CN-121997131-A - Ecological restoration area intelligent identification method based on multi-source data and model integration

CN121997131ACN 121997131 ACN121997131 ACN 121997131ACN-121997131-A

Abstract

The invention relates to the technical field of ecological restoration, and discloses an intelligent ecological restoration area identification method based on multi-source data and model integration. The method comprises the steps of obtaining multisource original data, and obtaining ecological data in a standard format through cleaning and checking. And forming a grid characteristic sequence containing ecological elements and industrial activity characteristics by carrying out space-time grid division and characteristic extraction on the standard data. And carrying out restoration potential evaluation on the sequence by using the pre-constructed ecological restoration knowledge graph to generate a preliminary restoration region set. And inputting the preliminary result into a dynamic identification model, generating an iterative optimization process of benefit simulation by the model through a repair strategy, and outputting an optimized repair strategy set. And finally finishing regional boundary refinement and priority ordering based on the strategy set, and generating a final ecological restoration regional list and a matched strategy. The method improves the accuracy of ecological restoration area identification and the adaptability of restoration strategies.

Inventors

SI HONGTAO
Yao Cesen
ZHAO XIANJUN
ZHOU JINGYI
Xiang Lezhong
WANG CHEN
Huang Xuepiao
LIU DIE
FENG FAN
LIU JIANG
JIAN HENG
ZHOU XUAN
CHEN GUO

Assignees

重庆华地资环科技有限公司
重庆地质矿产研究院

Dates

Publication Date: 20260508
Application Date: 20260120
Priority Date: 20260119

Claims (10)

1. An intelligent ecological restoration area identification method based on multi-source data and model integration is characterized by comprising the following steps: Acquiring an original multi-source data set of an ecological restoration area; performing data cleaning and consistency verification on the original multi-source data set to generate ecological data in a standard format; Performing space-time meshing and feature extraction on the standard format ecological data to form a mesh feature sequence containing ecological element features and industrial activity features; Based on the grid feature sequence, performing restoration potential evaluation through a pre-constructed ecological restoration knowledge graph to generate a preliminary restoration area set; Inputting the preliminary repair region set and the grid feature sequence into a dynamic identification model, generating an iterative process of benefit simulation by fusing a repair strategy by the dynamic identification model, and outputting an optimized repair strategy set; and carrying out boundary refinement and priority ranking on the preliminary restoration area set according to the optimized restoration strategy set to generate a final ecological restoration area list and a matched restoration strategy thereof.
2. The method for intelligently identifying an ecological restoration area based on multi-source data and model integration according to claim 1, wherein the method for performing data cleaning and consistency verification on an original multi-source data set to generate ecological data in a standard format comprises the following steps: Identifying abnormal values and missing values in the original multi-source data set, and filling and correcting by adopting a data restoration method based on adjacent space-time context; Uniformly converting the filled and corrected data into a preset space-time coordinate system and a data format; and carrying out dimension normalization processing on the data under the unified coordinate system to generate the ecological data in the standard format.
3. The method for intelligently identifying an ecological restoration area based on multi-source data and model integration according to claim 2, wherein performing space-time meshing and feature extraction on the standard format ecological data to form a mesh feature sequence containing ecological element features and industrial activity features comprises: dividing the ecological data in the standard format into a plurality of space-time grid units according to the preset space-time resolution; for each space-time grid unit, respectively extracting quantitative characteristic indexes from the dimension of the ecological element and the dimension of the industrial activity; and combining all quantized feature indexes of each space-time grid unit into feature vectors of the space-time grid unit, and arranging the feature vectors of all the space-time grid units according to a time-space sequence to form a grid feature sequence.
4. The method for intelligently identifying an ecological restoration area based on multi-source data and model integration according to claim 3, wherein the method for evaluating restoration potential through a pre-constructed ecological restoration knowledge graph based on a grid feature sequence, and generating a preliminary restoration area set comprises the following steps: carrying out semantic matching and association degree calculation on each feature vector in the grid feature sequence and the nodes in the ecological restoration knowledge graph; for each space-time grid unit, according to the correlation degree calculation result, retrieving an applicable repair mode and constraint conditions from the ecological repair knowledge graph; calculating a repair potential comprehensive score of each space-time grid unit based on the retrieved repair mode and constraint conditions; And screening space-time grid units with the comprehensive score of the repair potential exceeding a preset threshold value, merging the units adjacent in space, and generating a preliminary repair area set.
5. The intelligent recognition method for the ecological restoration area based on the multi-source data and model integration according to claim 4, wherein the initial restoration area set and the grid feature sequence are input into a dynamic recognition model, the dynamic recognition model outputs an optimized restoration strategy set through an iterative process of generating and benefit simulation by fusing restoration strategies, and the method comprises the following steps: the dynamic identification model generates one or more candidate repair strategies according to the grid characteristic sequence segments of each region in the preliminary repair region set; Aiming at each candidate restoration strategy, the dynamic identification model simulates the evolution track of the dynamic identification model, which is generated by the dynamic identification model for ecological elements and industrial activities in a preset time scale; calculating a long-term comprehensive benefit estimated value of each candidate restoration strategy according to the evolution track obtained by simulation; And based on the long-term comprehensive benefit estimation, adopting a strategy optimization mechanism to iteratively adjust and update the candidate restoration strategies until convergence conditions are met, and outputting an optimized restoration strategy set.
6. The method for intelligently identifying an ecological restoration area based on multi-source data and model integration according to claim 5, wherein the dynamic identification model generates one or more candidate restoration strategies according to the grid feature sequence segments of each area in the preliminary restoration area set, and the method comprises the following steps: Analyzing the grid characteristic sequence segments, and identifying key ecological restriction factors and industry coordination opportunity points; combining a predefined repair measure library, and matching a repair technology combination aiming at a key ecological constraint factor; Designing an industry operation adjustment scheme integrated with an industry coordination opportunity point; binding the repair technology combination with an industrial operation adjustment scheme to form a candidate repair strategy.
7. The method for intelligently identifying an ecological restoration area based on multi-source data and model integration according to claim 6, wherein for each candidate restoration strategy, the dynamic identification model simulates the evolution track generated by the dynamic identification model on ecological elements and industrial activities within a preset time scale, and comprises the following steps: constructing a system dynamics model reflecting the interaction mechanism of the ecological elements and the industrial activities; Inputting the candidate repair strategy as an external intervention variable into a system dynamics model; running a system dynamics model, deducing a dynamic process of the change of ecological element indexes and industrial activity indexes along with time under the action of a candidate restoration strategy, and generating an evolution track; The construction of a system dynamics model reflecting the interaction mechanism of ecological elements and industrial activities comprises the following steps: Determining a system boundary, wherein the ecological element index comprises at least one of vegetation coverage, biodiversity index and soil organic matter content, and the industrial activity index comprises at least one of agricultural irrigation water consumption, industrial carbon emission intensity and travel infrastructure density; Based on historical time sequence data, identifying a causal relationship chain between the ecological element index and the industrial activity index by adopting a Granges causal inspection method, and drawing a causal loop graph to visualize a feedback mechanism; defining a state variable, a flow variable and an auxiliary variable according to a causal loop chart, and establishing a system dynamics equation set based on mathematical relations among the variables, wherein the change rate of the state variable is controlled by the flow variable, and the auxiliary variable is used for describing external intervention; calibrating parameters of a system dynamics equation set by utilizing a least square method or Bayesian reasoning, and verifying model simulation accuracy by calculating Nash-Suterkov efficiency coefficients; after model verification, the system dynamics model is integrated into a dynamic recognition model for modeling the long-term impact of candidate repair strategies.
8. The method for intelligently identifying an ecological restoration area based on multi-source data and model integration according to claim 7, wherein calculating a long-term comprehensive benefit estimation of each candidate restoration strategy according to an evolution track obtained by simulation comprises: extracting an ecological state index value and an industrial economic index value at the tail end of a preset time scale from the evolution track; Converting the ecological state index value into a ecological benefit quantification value by adopting an ecological service value evaluation method; converting the industrial economic index value into an economic benefit quantification value by adopting a cost benefit analysis method; and carrying out weighted fusion on the ecological benefit quantized value and the economic benefit quantized value to generate a long-term comprehensive benefit estimated value.
9. The method for intelligently identifying an ecological restoration area based on multi-source data and model integration according to claim 8, wherein iteratively adjusting and updating candidate restoration strategies by a strategy optimization mechanism based on long-term comprehensive benefit estimation until convergence conditions are satisfied, comprises: comparing long-term comprehensive benefit estimation values of all candidate repair strategies in the current iteration; Eliminating partial candidate restoration strategies after the long-term comprehensive benefit estimation ranking; Performing parameter fine adjustment or structure recombination on the reserved candidate restoration strategies to generate a new generation candidate restoration strategy; And repeatedly executing the process from strategy simulation to generation of a new generation candidate repair strategy until the long-term comprehensive benefit estimation value of the optimal candidate repair strategy is not significantly improved after a plurality of continuous iterations, and judging that the convergence condition is met.
10. The intelligent recognition method of the ecological restoration area based on the multi-source data and model integration according to claim 9, wherein the steps of performing boundary refinement and priority ranking on the preliminary restoration area set according to the optimized restoration strategy set to generate a final ecological restoration area list and a matched restoration strategy thereof include: mapping each optimization repair strategy in the optimization repair strategy set back to the corresponding preliminary repair area; Based on engineering implementation range included in the optimized repair strategy, carrying out fine investigation on the space boundary of the preliminary repair area; Performing priority assignment on the ecological restoration area after the fine investigation according to the long-term comprehensive benefit estimation of the optimized restoration strategy; integrating all the ecological restoration areas subjected to boundary refinement and priority assignment and matched optimization restoration strategies thereof to generate a final ecological restoration area list.

Description

Ecological restoration area intelligent identification method based on multi-source data and model integration Technical Field The invention relates to the technical field of ecological restoration, in particular to an intelligent ecological restoration area identification method based on multi-source data and model integration. Background Currently, identification of areas where ecological restoration is needed is mainly dependent on remote sensing image interpretation combined with limited field investigation. This approach typically makes a threshold decision based on a single ecological element indicator, such as vegetation coverage or water pollution level. The common technical means comprise spatial superposition analysis by using GIS software or construction of a simple linear evaluation model, assigning fixed weights to each ecological factor for calculation and scoring, and defining a potential restoration range according to the final score. Such solutions rely to a large extent on the prior knowledge of the expert to determine the evaluation factors and the weighting system. The prior art solutions have drawbacks. The evaluation model based on the static index and the fixed weight is difficult to accurately reflect the complex nonlinear interaction among the elements in the ecological system and the dynamic pressure caused by the human industrial activity. The model evaluation results are often disposable, and prospective simulation cannot be performed on the linkage benefits possibly generated by the repair measures to be taken, so that the recommended repair area may not be selected with optimal benefits, or the formulated repair strategy lacks adaptability. Due to the failure to deeply analyze the inherent correlation between ecological degradation and industrial activities, the existing method often ignores the continuous influence of economic activities on the ecological system when identifying the repair potential, so that the repair effect is difficult to last, and even the phenomenon of re-degradation after repair can occur. The static assessment framework cannot adapt to the natural change of an ecological system along with the time and the dynamic response of artificial interference prognosis, so that the scientificity and practicability of the identification result are reduced. Disclosure of Invention The invention aims to provide an intelligent ecological restoration area identification method based on multi-source data and model integration, so as to solve the problems in the background technology. In order to achieve the above object, the present invention provides an intelligent ecological restoration area identification method based on multi-source data and model integration, the method comprising: Acquiring an original multi-source data set of an ecological restoration area; performing data cleaning and consistency verification on the original multi-source data set to generate ecological data in a standard format; Performing space-time meshing and feature extraction on the standard format ecological data to form a mesh feature sequence containing ecological element features and industrial activity features; Based on the grid feature sequence, performing restoration potential evaluation through a pre-constructed ecological restoration knowledge graph to generate a preliminary restoration area set; Inputting the preliminary repair region set and the grid feature sequence into a dynamic identification model, generating an iterative process of benefit simulation by fusing a repair strategy by the dynamic identification model, and outputting an optimized repair strategy set; and carrying out boundary refinement and priority ranking on the preliminary restoration area set according to the optimized restoration strategy set to generate a final ecological restoration area list and a matched restoration strategy thereof. Preferably, the data cleaning and consistency verification are performed on the original multi-source data set to generate ecological data in a standard format, which comprises the following steps: Identifying abnormal values and missing values in the original multi-source data set, and filling and correcting by adopting a data restoration method based on adjacent space-time context; Uniformly converting the filled and corrected data into a preset space-time coordinate system and a data format; and carrying out dimension normalization processing on the data under the unified coordinate system to generate the ecological data in the standard format. Preferably, performing space-time meshing and feature extraction on the standard format ecological data to form a mesh feature sequence including ecological element features and industrial activity features, comprising: dividing the ecological data in the standard format into a plurality of space-time grid units according to the preset space-time resolution; for each space-time grid unit, respectively extracting quantitative char