CN-121859276-B - Micro-plastic ecological risk early warning method and system based on knowledge graph and machine learning

CN121859276BCN 121859276 BCN121859276 BCN 121859276BCN-121859276-B

Abstract

The invention provides a micro-plastic ecological risk early warning method and a micro-plastic ecological risk early warning system based on knowledge graph and machine learning, wherein the micro-plastic ecological risk early warning method comprises the steps of collecting micro-plastic ecological related data under different data sources, and preprocessing the micro-plastic ecological related data to form a standardized data set with a unified space-time frame and association relation; the method comprises the steps of establishing a micro-plastic ecological risk knowledge graph which takes micro-plastics, marine organisms, exposure ways and toxic effects as core entities by utilizing the standardized data set with unified space-time frames and incidence relations, generating a mechanism early warning threshold value of micro-plastic concentration according to input micro-plastic concentration data of a target sea area and target organism species, acquiring a micro-plastic concentration predicted value and an optimal early warning threshold value corresponding to the micro-plastic type by utilizing a machine learning early warning model which is trained, and acquiring a risk level of the target sea area aiming at the target organism species and a dynamic early warning threshold value of the micro-plastic concentration.

Inventors

CAI MINGGANG
CUI BOWEN
CHEN KAI

Assignees

厦门大学

Dates

Publication Date: 20260508
Application Date: 20260318

Claims (9)

1. The micro-plastic ecological risk early warning method based on knowledge graph and machine learning is characterized by comprising the following steps of: Collecting the micro-plastic ecological related data under different data sources, and preprocessing the micro-plastic ecological related data to form a standardized data set with a unified space-time frame and association relation; Constructing a micro-plastic ecological risk knowledge graph which takes micro-plastics, marine organisms, exposure paths and toxic effects as core entities and associates space-time dynamic data by utilizing the standardized data set with the unified space-time frame and association relation; Generating a mechanism early warning threshold value of the micro plastic concentration by utilizing historical evidence of the micro plastic concentration data and the target biological species in the micro plastic ecological risk knowledge graph about the target sea area according to the input micro plastic concentration data and the target biological species of the target sea area, wherein the mechanism early warning threshold value comprises the steps of positioning entity nodes corresponding to the target biological species in the micro plastic ecological risk knowledge graph based on the input micro plastic concentration data and the target biological species, taking the entity nodes as core biological species nodes, taking the core biological species nodes as the centers, searching and extracting entity nodes corresponding to the target biological species corresponding to the core biological species nodes, taking the entity nodes as candidate micro plastic nodes, taking attribute information corresponding to the candidate micro plastic nodes, acquiring micro plastic characteristic parameters from the attribute information corresponding to the candidate micro plastic nodes, wherein the micro plastic characteristic parameters comprise micro plastic particle size, micro plastic concentration data and micro plastic distribution range area, comparing the micro plastic concentration data corresponding to the candidate micro plastic nodes with the input target sea area, taking the core biological species concentration data and the input micro plastic species corresponding to the core biological species nodes as core plastic species area, comparing the core plastic material concentration data and the candidate plastic area in the target sea area, and the core plastic area data is obtained by combining the difference between the candidate micro plastic species concentration data and the target plastic area in the preset range, and the core plastic area is obtained according to the difference between the candidate plastic area and the target micro plastic area and the target plastic area, generating a mechanism early warning threshold value of the concentration of the microplastic; According to the input microplastic concentration data of the target sea area and the target biological species, acquiring a microplastic concentration predicted value and an optimal early-warning threshold value corresponding to the microplastic type by using a machine learning early-warning model which is trained, and carrying out fusion judgment by using the optimal early-warning threshold value and a mechanism early-warning threshold value to acquire the risk level of the target sea area aiming at the target biological species and a dynamic early-warning threshold value of the microplastic concentration.
2. The method for early warning of the ecological risk of the micro plastic according to claim 1, wherein the steps of collecting the related data of the micro plastic ecology under different data sources, and preprocessing the related data of the micro plastic ecology to form a standardized data set with a unified space-time frame and association relation, comprise the following steps: Collecting micro-plastic ecology related data under different data sources, wherein the different data sources comprise an academic literature database, an environment monitoring database, a biological species database and a chemical toxicity database; Performing data cleaning, abnormal data eliminating and data normalization on the micro-plastic ecology related data to obtain micro-plastic ecology related data after initial pretreatment; and carrying out data association and space-time frame unified processing on the initially preprocessed micro-plastic ecological related data to obtain a standardized data set with unified space-time frame and association relation corresponding to the micro-plastic ecological related data.
3. The method for early warning of the ecological risk of the micro plastic according to claim 2, wherein the method for carrying out unified processing of data association and space-time frames on the initially preprocessed micro plastic ecological related data to obtain a standardized data set corresponding to the micro plastic ecological related data and having unified space-time frames and association relations comprises the following steps: The data information with the time attribute is called from the initially preprocessed micro-plastic ecological related data, and standard time stamps are added to all the data information with the time attribute; The data information with the spatial attribute is extracted from the initially preprocessed micro-plastic ecological related data, and standard spatial coordinates are added for all the data information with the spatial attribute; And generating a standardized data set which is corresponding to the micro-plastic ecological related data and has a unified space-time frame and association relation by combining the standard time stamp and the standard space coordinate.
4. The method for early warning of the ecological risk of the micro plastic according to claim 1, wherein the construction of the knowledge graph of the ecological risk of the micro plastic by taking the micro plastic, the marine organism, the exposure path and the toxic effect as core entities and correlating the space-time dynamic data by utilizing the standardized data set with the unified space-time frame and the correlation relation comprises the following steps: Extracting entity data from the standardized data set with the unified space-time frame and the association relation, wherein the types of the entity data comprise microplastic, marine organisms, exposure paths and toxic effects; Creating entity nodes corresponding to each type of entity data, calling the relation among each type of entity data, and constructing an initial micro-plastic ecological risk knowledge graph by combining the entity nodes corresponding to each type of entity data and the relation; Collecting and de-duplicating attribute information of the same entity in different data sources, obtaining attribute information corresponding to each entity node, and filling the attribute information into an initial micro-plastic ecological risk knowledge graph; And (3) invoking concentration detection events in a standardized data set with a unified space-time frame and association relation, and generating a micro-plastic ecological risk knowledge graph of the associated space-time dynamic data by combining the initial micro-plastic ecological risk knowledge graph with the concentration detection events.
5. The method of claim 4, wherein generating the micro-plastic ecological risk knowledge graph of the associated space-time dynamic data by combining the concentration detection event with the initial micro-plastic ecological risk knowledge graph comprises: for each concentration detection event, calling event data information corresponding to the concentration detection event, wherein the event data information at least comprises concentration detection time, concentration detection position and concentration detection value; determining space-time coordinates corresponding to each concentration detection event based on the event data information corresponding to each concentration detection event, wherein the space-time coordinates are generated in an associated manner according to the concentration detection time and the concentration detection position; Calling entity nodes corresponding to the detection time of each concentration detection event in the initial micro-plastic ecological risk knowledge graph; And respectively establishing association connection between each concentration detection event with space-time coordinates and corresponding entity nodes to represent a specific environment carrier corresponding to each entity node and concentration detection events occurring on the carrier, and generating a micro-plastic ecological risk knowledge graph of associated space-time dynamic data.
6. The method for early warning of the ecological risk of the micro plastic according to claim 1, wherein the method for generating the mechanism early warning threshold of the micro plastic concentration by combining the species concentration and the distribution range area of the target biological species obtained at each historical monitoring moment based on the micro plastic characteristic parameters corresponding to the core micro plastic nodes comprises the following steps: The method comprises the steps of calling a standard time stamp corresponding to each historical monitoring moment of a target biological species, and extracting a micro-plastic characteristic parameter corresponding to each core micro-plastic node under the standard time stamp according to the standard time stamp corresponding to the historical monitoring moment; and calculating a mechanism early warning threshold value for generating the concentration of the micro plastic by utilizing the micro plastic characteristic parameters of each core micro plastic node and the species concentration and the distribution range area of the target biological species at each historical monitoring moment.
7. The method for early warning of the ecological risk of the micro plastic according to claim 1, wherein the method for acquiring the predicted value and the optimal early warning threshold value of the micro plastic concentration corresponding to the micro plastic type by utilizing the trained machine learning early warning model according to the input micro plastic concentration data of the target sea area and the target biological species comprises the following steps: inputting the microplastic concentration data and the target biological species of the target sea area into a machine learning early warning model which is trained; And acquiring a microplastic concentration predicted value and an optimal early warning threshold value which correspond to each microplastic type respectively through the machine learning early warning model after training is completed.
8. The method for early warning the ecological risk of the micro plastic according to claim 1, wherein the method for acquiring the dynamic early warning threshold of the risk level and the micro plastic concentration of the target sea area aiming at the target biological species by utilizing the fusion judgment of the optimal early warning threshold and the mechanism early warning threshold comprises the following steps: An optimal early warning threshold and a mechanism early warning threshold are called, and threshold fusion is carried out on the optimal early warning threshold and the mechanism early warning threshold according to a fusion strategy to generate a dynamic early warning threshold; And comparing the micro plastic concentration predicted value corresponding to the micro plastic type with a dynamic early warning threshold, and judging that the risk level of the target sea area for the target biological species is high when the micro plastic concentration predicted value is not lower than the dynamic early warning threshold, otherwise, judging that the risk level of the target sea area for the target biological species is low.
9. A micro-plastic ecological risk early warning system based on knowledge graph and machine learning for executing the method of any one of claims 1 to 8, characterized in that the micro-plastic ecological risk early warning system comprises: The data acquisition and preprocessing module is used for acquiring the micro-plastic ecological related data under different data sources, and preprocessing the micro-plastic ecological related data to form a standardized data set with a unified space-time frame and association relation; the knowledge graph construction module is used for constructing a micro-plastic ecological risk knowledge graph which takes micro-plastics, marine organisms, exposure paths and toxic effects as core entities and is related with space-time dynamic data by utilizing the standardized data set with the unified space-time frame and the incidence relation; The mechanism early warning module is used for generating a mechanism early warning threshold of the micro-plastic concentration by utilizing historical evidence about the micro-plastic concentration data of the target sea area and the target biological species in the micro-plastic ecological risk knowledge graph according to the input micro-plastic concentration data of the target sea area and the target biological species, and comprises the steps of positioning entity nodes corresponding to the target biological species in the micro-plastic ecological risk knowledge graph based on the input micro-plastic concentration data of the target sea area and the target biological species, taking the entity nodes as core biological species nodes, taking the core biological species nodes as the center, searching and extracting entity nodes corresponding to the micro-plastic types corresponding to the core biological species nodes, which have association relation with the input target sea area, as candidate micro-plastic nodes, taking attribute information corresponding to each candidate micro-plastic node, acquiring micro-plastic characteristic parameters from the attribute information corresponding to the candidate micro-plastic nodes, wherein the micro-plastic characteristic parameters comprise micro-plastic particle size, micro-plastic concentration data and micro-plastic distribution area, comparing the micro-plastic concentration data corresponding to each candidate micro-plastic node with the input micro-plastic species corresponding to the input sea area, taking the core biological species nodes as core biological species corresponding to the target biological species, comparing the candidate plastic concentration data and the candidate plastic nodes, and the core plastic nodes are compared according to the difference between the core plastic concentration data and the corresponding to the candidate plastic concentration data, and the core plastic nodes are obtained, combining species concentration and distribution range area of the target biological species obtained at each historical monitoring moment to generate a mechanism early warning threshold value of the microplastic concentration; And the comprehensive early warning module is used for acquiring a micro plastic concentration predicted value and an optimal early warning threshold value corresponding to the micro plastic type by utilizing a machine learning early warning model which is trained according to the input micro plastic concentration data of the target sea area and the target biological species, and carrying out fusion judgment by utilizing the optimal early warning threshold value and the mechanism early warning threshold value to acquire the risk grade of the target sea area aiming at the target biological species and the dynamic early warning threshold value of the micro plastic concentration.

Description

Micro-plastic ecological risk early warning method and system based on knowledge graph and machine learning Technical Field The invention provides a micro-plastic ecological risk early warning method and system based on knowledge graph and machine learning, and belongs to the technical field of ecological risk prediction. Background With the aggravation of marine pollution problems, the micro-plastics have serious threat to marine ecosystems and biosafety due to the characteristics of wide distribution, difficult degradation, strong toxicity and the like, and the development of micro-plastic ecological risk early warning has become a key requirement for marine environment management. The existing micro-plastic ecological risk early warning method is mainly based on a machine learning model, and concentration prediction and risk level judgment are achieved by utilizing a monitoring data training model. However, the existing micro-plastic ecological risk early warning method in the prior art has low early warning precision and weak dynamic adaptability, and cannot meet the accurate early warning requirement under the complex marine environment. Disclosure of Invention The invention provides a micro-plastic ecological risk early warning method and a micro-plastic ecological risk early warning system based on a knowledge graph and machine learning, which are used for solving the technical problems in the prior art, and the adopted technical scheme is as follows: the micro-plastic ecological risk early warning method based on knowledge graph and machine learning comprises the following steps: Collecting the micro-plastic ecological related data under different data sources, and preprocessing the micro-plastic ecological related data to form a standardized data set with a unified space-time frame and association relation; Constructing a micro-plastic ecological risk knowledge graph which takes micro-plastics, marine organisms, exposure paths and toxic effects as core entities and associates space-time dynamic data by utilizing the standardized data set with the unified space-time frame and association relation; Generating a mechanism early warning threshold value of the micro plastic concentration by utilizing the micro plastic concentration data of the target sea area and historical evidence of the target biological species in the micro plastic ecological risk knowledge graph according to the input micro plastic concentration data of the target sea area and the target biological species; According to the input microplastic concentration data of the target sea area and the target biological species, acquiring a microplastic concentration predicted value and an optimal early-warning threshold value corresponding to the microplastic type by using a machine learning early-warning model which is trained, and carrying out fusion judgment by using the optimal early-warning threshold value and a mechanism early-warning threshold value to acquire the risk level of the target sea area aiming at the target biological species and a dynamic early-warning threshold value of the microplastic concentration. Further, the collecting the micro-plastic ecological related data under different data sources and preprocessing the micro-plastic ecological related data to form a standardized data set with unified space-time frame and association relation, comprising: Collecting micro-plastic ecology related data under different data sources, wherein the different data sources comprise an academic literature database, an environment monitoring database, a biological species database and a chemical toxicity database; Performing data cleaning, abnormal data eliminating and data normalization on the micro-plastic ecology related data to obtain micro-plastic ecology related data after initial pretreatment; and carrying out data association and space-time frame unified processing on the initially preprocessed micro-plastic ecological related data to obtain a standardized data set with unified space-time frame and association relation corresponding to the micro-plastic ecological related data. Further, carrying out data association and space-time frame unified processing on the initially preprocessed micro-plastic ecological related data to obtain a standardized data set corresponding to the micro-plastic ecological related data and having a unified space-time frame and association relation, wherein the standardized data set comprises: The data information with the time attribute is called from the initially preprocessed micro-plastic ecological related data, and standard time stamps are added to all the data information with the time attribute; The data information with the spatial attribute is extracted from the initially preprocessed micro-plastic ecological related data, and standard spatial coordinates are added for all the data information with the spatial attribute; And generating a standardized data set which is corresponding to the mic