Ecological mechanism and causal enhancement integrated blue algae community succession prediction method and device

Abstract

The application discloses a cyanobacterial community succession prediction method and device fusing ecological mechanism and causal enhancement. The method comprises the following steps: collecting multi-source high-frequency monitoring data of a target water area, and constructing an environmental driving factor characteristic matrix; constructing a causal weight matrix through causal relationship aggregation analysis, and outputting a causal enhancement characteristic matrix; constructing a species ecological strategy embedding matrix; performing multi-scale decomposition on algal density time series data; constructing a multi-species interaction prediction model based on a multi-head self-attention mechanism, fusing environmental characteristics, causal enhancement characteristics, ecological strategy embedding and multi-scale time series characteristics, and predicting the biomass and relative abundance of each dominant cyanobacterial species; and outputting a cyanobacterial community succession process and a water bloom risk grade. Through deep fusion of causal enhancement, ecological strategy embedding and multi-head attention mechanism, the precision, stability and ecological interpretability of cyanobacterial community succession prediction are significantly improved, and the application is suitable for water bloom early warning and ecological management of lakes, reservoirs and rivers.

Description

Technical Field

[0001] This invention relates to the field of aquatic ecological environment prediction and ecological risk early warning technology, specifically a method for predicting cyanobacterial community succession that integrates ecological mechanisms and causal enhancement. It is applicable to the simulation prediction of the cyanobacterial community structure succession process and the assessment of algal bloom risk in water bodies such as lakes, reservoirs and rivers. Background Technology

[0002] Cyanobacterial blooms are a significant ecological and environmental problem facing freshwater ecosystems such as lakes, reservoirs, and rivers worldwide. The changes in cyanobacterial community structure are usually manifested as succession between different dominant species. For example, species such as Microcystis, Dolichospermum, and Pseudanabaena exhibit phased dominance under different environmental conditions.

[0003] Existing methods for predicting cyanobacteria mainly include: empirical models based on statistical regression, black-box prediction models based on machine learning, and mechanism models based on ecological dynamics equations. However, the above methods have the following shortcomings: (1) lack a clear characterization of the causal relationship between environmental factors and different species; (2) fail to effectively integrate the interspecific interaction mechanisms among cyanobacteria species; (3) ignore the differences in ecological strategies of different species in terms of nutrient use, light adaptation, and temperature response; (4) time series prediction models are difficult to take into account trend, seasonality, and random disturbance components; and (5) deep learning models have insufficient interpretability.

[0004] Therefore, it is necessary to propose a method for predicting cyanobacterial community succession that integrates ecological mechanisms and causal enhancement structures in order to improve the model's interpretability and prediction accuracy. Summary of the Invention

[0005] The purpose of this invention is to provide a method and device for predicting cyanobacterial community succession that integrates ecological mechanisms and causal enhancement. By constructing a causal relationship aggregation structure between environmental factors and dominant cyanobacterial species, introducing a species ecological strategy embedding mechanism, and combining time series multi-scale analysis, a multi-species interaction-enhanced artificial prediction model is constructed. This enables high-precision simulation and prediction of the biomass, relative abundance, and community structure succession process of different dominant cyanobacterial species, thereby improving the interpretability and ecological applicability of the model.

[0006] The present invention adopts the following technical solution.

[0007] A method for predicting cyanobacterial community succession that integrates ecological mechanisms and causal enhancement includes the following steps: S1. Collecting multi-source high-frequency monitoring data from multiple key locations in the target water area, including hydrological data, water quality data, and aquatic ecological data, and obtaining time-series data on the algal density and relative abundance of different dominant cyanobacterial species; S2. Preprocessing and standardizing the multi-source high-frequency monitoring data to construct an environmental driving factor feature matrix X; S3. Performing causal relationship aggregation analysis based on the dependency relationship between the environmental driving factor feature matrix X and the biomass changes of each dominant cyanobacterial species, constructing a causal weight matrix W to characterize the degree of influence of different environmental factors on different cyanobacterial species, and merging the environmental driving factor feature matrix X and the causal weight matrix W to output a causal enhancement feature matrix X_c; S4. Constructing a species ecological strategy embedding matrix E to characterize the differences in resource utilization and environmental adaptation among species based on the ecological response characteristics of different dominant cyanobacterial species; S5. Time series decomposition was performed on the time series data of densities and relative abundances of different dominant cyanobacteria to extract trend components, periodic components, and random perturbation components, and a multi-scale time series feature set was constructed; S6. Based on the multi-head self-attention mechanism, the environmental driving factor feature matrix X, the causal enhancement feature matrix X_c, the species ecological strategy embedding matrix E, and the multi-scale time series feature set were integrated to construct a multi-species interaction time series prediction model to characterize the interaction and influence relationships among different cyanobacteria species, and to predict the biomass and relative abundance of each dominant cyanobacteria species at future times; S7. Based on the prediction results, the composition structure of the cyanobacteria community, the succession process of dominant species, and the risk level of algal blooms were simulated and output.

[0008] Furthermore, the hydrological data includes water level, flow velocity, and hydrodynamic parameters; the water quality data includes water temperature, pH, dissolved oxygen, total nitrogen, total phosphorus, ammonia nitrogen, nitrate nitrogen, dissolved reactive phosphorus, and transparency; and the aquatic ecological data includes chlorophyll a concentration, density of different cyanobacteria species, and their relative abundance.

[0009] Furthermore, the causal relationship aggregation analysis method in step S3 includes at least one of the following: a time series-based causal testing method; a structural equation model-based causal path analysis method; a graph model-based causal structure learning method; and a machine learning-based causal inference method.

[0010] Furthermore, the causal enhancement feature matrix X_c in step S3 is formed by weighted integration of the causal weights between environmental factors and each dominant cyanobacterial species, and is used to enhance the model's ability to explain environmental driving factors.

[0011] Furthermore, the time series decomposition in step S5 is implemented using one of the following methods: trend-seasonal-residual decomposition method, empirical mode decomposition method, or wavelet decomposition method.

[0012] Furthermore, the biological characteristic parameters included in the species ecological strategy embedding matrix E in step S4 include at least one of the following: maximum growth rate, half-saturated nutrient constant, light compensation point, suitable temperature range, nitrogen and phosphorus use efficiency ratio, and buoyancy regulation capability.

[0013] Furthermore, the multi-species interaction prediction model based on the multi-head self-attention mechanism described in step S6 includes: an environmental factor feature input layer, a causal enhancement fusion layer, an ecological strategy embedding layer, a multi-head self-attention structure layer, and a time-series prediction output layer; wherein, the multi-head self-attention structure layer is used to characterize the response weights of different species to environmental factors and the interaction effects between species.

[0014] Furthermore, the algal bloom risk level mentioned in step S7 is an early warning signal generated after evaluating the prediction results based on a preset risk threshold.

[0015] A device for predicting cyanobacterial community succession by integrating ecological mechanisms and causal enhancement includes: a data acquisition module for collecting multi-source high-frequency monitoring data from multiple key locations in a target water area, including hydrological data, water quality data, and aquatic ecological data, and acquiring time-series data on the algal density and relative abundance of different dominant cyanobacterial species; a data preprocessing module for preprocessing and standardizing the multi-source high-frequency monitoring data to construct an environmental driving factor feature matrix X; a causal enhancement analysis module for performing causal relationship aggregation analysis based on the dependency relationship between the environmental driving factor feature matrix X and the biomass changes of each dominant cyanobacterial species, constructing a causal weight matrix W to characterize the degree of influence of different environmental factors on different cyanobacterial species, and outputting a causal enhancement feature matrix X_c based on the fusion of the environmental driving factor feature matrix X and the causal weight matrix W; and an ecological strategy embedding module. The system comprises four modules: a species ecological strategy embedding matrix E, which is used to construct a species ecological strategy embedding matrix E to characterize the differences in resource utilization and environmental adaptation among different dominant cyanobacterial species based on their ecological response characteristics; a time series decomposition module, which decomposes the time series data of density and relative abundance of different dominant cyanobacteria into trend components, periodic components, and random perturbation components to construct a multi-scale time series feature set; a multi-species interaction prediction model module, which, based on a multi-head self-attention mechanism, integrates the environmental driving factor feature matrix X, the causal enhancement feature matrix X_c, the species ecological strategy embedding matrix E, and the multi-scale time series feature set to obtain a multi-species interaction prediction model that characterizes the interaction and influence relationships among different cyanobacterial species, and predicts the biomass and relative abundance of each dominant cyanobacterial species at future times; and a risk assessment and output module, which simulates and outputs the composition structure of cyanobacterial communities, the succession process of dominant species, and the risk level of algal blooms based on the prediction results.

[0016] A non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described method for predicting cyanobacterial community succession by integrating ecological mechanisms and causal enhancement.

[0017] The present invention has the following beneficial effects.

[0018] 1. Improve prediction accuracy and long-term stability.

[0019] By using causal aggregation analysis, the dependence between environmental factors and cyanobacterial species is upgraded from traditional correlation modeling to causal modeling. A causal enhancement feature matrix X_c is constructed to effectively filter out non-causal noise and enhance the model's response accuracy to environmental driving signals. Multi-scale time series decomposition is introduced to extract trend, periodic and random disturbance components, improve the model's ability to fit complex time series dynamics and enhance the stability of long-term predictions.

[0020] 2. Enhance the interpretability of the model ecosystem.

[0021] A species ecological strategy embedding matrix E is constructed, and the biological characteristics of species (such as light compensation point, temperature adaptation range, etc.) are parameterized and embedded into the model. This enables the model to distinguish the differences in resource utilization strategies and environmental adaptation among different species, overcoming the shortcomings of traditional black box models that lack ecological meaning. The prediction results are more in line with ecological laws, making it easier for ecological managers to understand and trust the model output.

[0022] 3. Accurately depict the dynamics of cyanobacterial community succession.

[0023] By leveraging multi-head self-attention mechanisms, this model simultaneously models competition, substitution, and synergy among multiple species, accurately capturing the succession process of dominant species and the evolution of community structure. Combining multi-scale temporal features and causal enhancement features, the model can identify species response patterns at different time scales, improving the simulation accuracy of succession processes.

[0024] 4. Achieve end-to-end intelligent early warning and risk assessment.

[0025] Based on the model's prediction results, the system automatically identifies the trend of dominant species replacement and assesses the risk of algal blooms in a graded manner, forming a complete technical chain from data input to early warning output. The output results include changes in species density, relative abundance, community structure evolution path, and risk level, providing actionable decision support for aquatic ecological management.

[0026] 5. Improve the model's generalization ability and applicability.

[0027] It supports various causal relationship analysis methods, time series decomposition methods, and ecological parameter construction methods, and has good algorithm scalability and scenario adaptability. It is suitable for various water types such as lakes, reservoirs, and rivers, and supports hourly or daily high-frequency data input to meet different monitoring conditions and early warning needs. Attached Figure Description

[0028] Figure 1 This is a flowchart illustrating a method for predicting cyanobacterial community succession that integrates ecological mechanisms and causal enhancement, as described in an embodiment of the present invention.

[0029] Figure 2 This is a schematic diagram of causal relationships in an embodiment of the present invention.

[0030] Figure 3 This is a schematic diagram of the ecological strategy embedding and prediction model architecture in an embodiment of the present invention. Detailed Implementation

[0031] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0032] Please see Figure 1 The first aspect of the present invention provides a method for predicting cyanobacterial community succession that integrates ecological mechanisms and causal enhancement, comprising the following steps.

[0033] Step 1: Acquisition of multi-source high-frequency monitoring data.

[0034] Acquire multi-source high-frequency monitoring data from multiple key monitoring points in the target water area. The multi-source high-frequency monitoring data includes: hydrological data, including water level, flow velocity, and hydrodynamic parameters; water quality data, including water temperature, pH, dissolved oxygen, total nitrogen, total phosphorus, ammonia nitrogen, nitrate nitrogen, dissolved reactive phosphorus, and transparency; and aquatic ecological data, including chlorophyll a concentration, density of different cyanobacteria species, and their relative abundance.

[0035] The time resolution of the data is on the hourly or daily level. The above data is organized according to time series to construct the original dataset of environmental driving factors.

[0036] Step 2: Environmental driving factor preprocessing and feature construction.

[0037] The multi-source high-frequency monitoring data obtained in step one are preprocessed, including: missing value imputation; outlier removal; and data standardization or normalization. Key feature variables are selected based on correlation analysis or dimensionality reduction methods, and an environmental driving factor feature matrix X is constructed: X ∈ R^(T × N); where: T is the time length; and N is the number of environmental features.

[0038] Step 3: Causal relationship aggregation analysis and causal enhancement feature construction.

[0039] Based on time-series data of environmental factors and biomass of dominant cyanobacterial species, time-series causal analysis or structural equation modeling is used to identify the dependence-response relationships of environmental factors on different cyanobacterial species (e.g., Figure 2 (As shown).

[0040] Construct a causal weight matrix W: W ∈ R^(N × M); where: M is the number of dominant cyanobacterial species; W_ij represents the causal influence weight of the i-th environmental factor on the j-th species.

[0041] Based on the mapping between the environmental feature matrix X and the causal weight matrix W, a causal enhancement feature matrix X_c is formed: X_c = X × W; which is used to enhance the expressive power of environmental driving signals in the feature space.

[0042] Step 4: Construction of species ecological strategies.

[0043] To characterize the differences in ecological resource utilization strategies among different cyanobacterial species, species ecological strategy embedding vectors are constructed.

[0044] The species ecological strategy embedding vector includes, but is not limited to, the following ecological parameters: maximum growth rate; half-saturated nutrient constant; light compensation point; suitable temperature range; nitrogen and phosphorus use efficiency ratio; buoyancy regulation capacity (e.g., ... Figure 3 (As shown).

[0045] Construct an ecological strategy embedding matrix: E ∈ R^(M × K); where K is the ecological strategy feature dimension; each row represents the ecological strategy vector of a species.

[0046] Step 5: Time series decomposition and multi-scale feature extraction.

[0047] The time series data of cyanobacterial species density and relative abundance are decomposed to extract multi-scale temporal features. The decomposition methods include, but are not limited to: trend-seasonal-residual decomposition and empirical mode decomposition. The original sequence is decomposed into: X_s(t) = Trend(t) + Season(t) + Residual(t); a multi-scale temporal feature set containing long-term trend features, periodic features, and random perturbation features is constructed.

[0048] Step 6: Based on the multi-head self-attention mechanism, the environmental driving factor feature matrix X, the causal enhancement feature matrix X_c, the species ecological strategy embedding matrix E, and the multi-scale temporal feature set are integrated to construct a multi-species interaction temporal prediction model to characterize the interaction and influence relationship between different cyanobacterial species and predict the biomass and relative abundance of each dominant cyanobacterial species in the future.

[0049] The multi-species interaction time-series prediction model includes: an environmental factor feature input layer; a causal enhancement fusion layer; an ecological strategy embedding fusion layer; a multi-head self-attention structure layer; and a time-series prediction output layer.

[0050] The input layer receives an environmental feature matrix, a causal enhancement feature matrix, and an ecological strategy embedding matrix.

[0051] The competitive, substitution, and synergistic relationships among different cyanobacterial species were characterized by a multi-head self-attention mechanism.

[0052] The output layer generates predicted density and relative abundance values ​​for each dominant cyanobacterial species within a future time window.

[0053] Step 7: Succession process simulation and algal bloom risk assessment.

[0054] Based on the model's predictions, the path of cyanobacterial community structure change within a future time window is simulated.

[0055] The output includes: the density change trend of each dominant species; the relative abundance change; and the process of community dominant species succession.

[0056] Based on the preset risk threshold, the prediction results are used to assess the risk level of algal blooms and an early warning signal is output.

[0057] Compared with existing technologies, this invention has the following characteristics and effects: Improved prediction accuracy and stability: Through causal relationship aggregation analysis, the dependence between environmental factors and cyanobacterial species is elevated from traditional correlation to a causal level. A causal enhancement feature matrix is ​​constructed, effectively filtering out non-causal noise, making the model's response to environmental driving signals more accurate, thereby significantly improving the prediction accuracy and long-term stability of biomass and relative abundance. Enhanced model ecological interpretability: Ecological strategy embedding vectors based on species biological characteristics (such as nutrient use efficiency, light compensation point, temperature response parameters, etc.) are introduced, enabling the model to explicitly distinguish the resource utilization strategies and environmental adaptation differences of different dominant species. This avoids the shortcomings of traditional black-box models that lack ecological meaning, and the prediction results are more consistent with ecological laws. Accurate characterization of community succession dynamics: Multi-scale time series decomposition extracts trend, periodic, and random perturbation components. Combined with the multi-species interaction prediction model based on multi-head self-attention in step S6, it can simultaneously capture the competition, substitution, and synergistic relationships between species, accurately simulating the process of dominant species succession and community structure evolution, providing a scientific basis for the identification of dominant species in algal blooms. Achieve end-to-end intelligent early warning: Automatically identify the trend of dominant species replacement and assess the risk of algal blooms in a graded manner based on the prediction results, forming a complete technology chain from data input to early warning output, and providing operable decision support for aquatic ecological management.

[0058] In summary, this invention solves the common problems of low prediction accuracy, weak interpretability, and difficulty in simulating interspecies interactions in existing methods through a deep fusion framework of "causal enhancement + ecological embedding + multi-head attention," and has outstanding substantive features and significant technological progress.

[0059] The second aspect of this invention provides a device for predicting cyanobacterial community succession by integrating ecological mechanisms and causal enhancement, comprising: a data acquisition module for acquiring multi-source high-frequency monitoring data from multiple key locations in a target water area, the multi-source high-frequency monitoring data including hydrological data, water quality data, and aquatic ecological data, and acquiring time-series data on the algal density and relative abundance of different dominant cyanobacterial species; a data preprocessing module for preprocessing and standardizing the multi-source high-frequency monitoring data to construct an environmental driving factor feature matrix X; a causal enhancement analysis module for performing causal relationship aggregation analysis based on the dependency relationship between the environmental driving factor feature matrix X and the biomass changes of each dominant cyanobacterial species, constructing a causal weight matrix W to characterize the degree of influence of different environmental factors on different cyanobacterial species, and outputting a causal enhancement feature matrix X_c based on the fusion of the environmental driving factor feature matrix X and the causal weight matrix W; and ecological strategies. The system comprises the following modules: an embedding module for constructing a species ecological strategy embedding matrix E based on the ecological response characteristics of different dominant cyanobacterial species, which characterizes differences in resource utilization and environmental adaptation among species; a time-series decomposition module for performing time-series decomposition on the density and relative abundance of different dominant cyanobacteria, extracting trend components, periodic components, and random perturbation components, and constructing a multi-scale time-series feature set; a multi-species interaction prediction model module for fusing the environmental driving factor feature matrix X, the causal enhancement feature matrix X_c, the species ecological strategy embedding matrix E, and the multi-scale time-series feature set based on a multi-head self-attention mechanism, to obtain a multi-species interaction prediction model that characterizes the interaction relationships among different cyanobacterial species, and predicts the biomass and relative abundance of each dominant cyanobacterial species at future times; and a risk assessment and output module for simulating and outputting the composition structure of cyanobacterial communities, the succession process of dominant species, and the risk level of algal blooms based on the prediction results.

[0060] Another aspect of the present invention provides a cyanobacterial community succession prediction system that integrates ecological mechanisms and causal enhancement, comprising: a computer-readable storage medium and a processor; the computer-readable storage medium is used to store executable instructions; the processor is used to read the executable instructions stored in the computer-readable storage medium and execute the cyanobacterial community succession prediction method that integrates ecological mechanisms and causal enhancement described in the first aspect.

[0061] In another aspect, the present invention provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method for predicting cyanobacterial community succession that integrates ecological mechanisms and causal enhancement as described in the first aspect.

[0062] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied 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.

[0063] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0064] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0065] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0066] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A cyanobacterial community succession prediction method that combines ecological mechanisms with causal enhancement, characterized in that, Includes the following steps: S1. Collect multi-source high-frequency monitoring data from multiple key points in the target water area. The multi-source high-frequency monitoring data includes hydrological data, water quality data, and aquatic ecological data. Also, obtain time-series data on the algal density and relative abundance of different dominant cyanobacterial species. S2. Preprocess and standardize the multi-source high-frequency monitoring data to construct the environmental driving factor feature matrix X; S3. Based on the dependence between the environmental driving factor feature matrix X and the biomass changes of each dominant cyanobacterial species, a causal relationship aggregation analysis is performed to construct a causal weight matrix W to characterize the degree of influence of different environmental factors on different cyanobacterial species. Based on the fusion of the environmental driving factor feature matrix X and the causal weight matrix W, a causal enhancement feature matrix X_c is output. S4. Based on the ecological response characteristics of different dominant cyanobacterial species, construct a species ecological strategy embedding matrix E to characterize the differences in resource utilization and environmental adaptation among species; S5. Perform time series decomposition on time series data of different dominant cyanobacteria densities and relative abundances, extract trend components, periodic components and random perturbation components, and construct a multi-scale time series feature set; S6. Based on the multi-head self-attention mechanism, the environmental driving factor feature matrix X, the causal enhancement feature matrix X_c, the species ecological strategy embedding matrix E, and the multi-scale temporal feature set are integrated to construct a multi-species interaction temporal prediction model to characterize the interaction and influence relationship between different cyanobacterial species and predict the biomass and relative abundance of each dominant cyanobacterial species in the future. S7. Based on the prediction results, simulate and output the composition structure of the cyanobacterial community, the succession process of dominant species, and the risk level of algal bloom.

2. The method of claim 1, wherein, The hydrological data includes water level, flow velocity, and hydrodynamic parameters; the water quality data includes water temperature, pH, dissolved oxygen, total nitrogen, total phosphorus, ammonia nitrogen, nitrate nitrogen, dissolved reactive phosphorus, and transparency; and the aquatic ecological data includes chlorophyll a concentration, density of different cyanobacteria species, and their relative abundance.

3. The method according to claim 1, characterized in that, The causal relationship aggregation analysis method in step S3 includes at least one of the following: Causality testing methods based on time series; A causal path analysis method based on structural equation modeling; A causal structure learning method based on graph models; Causal inference methods based on machine learning.

4. The method according to claim 1, characterized in that, The causal enhancement feature matrix X_c in step S3 is formed by weighted integration of the causal weights between environmental factors and each dominant cyanobacterial species.

5. The method according to claim 1, characterized in that, The time series decomposition in step S5 is achieved using one of the following methods: trend-seasonal-residual decomposition method, empirical mode decomposition method, or wavelet decomposition method.

6. The method according to claim 1, characterized in that, The biological characteristic parameters included in the species ecological strategy embedding matrix E in step S4 include at least one of the following: maximum growth rate, half-saturated nutrient constant, light compensation point, suitable temperature range, nitrogen and phosphorus use efficiency ratio, and buoyancy regulation capability.

7. The method according to claim 1, characterized in that, The multi-species interaction prediction model based on the multi-head self-attention mechanism mentioned in step S6 includes: Environmental factor feature input layer, causal enhancement fusion layer, ecological strategy embedding layer, multi-head self-attention structure layer, and time series prediction output layer; The multi-head self-attention structure layer is used to characterize the response weights of different species to environmental factors and the interaction effects between species.

8. The method according to claim 1, characterized in that, The algal bloom risk level mentioned in step S7 is an early warning signal generated after evaluating the prediction results based on a preset risk threshold.

9. A device for predicting cyanobacterial community succession that integrates ecological mechanisms and causal enhancement, characterized in that, include: The data acquisition module is used to collect multi-source high-frequency monitoring data from multiple key points in the target water area. The multi-source high-frequency monitoring data includes hydrological data, water quality data, and aquatic ecological data, and acquires time-series data on the algal density and relative abundance of different dominant cyanobacterial species. The data preprocessing module is used to preprocess and standardize the multi-source high-frequency monitoring data to construct the environmental driving factor feature matrix X. The causal enhancement analysis module is used to perform causal relationship aggregation analysis based on the dependency relationship between the environmental driving factor feature matrix X and the biomass changes of each dominant cyanobacterial species, construct a causal weight matrix W to characterize the degree of influence of different environmental factors on different cyanobacterial species, and output the causal enhancement feature matrix X_c based on the fusion of the environmental driving factor feature matrix X and the causal weight matrix W. The ecological strategy embedding module is used to construct a species ecological strategy embedding matrix E based on the ecological response characteristics of different dominant cyanobacterial species to characterize the differences in resource utilization and environmental adaptation among species. The temporal decomposition module is used to decompose the temporal data of different dominant cyanobacteria densities and relative abundances into time series, extracting trend components, periodic components and random perturbation components, and constructing a multi-scale temporal feature set. The multi-species interaction prediction model module is used to integrate the environmental driving factor feature matrix X, the causal enhancement feature matrix X_c, the species ecological strategy embedding matrix E, and the multi-scale temporal feature set based on the multi-head self-attention mechanism to obtain a multi-species interaction prediction model that characterizes the interaction and influence relationship between different cyanobacterial species, and to predict the biomass and relative abundance of each dominant cyanobacterial species at future times. The risk assessment and output module is used to simulate and output the composition and structure of cyanobacterial communities, the succession process of dominant species, and the risk level of algal blooms based on the prediction results.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method for predicting cyanobacterial community succession by integrating ecological mechanisms and causal enhancement as described in any one of claims 1-8.

Images