Intelligent evaluation method and system for ecological restoration engineering effect

Abstract

The application provides an ecological restoration engineering effect intelligent evaluation method and system, constructs a comprehensive data set based on multi-source heterogeneous data of a to-be-evaluated region, extracts entities and relationships between the entities in the comprehensive data set, and further constructs a spatio-temporal knowledge graph. Taking a target restoration engineering entity as a query condition, entity information of an engineering management area and entity information of a virtual control area are extracted in the spatio-temporal knowledge graph, and then vegetation state variable information of the engineering management area and the virtual control area is calculated respectively, and an engineering net effect is obtained based on the vegetation state variable information of the engineering management area and the virtual control area. The engineering net effect is applied to represent the restoration effect of the target restoration engineering after external climate interference is stripped. An evaluation result of the target restoration engineering is obtained based on the engineering net effect, and the spatio-temporal knowledge graph is updated. According to the scheme, the problem that an existing evaluation method cannot construct an anti-fact benchmark is solved through virtual control area construction based on knowledge graph matching, and the attribution accuracy of the engineering net effect is improved.

Description

Technical Field

[0001] This invention relates to the field of ecological environment monitoring and assessment technology, and more specifically, to an intelligent assessment method and system for the effectiveness of ecological restoration projects. Background Technology

[0002] Alpine grasslands are the dominant ecosystem type on the Qinghai-Tibet Plateau, and are vulnerable and sensitive to climate change. In recent years, with global warming and intensified human activities, the degradation of alpine grasslands has become increasingly severe.

[0003] Currently, the evaluation of the effectiveness of alpine grassland ecological restoration projects mainly relies on two types of technical methods: First, ground quadrat surveys, which assess the restoration effect by artificially setting up fixed plots and regularly observing indicators such as vegetation cover, biomass, and species diversity. Second, time-series analysis of remote sensing vegetation indices, which uses satellite remote sensing images to invert and retrieve greenness indicators such as the Normalized Difference Vegetation Index (NDVI), and evaluates the project's effectiveness by comparing changes in vegetation greenness before and after the project's implementation.

[0004] However, the above methods all use the pre-project state as the sole reference, failing to isolate the impact of concurrent climate fluctuations on grassland restoration and easily misinterpreting climate effects as project outcomes. Evaluation indicators are mostly limited to changes in vegetation greenness, making it difficult to comprehensively characterize the overall restoration status of grassland ecosystems. Furthermore, the selection of control areas and indicator calculations rely on manual experience, making it difficult to support large-scale, multi-type project evaluations.

[0005] The aforementioned technical bottlenecks manifest themselves in the following three aspects: First, the lack of counterfactual benchmarks prevents the separation of climate disturbances. Existing methods generally employ simple before-and-after comparisons of the engineering area itself, failing to construct a counterfactual reference of how the ecosystem would have evolved under the same climatic conditions if the project had not been implemented. Given that interannual precipitation fluctuations in alpine grasslands can exceed ±30%, the engineering effects and climate effects are completely intertwined, making it difficult to accurately attribute the net effects of the project. Second, weak knowledge support and insufficient intelligence. The selection of control areas, indicator calculations, and grade determination during the evaluation process heavily rely on human experience, exhibiting strong subjectivity and poor repeatability, making it difficult to support large-scale, multi-type, batch-based, and standardized evaluations. Third, the evaluation process ends abruptly, lacking a mechanism for knowledge accumulation and feedback evolution. Existing evaluation results are not systematically stored and reused, resulting in the waste of knowledge regarding the relationships between measures, conditions, and effects contained in numerous engineering cases. The evaluation system cannot continuously learn and optimize from accumulated cases.

[0006] With the rapid development of Earth observation big data and artificial intelligence technologies, the technical approach for evaluating the effectiveness of alpine grassland ecological restoration projects is undergoing profound changes. Knowledge engineering methods, represented by knowledge graphs, can achieve the systematic organization, intelligent retrieval, and associative reasoning of multi-source heterogeneous knowledge required for evaluation, providing a new approach to overcoming bottlenecks in traditional evaluations such as information fragmentation, weak knowledge connections, and insufficient reasoning capabilities. Against this backdrop, there is an urgent need to construct an intelligent evaluation method for the effectiveness of alpine grassland ecological restoration projects, using knowledge graphs as the core knowledge foundation to overcome key technical bottlenecks such as the lack of counterfactual benchmarks, difficulty in removing climate interference, single evaluation dimensions, and low levels of intelligence, providing technical support for the refined management and scientific decision-making of ecological restoration projects. Summary of the Invention

[0007] The purpose of this invention is to provide an intelligent evaluation method and system for the effects of ecological restoration projects, in order to solve the technical problems in the prior art of being unable to effectively isolate external climate interference and having difficulty in quantifying the net effects of projects.

[0008] In a first aspect, the present invention provides an intelligent evaluation method for the effectiveness of ecological restoration projects, the method comprising: Obtain multi-source heterogeneous data of the region to be evaluated, and construct a comprehensive dataset based on the multi-source heterogeneous data; Extract the entities and relationships between entities from the comprehensive dataset, wherein the entities include repair project entities; Construct a spatiotemporal knowledge graph based on the entities and the relationships between them; Using the target repair project entity as the query condition, extract entity information of the project treatment area and entity information of the virtual comparison area from the spatiotemporal knowledge graph; Based on the entity information of the engineering treatment area and the entity information of the virtual control area, the vegetation state variable information of the engineering treatment area and the virtual control area are calculated respectively. The net effect of the project is obtained based on the vegetation state variable information of the engineering treatment area and the virtual control area. The net effect of the project is used to characterize the restoration effect of the target restoration project after removing external climate interference. The evaluation results of the target repair project are obtained based on the net effect of the project. The spatiotemporal knowledge graph is updated based on the evaluation results.

[0009] In an optional implementation, the step of acquiring multi-source heterogeneous data of the region to be evaluated and constructing a comprehensive dataset based on the multi-source heterogeneous data includes: Acquire multi-source heterogeneous data of the area to be evaluated, including remote sensing image data, aerial remote sensing data, IoT monitoring data, and thematic operational data; The multi-source heterogeneous data is subjected to spatiotemporal registration processing to unify the geographic coordinate system and time scale, resulting in a processed comprehensive dataset.

[0010] In an optional implementation, the entity includes a monitoring carrier entity, an interference factor entity, and a repair engineering entity; The monitoring carrier entities include alpine meadow entities, alpine grassland entities, and alpine desert grassland entities; The interference factor entities include environmental background factors and exogenous driving factors. The environmental background factors include topographic entities, soil entities, and hydrological entities. The exogenous driving factors include climate factor entities and human activity entities. The restoration project entities include grazing land restoration projects, black soil beach management projects, and artificial grassland planting projects. The relationships between entities include spatial relationships, temporal relationships, logical relationships, association relationships, process relationships, and causal relationships. Among them, the causal relationships include direct causal relationships, cascade causal relationships, and cumulative effects. Each level of causal relationship is associated with a time delay parameter, a spatial transmission scale parameter, and a causal strength coefficient.

[0011] In an optional implementation, the step of extracting entities and relationships between entities from the comprehensive dataset includes: A deep learning semantic segmentation model is used to extract the spatial boundaries of the monitoring carrier entities from the comprehensive dataset and obtain attribute features; A pre-trained language model is used to extract entities and relationships between entities from the text data included in the comprehensive dataset to form text features; By using a cross-modal attention mechanism, the semantic alignment and interactive fusion of the attribute features and the text features are achieved, resulting in entities containing attribute features and the relationships between entities.

[0012] In an optional implementation, the step of constructing a spatiotemporal knowledge graph based on the entities and the relationships between entities includes: Entities containing attribute features and the relationships between entities are organized using an attribute graph model; The spatiotemporal knowledge graph obtained by the organization is stored in a graph database, wherein the graph database is used to provide a unified knowledge access interface.

[0013] In an optional implementation, the step of extracting entity information of the engineering treatment area and entity information of the virtual control area from the spatiotemporal knowledge graph using the target repair engineering entity as the query condition includes: Using the target repair project entity as the query condition, extract the engineering governance area entity located within the implementation space of the target repair project and the attribute features within the evaluation period from the spatiotemporal knowledge graph. The attribute features include the attribute features within a first set period before the implementation of the target repair project and a second set period after the implementation. Around the entity of the engineering treatment area or within the same ecological geographical unit, a virtual control area with similar ecological background conditions to the entity of the engineering treatment area before the implementation of the restoration project but without the implementation of the target restoration project is retrieved from the spatiotemporal knowledge graph, and the attribute characteristics of the virtual control area are obtained.

[0014] In an optional implementation, the step of obtaining the net engineering effect based on the vegetation state variable information of the engineering treatment area and the virtual control area includes: Based on the vegetation status variable information of the engineering treatment area, the comprehensive ecological index of the engineering treatment area is calculated in the first set period before the implementation of the target restoration project and in the second set period after the implementation of the target restoration project. A time base value is obtained based on the comprehensive ecological index in the first set period and the second set period. Based on the vegetation state variable information of the virtual control area, the comprehensive ecological index of the virtual control area during the first and second set time periods is calculated to obtain the spatial benchmark value; The difference between the time reference value and the spatial reference value is calculated to obtain the net repair effect value; The net restoration effect value, the comprehensive ecological index of the engineering treatment area during the first set time period, and the ecological index reference value are used to calculate the net recovery rate.

[0015] In an optional implementation, the step of obtaining the evaluation result of the target repair project based on the net effect of the project includes: The recovery speed and duration of effect are obtained based on the net recovery rate. The evaluation results of the target repair project are obtained by combining the net recovery rate, recovery speed and effect persistence. The evaluation results are also used to update the attribute features of entities related to the target repair project in the spatiotemporal knowledge graph.

[0016] In an optional implementation, the step of obtaining the evaluation result of the target repair project by combining the net recovery rate, recovery speed, and effect persistence includes: Assess the respective sub-levels of the grading system to which the net recovery rate, recovery speed, and duration of effect belong; The evaluation level of the target repair project is obtained by combining the sub-levels of net recovery rate, recovery speed and effect persistence.

[0017] Secondly, this invention provides an intelligent evaluation system for the effectiveness of ecological restoration projects, the system comprising: The acquisition module is used to acquire multi-source heterogeneous data of the region to be evaluated, and to construct a comprehensive dataset based on the multi-source heterogeneous data; An extraction module is used to extract entities and relationships between entities from the comprehensive dataset, wherein the entities include repair project entities; The construction module is used to construct a spatiotemporal knowledge graph based on the entities and the relationships between them; The query module is used to extract entity information of the engineering treatment area and entity information of the virtual comparison area from the spatiotemporal knowledge graph, using the target repair engineering entity as the query condition. The calculation module is used to calculate the vegetation state variable information of the engineering treatment area and the virtual control area based on the entity information of the engineering treatment area and the entity information of the virtual control area, respectively, and to obtain the net effect of the project based on the vegetation state variable information of the engineering treatment area and the virtual control area. The net effect of the project is used to characterize the restoration effect of the target restoration project after removing external climate interference. An evaluation module is used to obtain an evaluation result of the target repair project based on the net effect of the project; An update module is used to update the spatiotemporal knowledge graph based on the evaluation results.

[0018] This invention provides an intelligent evaluation method and system for the effectiveness of ecological restoration projects. It constructs a comprehensive dataset based on multi-source heterogeneous data of the area to be evaluated, and extracts entities and relationships between them. A spatiotemporal knowledge graph is then constructed based on these entities and relationships. Using the entities of the target restoration project as query conditions, entity information for the engineering treatment area and the virtual control area is extracted from the spatiotemporal knowledge graph. Vegetation state variables for both the engineering treatment area and the virtual control area are calculated based on these entity information. The net effect of the project is then obtained based on these vegetation state variables, characterizing the restoration effect of the target project after removing external climate interference. The evaluation result of the target restoration project is obtained based on the net effect, and this evaluation result can be used to update the spatiotemporal knowledge graph. In this scheme, the construction of a virtual control area based on knowledge graph matching solves the problem that existing evaluation methods cannot construct counterfactual benchmarks, thus improving the attribution accuracy of the project's net effect. Attached Figure Description

[0019] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments of the present invention will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 A flowchart of the intelligent evaluation method for the effectiveness of ecological restoration projects provided in this embodiment of the invention; Figure 2 for Figure 1 A flowchart of the sub-steps included in S11; Figure 3 for Figure 1 A flowchart of the sub-steps included in S12; Figure 4 for Figure 1 A flowchart of the sub-steps included in S13; Figure 5 for Figure 1 A flowchart of the sub-steps included in S14; Figure 6 for Figure 1 A flowchart of the sub-steps included in S16; Figure 7 for Figure 1 A flowchart of the sub-steps included in S17; Figure 8 This is a functional module block diagram of the intelligent evaluation system for ecological restoration engineering effects provided in an embodiment of the present invention; Figure 9 This is a structural block diagram of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0021] The technical solutions of the present invention will now be described with reference to the accompanying drawings in the embodiments of the present invention.

[0022] This invention provides an intelligent evaluation method for the effectiveness of ecological restoration projects, which can be executed by a server or cloud platform. Please refer to the following references. Figure 1 The flowchart below shows the intelligent evaluation method for ecological restoration engineering effects provided in this embodiment of the invention. The method specifically includes the following steps: S11, Obtain multi-source heterogeneous data of the region to be evaluated, and construct a comprehensive dataset based on the multi-source heterogeneous data; S12, extract the entities and relationships between entities from the comprehensive dataset, where the entities include the repair project entities; S13, Construct a spatiotemporal knowledge graph based on entities and the relationships between entities; S14: Using the target repair project entity as the query condition, extract entity information of the project treatment area and entity information of the virtual comparison area from the spatiotemporal knowledge graph; S15, calculate vegetation state variable information for the engineering treatment area and the virtual control area based on the entity information of the engineering treatment area and the entity information of the virtual control area, respectively; S16. The net effect of the project is obtained based on the vegetation state variable information of the engineering treatment area and the virtual control area. The net effect of the project is used to characterize the restoration effect of the target restoration project after removing external climate interference. S17, The evaluation results of the target repair project are obtained based on the net effect of the project; S18, Update the spatiotemporal knowledge graph based on the evaluation results.

[0023] The intelligent evaluation method for ecological restoration projects provided in this embodiment constructs a spatiotemporal knowledge graph and extracts information from the project treatment area and the virtual control area using the target restoration project entity as the query condition. Based on both, vegetation state variables are calculated to obtain the net effect of the project, achieving effective isolation of climate disturbance and accurate quantification of the net effect of the project. This method fundamentally solves the technical problems of lacking counterfactual benchmarks and being unable to isolate climate disturbance in existing technologies, providing a scientific basis for the objective and accurate evaluation of the effects of ecological restoration projects.

[0024] Specifically, this solution fundamentally solves the problem of missing counterfactual benchmarks by constructing a virtual control area and intelligently extracting the evaluation benchmark based on a spatiotemporal knowledge graph. This solution does not simply compare the engineering area itself before and after; instead, it first constructs a spatiotemporal knowledge graph containing rich entities and their relationships. Then, using the target restoration project entities as query conditions, it extracts entity information of the engineering treatment area from the knowledge graph, while simultaneously retrieving and obtaining information from the virtual control area. Vegetation state variables are calculated based on both the engineering treatment area and the virtual control area, and the net effect of the project is obtained through comparative analysis. This net effect represents the actual restoration effect brought about by the target restoration project itself after removing concurrent external climate disturbances. Thus, by establishing a scientific counterfactual reference benchmark, the effective separation of engineering effects and climate effects is achieved, solving the technical problem of the two being completely mixed and unable to be accurately attributed in traditional methods.

[0025] Furthermore, this solution achieves intelligent integration and knowledge mining of multi-source heterogeneous data by constructing a comprehensive dataset, automating entity and relationship extraction, and building a spatiotemporal knowledge graph, significantly improving the intelligence level of the assessment. First, the solution acquires multi-source heterogeneous data from the region to be assessed and constructs a comprehensive dataset, unifying data from different sources and with varying formats. Second, it extracts entities and relationships between entities from the comprehensive dataset, transforming fragmented data into structured knowledge units. Finally, based on the extracted entities and relationships, a spatiotemporal knowledge graph is constructed, organizing the complex spatiotemporal relationships and causal logic between monitored objects, interference factors, and remediation measures into an organic whole. This eliminates reliance on human experience in the assessment process, instead enabling standardized and automated processing of steps such as control area selection, information extraction, and indicator calculation based on the unified knowledge access interface provided by the knowledge graph, significantly improving the objectivity, repeatability, and batch processing capabilities of the assessment.

[0026] Furthermore, this solution uses a spatiotemporal knowledge graph as its core knowledge foundation, inherently possessing the ability for continuous knowledge accumulation and dynamic updates. In each evaluation process, newly acquired multi-source heterogeneous data, newly extracted entities and relationships between entities, newly calculated vegetation state variables, and the final net engineering effect evaluation results can all be continuously integrated into the spatiotemporal knowledge graph as new knowledge increments. Entity attributes, relationship strengths, and causal chains in the knowledge graph will be continuously enriched and optimized with the accumulation of evaluation cases, enabling the evaluation system to continuously learn from historical cases and achieve self-evolution in evaluation capabilities and accuracy. This mechanism ensures that the knowledge of the relationships between measures, conditions, and effects contained in a large number of engineering cases can be systematically accumulated and reused, completely changing the limitation of traditional methods where evaluation ends immediately, and forming a continuously evolving intelligent evaluation closed loop.

[0027] The specific implementation process of each of the above steps will be explained in detail below.

[0028] In this embodiment, firstly, multi-source heterogeneous data of the region to be evaluated are acquired to construct a comprehensive dataset. For details, please refer to [link to relevant documentation]. Figure 2 This step can be achieved in the following ways: S111, acquire multi-source heterogeneous data of the area to be evaluated. The multi-source heterogeneous data includes remote sensing image data, aerial remote sensing data, IoT monitoring data, and thematic business data. S112 performs spatiotemporal registration processing on multi-source heterogeneous data to unify the geographic coordinate system and time scale, resulting in a processed integrated dataset.

[0029] First, acquire data from multiple sources and in multiple formats for the area to be assessed (e.g., the Sanjiangyuan Nature Reserve). This multi-source heterogeneous data includes remote sensing imagery (such as Landsat and Sentinel satellite imagery), aerial remote sensing data (such as drone aerial imagery), IoT monitoring data (such as real-time soil temperature and humidity data and meteorological data transmitted from ground-based ecological monitoring stations), and thematic operational data (such as historical grassland survey records, ecological restoration project archives, and human activity statistics).

[0030] These data encompass various formats, including structured, semi-structured, and unstructured data, collectively forming the data foundation for the evaluation. After acquiring the aforementioned multi-source heterogeneous data, spatiotemporal registration processing was performed on it.

[0031] Specifically, all spatial data are unified to the same geographic coordinate system (such as WGS84 or CGCS2000), and data with different temporal resolutions are unified to the same time scale (for example, daily remote sensing observation data are synthesized into monthly or annual scale data) to eliminate the deviation caused by the inconsistency of spatial benchmarks and temporal granularity between different data sources, forming a spatiotemporally aligned comprehensive dataset. This ensures the consistency and comparability of data in subsequent knowledge extraction and map construction stages, and improves the reliability of the evaluation results.

[0032] Based on this, the entities and relationships between entities in the comprehensive dataset are extracted. This scheme abstracts the core elements involved in the ecological restoration process into three types of entities: monitoring carrier entities (such as alpine meadow entities, alpine grassland entities, alpine desert grassland entities, etc.), disturbance factor entities (including environmental background factors such as topography, soil, hydrology entities, etc., as well as exogenous driving factors such as climate factor entities, human activity entities, etc.), and restoration project entities (such as grazing land restoration project entities, black soil beach management project entities, artificial grassland planting project entities, etc.).

[0033] At the same time, a system of relationship types between entities is defined, including six major categories: spatial relationships (such as being located or adjacent), temporal relationships (such as occurring before), logical relationships, association relationships, process relationships, and causal relationships.

[0034] Among them, the causal relationship is further refined into direct causal relationship, cascade causal relationship and cumulative effect. Each level of causal relationship is also associated with time delay parameter ΔT, spatial transfer scale parameter ΔS and causal intensity coefficient, so as to accurately characterize the spatiotemporal lag and intensity decay characteristics of the ecological response process.

[0035] In this embodiment, by constructing a three-layer entity system of monitoring carrier, interference factor, and restoration project, and defining six major types of relationships: space, time, logic, correlation, process, and causality, and in particular, refining the causal relationship into direct causality, cascade causality, and cumulative effect, and associating it with time delay parameters, spatial transmission scale parameters, and causal intensity coefficients, it is possible to finely characterize the complex spatiotemporal coupling and causal transmission mechanism among various elements in the ecological restoration process, and provide a structured knowledge framework for a deeper understanding of the formation mechanism of the project effect.

[0036] For details on extracting entities and relationships between entities from a comprehensive dataset, please refer to [link / reference needed]. Figure 3 This can be achieved in the following ways: S121, a deep learning semantic segmentation model is used to extract the spatial boundaries of the monitoring carrier entities from the comprehensive dataset and obtain attribute features; S122 uses a pre-trained language model to extract entities and relationships between entities from the text data included in the comprehensive dataset to form text features; S123, through a cross-modal attention mechanism, achieves semantic alignment and interactive fusion of attribute features and text features, and obtains entities containing attribute features and relationships between entities.

[0037] In this embodiment, a cross-modal information extraction framework is adopted. For image data such as remote sensing images, a deep learning semantic segmentation model (such as U-Net and DeepLab series) is used to automatically extract the precise spatial boundaries of monitoring carrier entities such as alpine meadows and alpine grasslands, and calculate and obtain their attribute features such as vegetation index (NDVI) and biomass.

[0038] For textual data such as project reports, scientific literature, and engineering archives, pre-trained language models (such as BERT and its variants) are used for named entity recognition and relation extraction. The system automatically extracts entities related to the grassland restoration project, their interactions with alpine meadows, and causal relationships between different entities, which constitute textual features.

[0039] Finally, a cross-modal attention mechanism is used to align and interactively fuse the attribute features of the image modality with the semantic features of the text modality. For example, the implementation area of ​​a black soil beach management project mentioned in the text is precisely matched with the corresponding management patches segmented from the remote sensing image, thereby obtaining entities containing both spatial attributes and semantic descriptions, as well as the relationships between entities.

[0040] Furthermore, in this embodiment, for extraction results with a knowledge extraction confidence level below a preset threshold, the RAG retrieval enhancement process is triggered. Similar high-confidence knowledge fragments are retrieved from a dedicated vector knowledge base for alpine grasslands as contextual enhancement information to regenerate refined extraction results. A three-dimensional weighted voting strategy based on spatial matching, temporal matching, and semantic matching is employed to achieve cross-data source entity alignment and unique identification.

[0041] Based on this, a spatiotemporal knowledge graph is constructed based on entities and the relationships between entities. For details, please refer to [link to relevant documentation]. Figure 4 This step can be achieved in the following way: S131, Entities containing attribute features and the relationships between entities are organized using an attribute graph model; S132, the spatiotemporal knowledge graph obtained by the organization is stored in a graph database, where the graph database is used to provide a unified knowledge access interface.

[0042] In this embodiment, the entities containing attribute features and the relationships between entities extracted in the previous step are organized using an attribute graph model. The attribute graph model represents entities with nodes and relationships between entities with edges. Attribute key-value pairs can be attached to both nodes and edges, making it very suitable for expressing rich spatiotemporal attributes and semantic information.

[0043] The spatiotemporal knowledge graph obtained from the organization is ultimately stored in a graph database (such as Neo4j). The graph database provides a unified knowledge access interface for subsequent evaluation steps, supporting efficient graph traversal, pattern matching, and path query operations.

[0044] In this solution, entities and relationships are organized using an attribute graph model, and the spatiotemporal knowledge graph is stored in a graph database, providing a unified knowledge access interface. This enables integrated and efficient storage and flexible querying of multi-dimensional information such as space, time, and attributes, providing a stable and reliable knowledge foundation for subsequent evaluation and analysis.

[0045] Based on this, using the target repair project entity as the query condition, entity information of the project treatment area and virtual control area is extracted from the spatiotemporal knowledge graph. For details, please refer to [link to relevant documentation]. Figure 5 This step can be achieved in the following ways: S141, using the target repair project entity as the query condition, extract the engineering treatment area entity located within the implementation space of the target repair project and the attribute features within the evaluation period from the spatiotemporal knowledge graph; The attribute characteristics include those within a first set time period before the implementation of the target repair project and within a second set time period after implementation; S142. Around the entity of the engineering treatment area or within the same ecological geographical unit, retrieve virtual control areas from the spatiotemporal knowledge graph that have similar ecological background conditions to the entity of the engineering treatment area before the implementation of the restoration project but have not implemented the target restoration project, and obtain the attribute characteristics of the virtual control areas.

[0046] When it is necessary to evaluate the effect of a specific restoration project (such as a grazing land restoration project in a certain area), the entity of the target restoration project is used as the query condition to perform a graph query in the constructed spatiotemporal knowledge graph.

[0047] First, all entities within the engineering treatment area located within the spatial scope of the target restoration project are extracted from the spatiotemporal knowledge graph, and the attribute characteristics of these entities during the evaluation period are obtained. These attribute characteristics cover the time series of indicators such as vegetation index, aboveground biomass, and NDVI during the first set period before the implementation of the target restoration project (e.g., 5 years before the project implementation) and the second set period after the implementation (e.g., 5 years after the project implementation).

[0048] Then, around the entity in the engineering remediation area or within the same ecological geographic unit, a graph algorithm is used to retrieve from the spatiotemporal knowledge graph areas that have similar ecological background conditions (including grassland type, altitude range, slope aspect, coverage level, soil type, historical degradation level, etc.) to the entity in the engineering remediation area before the implementation of the remediation project, but have never implemented the target remediation project, as virtual control areas, and obtain the attribute characteristics of the virtual control areas in the same first and second set time periods.

[0049] There can be multiple virtual control regions. The N candidate virtual control regions with the highest comprehensive similarity can be selected as the final virtual control regions through multi-attribute weighted similarity matching, where N≥3.

[0050] These virtual control zones represent the most likely natural evolution path of the treated area under the same climatic conditions without engineering intervention, realizing the intelligent construction of counterfactual assessment benchmarks and ensuring the scientific, objective and repeatable nature of the selection of control zones.

[0051] In addition, in this embodiment, historical reference cases can be extracted. Historical restoration project entities with the same type of target project, the same ecological and geographical zoning, and similar degradation level before implementation can be retrieved from the knowledge graph. Their project attributes, ecological background characteristics, and restoration effect data (restoration rate, restoration cycle) can be extracted as prior references for effect evaluation.

[0052] The net effect of the project was obtained based on vegetation state variable information from the engineering treatment area and the virtual control area. For details, please refer to [link to relevant documentation]. Figure 6 This step can be achieved in the following ways: S161, based on the vegetation state variable information of the engineering treatment area, calculate the comprehensive ecological index of the engineering treatment area in the first set period before the implementation of the target restoration project and in the second set period after the implementation of the target restoration project, and obtain the time base value based on the comprehensive ecological index in the first set period and the second set period. S162, Based on the vegetation state variable information of the virtual control area, the comprehensive ecological index of the virtual control area in the first and second set time periods is calculated to obtain the spatial benchmark value; S163, calculate the difference between the time reference value and the spatial reference value to obtain the net repair effect value; S164. The net restoration rate is calculated based on the net restoration effect value, the comprehensive ecological index of the engineering treatment area during the first set time period, and the ecological index reference value.

[0053] In this embodiment, a comprehensive ecological index is constructed, and a dual-benchmark evaluation framework is established with a time benchmark (comparison of the project area before and after) and a spatial benchmark (comparison of the project area and the virtual control area) to calculate the net restoration effect of the restoration project after removing the interference of climate fluctuations.

[0054] First, based on the vegetation state variable information of the engineering treatment area, the comprehensive ecological index of the engineering treatment area was calculated for the first set period before the project implementation and the second set period after the project implementation. Specifically, the time series values ​​of indicators such as coverage (C, unit: %), aboveground biomass (B, unit: g / m²), and NDVI (N, dimensionless) were extracted from the spatiotemporal knowledge graph. Z-score standardization was performed first, followed by principal component analysis.

[0055] Based on grassland patch samples marked with degradation levels within the area to be assessed (referring to the GB / T 19377 grading standard), the distribution range of M values ​​corresponding to each degradation level was statistically analyzed, and the correspondence between M values ​​and degradation levels was established.

[0056] The comprehensive ecological index of the second set time period is compared with that of the first set time period to obtain a time baseline value. This time baseline value reflects the total change under the combined effects of engineering intervention and natural fluctuations. Specifically, the time baseline value ΔTemporal is calculated using the following formula: ΔTemporal = M1 - M2 Among them, M2 is the average comprehensive ecological index of the engineering treatment area in the first set period before the implementation of the target restoration project (such as 3 consecutive years before implementation), and M1 is the average comprehensive ecological index of the engineering treatment area in the second set period after the implementation of the target restoration project (such as 3 consecutive years after implementation).

[0057] Secondly, based on the vegetation state variable information of the virtual control area, the comprehensive ecological index of the virtual control area within the same first and second set time periods is calculated. A comparison is then made before and after to obtain a spatial baseline value. This spatial baseline value reflects the amount of ecological change caused purely by natural fluctuations (including interannual climate fluctuations). Specifically, the spatial baseline value (Δ baseline) is calculated using the following formula: Δbaseline = (1 / k) × Σ(M3 - M4) Where k is the number of virtual control areas (k≥3), and M3 and M4 are the average comprehensive ecological indices of the i-th virtual control area in the first and second time periods, respectively.

[0058] Then, the net repair effect value is calculated as follows: ΔNet = ΔTemporal - ΔBenchmark Wherein, ΔNet represents the net repair effect of the project after removing climate fluctuations.

[0059] The net recovery rate, Recovery_Rate, is calculated as follows: Recovery_Rate = ΔNet / (M_reference - M2) × 100% M_reference is a comprehensive ecological index reference value for non-degraded grassland patches of the same type retrieved from the spatiotemporal knowledge graph. M_reference can be calculated by retrieving grassland patch entities of the same type and non-degraded level (Level I in GB / T 19377) from the spatiotemporal knowledge graph, using the ecological geographical zone where the target project is located as the spatial range, and extracting their multi-year average M value.

[0060] Based on the above, the evaluation results of the target repair project are obtained based on the net effect of the project. For details, please refer to [link / reference needed]. Figure 7 This step can be achieved in the following ways: S171, based on the net recovery rate, determines the recovery speed and the duration of the effect; S172, combining net recovery rate, recovery speed and effect persistence to obtain the evaluation results of the target repair project, wherein the evaluation results are also used to update the attribute characteristics of entities related to the target repair project in the spatiotemporal knowledge graph.

[0061] In this embodiment, based on the calculated net recovery rate, its trend over time is further analyzed to obtain indicators such as recovery speed and duration of effect.

[0062] The net recovery rate (Recovery_Rate) reflects the degree of recovery achieved by the current project. A negative value indicates increased degradation. 0-30% is low recovery, 30%-60% is medium recovery, and 60%-100% is high recovery.

[0063] Recovery speed (Recovery_Speed) = ΔRecovery_Rate / Δt, where ΔRecovery_Rate is the change in net recovery rate per unit time, and Δt is the time interval (years), reflecting the speed of recovery.

[0064] Persistence is the coefficient of variation (CV) of the net recovery rate over several consecutive years after the project is implemented. CV = standard deviation of net recovery rate / mean of net recovery rate. The lower the CV, the more stable and sustainable the recovery effect. A CV higher than 0.3 is marked as significant fluctuation in effect.

[0065] In the specific assessment, the three dimensions of net recovery rate, recovery speed, and effect persistence are graded separately and divided into their respective sub-levels (e.g., excellent, good, average, and poor). Then, by combining the sub-levels of these three dimensions and using preset assessment rules or weighting strategies, the final assessment level of the target restoration project is obtained.

[0066] Furthermore, the evaluation results will also be used as new attribute information to be written back and updated to the attribute features of entities related to the target repair project in the spatiotemporal knowledge graph, realizing dynamic iteration of the knowledge graph and system self-optimization, so that the evaluation system can continuously learn and accumulate knowledge from historical cases.

[0067] Specifically, the effect level attribute and evaluation status attribute of the repair project entity can be updated. For example, if the project effect is Level I or Level II, the confidence of the causal relationship of the repair and improvement on the path of the project measure type, the grassland type, and the ecological geographical region in the spatiotemporal knowledge graph can be increased (confidence increase Δ=0.05).

[0068] If the engineering effect is level IV, reduce the confidence level of the causal relationship (confidence level reduction Δ=0.10) or mark it as pending further verification.

[0069] The complete evaluation record of the target project is stored as a new historical case entity in the spatiotemporal knowledge graph for use in similarity matching of subsequent project evaluations. When the cumulative number of evaluation cases reaches a preset threshold, the statistical benchmark for the grading judgment threshold is automatically updated.

[0070] The intelligent evaluation method for ecological restoration projects provided in this embodiment innovatively introduces a spatiotemporal knowledge graph of alpine grasslands into the field of ecological restoration project evaluation. Existing technologies have not utilized knowledge graph technology to systematically organize the multi-source heterogeneous knowledge required for evaluating alpine grassland ecological restoration projects. This invention constructs a three-layer entity system of monitoring carriers, interference factors, and restoration projects, along with six types of semantic relationships (spatial, temporal, logical, relational, process, and causal relationships). This achieves structured storage and intelligent retrieval of entities in the engineering treatment area, candidate entities in the virtual control area, and entities in historical reference cases, as well as their attributes and relationships. This upgrades the acquisition of evaluation object information from manual querying to automatic extraction from a knowledge graph.

[0071] Furthermore, a dual-benchmark assessment framework with temporal and spatial benchmarks was constructed to effectively isolate the interference of climate fluctuations. Existing methods only perform simple before-and-after comparisons and cannot distinguish between engineering effects and climate effects. This invention automatically selects virtual control areas through multi-attribute similarity matching of knowledge graphs, using their contemporaneous average variation as a spatial benchmark to calculate the net engineering effect after removing climate fluctuations, thus solving the long-standing problem of the lack of counterfactual benchmarks in the assessment of ecological restoration effects.

[0072] This scheme introduces a comprehensive ecological index to replace single indicators, as existing technologies often use single indicators such as NDVI for assessment. This scheme extracts three core indicators—coverage, aboveground biomass, and NDVI—from a spatiotemporal knowledge graph. A comprehensive ecological index M is constructed through Z-score standardization and principal component analysis. Based on a grading standard, the distribution range of M values ​​corresponding to each degradation level is calibrated, making the assessment results more comprehensive and possessing intuitive ecological interpretability.

[0073] Furthermore, this invention innovatively upgrades the evaluation process into a closed-loop system of evaluation, feedback, and evolution. Existing methods terminate the evaluation process immediately, and the results are not reused. This invention feeds the evaluation results back to a spatiotemporal knowledge graph, and stores the evaluation records as new historical case entities, enabling the evaluation system to have self-evolution capabilities.

[0074] To verify the effectiveness of this solution, it was further applied to specific examples. The results show that the solution achieves good evaluation results; specifically, the evaluation efficiency is significantly improved after adopting this solution, enabling batch intelligent evaluation. Through automated information extraction and calculation using a spatiotemporal knowledge graph, the evaluation cycle is significantly shortened. Taking a single project as an example, traditional ground surveys and manual analysis take 3-6 months. This invention, relying on a knowledge graph, automatically completes entity information extraction, comparison area retrieval, indicator calculation, and hierarchical evaluation, shortening the evaluation cycle to within one month.

[0075] Furthermore, the assessment results are quantifiable, comparable, and traceable. Through the quantitative grading and effect level determination of the comprehensive ecological index, the assessment results of different projects, regions, and periods can be compared horizontally and vertically on a unified scale, providing a quantifiable scientific basis for the allocation of ecological compensation funds and project management decisions.

[0076] The knowledge system continuously evolves, and assessment capabilities improve with the accumulation of cases. Through a closed-loop mechanism of assessment, feedback, and evolution, the confidence level of repairing and improving causal relationships in the spatiotemporal knowledge graph is dynamically adjusted after each project assessment, and the historical case library is expanded. When the accumulated assessment cases exceed 30, the grading threshold will be automatically optimized based on the actual case distribution, making the assessment standards more aligned with regional realities. The accuracy and experience of the assessment system continuously improve with the increase in project cases.

[0077] Based on the same inventive concept, please refer to Figure 8 This invention also provides a functional module diagram of an intelligent evaluation system for the effectiveness of ecological restoration projects. This embodiment can divide the intelligent evaluation system for the effectiveness of ecological restoration projects into functional modules based on the above method embodiments. For example, each function can be divided into its own functional modules, or two or more functions can be integrated into one processing module. The integrated modules can be implemented in hardware or as software functional modules. It should be noted that the module division in this invention embodiment is illustrative and only represents one logical functional division; other division methods may be used in actual implementation.

[0078] For example, when dividing functional modules according to their respective functions, Figure 8 The illustrated intelligent evaluation system for ecological restoration projects is only a schematic diagram. This system may include modules for acquisition, extraction, construction, querying, calculation, evaluation, and updating. The functions of each module are described in detail below.

[0079] The acquisition module is used to acquire multi-source heterogeneous data of the region to be evaluated and to construct a comprehensive dataset based on the multi-source heterogeneous data; The extraction module is used to extract entities and relationships between entities from the comprehensive dataset, including repair project entities; The building module is used to construct spatiotemporal knowledge graphs based on entities and the relationships between entities; The query module is used to extract entity information of the engineering treatment area and virtual comparison area from the spatiotemporal knowledge graph, based on the target repair engineering entity as the query condition. The calculation module is used to calculate the vegetation state variable information of the engineering treatment area and the virtual control area based on the entity information of the engineering treatment area and the entity information of the virtual control area, respectively. Based on the vegetation state variable information of the engineering treatment area and the virtual control area, the net effect of the project is obtained. The net effect of the project is used to characterize the restoration effect of the target restoration project after removing external climate interference. The evaluation module is used to obtain the evaluation results of the target repair project based on the net effect of the project; An update module is used to update the spatiotemporal knowledge graph based on the evaluation results.

[0080] The intelligent evaluation system for ecological restoration engineering effects provided in this embodiment can be used to execute the intelligent evaluation method for ecological restoration engineering effects under any of the above embodiments. For details not covered in this embodiment, please refer to the corresponding descriptions in the above embodiments. This embodiment will not elaborate further here.

[0081] Please see Figure 9 This is a structural block diagram of an electronic device provided in an embodiment of the present invention. The electronic device can be a server, computer equipment, etc. The electronic device includes a memory, a processor, and a communication module. The memory, processor, and communication module are electrically connected directly or indirectly to each other to achieve data transmission or interaction. For example, these components can be electrically connected to each other through one or more communication buses or signal lines.

[0082] The memory is used to store computer programs or data. Memory can be, but is not limited to, Random Access Memory (RAM), Read Only Memory (ROM), Programmable Read-Only Memory (PROM), Erasable Programmable Read-Only Memory (EPROM), Electrically Erasable Programmable Read-Only Memory (EEPROM), etc.

[0083] The processor is used to read / write data or programs stored in the memory and execute the intelligent evaluation method for ecological restoration engineering effects provided in any embodiment of the present invention.

[0084] The communication module is used to establish communication connections between electronic devices and other communication terminals via a network, and to send and receive data via the network.

[0085] It should be understood that, Figure 9 The structure shown is only a schematic diagram of an electronic device; the electronic device may also include components that are larger than those shown. Figure 9 The more or fewer components shown, or having the same Figure 9 The different configurations shown.

[0086] Furthermore, embodiments of the present invention also provide a computer-readable storage medium storing machine-executable instructions, which, when executed, implement the intelligent evaluation method for the ecological restoration project effect provided in the above embodiments.

[0087] Specifically, the computer-readable storage medium can be a general-purpose storage medium, such as a removable disk or hard drive. When the computer program on the computer-readable storage medium is run, it can execute the aforementioned intelligent evaluation method for the effectiveness of ecological restoration projects. The processes involved in the execution of the executable instructions on the computer-readable storage medium can be referred to the relevant descriptions in the above method embodiments, and will not be detailed here.

[0088] In the embodiments provided by this invention, it should be understood that the disclosed apparatus and method can be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and there may be other division methods in actual implementation. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the coupling or direct coupling or communication connection shown or discussed may be through some communication interface; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0089] Furthermore, the units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0090] Furthermore, in the various embodiments of the present invention, the functional modules can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0091] It should be noted that if the functionality is implemented as a software module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0092] In this document, relational terms such as first and second are used only to distinguish one entity or operation from another entity or operation, without necessarily requiring or implying any such actual relationship or order between these entities or operations.

[0093] The above are merely embodiments of the present invention and are not intended to limit the scope of protection of the present invention. For those skilled in the art, the present invention can have various modifications and variations. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for intelligent evaluation of the effectiveness of ecological restoration projects, characterized in that, The method includes: Obtain multi-source heterogeneous data of the region to be evaluated, and construct a comprehensive dataset based on the multi-source heterogeneous data; Extract the entities and relationships between entities from the comprehensive dataset, wherein the entities include repair project entities; Construct a spatiotemporal knowledge graph based on the entities and the relationships between them; Using the target repair project entity as the query condition, extract entity information of the project treatment area and entity information of the virtual comparison area from the spatiotemporal knowledge graph; Based on the entity information of the engineering treatment area and the entity information of the virtual control area, the vegetation state variable information of the engineering treatment area and the virtual control area are calculated respectively. The net effect of the project is obtained based on the vegetation state variable information of the engineering treatment area and the virtual control area. The net effect of the project is used to characterize the restoration effect of the target restoration project after removing external climate interference. The evaluation results of the target repair project are obtained based on the net effect of the project. The spatiotemporal knowledge graph is updated based on the evaluation results.

2. The intelligent evaluation method for the effectiveness of ecological restoration projects according to claim 1, characterized in that, The step of acquiring multi-source heterogeneous data of the region to be evaluated and constructing a comprehensive dataset based on the multi-source heterogeneous data includes: Acquire multi-source heterogeneous data of the area to be evaluated, including remote sensing image data, aerial remote sensing data, IoT monitoring data, and thematic operational data; The multi-source heterogeneous data is subjected to spatiotemporal registration processing to unify the geographic coordinate system and time scale, resulting in a processed comprehensive dataset.

3. The intelligent evaluation method for the effectiveness of ecological restoration projects according to claim 1, characterized in that, The entities include monitoring carrier entities, interference factor entities, and repair engineering entities; The monitoring carrier entities include alpine meadow entities, alpine grassland entities, and alpine desert grassland entities; The interference factor entities include environmental background factors and exogenous driving factors. The environmental background factors include topographic entities, soil entities, and hydrological entities. The exogenous driving factors include climate factor entities and human activity entities. The restoration project entities include grazing land restoration projects, black soil beach management projects, and artificial grassland planting projects. The relationships between entities include spatial relationships, temporal relationships, logical relationships, association relationships, process relationships, and causal relationships. Among them, the causal relationships include direct causal relationships, cascade causal relationships, and cumulative effects. Each level of causal relationship is associated with a time delay parameter, a spatial transmission scale parameter, and a causal strength coefficient.

4. The intelligent evaluation method for the effectiveness of ecological restoration projects according to claim 3, characterized in that, The step of extracting entities and relationships between entities from the comprehensive dataset includes: A deep learning semantic segmentation model is used to extract the spatial boundaries of the monitoring carrier entities from the comprehensive dataset and obtain attribute features; A pre-trained language model is used to extract entities and relationships between entities from the text data included in the comprehensive dataset to form text features; By using a cross-modal attention mechanism, the semantic alignment and interactive fusion of the attribute features and the text features are achieved, resulting in entities containing attribute features and the relationships between entities.

5. The intelligent evaluation method for the effectiveness of ecological restoration projects according to claim 4, characterized in that, The steps for constructing a spatiotemporal knowledge graph based on the entities and the relationships between entities include: Entities containing attribute features and the relationships between entities are organized using an attribute graph model; The spatiotemporal knowledge graph obtained by the organization is stored in a graph database, wherein the graph database is used to provide a unified knowledge access interface.

6. The intelligent evaluation method for the effectiveness of ecological restoration projects according to claim 1, characterized in that, The step of extracting entity information of the engineering treatment area and entity information of the virtual comparison area from the spatiotemporal knowledge graph using the target repair engineering entity as the query condition includes: Using the target repair project entity as the query condition, extract the engineering governance area entity located within the implementation space of the target repair project and the attribute features within the evaluation period from the spatiotemporal knowledge graph. The attribute features include the attribute features within a first set period before the implementation of the target repair project and a second set period after the implementation. Around the entity of the engineering treatment area or within the same ecological geographical unit, a virtual control area with similar ecological background conditions to the entity of the engineering treatment area before the implementation of the restoration project but without the implementation of the target restoration project is retrieved from the spatiotemporal knowledge graph, and the attribute characteristics of the virtual control area are obtained.

7. The intelligent evaluation method for the effectiveness of ecological restoration projects according to claim 6, characterized in that, The step of obtaining the net effect of the project based on the vegetation state variable information of the engineering treatment area and the virtual control area includes: Based on the vegetation status variable information of the engineering treatment area, the comprehensive ecological index of the engineering treatment area is calculated in the first set period before the implementation of the target restoration project and in the second set period after the implementation of the target restoration project. A time base value is obtained based on the comprehensive ecological index in the first set period and the second set period. Based on the vegetation state variable information of the virtual control area, the comprehensive ecological index of the virtual control area during the first and second set time periods is calculated to obtain the spatial benchmark value; The difference between the time reference value and the spatial reference value is calculated to obtain the net repair effect value; The net restoration effect value, the comprehensive ecological index of the engineering treatment area during the first set time period, and the ecological index reference value are used to calculate the net recovery rate.

8. The intelligent evaluation method for the effectiveness of ecological restoration projects according to claim 7, characterized in that, The step of obtaining the evaluation result of the target repair project based on the net effect of the project includes: The recovery speed and duration of effect are obtained based on the net recovery rate. The evaluation results of the target repair project are obtained by combining the net recovery rate, recovery speed and effect persistence. The evaluation results are also used to update the attribute features of entities related to the target repair project in the spatiotemporal knowledge graph.

9. The intelligent evaluation method for the effectiveness of ecological restoration projects according to claim 8, characterized in that, The step of obtaining the evaluation result of the target repair project by combining the net recovery rate, recovery speed, and effect persistence includes: Assess the respective sub-levels of the grading system to which the net recovery rate, recovery speed, and duration of effect belong; The evaluation level of the target repair project is obtained by combining the sub-levels of net recovery rate, recovery speed and effect persistence.

10. An intelligent evaluation system for the effectiveness of ecological restoration projects, characterized in that, The system includes: The acquisition module is used to acquire multi-source heterogeneous data of the region to be evaluated, and to construct a comprehensive dataset based on the multi-source heterogeneous data; An extraction module is used to extract entities and relationships between entities from the comprehensive dataset, wherein the entities include repair project entities; The construction module is used to construct a spatiotemporal knowledge graph based on the entities and the relationships between them; The query module is used to extract entity information of the engineering treatment area and entity information of the virtual comparison area from the spatiotemporal knowledge graph, using the target repair engineering entity as the query condition. The calculation module is used to calculate the vegetation state variable information of the engineering treatment area and the virtual control area based on the entity information of the engineering treatment area and the entity information of the virtual control area, respectively, and to obtain the net effect of the project based on the vegetation state variable information of the engineering treatment area and the virtual control area. The net effect of the project is used to characterize the restoration effect of the target restoration project after removing external climate interference. An evaluation module is used to obtain an evaluation result of the target repair project based on the net effect of the project; An update module is used to update the spatiotemporal knowledge graph based on the evaluation results.

Images