Self-adaptive regulation and control method for biodiversity of water body

Abstract

The invention discloses a self-adaptive regulation and control method for water body biodiversity, which comprises the following steps: monitoring the number and distribution condition of water body organisms in a water body in real time, comparing with a preset standard, and then establishing a water quality and water quantity scheduling model; controlling the water quality of the water quality and water quantity scheduling model in a circulating pump linkage scheduling form, and establishing a water quality purification model based on the biofilm carrier; based on the water quality purification culture model, a deep learning algorithm is adopted to detect the quantity and the variety of the water body organisms, a biomass evaluation module is constructed, and real-time diversity data of the water body organisms are obtained; diversity data of the water body organisms are calculated by adopting a multiple linear regression equation, optimal generation environment data of the water body organisms are adaptively predicted, specific environment conditions of different water bodies are monitored and analyzed in real time by adopting a multi-source data fusion and deep learning algorithm, and a regulation and control strategy is automatically adjusted to adapt to changing environment conditions. And the effectiveness and continuity of regulation and control measures are ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of biodiversity, and particularly relates to an adaptive regulation method for water body biodiversity. Background Art

[0002] Water body biodiversity is one of the important indicators for evaluating the health status of water ecosystem. It is directly related to water quality, ecological balance and the sustainability of human activities. In order to address these issues, various regulation methods have been proposed to protect and restore water body biodiversity.

[0003] Traditional regulation methods for water body biodiversity mainly include physical restoration, chemical restoration and biological restoration, etc. Physical restoration usually involves means such as dredging and constructing artificial wetlands, aiming to improve water quality and restore habitats; Chemical restoration removes pollutants by adding specific chemical agents, but may cause secondary pollution problems; Biological restoration uses microorganisms or plants to absorb and degrade pollutants, promoting the self-restoration ability of the ecosystem. In addition, it also includes formulating strict environmental protection regulations and management measures to limit pollutant emissions and protect endangered species. Although these methods help to alleviate the deterioration trend of water body biodiversity to a certain extent, they still face many challenges in practical applications and there are also the following problems:

[0004] (1) Traditional methods often focus on solving problems in a certain aspect, such as only focusing on water quality purification or specific species protection, lacking an overall solution that comprehensively considers the interactions of various components of the ecosystem. Moreover, different water bodies have unique geographical, climatic and ecological conditions, and it is difficult for existing technologies to be flexibly adjusted according to specific environmental conditions, resulting in unstable or unsatisfactory effects;

[0005] (2) When facing sudden environmental pollution incidents or new ecological threats, existing technologies cannot respond quickly and take effective measures to deal with them. Moreover, the current solutions lack a real-time monitoring system and cannot obtain key data in a timely manner to evaluate the regulation effect and dynamically adjust strategies accordingly. Summary of the Invention

[0006] The purpose of the present invention is to provide an adaptive regulation method for water body biodiversity to solve the technical problems in the prior art that lack an overall solution that comprehensively considers the interactions of various components of the ecosystem, and lack a real-time monitoring system, cannot obtain key data in a timely manner to evaluate the regulation effect, and cannot dynamically adjust strategies accordingly.

[0007] To solve the above technical problems, the present invention specifically provides the following technical solutions:

[0008] The present invention provides an adaptive regulation method for water body biodiversity, including the following steps:

[0009] Monitor the quantity and distribution of aquatic organisms in the water body in real time, evaluate the current status of aquatic biodiversity, and establish a water quality and quantity scheduling model after comparing with the preset standards;

[0010] Control the water quality by using a circulating pump linkage scheduling form for the water quality and quantity scheduling model. Use the nutrients of the aquatic organisms as the biofilm carrier to establish a water quality purification model based on the biofilm carrier;

[0011] Detect the quantity and species of the aquatic organisms based on the water quality purification and breeding model by using a deep learning algorithm, construct a biomass evaluation module, and obtain the real-time biodiversity data of the aquatic organisms;

[0012] Calculate the biodiversity data of the aquatic organisms by using a multiple linear regression equation, construct the relationship between water quality indicators and biomass, and adaptively predict the optimal growth environment data of the aquatic organisms.

[0013] As a preferred embodiment of the present invention, monitoring the quantity and distribution of aquatic organisms in the water body in real time, evaluating the current status of aquatic biodiversity, and establishing a water quality and quantity scheduling model after comparing with the preset standards, including:

[0014] Use an optical imaging device to monitor the quantity and species distribution of the aquatic organisms in the water body in real time, regularly collect water sample data at different depths and positions, and use the screened water sample data as the collection data of the organisms;

[0015] Combine the collection data with the historical data of aquatic organisms to determine the biodiversity index suitable for the local ecological environment. The biodiversity index includes but is not limited to the quantity range, distribution pattern, and ecological function of different species;

[0016] Compare the real-time monitored biodiversity data with the biodiversity index to identify the problem data of the aquatic organisms in the current state;

[0017] Analyze the problem data to obtain the reasons for the differences, and establish a water quality and quantity scheduling model to predict the change trend of biodiversity under different management modes, and obtain the real-time water quality and quantity scheduling data.

[0018] As a preferred embodiment of the present invention, control the water quality by using a circulating pump linkage scheduling form for the water quality and quantity scheduling model, including:

[0019] Set an adjustable-speed circulating pump in the water body to circulate and expose the air intake flow of the water body, schedule an intelligent microfilter to purify the water body according to the water quality and quantity scheduling data, and monitor the water quality parameters in real time;

[0020] During the water body circulation purification process, use the water quality parameters as independent variables and the filtration times of the intelligent microfilter as dependent variables, and obtain a water quality and water volume scheduling model by fitting a curve.

[0021] According to the water quality and water volume scheduling model, count the factors affecting water quality, and use the support vector machine SVM to process the corresponding factor data to predict the decision data points affecting water quality and water volume.

[0022] As a preferred embodiment of the present invention, use the nutrients of the water body organisms as the biofilm carrier, and establish a water quality purification model based on the biofilm carrier, including:

[0023] Select a carrier material suitable for the growth of water body organisms, inoculate microorganisms to attach and grow to form a biofilm, and place the biofilm on the side of the adjustable-speed circulation pump to ensure that the water flow can fully contact the carrier surface;

[0024] Use the Monod equation to establish the rate equation for microorganisms to remove organic pollutants, and obtain the water quality dilution rate, filtration cycle, and pollutant degradation efficiency;

[0025] Use the support vector machine model to classify the decision data points for the rate equation, obtain the optimal position points of the decision data points, and perform cross-validation on the optimal position points through a kernel function. The expression of the kernel function is:

[0026] K(a,b)=exp(-γ‖a - b‖ 2 )

[0027] Where a and b respectively represent the position coordinates of the decision data points, and γ represents a positive parameter of linear correlation;

[0028] Use GridSearchCV to optimize the parameter γ, predict the optimal data points through cross-validation, and establish a water quality purification model.

[0029] As a preferred embodiment of the present invention, use a deep learning algorithm to detect the quantity and species of the water body organisms based on the water quality purification and breeding model, including:

[0030] Collect water quality parameters, pH values, temperature and other parameters as training data based on the optimal position points of the decision data points. At the same time, take real-time water body video data at the optimal position points of the decision data points. Use the training data as water body sample parameters, and combine the known quantity and species of organisms to perform data preprocessing on the water body sample parameters;

[0031] Use the preprocessed water body sample parameters to construct a deep learning algorithm detection model with a long short-term memory network to extract data features;

[0032] Analyze the water body video data using a convolutional neural network (CNN) to identify different biological data;

[0033] Use the deep learning algorithm detection model to train the different biological data and data features, and predict the training data in the form of labels through supervised learning to obtain the number and types of organisms in a given sample;

[0034] Monitor the changes in water quality and biological populations in real time, and regularly predict the types and numbers of organisms by continuously optimizing the detection model.

[0035] As a preferred embodiment of the present invention, construct a biomass assessment module based on the types and numbers of organisms to obtain real-time diversity data of the water body organisms, including:

[0036] Obtain the types and numbers of organisms in the water body by monitoring the changes in water quality and biological populations in real time, and use redundancy analysis to monitor the correlation between species data and environmental factors for the types and numbers of organisms;

[0037] Use a correlation heat map to construct a correlation analysis with environmental factors from two dimensions of the types and numbers of organisms respectively to obtain the influence weights of the environmental factors on the types and numbers of organisms;

[0038] Establish a biomass assessment module in the water body based on the magnitudes of the influence weights to obtain real-time diversity data of the water body organisms.

[0039] As a preferred embodiment of the present invention, the biomass assessment module uses the Beta diversity analysis method to compare the diversity between different biological communities, including:

[0040] Obtain the species types at different times and locations in the water body, quantify the species numbers between different samples using the Jaccard index, and construct an index similarity matrix by calculating the similarity of the species numbers between different samples;

[0041] Use principal coordinate analysis on the index similarity matrix to analyze the influence degree of the influence weights of the environmental factors on species composition;

[0042] Monitor the changes in species composition in real time to obtain real-time diversity data of the water body organisms.

[0043] As a preferred embodiment of the present invention, use a multiple linear regression equation to calculate the diversity data of the water body organisms and construct the relationship between water quality indicators and biomass, including:

[0044] The stepwise regression of the multiple linear regression equation is used to introduce the environmental factors into the model one by one as variables. Each of the introduced variables is subjected to an F-test, and when the interpretation of the introduced variable for the multiple linear regression equation is not significant, the corresponding variable is deleted;

[0045] The introduced variables are repeatedly and gradually introduced in the form of permutations and combinations to determine that the multiple linear regression equation has the optimal explanatory ability;

[0046] Taking the introduced variables as independent variables, the relationship between water quality indicators and biomass is constructed.

[0047] As a preferred embodiment of the present invention, the multiple linear regression equation takes the environmental factors as independent variables, selects biomass as the target dependent variable and conducts a correlation analysis, establishes a prediction model between water quality indicators and biomass, obtains biomass prediction data, and the specific expression of the multiple linear regression equation is:

[0048] G = σ0 + σ1x1 + σ2x2 + …… + σ n x n

[0049] Among them, G represents the biomass in the water body, σ0 represents a constant, x1, x2…x n represents the environmental factors affecting water quality parameters in the water body, and σ1, σ2…σ n represents the regression coefficient.

[0050] As a preferred embodiment of the present invention, the support vector regression algorithm is used to optimize the parameters of the biomass prediction data, calculate the mean square error of the parameters as the influencing factor of the fitness function, and adaptively predict the optimal generation environment data of the water body organisms through a cyclic parameter optimization process.

[0051] The present invention has the following beneficial effects compared with the prior art:

[0052] The present invention adopts multi-source data fusion and deep learning algorithms to accurately identify key influencing factors and their interaction relationships, improves resource utilization efficiency, monitors and analyzes the specific environmental conditions of different water bodies in real time, automatically adjusts control strategies to adapt to changing environmental conditions, ensures the effectiveness and sustainability of control measures, has a monitoring and early warning mechanism, can discover and respond to sudden environmental pollution events or new ecological threats in the first time, quickly take effective countermeasures, prevent the further expansion of ecological damage, and ensure the healthy and stable ecosystem. Brief Description of the Drawings

[0053] To more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only exemplary. For those of ordinary skill in the art, without creative efforts, other implementation drawings can be obtained based on the provided drawings.

[0054] Figure 1 It is a flowchart of the adaptive regulation method for water body biodiversity provided by the embodiment of the present invention. Specific embodiments

[0055] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of them. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0056] As Figure 1 shown, the present invention provides an adaptive regulation method for water body biodiversity, including the following steps:

[0057] Real-time monitor the quantity and distribution of water body organisms in the water body, evaluate the current water body biodiversity status, and establish a water quality and quantity scheduling model after comparison with the preset standard;

[0058] In this embodiment, according to the characteristics of the target water body, organisms suitable for survival in the water body are selected for monitoring. Through on-site sampling, automatic monitoring equipment and other means, the quantity and distribution of the selected organisms are collected in real time.

[0059] Control the water quality of the water quality and quantity scheduling model in the form of a circulating pump linkage scheduling, use the nutrients of the water body organisms as the biofilm carrier, and establish a water quality purification model based on the biofilm carrier;

[0060] In this embodiment, the circulating pump linkage scheduling form is used to repeatedly detect the water quality cleanliness in the water body per unit time, construct a circulating volume regulation model based on machine learning, so as to control the water quality and quantity scheduling model to realize the overall circulation of the water volume in the water body.

[0061] Based on the water quality purification and breeding model, use the deep learning algorithm to detect the quantity and types of the water body organisms, construct a biomass evaluation module, and obtain the real-time biodiversity data of the water body organisms;

[0062] In this embodiment, video frames are obtained by real-time shooting in the water body. A biological classification and detection model is constructed through a convolutional neural network using a deep learning algorithm, enabling the constructed biomass evaluation module to have good adaptability and realizing the function of real-time recognition.

[0063] The diversity data of the water body organisms is calculated using a multiple linear regression equation, the relationship between water quality indicators and biomass is constructed, and the optimal growth environment data of the water body organisms is adaptively predicted.

[0064] In this embodiment, a multiple linear regression equation is used to statistically analyze the changes in the types and quantities of organisms under the influence of multiple factors, and a linear relationship between water quality indicators and biomass is constructed, enabling the real-time detection of the impact of water quality changes on biomass.

[0065] The quantity and distribution of water body organisms in the water body are monitored in real time, the current state of water body biodiversity is evaluated, and a water quality and quantity scheduling model is established after comparison with a preset standard, including:

[0066] An optical imaging device is used to monitor in real time the quantity and species distribution of the water body organisms in the water body, and water sample data at different depths and positions are collected regularly. The screened water sample data is used as the collection data of the organisms.

[0067] The collection data is combined with the historical data of water body organisms to determine biodiversity indicators suitable for the local ecological environment. The biodiversity indicators include but are not limited to the quantity range, distribution pattern, and ecological functions of different species.

[0068] The biodiversity data monitored in real time is compared with the biodiversity indicators to identify the problem data of the water body organisms in the current state.

[0069] The problem data is analyzed to obtain the reasons for the differences, and a water quality and quantity scheduling model is established to predict the change trends of biodiversity under different management modes and obtain real-time water quality and quantity scheduling data.

[0070] In this embodiment, an optical imaging device is used to take real-time biological data in the water body. The real-time monitored biological data is compared with the local biodiversity indicators to obtain the biological problem data under the current water quality. The problem data is used as the influencing factors affecting the water quality indicators, and the completely mixed water quality model is selected as the water quality and quantity scheduling model by comprehensively considering the influencing factors.

[0071] In this embodiment, it is assumed that the substances in the water body are completely mixed, that is, the influence of the internal concentration gradient change is ignored, and the basic equation for water quality migration and transformation is derived based on the water volume balance equation of the water body unit and the law of conservation of mass of water quality indicators.

[0072] In this embodiment, in the completely mixed water quality model, a water body unit such as a pool, a lake or a reservoir is regarded as a completely mixed reaction unit, that is, it is assumed that the concentrations of the same water quality indicators are basically the same everywhere inside the water body unit. During the micro-time period dt, the inflow rate of the water body unit is Q t , and the concentration of the water quality indicator is C t . After entering this reaction unit, under the action of external stirring and mixing, the water quality indicators are instantly and evenly dispersed throughout the reaction unit, and its average concentration is C. The outflow rate is Q. At this time, based on the principles of water volume balance and mass conservation, the basic equation for water quality migration and transformation can be established. The expression of the water volume balance equation is:

[0073]

[0074] where dt is the length of the time period; Q t and Q are the average inflow and outflow rates of the water body unit during the t time period respectively, and dV is the change in the water volume of the water body unit during the t time period, that is, the change in the reservoir water storage.

[0075] The expression of the basic equation for water quality migration and transformation is:

[0076]

[0077] where C is the average concentration of the water quality indicator in the water body unit during the t time period, C t is the average concentration of the water quality indicator flowing into the water body unit during the t time period, ∑D i is the mass increment per unit volume caused by biochemical reaction in the unit water body of the water quality indicator, and V represents the change in water volume.

[0078] In this embodiment, based on the basic equation for water quality migration and transformation as the data basis of the water quality and water volume scheduling model, the prediction accuracy of the water quality model for the change law of water quality indicators is improved.

[0079] The water quality and water volume scheduling model is controlled in the form of a circulating pump linkage scheduling, including:

[0080] Set an adjustable-speed circulating pump in the water body to circulate and expose the air intake flow rate of the water body, and dispatch an intelligent microfilter to purify the water body according to the water quality and water volume scheduling data, and monitor the water quality parameters in real time;

[0081] During the water body circulation and purification process, use the water quality parameters as independent variables and the filtration times of the intelligent microfilter as dependent variables to obtain the water quality and water volume scheduling model by fitting curves;

[0082] According to the water quality and water volume scheduling model, count the factors affecting the water quality, and use the support vector machine SVM to process the corresponding factor data to predict the decision data points affecting the water quality and water volume.

[0083] In this embodiment, using the support vector machine (SVM) to process the corresponding factor data helps improve the prediction ability of the model and minimize the errors that may randomly occur when classifying statistical data.

[0084] Taking the nutrients of the aquatic organisms as the biofilm carrier, a water quality purification model based on the biofilm carrier is established, including:

[0085] Select a carrier material suitable for the growth of aquatic organisms, inoculate microorganisms to attach and grow to form a biofilm, place the biofilm on the side of the adjustable-speed circulation pump to ensure that the water flow can fully contact the surface of the carrier;

[0086] Use the Monod equation to establish the rate equation for microorganisms to remove organic pollutants, and obtain the water quality dilution rate, filtration cycle, and pollutant degradation efficiency;

[0087] Use the support vector machine model to classify the decision data points for the rate equation, obtain the optimal position points of the decision data points, and perform cross-validation on the optimal position points through the kernel function. The expression of the kernel function is:

[0088] K(a,b) = exp(-γ‖a - b‖ 2 )

[0089] where a and b respectively represent the position coordinates of the decision data points, and γ represents a positive parameter of linear correlation;

[0090] Use GridSearchCV to optimize the parameter γ, predict the optimal data points through cross-validation, and establish a water quality purification model.

[0091] In this embodiment, use the support vector machine model to obtain the data points closest to the decision surface, generate the optimal classification hyperplane through cross-validation, perform linear programming on the optimal classification hyperplane to obtain the training vector set that separates two separate classes, use the training vector set as the optimal position points, and when the optimal classification surface is generated, the vectors will be classified correctly. When redundancy occurs, use the kernel function to constrain and obtain the hyperplane that optimizes the data to obtain the water quality purification data.

[0092] Based on the water quality purification and breeding model, use the deep learning algorithm to detect the quantity and species of the aquatic organisms, including:

[0093] Collect water quality parameters, pH values, temperature and other parameters as training data according to the optimal position points of the decision data points. At the same time, take real-time water body video data at the optimal position points of the decision data points, use the training data as the water body sample parameters, and combine the known quantity and species of organisms to perform data preprocessing on the water body sample parameters;

[0094] In this embodiment, by monitoring the changes in water quality parameters and biological populations in real time, the system can dynamically adapt to environmental changes. Even when environmental conditions change, it can accurately monitor and predict the quantity and types of organisms.

[0095] The water sample parameters after data preprocessing are used to construct a deep learning algorithm detection model with a long short-term memory network to extract data features.

[0096] In this embodiment, a deep learning algorithm detection model is constructed with a long short-term memory network. As more data is collected and the model is continuously trained, the accuracy of the detection model will gradually improve. Using supervised learning to predict the training data in the form of labels helps the model to more accurately identify the data features of different organisms.

[0097] The water body video data is analyzed using a convolutional neural network (CNN) to identify different biological data.

[0098] The deep learning algorithm detection model is used to train the different biological data and data features. Through supervised learning, the training data is predicted in the form of labels to obtain the quantity and types of organisms in a given sample.

[0099] Monitor the changes in water quality and biological populations in real time, and regularly predict the types and quantities of organisms by continuously optimizing the detection model.

[0100] In this embodiment, by precisely monitoring the quantity, types of organisms, and water quality conditions, water resources can be managed more efficiently, ensuring that the aquaculture environment is in the best state and reducing unnecessary resource waste. Potential problems, such as water quality deterioration or an abnormal increase in the quantity of a certain organism, can be quickly identified, allowing managers to take timely measures to solve the problems and protect the ecological environment balance.

[0101] Construct a biomass assessment module based on the types and quantities of organisms to obtain the real-time biodiversity data of the water body organisms, including:

[0102] Obtain the types and quantities of organisms in the water body by monitoring the changes in water quality and biological populations in real time, and use redundancy analysis to monitor the correlation between species data and environmental factors.

[0103] In this embodiment, by using redundancy analysis to monitor the correlation between species data and environmental factors, it is possible to deeply understand how different environmental factors affect the types and quantities of organisms in the water body, providing a more comprehensive perspective to observe the entire ecosystem.

[0104] Construct a correlation analysis with environmental factors from two dimensions of biological species and quantity respectively using a correlation heatmap to obtain the influence weights of the environmental factors on the biological species and quantity;

[0105] In this embodiment, constructing a correlation analysis with environmental factors from two dimensions of biological species and quantity respectively using a correlation heatmap can more accurately quantify the influence degree of each environmental factor on organisms, thereby improving the accuracy of biomass assessment. Moreover, the correlation heatmap provides an intuitive data display method, enabling managers to more easily identify which environmental factors have a significant impact on specific biological species or quantity, which helps to make data-based scientific decisions.

[0106] Establish a biomass assessment module in the water body according to the magnitude of the influence weights, and obtain the real-time diversity data of the water body organisms.

[0107] In this embodiment, establishing a biomass assessment module according to the magnitude of the influence weights of environmental factors on biological species and quantity can realize the dynamic adjustment of water quality management and biological protection measures. When some key environmental factors change, corresponding measures can be taken promptly to maintain or improve the water body biodiversity, and resources can be more effectively allocated to those most critical factors, thereby optimizing management efficiency and reducing costs. By monitoring and evaluating the changes in water body biodiversity in real time, it helps to detect and solve potential problems in a timely manner, ensure the health and stability of the ecosystem, and further promote the sustainable development of aquaculture and other industries relying on water resources. Based on the change trend of the influence weights of environmental factors, it is possible to predict in advance possible biodiversity changes or water quality deterioration events, providing a valuable time window for taking preventive measures.

[0108] The biomass assessment module uses the Beta diversity analysis method to compare the diversity between different biological communities, including:

[0109] Obtain the species types at different times and locations in the water body, use the Jaccard index to quantify the species quantity between different samples, and construct an index similarity matrix by calculating the similarity of the species quantity between different samples;

[0110] Use the principal coordinate analysis method for the index similarity matrix to analyze the influence degree of the influence weights of the environmental factors on the species composition;

[0111] Monitor the change situation of the species composition in real time, and obtain the real-time diversity data of the water body organisms.

[0112] In this embodiment, the Jaccard index is used to quantify the similarity of species numbers between different samples, which can accurately measure the differences and similarities between biological communities at different times and locations, providing a quantitative basis for subsequent analysis. The constructed index similarity matrix can clearly display the similarities and differences between different samples, helping to identify which time periods or geographical locations have the most similar or significantly different biological communities. Using principal coordinate analysis, the spatial relationships between samples can be displayed from a multi-dimensional perspective. At the same time, by combining the influence weights of environmental factors, the specific impacts of these environmental factors on species composition can be more accurately evaluated, revealing potential ecological mechanisms, and the relative positions of samples and their community structure characteristics can be intuitively presented, making complex ecological data easier to understand and interpret, and facilitating managers to make data-based decisions.

[0113] The diversity data of the aquatic organisms are calculated using a multiple linear regression equation to establish the relationship between water quality indicators and biomass, including:

[0114] In the stepwise regression of the multiple linear regression equation, the environmental factors are introduced into the model one by one as variables. Each introduced variable passes the F-test. When the explanation of the introduced variable for the multiple linear regression equation is no longer significant, the corresponding variable is deleted;

[0115] The introduced variables are repeatedly and gradually introduced in the form of permutations and combinations to determine that the multiple linear regression equation has the optimal explanatory ability;

[0116] Taking the introduced variables as independent variables, the relationship between water quality indicators and biomass is established.

[0117] In this embodiment, the stepwise introduction and deletion of variables based on the F-test can ensure that the final model only contains environmental factors that have a significant explanatory ability for response variables such as biomass or diversity, helping to avoid overfitting and improving the generalization ability of the model. By repeatedly and gradually introducing variables in the form of permutations and combinations, a variable combination with the best explanatory ability can be found, thus constructing a more accurate and reliable prediction model.

[0118] The multiple linear regression equation takes the environmental factors as independent variables, selects biomass as the target dependent variable and conducts a correlation analysis to establish a prediction model between water quality indicators and biomass, and obtains biomass prediction data. The specific expression of the multiple linear regression equation is:

[0119] G = σ0 + σ1x1 + σ2x2 + …… + σ n x n

[0120] where G represents the biomass in the water body, σ0 represents a constant, and x1, x2…x nRepresent environmental factors affecting water quality parameters in water bodies, σ1, σ2…σ n Represent regression coefficients.

[0121] In this embodiment, a multiple linear regression equation is adopted and the relationship between water quality indicators and biomass is constructed through the stepwise regression method. This can not only improve the accuracy and reliability of the model, but also provide strong support for actual management and scientific research, which is of great significance for effectively protecting and restoring aquatic ecosystems.

[0122] The support vector regression algorithm is used to optimize the parameters of the biomass prediction data, and the mean square error of the parameters is calculated as the influencing factor of the fitness function. Through the cyclic parameter optimization process, the optimal generation environment data of the water body organisms is predicted adaptively.

[0123] In this embodiment, using support vector regression can effectively handle non-linear relationships and is suitable for complex ecosystem modeling, making biomass prediction more accurate. By optimizing the parameters to minimize the mean square error, the prediction accuracy of the model can be significantly improved, ensuring that the deviation between the prediction result and the actual observation value is as small as possible, enabling the model to better adapt to the characteristics of different data sets and ensuring that the model is always in the best state to adapt to the changing ecological environment.

[0124] The present invention adopts multi-source data fusion and deep learning algorithms to accurately identify key influencing factors and their interaction relationships, improves resource utilization efficiency, monitors and analyzes the specific environmental conditions of different water bodies in real time, automatically adjusts control strategies to adapt to changing environmental conditions, ensures the effectiveness and sustainability of control measures, has a monitoring and early warning mechanism, can discover and respond to sudden environmental pollution events or new ecological threats in a timely manner, quickly take effective countermeasures, and prevent the further expansion of ecological damage, ensuring the healthy and stable ecosystem.

[0125] The above embodiments are only exemplary embodiments of the present application and are not used to limit the present application. The protection scope of the present application is defined by the claims. Those skilled in the art can make various modifications or equivalent replacements within the essence and protection scope of the present application, and such modifications or equivalent replacements should also be regarded as falling within the protection scope of the present application.

Claims

1. A method for adaptively controlling biodiversity of aquatic bodies, characterized in that: The following steps are involved: Real-time monitoring of the number and distribution of aquatic organisms in water bodies, assessment of the current state of biodiversity in water bodies, and establishment of a water quality and quantity scheduling model after comparison with preset standards; The water quality and water quantity scheduling model adopts the form of circulating pump linkage scheduling to control the water quality, and uses the nutrients of the water organisms as biofilm carriers to establish a water quality purification model based on the biofilm carriers; Based on the water purification breeding model, a deep learning algorithm is used to detect the number and type of the water organisms, a biomass assessment module is constructed, and real-time diversity data of the water organisms is obtained; The multivariate linear regression equation is used to calculate the diversity data of the aquatic organisms, to construct the relationship between water quality indicators and biomass, and to adaptively predict the optimal generation environment data of the aquatic organisms.

2. The method for adaptively controlling water biodiversity according to claim 1, characterized in that: Real-time monitoring of the number and distribution of aquatic organisms in water bodies, assessment of the current state of biodiversity in water bodies, and establishment of a water quality and quantity scheduling model after comparison with preset standards, including: Using optical imaging equipment to monitor the number and species distribution of the aquatic organisms in the water body in real time, and regularly collecting water sample data at different depths and locations, and using the screened and processed water sample data as the collected data of the organisms; Combining the collected data with historical data of aquatic organisms to determine biodiversity indicators suitable for the local ecological environment, the biodiversity indicators include but are not limited to the number range, distribution pattern and ecological function of different species; Compare the real-time monitored biodiversity data with the biodiversity indicators to identify problematic data of the aquatic organisms in the current state; Analyze the reasons for the differences in data acquisition, establish a water quality and quantity scheduling model, predict the changing trend of biodiversity under different management models, and obtain real-time water quality and quantity scheduling data.

3. The method for adaptively controlling water biodiversity according to claim 2, characterized in that: The water quality and water quantity scheduling model adopts the circulating pump linkage scheduling form to control the water quality, including: An adjustable speed circulating pump is arranged in the water body to circulate and expose the air intake flow of the water body, and an intelligent microfilter is dispatched to purify the water body according to the water quality and water quantity dispatching data, and water quality parameters are monitored in real time; In the process of water circulation purification, the water quality parameter is used as an independent variable, the filtering times of the intelligent microfilter is used as a dependent variable, and a water quality and water quantity scheduling model is obtained by fitting a curve; The factors affecting water quality are statistically analyzed according to the water quality and quantity scheduling model, and the corresponding factor data are processed using a support vector machine (SVM) to predict decision data points affecting water quality and quantity.

4. The method for adaptively controlling water biodiversity according to claim 3, characterized in that: Using the nutrients of the aquatic organisms as biofilm carriers, a water purification model based on the biofilm carriers is established, including: Select a carrier material suitable for the growth of aquatic organisms, form a biofilm by inoculating microorganisms to attach and grow, and place the biofilm on the side of the adjustable speed circulation pump to ensure that the water flow can fully contact the carrier surface; The Monod equation was used to establish the rate equation for microbial removal of organic pollutants, and the water quality dilution rate, filtration cycle and pollutant removal efficiency were obtained; The rate equation is classified by using a support vector machine model to classify the decision data points, and the optimal position point of the decision data point is obtained. The optimal position point is cross-validated by a kernel function, and the kernel function expression is: K(a,b)=exp(-γ‖a-b‖ 2 ) Among them, a and b represent the location coordinates of the decision data points, and γ represents a linearly related positive parameter; GridSearchCV was used to optimize the positive parameter γ, and the optimal data points were predicted through cross-validation to establish a water purification model.

5. The method for adaptively controlling water biodiversity according to claim 3, characterized in that: Based on the water purification breeding model, a deep learning algorithm is used to detect the number and type of water organisms, including: Collect water quality parameters, pH value, temperature and other parameters as training data according to the optimal position of the decision data point, and shoot water body video data in real time at the optimal position of the decision data point, use the training data as water body sample parameters, and perform data preprocessing on the water body sample parameters in combination with the known number and type of organisms; The water sample parameters after the data preprocessing are subjected to the deep learning algorithm detection model constructed by using the long short-term memory network to extract data features; The water body video data is analyzed using a convolutional neural network (CNN) to identify different biological data; The deep learning algorithm detection model is used to train the different biological data and data features, and the training data is predicted in the form of labels through supervised learning to obtain the number and type of organisms in a given sample; Monitor changes in water quality and biological populations in real time, and regularly predict biological species and quantities by continuously optimizing detection models.

6. The method for adaptively controlling water biodiversity according to claim 5, characterized in that: A biomass assessment module is constructed based on the species and quantity of the organisms to obtain real-time diversity data of the aquatic organisms, including: By real-time monitoring of water quality and biological population changes, the species and quantity of organisms in the water body are obtained, and the correlation between the species data and environmental factors is monitored using redundancy analysis on the species and quantity of the organisms; A correlation heat map is used to construct a correlation analysis with environmental factors from two dimensions: species and quantity, to obtain the influence weights of the environmental factors on the species and quantity of the organisms; A biomass assessment module in the water body is established according to the size of the impact weight to obtain real-time diversity data of the water body organisms.

7. The method for adaptively controlling water biodiversity according to claim 6, characterized in that: The biomass assessment module uses the Beta diversity analysis method to compare the diversity between different biomes, including: Obtain species at different times and locations in the water body, use the Jaccard index to quantify the number of species between different samples, and construct an index similarity matrix by calculating the similarity of the number of species between the different samples; The principal coordinate analysis method is used to analyze the influence degree of the influence weight of the environmental factors on the species composition on the index similarity matrix; Monitor the changes in species composition in real time and obtain real-time aquatic biological diversity data.

8. The method for adaptively controlling water biodiversity according to claim 7, characterized in that: The multivariate linear regression equation is used to calculate the diversity data of the water body organisms and to construct the relationship between water quality indicators and biomass, including: The environmental factors are introduced into the model as variables one by one by using the stepwise regression of the multiple linear regression equation, and each of the introduced variables passes the F test. When the introduced variables are not significantly explained in the multiple linear regression equation, the corresponding variables are deleted; The introduced variables are repeatedly and stepwise introduced in the form of permutations and combinations to determine that the multivariate linear regression equation has the best explanatory power; The introduced variables are used as independent variables to construct the relationship between water quality indicators and biomass.

9. The method for adaptively controlling water biodiversity according to claim 8, characterized in that: The multivariate linear regression equation takes the environmental factors as independent variables, selects biomass as the target dependent variable and performs correlation analysis, establishes a prediction model between water quality indicators and biomass, and obtains biomass prediction data. The specific expression of the multivariate linear regression equation is: G=σ0+σ1x1+σ2x2+……+σ n x n Among them, G represents the biomass in the water body, σ0 represents a constant, x1, x2…x n Represents the environmental factors that affect water quality parameters in water bodies, σ1, σ2…σ n represents the regression coefficient.

10. The method for adaptively controlling water biodiversity according to claim 9, characterized in that: The support vector regression algorithm is used to optimize the parameters of the biomass prediction data, and the mean square error of the parameters is calculated as the influencing factor of the fitness function. Through a cyclic parameter optimization process, the optimal generation environment data of the aquatic organisms is adaptively predicted.

Images