A mixed-nutrient dinoflagellate bloom outbreak evaluation system based on big data analysis

Abstract

The application discloses a mixed-nutrient dinoflagellate bloom outbreak evaluation system based on big data analysis, which comprises the following units: a dinoflagellate metabolic activity analysis unit, which collects water body environmental parameters, carries out metabolic activity correlation analysis on the mixed-nutrient dinoflagellate based on the water body environmental parameters, and obtains dinoflagellate metabolic activity parameters; and according to the dinoflagellate metabolic activity parameters, algal population spatiotemporal distribution characteristics are extracted to obtain characteristic data of the algal population spatiotemporal distribution. The application relates to the technical field of marine ecological early warning. The mixed-nutrient dinoflagellate bloom outbreak evaluation system based on big data analysis achieves the following effects: reliable basic data are obtained through comprehensive and accurate data collection and analysis; dinoflagellate dynamics are intuitively presented through accurate trajectory reconstruction and density modeling; the accuracy of risk evaluation is improved through scientific critical threshold prediction; the evaluation is more practical through effective environmental factor correction; and the prediction accuracy is improved and timely warning is realized through intelligent prediction and warning by using a fusion algorithm.

Description

Technical Field

[0001] This invention relates to the field of marine ecological early warning, and more specifically, to an assessment system for mixed trophic dinoflagellate blooms based on big data analysis. Background Technology

[0002] With changes in the marine ecological environment and the impact of human activities, marine algal blooms occur frequently. Among them, mixed-trophic dinoflagellate blooms, due to their complex physiological characteristics and ecological impacts, have affected marine ecosystems, fishery resources and human health.

[0003] Traditional methods for monitoring and assessing algal blooms mainly rely on manual sampling and laboratory analysis, which suffer from limitations such as limited monitoring scope, poor timeliness, and inability to reflect the dynamic changes of algal blooms in real time. Furthermore, for dynotrophic dinoflagellates, their metabolic activity is influenced by multiple environmental factors, and traditional assessment methods struggle to comprehensively consider these complex factors, resulting in low accuracy in predicting algal blooms and failing to meet the needs of practical ecological monitoring and early warning.

[0004] In recent years, the rapid development of big data technology, sensor technology, and artificial intelligence algorithms has provided new means and methods for marine ecological monitoring. Through multi-source data fusion and intelligent algorithm analysis, more comprehensive and accurate information on the marine environment and the growth status of dinoflagellates can be obtained, thereby enabling precise assessment and timely early warning of mixed-trophic dinoflagellate blooms. However, currently, there is no complete big data-based assessment system for mixed-trophic dinoflagellate blooms that can comprehensively consider multiple aspects of information, including aquatic environmental parameters, dinoflagellate metabolic activity, hydrodynamic environmental factors, and meteorological factors, to achieve dynamic assessment and prediction of blooms. Summary of the Invention

[0005] To address the lack of a comprehensive big data-based assessment system for mixed-trophic dinoflagellate blooms that can dynamically assess and predict blooms, this invention aims to provide such a system. This system achieves robust data collection and analysis to acquire reliable baseline data, accurately reconstruct trajectories and density modeling to visually represent dinoflagellate dynamics, scientifically predict critical thresholds to improve risk assessment accuracy, effectively correct environmental factors to make assessments more realistic, and utilize intelligent prediction and early warning algorithms to enhance prediction precision and provide timely warnings. This comprehensive and powerful system offers robust support for marine ecological protection and fisheries resource management.

[0006] To solve the above problems, the present invention adopts the following technical solution.

[0007] A big data analysis-based assessment system for mixed-trophic dinoflagellate blooms includes the following units: The dinoflagellate metabolic activity analysis unit collects aquatic environmental parameters and performs metabolic activity correlation analysis between these parameters and mixed-trophic dinoflagellates to obtain dinoflagellate metabolic activity parameters. Based on these parameters, it extracts the spatiotemporal distribution characteristics of the algal community to obtain characteristic data of the spatiotemporal distribution of the algal community. The density dynamic modeling unit performs three-dimensional reconstruction of the dinoflagellate movement trajectory based on the feature data to obtain the trajectory data of the dinoflagellate movement, and establishes an evolution model of the algal community aggregation density in combination with the trajectory data. The hydrodynamic environmental factor coupling correction unit acquires water flow field dynamics data and uses the water flow field dynamics data to perform spatial diffusion correction on the evolution model to obtain the corrected algal community diffusion model. The algal bloom critical threshold prediction unit predicts the algal bloom critical threshold based on the modified algal community diffusion model to obtain initial bloom risk assessment data; it acquires meteorological forecast data and combines the meteorological forecast data with the initial bloom risk assessment data to perform environmental stress factor superposition analysis to obtain multi-factor coupled algal bloom risk assessment data. The intelligent prediction unit uses a long short-term memory network convolutional neural network fusion algorithm to predict the multi-factor coupled outbreak risk assessment data and construct an algal bloom outbreak assessment model; the algal bloom outbreak assessment model is deployed to a cloud platform for algal bloom outbreak early warning.

[0008] Furthermore, the water environment parameters include light intensity, nutrient concentration, and dissolved oxygen content.

[0009] Furthermore, based on the metabolic activity parameters of the dinoflagellates, the spatiotemporal distribution characteristics of the algal community are extracted to obtain characteristic data of the spatiotemporal distribution of the algal community, including the following steps: S1. A multi-scale spatiotemporal filtering algorithm is used to preprocess the metabolic activity parameters of dinoflagellates to eliminate noise interference and highlight key information. S11. Classify dinoflagellates with similar metabolic activity characteristics through cluster analysis to determine the spatiotemporal distribution range of different categories of dinoflagellates. S12. Extract key characteristic parameters of the distribution center, diffusion range, and distribution density of various dinoflagellates to form spatiotemporal distribution characteristic data of algal communities.

[0010] Furthermore, based on the aforementioned feature data, a three-dimensional reconstruction of the dinoflagellate movement trajectory is performed to obtain the trajectory data of the dinoflagellate movement. An evolutionary model of algal community aggregation density is then established using this trajectory data, including the following steps: S2. Using high-precision three-dimensional positioning technology, combined with the spatiotemporal distribution feature data, the individual dinoflagellates are tracked and located in real time to obtain their motion coordinates in three-dimensional space. S21. A trajectory interpolation algorithm based on fluid dynamics is used to optimize the positioning data, fill in the missing data parts, and make the trajectory more continuous and smooth. S22. Based on the dinoflagellate movement trajectory data, spatial statistical methods are used to analyze the degree of aggregation and movement direction of dinoflagellates in different regions, and an evolution model of algal community aggregation density is constructed.

[0011] Furthermore, water flow field dynamics data are acquired, and the spatial diffusion correction is performed on the algal community aggregation density evolution model established by the density dynamic modeling unit using the water flow field dynamics data to obtain the corrected algal community diffusion model, including the following steps: S3. Using an underwater acoustic Doppler current meter and remote sensing technology, acquire water flow field dynamics data, including flow velocity, flow direction, and vorticity parameters. S31. Use computational fluid dynamics to numerically simulate the water flow field and construct a refined water flow model. S32. Couple the algal community diffusion process in the algal community aggregation density evolution model established by the density dynamic modeling unit with the water flow model to obtain the modified algal community diffusion model.

[0012] Furthermore, based on the modified algal community diffusion model, the critical threshold for algal bloom is predicted to obtain initial bloom risk assessment data, including the following steps: S4. Collect historical algal bloom data and corresponding environmental parameters to construct an algal bloom database; S41. Use data mining techniques to extract key feature indicators related to the critical state of algal blooms from the database. S42. Compare and analyze the output results of the modified algal bloom diffusion model with these key characteristic indicators, and use the fuzzy comprehensive evaluation method to determine the contribution of different characteristic indicators to the algal bloom. S43. Establish a dynamic prediction model for the critical threshold of algal bloom based on contribution. S44. Substitute the current modified algal bloom diffusion model data into the dynamic prediction model to calculate the initial risk assessment data for the algal bloom outbreak.

[0013] Furthermore, after obtaining the initial outbreak risk assessment data, meteorological forecast data is acquired, and environmental stress factor overlay analysis is performed on the initial risk assessment data of algal bloom obtained in S44 in conjunction with the meteorological forecast data, including the following steps: S5. The meteorological forecast data includes temperature, light intensity, precipitation and air pressure, and the mechanism by which each meteorological element acts as an environmental stress factor is determined. S51. Construct a Bayesian network model of the effects of environmental stress factors; S52. Input the meteorological forecast data into the action model to calculate the environmental stress index for dinoflagellate growth under different meteorological conditions; S53. Overlay and analyze the environmental stress index with the initial risk assessment data; S54. Through a multi-factor comprehensive evaluation method, multi-factor coupled algal bloom risk assessment data are obtained.

[0014] Furthermore, an algal bloom assessment model is constructed by predicting the multi-factor coupled outbreak risk assessment data using a long short-term memory network convolutional neural network fusion algorithm, including the following steps: S6. Preprocess the algal bloom risk assessment data and use data standardization and normalization methods to eliminate dimensional differences between different data. S61. A long short-term memory network-convolutional neural network fusion architecture is adopted, which combines the feature extraction capability of convolutional neural networks with the time series processing capability of long short-term memory networks to automatically learn the spatial features and temporal dependencies in the processed algal bloom risk assessment data. S62. The fusion model is trained using a large amount of historical algal bloom data and corresponding multi-factor coupling data. The model parameters are adjusted using optimization algorithms to improve the prediction accuracy and generalization ability of the model. S63. Validate and test the trained model, and evaluate its performance metrics.

[0015] Compared with the prior art, the advantages of this invention are: 1. By collecting water environment parameters and conducting correlation analysis with dinoflagellates' metabolic activity, we obtained dinoflagellate metabolic activity parameters, and then extracted spatiotemporal distribution characteristics of algal communities. This comprehensive approach takes into account various factors affecting dinoflagellate growth and algal blooms, making the data sources more comprehensive and the analysis results more accurate, thus providing a reliable foundation for subsequent assessment and prediction.

[0016] 2. Based on the spatiotemporal distribution characteristics of algal communities, the movement trajectory of dinoflagellates is reconstructed in three dimensions to obtain the trajectory data of dinoflagellates. Combined with the trajectory data, an evolutionary model of algal community aggregation density is established, which can intuitively reflect the movement and aggregation degree of dinoflagellates in three-dimensional space, providing an important basis for the dynamic assessment of algal blooms.

[0017] 3. Based on the algal community aggregation density evolution model, the critical threshold for algal blooms is predicted. By collecting historical data to build a database, data mining technology and fuzzy comprehensive evaluation method are used to determine the contribution of key characteristic indicators and establish a dynamic prediction model. This model can scientifically predict the critical threshold for algal blooms by combining historical experience and actual conditions, thereby improving the accuracy of initial outbreak risk assessment.

[0018] 4. Obtain water flow field dynamics data and meteorological forecast data. Use water flow field dynamics data to perform spatial diffusion correction on the evolution model. Combine meteorological forecast data to perform superposition analysis of environmental stress factors, and obtain multi-factor coupled outbreak risk assessment data. Consider the impact of environmental factors such as hydrodynamics and meteorology on dinoflagellate growth and algal blooms, making the assessment results more consistent with the actual situation and improving the reliability of the assessment.

[0019] 5. Combining the feature extraction capabilities of convolutional neural networks and the time series processing capabilities of long short-term memory networks, it can automatically learn the spatial features and temporal dependencies in data, improve the accuracy and generalization ability of prediction, achieve timely early warning of algal blooms, and provide strong support for marine ecological protection and fishery resource management. Attached Figure Description

[0020] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments of this application and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1 This is a system block diagram of the present invention; Figure 2 This is a flowchart of the dinoflagellate metabolic activity analysis unit of the present invention; Figure 3 This is a flowchart of the intelligent prediction unit of the present invention.

[0021] Explanation of the labels in the diagram: 1. Dinoflagellate metabolic activity analysis unit; 2. Density dynamic modeling unit; 3. Algal bloom critical threshold prediction unit; 4. Hydrodynamic environmental factor coupling correction unit; 5. Intelligent prediction unit. Detailed Implementation

[0022] 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 a part of the embodiments of the present invention, and not all of them. 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. Example

[0023] Please see Figure 1-3This invention provides a technical solution: a system for assessing mixed-trophic dinoflagellate blooms based on big data analysis, characterized by comprising the following units: Unit 1, which analyzes the metabolic activity of dinoflagellates, collects water environment parameters and performs a correlation analysis between these parameters and the metabolic activity of mixed-trophic dinoflagellates to obtain metabolic activity parameters. Based on these parameters, the spatiotemporal distribution characteristics of the algal community are extracted to obtain characteristic data of the spatiotemporal distribution of the algal community. Density dynamic modeling unit 2 reconstructs the three-dimensional movement trajectory of dinoflagellates based on feature data to obtain the trajectory data of dinoflagellates movement, and establishes an evolutionary model of algal community aggregation density by combining the trajectory data. Hydrodynamic environmental factor coupling correction unit 4 acquires water flow field dynamics data, and uses the water flow field dynamics data to perform spatial diffusion correction on the evolution model to obtain the corrected algal community diffusion model. Algal bloom critical threshold prediction unit 3 predicts the critical threshold of algal bloom based on the modified algal community diffusion model to obtain initial bloom risk assessment data; it acquires meteorological forecast data and combines the meteorological forecast data with the initial bloom risk assessment data to perform environmental stress factor superposition analysis to obtain multi-factor coupled algal bloom risk assessment data. The intelligent prediction unit 5 uses a long short-term memory network convolutional neural network fusion algorithm to predict multi-factor coupled outbreak risk assessment data and construct an algal bloom outbreak assessment model; the algal bloom outbreak assessment model is deployed to the cloud platform for algal bloom outbreak early warning.

[0024] It should be noted that, during the analysis, dinoflagellates, being phytoplankton, are typically an important component of aquatic ecosystems, capable of photosynthesis. Metabolic activity refers to the growth, division, and metabolic activities of dinoflagellates under different environmental conditions. Analyzing these metabolic activities allows for the assessment of their growth trends and potential impacts on water bodies. Aquatic environmental parameters, including water temperature, pH, dissolved oxygen, salinity, nutrients, and light intensity, all influence the growth and metabolic activity of dinoflagellates. The spatiotemporal distribution characteristics of the algal community refer to the distribution of dinoflagellate communities across different times and spaces; these distribution characteristics are crucial for predicting outbreak risks. Three-dimensional reconstruction of dinoflagellate movement trajectories is performed by monitoring and calculating... Algal movement in water can reconstruct their trajectories in three-dimensional space, helping to assess their propagation patterns; evolutionary models of aggregation density describe the spatial aggregation and diffusion processes of dinoflagellates, helping to predict changes in dinoflagellate concentration in water bodies and their impact on the aquatic environment; algal blooms are a phenomenon of large-scale, rapid algal proliferation in water bodies, which usually leads to water quality deterioration, such as reduced dissolved oxygen, eutrophication, and impact on the health of aquatic organisms; the critical threshold refers to the value at which the density of dinoflagellates or other relevant parameters reaches a certain level, triggering an algal bloom; predicting this critical value can help provide early warning; initial outbreak risk assessment data is based on the critical value predicted by the model to assess the potential for an outbreak. In terms of risk; water flow dynamics data describes the internal flow state of water bodies, including information such as flow velocity, direction, and turbulence. This data helps understand how dinoflagellates spread and accumulate under the influence of water flow; meteorological forecast data includes meteorological factors such as temperature, precipitation, wind speed, and wind direction, which have a significant impact on the physical properties of water bodies and the growth of dinoflagellates; spatial diffusion correction is based on flow field data, adjusting the diffusion component in the evolution model to simulate the actual diffusion of dinoflagellates in water bodies; environmental stress factors refer to the pressure of external environmental factors on the growth of dinoflagellates; long short-term memory networks are a special type of recurrent neural network used to process time series data, effectively capturing dependencies over long time spans. Reliability is suitable for prediction tasks; Convolutional Neural Networks (CNNs) are a deep learning algorithm, commonly used for image recognition, but can also be used to process other types of structured data. Through convolution operations, local features can be extracted from the data; The fusion algorithm of CNNs and Long Short-Term Memory (LSTM) networks refers to the combination of these two algorithms, which can better handle complex data with both spatial and temporal features, improving the model's prediction accuracy; Algal bloom assessment models are mathematical models built by integrating various data and analysis methods, capable of predicting the probability and possible severity of algal blooms; A cloud platform is a cloud computing-based service platform that can store and process large amounts of data and provide real-time prediction and alert services.

[0025] In one embodiment, the aquatic environmental parameters include light intensity, nutrient concentration, and dissolved oxygen content.

[0026] The study investigated the effects of light intensity on dinoflagellates. Dinoflagellates synthesize organic matter through photosynthesis; higher light intensity leads to stronger photosynthesis and faster growth (before the light saturation point). Insufficient light slows down metabolism and halts growth; however, excessive light (beyond the light saturation point) can cause photoinhibition or even cell damage. Nutrient concentration, particularly nitrogen and phosphorus, is crucial for phytoplankton to synthesize proteins, nucleic acids, and membrane structures. A lack of nutrients inhibits algal reproduction, while excessive nutrients can trigger algal blooms. Dissolved oxygen is directly related to the metabolic activities of dinoflagellates. During photosynthesis, dinoflagellates release oxygen; however, they consume oxygen at night or during respiration.

[0027] In one embodiment, the spatiotemporal distribution characteristics of the algal community are extracted based on the metabolic activity parameters of dinoflagellates to obtain characteristic data of the spatiotemporal distribution of the algal community, including the following steps: S1. A multi-scale spatiotemporal filtering algorithm is used to preprocess the metabolic activity parameters of dinoflagellates to eliminate noise interference and highlight key information. S11. Classify dinoflagellates with similar metabolic activity characteristics through cluster analysis to determine the spatiotemporal distribution range of different categories of dinoflagellates. S12. Extract key characteristic parameters such as distribution center, diffusion range, and distribution density of various dinoflagellates to form spatiotemporal distribution characteristic data of algal communities.

[0028] Through design, S1 preprocesses the metabolic activity parameters of dinoflagellates using a multi-scale spatiotemporal filtering algorithm to remove noise interference from the observation data, retain effective information at different scales, and highlight the true changing characteristics of dinoflagellate activity. S11 uses cluster analysis to classify dinoflagellate populations with similar metabolic characteristics, which helps to identify their distribution patterns in the spatiotemporal dimension. S12 further extracts key parameters such as the distribution center, diffusion range, and density of each type of population to construct spatiotemporal distribution characteristic data of algal communities, providing data support for the study of dinoflagellate dynamic evolution and early warning models.

[0029] In one embodiment, a three-dimensional reconstruction of the dinoflagellate movement trajectory is performed based on feature data to obtain the trajectory data of the dinoflagellate movement. An evolutionary model of the algal community aggregation density is then established using this trajectory data, including the following steps: S2. Using high-precision three-dimensional positioning technology, combined with spatiotemporal distribution feature data, we can track and locate individual dinoflagellates in real time and obtain their motion coordinates in three-dimensional space. S21. A trajectory interpolation algorithm based on fluid dynamics is used to optimize the positioning data, fill in the missing data parts, and make the trajectory more continuous and smooth. Cubic spline interpolation: ; Among them, coefficient , , , Determined by boundary conditions, It is the position of the dinoflagellate at time t. These are adjacent sampling time points. , , , These are spline coefficients, ensuring the continuity of the first and second derivatives of the trajectory.

[0030] S22. Based on the dinoflagellate movement trajectory data, spatial statistical methods are used to analyze the degree of aggregation and movement direction of dinoflagellates in different regions, and an evolution model of algal community aggregation density is constructed.

[0031]

[0032] , These are the spatial coordinates of the i-th individual dinoflagellate; It is a bandwidth parameter (the larger the value, the smoother the density estimation); Estimated algal community density at spatial point (x,y).

[0033] By designing and utilizing high-precision 3D positioning technology, the 3D spatial coordinates of individual dinoflagellates were obtained. The positioning data was optimized using a trajectory interpolation algorithm based on fluid dynamics to make the trajectory continuous and smooth. Based on the trajectory data, spatial statistical methods were used to analyze the degree of aggregation and direction of movement of dinoflagellates, and an evolution model of algal community aggregation density was constructed. The 3D reconstruction can accurately present the movement trajectory of dinoflagellates, and the interpolation algorithm makes up for data gaps to make the trajectory complete, providing a reliable data foundation for subsequent research on algal community dynamics and algal blooms.

[0034] In one embodiment, water flow field dynamics data is acquired, and the spatial diffusion correction is performed on the algal community aggregation density evolution model established by the density dynamic modeling unit using the water flow field dynamics data to obtain a corrected algal community diffusion model, including the following steps: S3. Using an underwater acoustic Doppler current meter and remote sensing technology, acquire water flow field dynamics data, including flow velocity, flow direction, and vorticity parameters. S31. Use computational fluid dynamics to numerically simulate the water flow field and construct a refined water flow model. S32. Couple the algal community diffusion process in the algal community aggregation density evolution model established by the density dynamic modeling unit with the water flow model to obtain the modified algal community diffusion model.

[0035] ; C is the algal community concentration, u is the water flow velocity, D is the diffusion coefficient, r is the growth rate, and K is the environmental carrying capacity.

[0036] By designing and using underwater acoustic Doppler current meters and remote sensing technology to obtain parameters such as flow velocity, flow direction, and vorticity of the water flow field, a refined water flow model is constructed using computational fluid dynamics methods. The diffusion process in the algal community aggregation density evolution model is coupled with the water flow model. The water flow field affects the algal community diffusion, and by acquiring its dynamic data and constructing the model, the water flow can be accurately simulated. By coupling the algal community diffusion model with this model, the original model can be corrected, making the corrected algal community diffusion model more consistent with the actual situation and improving the accuracy of algal community diffusion prediction.

[0037] In one embodiment, the algal bloom critical threshold is predicted based on the modified algal community diffusion model to obtain initial bloom risk assessment data, including the following steps: S4. Collect historical algal bloom data and corresponding environmental parameters to construct an algal bloom database; S41. Use data mining techniques to extract key feature indicators related to the critical state of algal blooms from the database. S42. Compare and analyze the output results of the modified algal bloom diffusion model with these key characteristic indicators, and use the fuzzy comprehensive evaluation method to determine the contribution of different characteristic indicators to the algal bloom. S43. Establish a dynamic prediction model for the critical threshold of algal bloom based on contribution. S44. Substitute the current modified algal bloom diffusion model data into the dynamic prediction model to calculate the initial risk assessment data for the algal bloom outbreak.

[0038] By designing and collecting historical algal bloom data and environmental parameters to construct a database, data mining techniques are used to extract key characteristic indicators related to the critical state of algal blooms. The output results of the modified algal community diffusion model are compared with the key indicators, and the contribution of each indicator to the algal bloom is determined using the fuzzy comprehensive evaluation method. Then, a dynamic prediction model is established, and the initial risk assessment data is obtained by substituting the current data. Based on the analysis of historical data and model output, key influencing factors can be identified. The dynamic prediction model can adapt to different situations, making the initial risk assessment data more scientific and accurate, and providing a reliable basis for algal bloom prevention and control.

[0039] In one embodiment, after obtaining the initial outbreak risk assessment data, meteorological forecast data is then acquired, and environmental stress factor overlay analysis is performed on the initial outbreak risk assessment data in conjunction with the meteorological forecast data. The specific steps are as follows: S5. The meteorological forecast data includes temperature, light intensity, precipitation and air pressure, and the mechanism by which each meteorological element acts as an environmental stress factor is determined. S51. Construct a Bayesian network model of the effects of environmental stress factors; S52. Input the meteorological forecast data into the action model to calculate the environmental stress index for dinoflagellate growth under different meteorological conditions; S53. Overlay and analyze the environmental stress index with the initial risk assessment data; S54. Through a multi-factor comprehensive evaluation method, multi-factor coupled algal bloom risk assessment data are obtained.

[0040] After obtaining initial risk assessment data through design, meteorological forecast data including temperature, light intensity, precipitation, and air pressure are introduced to determine the role of each meteorological element as an environmental stress factor. A Bayesian network interaction model is constructed, and the environmental stress index is calculated by inputting meteorological data. This index is then superimposed on the initial data. Finally, a multi-factor coupled risk assessment data for algal blooms is obtained through multi-factor comprehensive evaluation. Meteorological factors have a significant impact on algal blooms. Combining them with the initial assessment data can comprehensively consider multiple factors. Multi-factor coupling makes the assessment results more accurately reflect the actual situation and provides more precise support for algal bloom prevention and control decisions.

[0041] In one embodiment, a multi-factor coupled outbreak risk assessment model is constructed by using a long short-term memory network convolutional neural network fusion algorithm to predict algal bloom outbreaks, including the following steps: S6. Preprocess the data for the risk assessment of algal blooms, and use data standardization and normalization methods to eliminate the dimensional differences between different data. S61. A long short-term memory network-convolutional neural network fusion architecture is adopted, which combines the feature extraction capability of convolutional neural networks with the time series processing capability of long short-term memory networks to automatically learn the spatial features and temporal dependencies in the processed algal bloom risk assessment data. Formulas related to Long Short-Term Memory Networks An LSTM cell contains input gates. Forgotten Gate Output gate and cell state and hidden state .

[0042] The input gate determines how much of the input information at the current moment can enter the cell state: ; in, It is the input at the current moment. It is the hidden state from the previous moment. , It is a weight matrix. It is a bias term. It is the sigmoid activation function.

[0043] The forgetting gate determines how much information from the cell's previous state needs to be forgotten: ; in, , It is a weight matrix. It is a bias term.

[0044] Cell state update formula:

[0045] in, This represents element-wise multiplication; tanh is the hyperbolic tangent activation function.

[0046] in, , It is a weight matrix. It is a bias term.

[0047] The output gate determines how much information about the cell's current state is output to the hidden state:

[0048] in, , It is a weight matrix. It is a bias term.

[0049] Hidden state update formula:

[0050] Formulas related to convolutional neural networks: Assuming the input data is X and the convolution kernel is K, the formula for calculating the output Y after the convolution operation is:

[0051] Where M and N are the length and width of the convolution kernel, respectively, and b is the bias term.

[0052] Pooling operations (taking max pooling as an example): Assume the pooling window size is p×p, the input feature map is F, and the output feature map contains G elements. The calculation formula is:

[0053] Output formula of the Long Short-Term Memory Network Convolutional Neural Network Fusion Algorithm Let the output of the LSTM part be The output of the CNN part is The two are then fused to obtain the fused feature vector. :

[0054] Then, a fully connected layer is used for the final prediction. Assume the weight matrix of the fully connected layer is... The bias term is The prediction result The calculation formula is:

[0055] S62. The fusion model is trained using a large amount of historical algal bloom data and corresponding multi-factor coupling data. The model parameters are adjusted using optimization algorithms to improve the prediction accuracy and generalization ability of the model. During model training, mean squared error (MSE) is used as the loss function L:

[0056] Where N is the number of samples, It is the actual value. These are predicted values. Optimization algorithms are used to adjust the model parameters based on the loss function to minimize the loss function value, thereby improving the model's prediction accuracy and generalization ability.

[0057] S63. Validate and test the trained model, and evaluate its performance metrics.

[0058] Through design, S6 preprocesses the algal bloom risk assessment data by standardizing and normalizing it to eliminate differences in data units and improve model processing efficiency. S61 constructs a fusion architecture, combining the spatial feature extraction capabilities of convolutional neural networks with the time series processing advantages of long short-term memory networks to achieve automatic learning of the spatiotemporal characteristics of algal bloom risk data. S62 trains the model using a large amount of historical algal bloom records and their multi-factor data, continuously adjusting parameters through algorithm optimization to improve prediction accuracy and model generalization ability. S63 validates and tests the trained model, evaluating its accuracy, stability, and other performance indicators to ensure its reliability and practicality in real-world applications.

[0059] Those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0060] The above embodiments provide a detailed description of the present invention. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A system for assessing mixed-trophic dinoflagellate blooms based on big data analysis, characterized in that, Includes the following units: The dinoflagellate metabolic activity analysis unit (1) collects water environment parameters, performs metabolic activity correlation analysis between the water environment parameters and mixed trophic dinoflagellates to obtain dinoflagellate metabolic activity parameters; and extracts the spatiotemporal distribution characteristics of the algal community based on the dinoflagellate metabolic activity parameters to obtain the spatiotemporal distribution characteristic data of the algal community. Density dynamic modeling unit (2) performs three-dimensional reconstruction of the movement trajectory of dinoflagellates based on the feature data to obtain the movement trajectory data of dinoflagellates, and establishes an evolution model of algal community aggregation density in combination with the trajectory data; The hydrodynamic environmental factor coupling correction unit (4) acquires water flow field dynamics data and uses the water flow field dynamics data to perform spatial diffusion correction on the evolution model to obtain the corrected algal community diffusion model. The algal bloom critical threshold prediction unit (3) predicts the algal bloom critical threshold according to the modified algal community diffusion model to obtain the initial bloom risk assessment data; obtains meteorological forecast data, and combines the meteorological forecast data with the initial bloom risk assessment data to perform environmental stress factor superposition analysis to obtain multi-factor coupled algal bloom risk assessment data. The intelligent prediction unit (5) uses a long short-term memory network convolutional neural network fusion algorithm to predict the multi-factor coupled outbreak risk assessment data and construct an algal bloom outbreak assessment model; the algal bloom outbreak assessment model is deployed to the cloud platform for algal bloom outbreak early warning, including the following steps: S6. Preprocess the algal bloom risk assessment data and use data standardization and normalization methods to eliminate dimensional differences between different data. S61. A long short-term memory network-convolutional neural network fusion architecture is adopted, which combines the feature extraction capability of convolutional neural networks with the time series processing capability of long short-term memory networks to automatically learn the spatial features and temporal dependencies in the processed algal bloom risk assessment data. S62. The fusion model is trained using a large amount of historical algal bloom data and corresponding multi-factor coupling data. The model parameters are adjusted using optimization algorithms to improve the prediction accuracy and generalization ability of the model. S63. Validate and test the trained model, and evaluate its performance metrics.

2. The system for assessing mixed-trophic dinoflagellate blooms based on big data analysis according to claim 1, characterized in that, The water environment parameters include light intensity, nutrient concentration, and dissolved oxygen content.

3. The system for assessing mixed-trophic dinoflagellate blooms based on big data analysis according to claim 2, characterized in that, Based on the metabolic activity parameters of the dinoflagellates, the spatiotemporal distribution characteristics of the algal community are extracted to obtain characteristic data of the spatiotemporal distribution of the algal community, including the following steps: S1. A multi-scale spatiotemporal filtering algorithm is used to preprocess the metabolic activity parameters of dinoflagellates to eliminate noise interference and highlight key information. S11. Classify dinoflagellates with similar metabolic activity characteristics through cluster analysis to determine the spatiotemporal distribution range of different categories of dinoflagellates. S12. Extract key characteristic parameters of the distribution center, diffusion range, and distribution density of various dinoflagellates to form spatiotemporal distribution characteristic data of algal communities.

4. The system for assessing mixed-trophic dinoflagellate blooms based on big data analysis according to claim 3, characterized in that, Based on the aforementioned feature data, a three-dimensional reconstruction of the dinoflagellate movement trajectory is performed to obtain the trajectory data of the dinoflagellate movement. An evolutionary model of algal community aggregation density is then established using this trajectory data, including the following steps: S2. Using high-precision three-dimensional positioning technology, combined with the spatiotemporal distribution feature data, the individual dinoflagellates are tracked and located in real time to obtain their motion coordinates in three-dimensional space. S21. A trajectory interpolation algorithm based on fluid dynamics is used to optimize the positioning data, fill in the missing data parts, and make the trajectory more continuous and smooth. S22. Based on the dinoflagellate movement trajectory data, spatial statistical methods are used to analyze the degree of aggregation and movement direction of dinoflagellates in different regions, and an evolution model of algal community aggregation density is constructed.

5. The system for assessing mixed-trophic dinoflagellate blooms based on big data analysis according to claim 4, characterized in that, Acquire water flow field dynamics data, and use the water flow field dynamics data to perform spatial diffusion correction on the algal community aggregation density evolution model established by the density dynamic modeling unit (2) to obtain the corrected algal community diffusion model, including the following steps: S3. Using an underwater acoustic Doppler current meter and remote sensing technology, acquire water flow field dynamics data, including flow velocity, flow direction, and vorticity parameters. S31. Use computational fluid dynamics to numerically simulate the water flow field and construct a refined water flow model. S32. Couple the algal community diffusion process in the algal community aggregation density evolution model established by the density dynamic modeling unit (2) with the water flow model to obtain the modified algal community diffusion model.

6. The system for assessing mixed-trophic dinoflagellate blooms based on big data analysis according to claim 5, characterized in that, Based on the modified algal community diffusion model, the critical threshold for algal blooms is predicted to obtain initial bloom risk assessment data, including the following steps: S4. Collect historical algal bloom data and corresponding environmental parameters to construct an algal bloom database; S41. Use data mining techniques to extract key feature indicators related to the critical state of algal blooms from the database. S42. Compare and analyze the output results of the modified algal bloom diffusion model with these key characteristic indicators, and use the fuzzy comprehensive evaluation method to determine the contribution of different characteristic indicators to the algal bloom. S43. Establish a dynamic prediction model for the critical threshold of algal bloom based on contribution. S44. Substitute the current modified algal bloom diffusion model data into the dynamic prediction model to calculate the initial risk assessment data for the algal bloom outbreak.

7. The system for assessing mixed-trophic dinoflagellate blooms based on big data analysis according to claim 6, characterized in that, After obtaining the initial outbreak risk assessment data, meteorological forecast data is acquired, and environmental stress factor overlay analysis is performed on the initial algal bloom outbreak risk assessment data obtained in S44 in conjunction with the meteorological forecast data, including the following steps: S5. The meteorological forecast data includes temperature, light intensity, precipitation and air pressure, and the mechanism by which each meteorological element acts as an environmental stress factor is determined. S51. Construct a Bayesian network model of the effects of environmental stress factors; S52. Input the meteorological forecast data into the action model to calculate the environmental stress index for dinoflagellate growth under different meteorological conditions; S53. Overlay and analyze the environmental stress index with the initial risk assessment data; S54. Through a multi-factor comprehensive evaluation method, multi-factor coupled algal bloom risk assessment data are obtained.

Images