal features and spatial features, the temporal features include parameter change trends, periodic fluctuation amplitude and mutation frequency, and the spatial features include spatial gradient difference, regional mean deviation and correlation degree between adjacent monitoring points; Anomaly pattern detection is performed on the multi-dimensional features, and anomalous feature fragments are marked; the temporal features, spatial features, and anomalous feature fragments are integrated to construct a feature sequence for each water quality parameter.

[0008] Preferably, based on the feature sequence, spatiotemporal correlation calculations between parameters and dynamic risk threshold adaptation are performed to obtain water quality risk early warning features, including: The feature sequences are subjected to pairwise spatiotemporal correlation calculations to obtain a parameter correlation matrix; Based on the parameter correlation matrix, the risk threshold is dynamically adapted to obtain a dynamic risk threshold table; The feature sequence is compared with the dynamic risk threshold table in real time to classify risk levels and mark risk trigger parameters. By fusing the parameter correlation matrix, the dynamic risk threshold table, and the risk level, water quality risk early warning characteristics are obtained.

[0009] Preferably, by combining the water quality risk warning characteristics and the aquaculture organism status data, a water quality-biology-ecology coupling model is constructed to analyze niche relationships, including: The acquired water quality risk warning characteristics and aquaculture organism status data are preprocessed to establish a spatiotemporal correspondence between the two. Based on the pretreated water quality risk warning characteristics and aquaculture organism status data, a water quality-biology-ecology coupling model is constructed. By analyzing the associated variables output by the water quality-biology-ecology coupling model, the niche relationships between cultured organisms and the water quality environment and ecosystem in the target aquaculture area are analyzed. Based on the analysis results of the niche association, the water quality-biology-ecology coupling model was verified and optimized to obtain a coupling model and niche association map suitable for the target aquaculture water area.

[0010] Preferably, the acquired water quality risk warning characteristics and aquaculture organism status data are preprocessed to establish a spatiotemporal correspondence between the two, including: The water quality risk warning features are classified and labeled according to monitoring sub-regions, monitoring levels, and collection times, and the aquaculture biological status data are classified and labeled according to biological sampling points, sampling times, and biological species. Numerical normalization was performed on the water quality risk warning characteristics and aquaculture organism status data after classification and labeling, and abnormal data exceeding the reasonable range were removed; Based on the spatial correlation between the monitoring sub-region and the biological sampling point, and the temporal synchronization between the water quality parameter collection time and the biological sample collection time, a corresponding mapping relationship between the water quality risk warning features and the aquaculture biological status data under the same spatiotemporal dimension is established to form a spatiotemporal matching dataset.

[0011] Preferably, based on the pretreated water quality risk warning characteristics and aquaculture organism status data, a water quality-biology-ecology coupling model is constructed, including: The water quality risk warning features are extracted from the spatiotemporal matching dataset as input layer variables of the model, the aquaculture organism status data are extracted as intermediate layer variables of the model, and the water temperature, light duration, and water flow velocity of the target aquaculture area are introduced as ecological regulation variables. Establish a correlation function between the input layer variables and the intermediate layer variables to quantify the impact weight of changes in water quality risk warning characteristics on the aquaculture organism status data; The ecological regulation variables are incorporated as feedback factors into the correlation function to construct a coupled model structure, which includes a water quality-biology direct correlation module, an ecology-water quality feedback module, and an ecology-biology regulation module. The interaction parameters of each module are set to complete the construction of the water quality-biology-ecology coupled model.

[0012] Preferably, based on the water quality-biology-ecology coupling model, time-series evolution prediction is performed to obtain multi-scenario control schemes. The effectiveness of these multi-scenario control schemes is then evaluated to obtain optimized control strategies, including: Based on the water quality-biology-ecology coupling model, the evolution trend of water quality parameters and changes in biological state under different environmental conditions within a preset time period are simulated to obtain a time-series evolution prediction curve. Based on the time-series evolution prediction curve, a multi-scenario control scheme is generated, wherein the control scheme includes aerator power adjustment, water exchange flow control, feed amount optimization, and probiotic addition amount adaptation. The aforementioned control schemes are evaluated for their benefits, including ecological benefits, economic benefits, and operating costs. The scheme with the best ecological and economic benefits is selected to obtain the optimized control strategy.

[0013] Preferably, the actuator is driven to perform adaptive control according to the optimized control strategy, and the controlled water quality data is fed back to the water quality parameter time series to form a closed-loop control, including: Based on the optimized control strategy, an execution instruction is generated to drive the corresponding actuator to perform the control operation; After the control operation is executed, water quality parameter data of the controlled area are collected once every preset collection cycle to obtain the water quality data after control. The adjusted water quality data is fed back to the water quality parameter time series, the sequence data is updated, and the time series sequence fitting is retried. Based on the updated water quality parameter time series, the operating parameters of the actuator are dynamically fine-tuned to form a closed-loop control.

[0014] A method for monitoring and adaptively controlling aquaculture water quality, applicable to the aforementioned aquaculture water quality monitoring and adaptive control platform, comprising: Water quality parameter data of aquaculture at multiple monitoring points and multiple time points are acquired, and the water quality parameter data are preprocessed and time-series sequenced for fitting to obtain a water quality parameter time series. Spatiotemporal semantic recognition and feature extraction are performed on the time series of water quality parameters to construct a feature sequence for each water quality parameter; Based on the feature sequence, the spatiotemporal correlation between parameters is calculated and dynamic risk threshold is adapted to obtain water quality risk early warning features; Acquire aquaculture organism status data, combine the water quality risk warning characteristics with the aquaculture organism status data to construct a water quality-biology-ecology coupling model, and analyze the niche correlation; Based on the water quality-biology-ecology coupling model, time-series evolution prediction is performed to obtain multi-scenario control schemes. The benefits of the multi-scenario control schemes are evaluated to obtain optimized control strategies. The optimized control strategy drives the actuator to perform adaptive control, and the controlled water quality data is fed back to the water quality parameter time series to form a closed-loop control.

[0015] Compared with the prior art, the beneficial effects of the present invention are: This invention acquires water quality parameter data from multiple monitoring points and time points, and processes it through time-series serialization fitting, enabling real-time monitoring of water quality changes in aquaculture. The system can perform precise adaptive regulation based on this data, improving the real-time performance and accuracy of water quality management and avoiding aquaculture risks caused by water quality changes. This invention utilizes multi-dimensional spatiotemporal semantic recognition and feature extraction. The platform can not only identify the temporal and spatial features of various water quality parameters, but also identify abnormal feature fragments, thereby providing a scientific basis for further risk warning and decision-making. Furthermore, by combining water quality parameters with aquaculture organism status data, the system can accurately assess the risk level of water quality, provide timely warnings of potential risk factors, and help aquaculture farmers take effective preventive measures to avoid the death or health problems of aquaculture organisms. This invention constructs a water quality-biology-ecology coupling model. The platform can analyze the interaction between water quality changes and the growth status of cultured organisms, optimize niche relationships, help achieve a balanced aquaculture environment, improve aquaculture efficiency, and reduce resource waste. Furthermore, the platform generates multi-scenario control schemes based on time-series evolution prediction curves, which can evaluate the benefits of control schemes under different environmental conditions and select the best control strategy based on ecological benefits, economic benefits, and operating costs, thereby optimizing resource utilization and aquaculture efficiency. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the system architecture of the control platform in one embodiment of the present invention; Figure 2 This is a schematic flowchart of the overall method in one embodiment of the present invention.

[0017] In the diagram: 1. Data processing module; 2. Feature extraction module; 3. Risk warning module; 4. Relationship analysis module; 5. Benefit evaluation module; 6. Strategy control module. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] Example 1, please refer to Figure 1 This invention provides a technical solution: an aquaculture water quality monitoring and adaptive control platform, comprising: Data processing module 1 is used to acquire water quality parameter data of aquaculture at multiple monitoring points and multiple time points, preprocess the water quality parameter data and perform time series fitting to obtain the water quality parameter time series. Feature extraction module 2 is used to perform spatiotemporal semantic recognition and feature extraction on the time series of water quality parameters, and to construct the feature sequence of each water quality parameter; Risk warning module 3 is used to calculate the spatiotemporal correlation between parameters and adapt dynamic risk thresholds based on feature sequences to obtain water quality risk warning features; Module 4, which is used to obtain aquaculture organism status data, combine water quality risk warning features and aquaculture organism status data to construct a water quality-biology-ecology coupling model and analyze niche relationships; The benefit assessment module 5 is used to predict the time-series evolution based on the water quality-biology-ecology coupling model, obtain multi-scenario control schemes, evaluate the benefits of the multi-scenario control schemes, and obtain optimized control strategies. The strategy control module 6 is used to drive the actuator to perform adaptive control according to the optimized control strategy, and feed back the controlled water quality data to the water quality parameter time series to form a closed-loop control.

[0020] It should be noted that the data processing module collects water quality parameter data from multiple monitoring points and at multiple time points, and preprocesses and fits this data to a time series sequence. Time series analysis yields the time series of water quality parameters, providing foundational data for subsequent analysis and prediction. The feature extraction module performs in-depth spatiotemporal semantic recognition and feature extraction on the time series of water quality parameters. This module can extract meaningful feature sequences from complex data, which helps to further understand the changing trends and correlations of water quality. The risk warning module calculates the spatiotemporal correlations between water quality parameters based on the extracted feature sequences and adapts them to dynamic risk thresholds, thereby generating water quality risk warning features. The goal of this module is to identify potential risks of water quality changes in advance, helping managers to make early warnings and responses. The relationship analysis module... Data on the status of aquaculture organisms is acquired and combined with water quality risk early warning characteristics to establish a water quality-biology-ecology coupling model. This model can analyze the correlation between water quality changes and the health of aquaculture organisms and the ecosystem, thus providing deeper insights for decision-making. The benefit assessment module uses the water quality-biology-ecology coupling model to predict time-series evolution and provides multiple control scenarios. Then, the benefits of these scenarios are evaluated to find the optimal control scheme and ultimately form an optimized control strategy. The goal of this module is to ensure the best control effect on water quality and the ecological environment through scientific evaluation. The strategy control module drives the actual implementing agencies (such as water pumps, air pumps, etc.) to perform adaptive control based on the optimized control strategy. The water quality data after control is fed back to the time series of water quality parameters, forming a closed-loop control system that continuously optimizes the water quality management and control effect.

[0021] In one alternative embodiment, the water quality parameters include water temperature, dissolved oxygen, ammonia nitrogen concentration, nitrite concentration, pH value, transparency, and plankton community density; Water quality parameter data are preprocessed and time-series sequenced for fitting to obtain a time-series sequence of water quality parameters, including: The raw data packets of water quality parameters within the monitoring area are acquired, and the raw data packets of water quality parameters are cleaned and standardized to obtain preprocessed data packets of water quality parameters. Dynamic and static features are extracted from the water quality parameter preprocessing data package. Dynamic features are water quality parameter features that change with monitoring time, while static features are water quality parameter inherent properties that do not change with monitoring time. The dynamic and static features are combined to obtain a water quality parameter feature set. The time-series segmentation threshold is determined based on the temporal correlation of the water quality parameter feature set, and the water quality parameter feature set is divided into multiple water quality parameter time-series segments according to the time-series segmentation threshold. The trend features of each water quality parameter in time series segments are extracted, and a time series fitting model is constructed based on the trend features. The time series fitting model is then used to fit each water quality parameter in time series segments to obtain the initial time series sequence of water quality parameters. The fitting error of the initial time series of water quality parameters is calculated, the parameters of the time series fitting model are adjusted according to the fitting error, and the initial time series of water quality parameters is corrected by the adjusted time series fitting model to obtain the optimized time series of water quality parameters. Determine the time alignment parameters for the optimized water quality parameter time series, and integrate all optimized water quality parameter time series based on the time alignment parameters to obtain the water quality parameter time series.

[0022] It should be noted that the system collects raw water quality data packets from multiple monitoring points. This data is typically in its raw, unprocessed state and may contain noise or errors. After obtaining the raw data, data cleaning is performed to remove outliers, fill in missing values, and standardize the data so that data from different parameters can be compared and analyzed within the same range. For example, if a monitoring point's temperature data has an outlier (e.g., suddenly jumping to 100°C), this data will be corrected or deleted, and the remaining data will be standardized to be between 0 and 1. Then, the system extracts dynamic features (such as temperature changes over time) and static features (such as water hardness, which often remains stable) from the processed water quality data. Dynamic features represent real-time changes in water quality, while static features represent inherent properties that do not change over time. For example, temperature changes (such as changes between morning and evening) are dynamic features, while water hardness is a static feature. Based on the temporal relationship between dynamic features, a "…" is determined. "Time series segmentation threshold" defines how data is divided according to time changes, ensuring that water quality changes in each time period can be analyzed independently. For example, assuming significant seasonal differences in water quality changes, the system can divide the data into two time series based on seasonal variations (such as spring and summer). By analyzing the trend of each time series (such as the pattern of water quality increase or decrease), a fitting model is constructed to simulate the trend of water quality changes. The time series fitting model can help predict water quality changes in the future. By calculating the model fitting error (i.e., the difference between the actual data and the fitted data), the parameters of the time series fitting model are adjusted. By adjusting the model, it can more accurately reflect the actual water quality data. For example, if the fitted temperature trend differs significantly from the actual situation, the model will adjust the parameters and recalculate the fitted curve to make the prediction more accurate. Finally, the system will perform time alignment based on the optimized time series data, integrating the time series data of all water quality parameters into a complete water quality time series on the same time line.

[0023] In an optional embodiment, spatiotemporal semantic recognition and feature extraction are performed on the time series of water quality parameters to construct a feature sequence for each water quality parameter, including: Spatiotemporal semantic recognition is performed on the time series of water quality parameters to identify the parameter distribution characteristics at different monitoring points and the change patterns over different time periods. The spatiotemporal semantic features of each water quality parameter are extracted in multiple dimensions to obtain multidimensional features. Among them, the spatiotemporal semantic features include time domain features and spatial domain features. The time domain features include parameter change trends, periodic fluctuation amplitude and mutation frequency, and the spatial domain features include spatial gradient difference, regional mean deviation and correlation between adjacent monitoring points. Anomaly patterns are initially detected in multi-dimensional features, and anomalous feature fragments are marked. Temporal features, spatial features, and anomalous feature fragments are integrated to construct a feature sequence for each water quality parameter.

[0024] It should be noted that the system will focus on water quality parameters at different monitoring points (spatial dimension) and at different time periods (temporal dimension), identifying their distribution characteristics and change patterns. For example, if the temperature in one area changes significantly over a period of time, while the temperature in another area remains almost constant, the system can identify this change pattern and use it as a spatiotemporal semantic feature. The system will extract features from the water quality data in multiple dimensions, including temporal and spatial features: Temporal features: analyzing the changes in water quality parameters over time, including: trend of change: whether the water quality parameter is rising or falling; amplitude of periodic fluctuations: how large the fluctuations in the water quality parameter are, and whether they exhibit periodic fluctuations; abrupt changes. Frequency: How frequently do water quality parameters change abruptly? Spatial characteristics: Analyzing spatial differences between different monitoring points, including: Spatial gradient difference: How much difference is there between water quality parameters at different monitoring points? Regional mean deviation: The degree to which water quality parameters in the entire monitoring area deviate from the mean? Correlation between adjacent monitoring points: Whether the changes in water quality parameters at two adjacent monitoring points are related? For example: Suppose that in a monitoring area, the temperature gradually rises over a certain period of time (trend of change) and fluctuates significantly (periodic fluctuation amplitude), but only some monitoring points experience abrupt changes (frequency of abrupt changes); These are all time-domain characteristics; while the large temperature difference between different monitoring points (spatial gradient difference) is a spatial-domain characteristic.

[0025] The system performs preliminary anomaly pattern detection on the extracted features; that is, it marks data segments that do not conform to the normal pattern, usually exhibiting sudden changes, periodic anomalies, or unexpected trends. For example, if the dissolved oxygen data at a monitoring point suddenly drops and the change is greater than the normal fluctuation range, this part of the data will be marked as an anomalous feature segment. Finally, the system integrates the temporal features, spatial features, and marked anomalous feature segments to construct a feature sequence for each water quality parameter. These feature sequences can provide a basis for subsequent water quality analysis, risk prediction, and decision-making.

[0026] In an optional embodiment, water quality risk early warning features are obtained by calculating the spatiotemporal correlation between parameters and adapting dynamic risk thresholds based on the feature sequence, including: The parameter correlation matrix is ​​obtained by performing pairwise parameter spatiotemporal correlation calculation on the feature sequences. The risk threshold is dynamically adapted based on the parameter correlation matrix to obtain a dynamic risk threshold table; The feature sequence is compared with the dynamic risk threshold table in real time to classify risk levels and mark risk trigger parameters. By integrating the parameter correlation matrix, dynamic risk threshold table, and risk level, water quality risk early warning characteristics are obtained.

[0027] It should be noted that the system calculates the spatiotemporal correlation between different water quality parameters; for example, whether temperature changes are related to changes in dissolved oxygen content; or whether water quality parameter changes at different monitoring points show similar trends. These correlations are represented as a "parameter correlation matrix," which describes the interrelationships between different parameters. The system determines the risk threshold for each water quality parameter by analyzing historical and current data. These thresholds may be dynamically adjusted over time and under different conditions. For example, the normal range of dissolved oxygen may differ in a particular season or region, thus requiring dynamic adjustment of the risk threshold based on different spatiotemporal characteristics. For instance, in summer when temperatures are high, dissolved oxygen content may be low, and the system will dynamically adjust the dissolved oxygen risk threshold based on this seasonal variation. The system compares the characteristic sequences of water quality data with the dynamic risk threshold table in real time. When a water quality parameter exceeds the current dynamic risk threshold, the system marks it as "risk triggered," indicating that the water quality is in a warning state.

[0028] Based on the risk comparison results, the system classifies water quality data into risk levels; for example, it may classify them into three levels: "low risk," "medium risk," and "high risk." Each water quality parameter's risk level is determined by whether it exceeds a set risk threshold. For example, if a water quality parameter fluctuates significantly and exceeds the set threshold, the system will mark that point as a "high-risk" area and conduct further analysis. Finally, the system integrates all spatiotemporal correlations (parameter correlation matrix), dynamic risk thresholds (dynamic risk threshold table), and risk levels (risk level table) to form a complete water quality risk early warning feature. These features can help identify potential risks in water quality in a timely manner, providing a scientific basis for decision-makers.

[0029] In an optional embodiment, a water quality-biology-ecology coupling model is constructed by combining water quality risk early warning characteristics and aquaculture organism status data to analyze niche relationships, including: The acquired water quality risk warning characteristics and aquaculture organism status data are preprocessed to establish a spatiotemporal correspondence between the two. Based on the pretreated water quality risk warning characteristics and aquaculture organism status data, a water quality-biology-ecology coupling model is constructed. By analyzing the associated variables output by the water quality-biology-ecology coupling model, the niche relationships between cultured organisms and the water quality environment and ecosystem in the target aquaculture area are analyzed. Based on the analysis results of niche association, the water quality-biology-ecology coupling model was verified and optimized to obtain a coupling model and niche association map suitable for the target aquaculture water area.

[0030] It should be noted that the system preprocesses water quality risk warning features and aquaculture organism status data to ensure they are matched and correlated. For example, water quality data may include parameters such as temperature and dissolved oxygen, while aquaculture organism status data may include growth rate and health status. The preprocessing process includes data cleaning, normalization, and time alignment to ensure that water quality data and biological data correspond within the same time frame. The water quality-biology-ecology coupling model combines water quality data, aquaculture organism status data, and ecosystem variables to form a complex coupling relationship. Changes in water quality (such as temperature and dissolved oxygen) directly affect the growth and health of aquaculture organisms, while the activities and status of aquaculture organisms (such as feed intake and growth rate) also feed back into changes in water quality and the ecosystem. For example, if the water temperature rises, it may cause some aquaculture organisms to accelerate their metabolism, thereby increasing oxygen consumption. This change will affect the dissolved oxygen level in the water, and changes in water quality may in turn affect the health of aquaculture organisms, forming a closed-loop coupling.

[0031] The coupled model analyzes the interactions between water quality, the state of cultured organisms, and the ecosystem, outputting a series of correlated variables. These variables help the system understand the relationship between cultured organisms, the water environment, and the ecosystem. For example, the model may reveal the correlation between water temperature, dissolved oxygen, and the growth rate of cultured organisms, or the interactive effect of stocking density and water quality changes. After obtaining the initial coupled model, the system will verify and optimize it based on actual aquaculture water data. This may include comparing the actual growth status of cultured organisms with the model's predictions and adjusting model parameters to improve its accuracy. The optimized model will be better adapted to the ecological environment of a specific water area. For example, assuming that low dissolved oxygen content in a certain aquaculture area leads to health problems in cultured organisms, the optimized model can predict and indicate similar problems that may occur in the next few days, allowing for proactive measures, such as increasing oxygen supply. Finally, the model will output a niche correlation map, which shows the complex relationships between cultured organisms, the environment (water quality), and the ecosystem. Through the map, it is clear which environmental factors have the greatest impact on cultured organisms and which variables need to be prioritized for management and optimization.

[0032] In an optional embodiment, the acquired water quality risk warning features and aquaculture organism status data are preprocessed to establish a spatiotemporal correspondence between the two, including: Water quality risk warning features are classified and labeled according to monitoring sub-regions, monitoring levels, and collection times; aquaculture biological status data are classified and labeled according to biological sampling points, sampling times, and biological species. Numerical normalization was performed on the water quality risk warning characteristics and aquaculture organism status data after classification and labeling, and abnormal data exceeding the reasonable range were removed; Based on the spatial correlation between the monitoring sub-region and the biological sampling point, and the temporal synchronization between the water quality parameter collection time and the biological sample collection time, a corresponding mapping relationship between water quality risk warning characteristics and aquaculture biological status data under the same spatiotemporal dimension is established to form a spatiotemporal matching dataset.

[0033] It should be noted that water quality risk warning characteristics are categorized according to different monitoring sub-regions (e.g., different aquaculture ponds or water areas) and labeled according to monitoring level (e.g., whether it is real-time monitoring or periodic monitoring) and collection time. This ensures that each data point corresponds to a specific water quality monitoring point and time range. Aquaculture organism status data are categorized and labeled according to sampling point (e.g., a specific location in an aquaculture area or water body), sampling time (e.g., a specific day or time period), and organism species (different types of aquaculture organisms). This ensures that each biological data point clearly corresponds to the corresponding sampling location and time. For example, suppose in an aquaculture area, water quality data is distributed across three sub-regions (A, B, C), with data collected at different time periods (morning, daytime, and evening); simultaneously, aquaculture organism data is collected at different locations (A1, B1, C1) and times (recorded once per hour); each set of data needs to be categorized and labeled according to this information.

[0034] When analyzing water quality risk warning characteristics and aquaculture organism status data, different data values ​​may have different dimensions and ranges. Therefore, numerical normalization is necessary to ensure that all data are compared and analyzed on the same scale. For example, data such as water temperature and dissolved oxygen should be converted to the same standard range (e.g., between 0 and 1). During data processing, some data may produce outliers due to sensor malfunctions, sampling errors, etc. If these outliers are not processed, they may affect the accuracy of the model. Therefore, it is necessary to remove outliers that exceed reasonable ranges to ensure the reliability of the analysis results. For example, suppose the normal range for a certain water quality parameter, dissolved oxygen, is 4-8 mg / L, but due to instrument malfunction, a certain data record shows 12 mg / L, which is obviously unreasonable and needs to be removed. At the same time, all dissolved oxygen data should be normalized according to certain rules so that they can be used with other parameters (such as temperature and pH).

[0035] By analyzing the spatial and temporal relationships between water quality data and aquaculture biological data, matching is ensured within the same spatiotemporal range. Water quality data is typically collected at specific locations and time intervals (e.g., hourly records), while aquaculture biological status data is recorded based on biological sampling points and time points. To establish a spatiotemporally matched dataset, it is necessary to ensure the correct correspondence of data within the same spatiotemporal dimension. Based on the spatial relationship between the monitoring sub-region and the biological sampling points, as well as the temporal synchronization of water quality data and aquaculture biological data, a corresponding spatiotemporally matched dataset is established. This means that water quality data and aquaculture biological status data within the same time and spatial range will be matched together to form a dataset that can be further analyzed and modeled.

[0036] In an optional embodiment, a water quality-biology-ecology coupling model is constructed based on pretreated water quality risk warning characteristics and aquaculture organism status data, including: Water quality risk warning features are extracted from the spatiotemporal matching dataset as input layer variables of the model, aquaculture biological status data are extracted as intermediate layer variables of the model, and water temperature, light duration, and water flow velocity of the target aquaculture area are introduced as ecological regulation variables. Establish a correlation function between input layer variables and intermediate layer variables to quantify the impact weight of changes in water quality risk warning characteristics on aquaculture organism status data; Ecological regulation variables are incorporated as feedback factors into the correlation function to construct a coupled model structure. The coupled model structure includes a water quality-biology direct correlation module, an ecology-water quality feedback module, and an ecology-biology regulation module. The interaction parameters of each module are set to complete the construction of the water quality-biology-ecology coupled model.

[0037] It should be noted that water quality risk warning features are selected from the spatiotemporally matched dataset as the model input; these are the main information used by the model to "perceive environmental conditions"; the status data of farmed organisms (such as the growth rate, activity level, and health status of fish or shellfish) are used as intermediate layer variables of the model to represent the actual impact of environmental changes on organisms; additional environmental regulation factors of the target water area (such as light duration, water flow velocity, and water mixing degree) are introduced. These variables affect the interaction between water quality and organisms, but are not directly used as inputs or outputs; the correlation function is used to quantify the degree of influence of water quality risk features on the status of farmed organisms; "If the water temperature rises by 1°C, how much might the activity level of fish decrease?" and "How much will an increase in ammonia nitrogen concentration affect the growth rate of fish?" Through this function, the model can transform the input water quality changes into predictions of the status of organisms.

[0038] Ecological regulation variables serve as "feedback factors" to regulate the relationship between water quality and organisms; they can enhance or weaken the impact of water quality changes on organisms. This step reflects the role of "ecological regulation," meaning that environmental factors not only affect water quality but also indirectly affect the state of organisms. For example, longer daylight hours may increase fish feeding, thus mitigating the negative impact of rising water temperature on growth rate; faster water flow may increase dissolved oxygen, also alleviating the impact of low dissolved oxygen on fish. The overall model structure includes three main modules: a water quality-organism direct correlation module, which directly describes the impact of water quality on the state of cultured organisms; an ecological-water quality feedback module, which describes how ecological factors regulate water quality parameters (such as the improvement of dissolved oxygen by water flow velocity); and an ecological-organism regulation module, which describes how ecological factors indirectly regulate the state of organisms (such as the impact of light on fish feeding and activity). The interaction parameters of each module need to be set, as these parameters determine the strength of each factor's effect, thus completing the construction of the overall coupled model.

[0039] In an optional embodiment, a time-series evolution prediction is performed based on a water quality-biology-ecology coupling model to obtain multi-scenario control schemes. The effectiveness of these multi-scenario control schemes is then evaluated to obtain an optimized control strategy, including: Based on the water quality-biology-ecology coupling model, the evolution trend of water quality parameters and changes in biological state under different environmental conditions within a preset time period are simulated to obtain the time-series evolution prediction curve. Based on the time-series evolution prediction curve, multi-scenario control schemes are generated, including aerator power adjustment, water exchange flow control, feed amount optimization, and probiotic addition amount adaptation. The effectiveness of the control schemes is evaluated, including ecological benefits, economic benefits, and operating costs. The scheme with the best ecological and economic benefits is selected to obtain the optimized control strategy.

[0040] It should be noted that, based on the previously established water quality-biology-ecology coupling model, the trends of water quality changes and changes in biological states under different time periods and environmental conditions are simulated. The model can predict how water quality parameters such as water temperature, dissolved oxygen, and ammonia nitrogen concentration will change over time, as well as how the growth and activity states of farmed organisms (such as fish) will change. After understanding the trends of water quality and biological changes under different environmental conditions, different control schemes can be designed based on these predicted curves to optimize the aquaculture environment. These control schemes include: increasing the power of aerators to increase the oxygen content in the water and improve low dissolved oxygen conditions; adjusting the frequency and flow rate of water exchanges according to water quality changes to ensure water quality stability; adjusting the amount of feed based on factors such as water temperature, light, and fish activity to avoid waste and promote healthy fish growth; and timely introduction of probiotics based on water quality changes to improve the aquatic environment, promote the growth of beneficial microorganisms, and reduce harmful substances. For example, assuming a hot summer season, rising water temperature leads to reduced fish activity and a decrease in dissolved oxygen; in this case, the power of aerators can be increased, along with the water exchange flow rate and the amount of feed adjusted, to help the fish maintain normal growth.

[0041] Each control scheme needs to be evaluated for its effectiveness, primarily measured from the following dimensions: Ecological benefits: Does the scheme effectively improve water quality and the ecological environment, promoting the health of farmed organisms? Economic benefits: Does the scheme increase biomass production while reducing farming costs or increasing profits? Operating costs: The complexity and cost of implementing the scheme, and whether it is feasible and economically reasonable. Through these evaluations, managers can understand which control schemes are most effective in practice and can produce the best ecological and economic benefits. For example, if increasing the power of aerators significantly improves the growth rate of fish with a small increase in electricity costs, then this measure may be an efficient scheme after considering economic benefits. On the other hand, if the addition of probiotics can effectively improve water quality and is simple and low-cost to operate, it may also be a scheme worth promoting. Based on the results of the benefit evaluation, the scheme that is most optimized in terms of ecological and economic benefits is selected, and the best control strategy is finally determined.

[0042] In an optional embodiment, the actuator is driven to perform adaptive control according to the optimized control strategy, and the controlled water quality data is fed back to the water quality parameter time series to form a closed-loop control, including: Based on the optimized control strategy, execution instructions are generated to drive the corresponding actuators to perform control operations. After the control operation is executed, water quality parameter data of the controlled area are collected once every preset collection cycle to obtain the water quality data after control. The adjusted water quality data is fed back to the water quality parameter time series, the sequence data is updated and the time series sequence fitting is retried; Based on the updated water quality parameter time series, the operating parameters of the actuator are dynamically fine-tuned to form a closed-loop control.

[0043] It should be noted that after selecting the optimal control strategy, the system generates a series of execution instructions based on these strategies. These instructions drive various control devices (such as aerators and water exchange equipment) to operate according to the predetermined plan. After executing the control operation, the system periodically collects water quality data of the controlled area and feeds this data back into the water quality parameter time series. This series records the trend of water quality changes over a period of time and is a key basis for analyzing water quality changes and evaluating the control effect. As newly collected water quality data is added to the time series, the original water quality evolution prediction model is updated. This is because water quality conditions may change due to the influence of control measures, and the system needs to refit the new water quality evolution trend to ensure the accuracy of subsequent decisions. For example, suppose the initial model predicts that the water quality will deteriorate and dissolved oxygen will drop to 2 mg / L in a certain period of time; but after the aerator power is increased, the new data shows that the dissolved oxygen will remain at 5 mg / L. mg / L; the system refits the water quality evolution trend based on this new data and adjusts future predictions; through real-time updated water quality data and model predictions, the system dynamically fine-tunes the operating parameters of the actuators; for example, if dissolved oxygen has recovered to the target level, the system may reduce the power of the aerator to avoid waste caused by excessive aeration.

[0044] Example 2, please refer to Figure 2 This invention provides a technical solution: a method for monitoring and adaptively controlling aquatic water quality, applicable to the aforementioned aquatic water quality monitoring and adaptive control platform, comprising: S1. Obtain water quality parameter data for aquaculture at multiple monitoring points and multiple time points, preprocess the water quality parameter data and perform time-series serialization fitting to obtain the water quality parameter time series. S2. Perform spatiotemporal semantic recognition and feature extraction on the time series of water quality parameters to construct a feature sequence for each water quality parameter; S3. Calculate the spatiotemporal correlation between parameters and adapt dynamic risk thresholds based on the feature sequence to obtain water quality risk early warning features; S4. Obtain aquaculture organism status data, combine water quality risk warning characteristics and aquaculture organism status data to construct a water quality-biology-ecology coupling model, and analyze the niche correlation. S5. Based on the water quality-biology-ecology coupling model, time-series evolution prediction is carried out to obtain multi-scenario control schemes. The benefits of the multi-scenario control schemes are evaluated to obtain optimized control strategies. S6. Drive the actuator to perform adaptive regulation based on the optimized regulation strategy, and feed back the regulated water quality data to the water quality parameter time series to form a closed-loop regulation.

[0045] The embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited thereto. Various changes can be made within the scope of knowledge possessed by those skilled in the art without departing from the spirit of the present invention.

Claims

1. A platform for monitoring and adaptive regulation of aquaculture water quality, characterized in that, include: The data processing module is used to acquire water quality parameter data of aquaculture at multiple monitoring points and multiple time points, and to preprocess and time-series sequence fitting the water quality parameter data to obtain a water quality parameter time-series sequence. The feature extraction module is used to perform spatiotemporal semantic recognition and feature extraction on the time series of water quality parameters, and to construct a feature sequence for each water quality parameter; The risk warning module is used to calculate the spatiotemporal correlation between parameters and adapt the dynamic risk threshold based on the feature sequence to obtain water quality risk warning features; The relationship parsing module is used to acquire aquaculture organism status data, combine the water quality risk warning features and the aquaculture organism status data to construct a water quality-organism-ecology coupling model, and parse the niche correlation relationship; The benefit assessment module is used to predict the time-series evolution based on the water quality-biology-ecology coupling model, obtain multi-scenario control schemes, evaluate the benefits of the multi-scenario control schemes, and obtain optimized control strategies. The strategy control module is used to drive the actuator to perform adaptive control according to the optimized control strategy, and feed back the controlled water quality data to the water quality parameter time series to form a closed-loop control.

2. The aquaculture water quality monitoring and adaptive control platform according to claim 1, characterized in that, The water quality parameters include water temperature, dissolved oxygen, ammonia nitrogen concentration, nitrite concentration, pH value, transparency, and plankton community density. The water quality parameter data is preprocessed and time-series sequenced for fitting to obtain a water quality parameter time-series sequence, including: Obtain raw data packets of water quality parameters within the monitoring area, and clean and standardize the raw data packets of water quality parameters to obtain preprocessed data packets of water quality parameters. Dynamic and static features are extracted from the water quality parameter preprocessing data package. The dynamic features are water quality parameter features that change with monitoring time, and the static features are water quality parameter inherent attribute features that do not change with monitoring time. The dynamic features and the static features are combined to obtain a water quality parameter feature set. The time-series segmentation threshold is determined based on the time correlation of the water quality parameter feature set, and the water quality parameter feature set is divided into multiple water quality parameter time-series segments according to the time-series segmentation threshold. The trend features of each water quality parameter time series segment are extracted, and a time series fitting model is constructed based on the trend features. The time series fitting model is used to fit each water quality parameter time series segment to obtain the initial time series sequence of water quality parameters. The fitting error of the initial time series sequence of the water quality parameters is calculated, the parameters of the time series fitting model are adjusted according to the fitting error, and the initial time series sequence of the water quality parameters is corrected by the adjusted time series fitting model to obtain the optimized time series sequence of the water quality parameters. Determine the time alignment parameter of the optimized water quality parameter time series, and integrate all the optimized water quality parameter time series according to the time alignment parameter to obtain the water quality parameter time series.

3. The aquaculture water quality monitoring and adaptive control platform according to claim 2, characterized in that, Spatiotemporal semantic recognition and feature extraction are performed on the time series of the water quality parameters to construct a feature sequence for each water quality parameter, including: Spatiotemporal semantic recognition is performed on the time series of water quality parameters to identify the parameter distribution characteristics at different monitoring points and the change patterns over different time periods; The spatiotemporal semantic features of each water quality parameter are extracted in multiple dimensions to obtain multidimensional features; wherein, the spatiotemporal semantic features include temporal features and spatial features, the temporal features include parameter change trends, periodic fluctuation amplitude and mutation frequency, and the spatial features include spatial gradient difference, regional mean deviation and correlation degree between adjacent monitoring points; Anomaly pattern detection is performed on the multi-dimensional features, and anomalous feature fragments are marked; the temporal features, spatial features, and anomalous feature fragments are integrated to construct a feature sequence for each water quality parameter.

4. The aquaculture water quality monitoring and adaptive control platform according to claim 3, characterized in that, Based on the aforementioned feature sequence, spatiotemporal correlation calculations and dynamic risk threshold adaptation are performed to obtain water quality risk early warning features, including: The feature sequences are subjected to pairwise spatiotemporal correlation calculations to obtain a parameter correlation matrix; Based on the parameter correlation matrix, the risk threshold is dynamically adapted to obtain a dynamic risk threshold table; The feature sequence is compared with the dynamic risk threshold table in real time to classify risk levels and mark risk trigger parameters. By fusing the parameter correlation matrix, the dynamic risk threshold table, and the risk level, water quality risk early warning characteristics are obtained.

5. The aquaculture water quality monitoring and adaptive control platform according to claim 4, characterized in that, Combining the aforementioned water quality risk warning characteristics and the aforementioned aquaculture organism status data, a water quality-biology-ecology coupling model is constructed to analyze niche relationships, including: The acquired water quality risk warning characteristics and aquaculture organism status data are preprocessed to establish a spatiotemporal correspondence between the two. Based on the pretreated water quality risk warning characteristics and aquaculture organism status data, a water quality-biology-ecology coupling model is constructed. By analyzing the associated variables output by the water quality-biology-ecology coupling model, the niche relationships between cultured organisms and the water quality environment and ecosystem in the target aquaculture area are analyzed. Based on the analysis results of the niche association, the water quality-biology-ecology coupling model was verified and optimized to obtain a coupling model and niche association map suitable for the target aquaculture water area.

6. The aquaculture water quality monitoring and adaptive control platform according to claim 5, characterized in that, The acquired water quality risk warning characteristics and aquaculture organism status data are preprocessed to establish a spatiotemporal correspondence between the two, including: The water quality risk warning features are classified and labeled according to monitoring sub-regions, monitoring levels, and collection times, and the aquaculture biological status data are classified and labeled according to biological sampling points, sampling times, and biological species. Numerical normalization was performed on the water quality risk warning characteristics and aquaculture organism status data after classification and labeling, and abnormal data exceeding the reasonable range were removed; Based on the spatial correlation between the monitoring sub-region and the biological sampling point, and the temporal synchronization between the water quality parameter collection time and the biological sample collection time, a corresponding mapping relationship between the water quality risk warning features and the aquaculture biological status data under the same spatiotemporal dimension is established to form a spatiotemporal matching dataset.

7. The aquaculture water quality monitoring and adaptive control platform according to claim 6, characterized in that, Based on the pretreated water quality risk warning characteristics and aquaculture organism status data, a water quality-biology-ecology coupling model is constructed, including: The water quality risk warning features are extracted from the spatiotemporal matching dataset as input layer variables of the model, the aquaculture organism status data are extracted as intermediate layer variables of the model, and the water temperature, light duration, and water flow velocity of the target aquaculture area are introduced as ecological regulation variables. Establish a correlation function between the input layer variables and the intermediate layer variables to quantify the impact weight of changes in water quality risk warning characteristics on the aquaculture organism status data; The ecological regulation variables are incorporated as feedback factors into the correlation function to construct a coupled model structure, which includes a water quality-biology direct correlation module, an ecology-water quality feedback module, and an ecology-biology regulation module. The interaction parameters of each module are set to complete the construction of the water quality-biology-ecology coupled model.

8. The aquaculture water quality monitoring and adaptive control platform according to claim 7, characterized in that, Based on the aforementioned water quality-biology-ecology coupling model, time-series evolution prediction is performed to obtain multi-scenario control schemes. The effectiveness of these multi-scenario control schemes is evaluated to obtain optimized control strategies, including: Based on the water quality-biology-ecology coupling model, the evolution trend of water quality parameters and changes in biological state under different environmental conditions within a preset time period are simulated to obtain a time-series evolution prediction curve. Based on the time-series evolution prediction curve, a multi-scenario control scheme is generated, wherein the control scheme includes aerator power adjustment, water exchange flow control, feed amount optimization, and probiotic addition amount adaptation. The aforementioned control schemes are evaluated for their benefits, including ecological benefits, economic benefits, and operating costs. The scheme with the best ecological and economic benefits is selected to obtain the optimized control strategy.

9. The aquaculture water quality monitoring and adaptive control platform according to claim 8, characterized in that, The optimized control strategy drives the actuator to perform adaptive control, and the controlled water quality data is fed back to the water quality parameter time series to form a closed-loop control, including: Based on the optimized control strategy, an execution instruction is generated to drive the corresponding actuator to perform the control operation; After the control operation is executed, water quality parameter data of the controlled area are collected once every preset collection cycle to obtain the water quality data after control. The adjusted water quality data is fed back to the water quality parameter time series, the sequence data is updated, and the time series sequence fitting is retried. Based on the updated water quality parameter time series, the operating parameters of the actuator are dynamically fine-tuned to form a closed-loop control.

10. A method for monitoring and adaptively controlling aquatic water quality, applicable to the aquatic water quality monitoring and adaptive control platform described in any one of claims 1-9, characterized in that, include: Water quality parameter data of aquaculture at multiple monitoring points and multiple time points are acquired, and the water quality parameter data are preprocessed and time-series sequenced for fitting to obtain a water quality parameter time series. Spatiotemporal semantic recognition and feature extraction are performed on the time series of water quality parameters to construct a feature sequence for each water quality parameter; Based on the feature sequence, the spatiotemporal correlation between parameters is calculated and dynamic risk threshold is adapted to obtain water quality risk early warning features; Acquire aquaculture organism status data, combine the water quality risk warning characteristics with the aquaculture organism status data to construct a water quality-biology-ecology coupling model, and analyze the niche correlation; Based on the water quality-biology-ecology coupling model, time-series evolution prediction is performed to obtain multi-scenario control schemes. The benefits of the multi-scenario control schemes are evaluated to obtain optimized control strategies. The optimized control strategy drives the actuator to perform adaptive control, and the controlled water quality data is fed back to the water quality parameter time series to form a closed-loop control.

Images